git.oblomov.eu Git - git/blob - diff.c

   1 /*
   2  * Copyright (C) 2005 Junio C Hamano
   3  */
   4 #include "cache.h"
   5 #include "config.h"
   6 #include "tempfile.h"
   7 #include "quote.h"
   8 #include "diff.h"
   9 #include "diffcore.h"
  10 #include "delta.h"
  11 #include "xdiff-interface.h"
  12 #include "color.h"
  13 #include "attr.h"
  14 #include "run-command.h"
  15 #include "utf8.h"
  16 #include "object-store.h"
  17 #include "userdiff.h"
  18 #include "submodule-config.h"
  19 #include "submodule.h"
  20 #include "hashmap.h"
  21 #include "ll-merge.h"
  22 #include "string-list.h"
  23 #include "argv-array.h"
  24 #include "graph.h"
  25 #include "packfile.h"
  26 #include "parse-options.h"
  27 #include "help.h"
  28 #include "fetch-object.h"
  29
  30 #ifdef NO_FAST_WORKING_DIRECTORY
  31 #define FAST_WORKING_DIRECTORY 0
  32 #else
  33 #define FAST_WORKING_DIRECTORY 1
  34 #endif
  35
  36 static int diff_detect_rename_default;
  37 static int diff_indent_heuristic = 1;
  38 static int diff_rename_limit_default = 400;
  39 static int diff_suppress_blank_empty;
  40 static int diff_use_color_default = -1;
  41 static int diff_color_moved_default;
  42 static int diff_color_moved_ws_default;
  43 static int diff_context_default = 3;
  44 static int diff_interhunk_context_default;
  45 static const char *diff_word_regex_cfg;
  46 static const char *external_diff_cmd_cfg;
  47 static const char *diff_order_file_cfg;
  48 int diff_auto_refresh_index = 1;
  49 static int diff_mnemonic_prefix;
  50 static int diff_no_prefix;
  51 static int diff_stat_graph_width;
  52 static int diff_dirstat_permille_default = 30;
  53 static struct diff_options default_diff_options;
  54 static long diff_algorithm;
  55 static unsigned ws_error_highlight_default = WSEH_NEW;
  56
  57 static char diff_colors[][COLOR_MAXLEN] = {
  58         GIT_COLOR_RESET,
  59         GIT_COLOR_NORMAL,       /* CONTEXT */
  60         GIT_COLOR_BOLD,         /* METAINFO */
  61         GIT_COLOR_CYAN,         /* FRAGINFO */
  62         GIT_COLOR_RED,          /* OLD */
  63         GIT_COLOR_GREEN,        /* NEW */
  64         GIT_COLOR_YELLOW,       /* COMMIT */
  65         GIT_COLOR_BG_RED,       /* WHITESPACE */
  66         GIT_COLOR_NORMAL,       /* FUNCINFO */
  67         GIT_COLOR_BOLD_MAGENTA, /* OLD_MOVED */
  68         GIT_COLOR_BOLD_BLUE,    /* OLD_MOVED ALTERNATIVE */
  69         GIT_COLOR_FAINT,        /* OLD_MOVED_DIM */
  70         GIT_COLOR_FAINT_ITALIC, /* OLD_MOVED_ALTERNATIVE_DIM */
  71         GIT_COLOR_BOLD_CYAN,    /* NEW_MOVED */
  72         GIT_COLOR_BOLD_YELLOW,  /* NEW_MOVED ALTERNATIVE */
  73         GIT_COLOR_FAINT,        /* NEW_MOVED_DIM */
  74         GIT_COLOR_FAINT_ITALIC, /* NEW_MOVED_ALTERNATIVE_DIM */
  75         GIT_COLOR_FAINT,        /* CONTEXT_DIM */
  76         GIT_COLOR_FAINT_RED,    /* OLD_DIM */
  77         GIT_COLOR_FAINT_GREEN,  /* NEW_DIM */
  78         GIT_COLOR_BOLD,         /* CONTEXT_BOLD */
  79         GIT_COLOR_BOLD_RED,     /* OLD_BOLD */
  80         GIT_COLOR_BOLD_GREEN,   /* NEW_BOLD */
  81 };
  82
  83 static const char *color_diff_slots[] = {
  84         [DIFF_CONTEXT]                = "context",
  85         [DIFF_METAINFO]               = "meta",
  86         [DIFF_FRAGINFO]               = "frag",
  87         [DIFF_FILE_OLD]               = "old",
  88         [DIFF_FILE_NEW]               = "new",
  89         [DIFF_COMMIT]                 = "commit",
  90         [DIFF_WHITESPACE]             = "whitespace",
  91         [DIFF_FUNCINFO]               = "func",
  92         [DIFF_FILE_OLD_MOVED]         = "oldMoved",
  93         [DIFF_FILE_OLD_MOVED_ALT]     = "oldMovedAlternative",
  94         [DIFF_FILE_OLD_MOVED_DIM]     = "oldMovedDimmed",
  95         [DIFF_FILE_OLD_MOVED_ALT_DIM] = "oldMovedAlternativeDimmed",
  96         [DIFF_FILE_NEW_MOVED]         = "newMoved",
  97         [DIFF_FILE_NEW_MOVED_ALT]     = "newMovedAlternative",
  98         [DIFF_FILE_NEW_MOVED_DIM]     = "newMovedDimmed",
  99         [DIFF_FILE_NEW_MOVED_ALT_DIM] = "newMovedAlternativeDimmed",
 100         [DIFF_CONTEXT_DIM]            = "contextDimmed",
 101         [DIFF_FILE_OLD_DIM]           = "oldDimmed",
 102         [DIFF_FILE_NEW_DIM]           = "newDimmed",
 103         [DIFF_CONTEXT_BOLD]           = "contextBold",
 104         [DIFF_FILE_OLD_BOLD]          = "oldBold",
 105         [DIFF_FILE_NEW_BOLD]          = "newBold",
 106 };
 107
 108 define_list_config_array_extra(color_diff_slots, {"plain"});
 109
 110 static int parse_diff_color_slot(const char *var)
 111 {
 112         if (!strcasecmp(var, "plain"))
 113                 return DIFF_CONTEXT;
 114         return LOOKUP_CONFIG(color_diff_slots, var);
 115 }
 116
 117 static int parse_dirstat_params(struct diff_options *options, const char *params_string,
 118                                 struct strbuf *errmsg)
 119 {
 120         char *params_copy = xstrdup(params_string);
 121         struct string_list params = STRING_LIST_INIT_NODUP;
 122         int ret = 0;
 123         int i;
 124
 125         if (*params_copy)
 126                 string_list_split_in_place(&params, params_copy, ',', -1);
 127         for (i = 0; i < params.nr; i++) {
 128                 const char *p = params.items[i].string;
 129                 if (!strcmp(p, "changes")) {
 130                         options->flags.dirstat_by_line = 0;
 131                         options->flags.dirstat_by_file = 0;
 132                 } else if (!strcmp(p, "lines")) {
 133                         options->flags.dirstat_by_line = 1;
 134                         options->flags.dirstat_by_file = 0;
 135                 } else if (!strcmp(p, "files")) {
 136                         options->flags.dirstat_by_line = 0;
 137                         options->flags.dirstat_by_file = 1;
 138                 } else if (!strcmp(p, "noncumulative")) {
 139                         options->flags.dirstat_cumulative = 0;
 140                 } else if (!strcmp(p, "cumulative")) {
 141                         options->flags.dirstat_cumulative = 1;
 142                 } else if (isdigit(*p)) {
 143                         char *end;
 144                         int permille = strtoul(p, &end, 10) * 10;
 145                         if (*end == '.' && isdigit(*++end)) {
 146                                 /* only use first digit */
 147                                 permille += *end - '0';
 148                                 /* .. and ignore any further digits */
 149                                 while (isdigit(*++end))
 150                                         ; /* nothing */
 151                         }
 152                         if (!*end)
 153                                 options->dirstat_permille = permille;
 154                         else {
 155                                 strbuf_addf(errmsg, _("  Failed to parse dirstat cut-off percentage '%s'\n"),
 156                                             p);
 157                                 ret++;
 158                         }
 159                 } else {
 160                         strbuf_addf(errmsg, _("  Unknown dirstat parameter '%s'\n"), p);
 161                         ret++;
 162                 }
 163
 164         }
 165         string_list_clear(&params, 0);
 166         free(params_copy);
 167         return ret;
 168 }
 169
 170 static int parse_submodule_params(struct diff_options *options, const char *value)
 171 {
 172         if (!strcmp(value, "log"))
 173                 options->submodule_format = DIFF_SUBMODULE_LOG;
 174         else if (!strcmp(value, "short"))
 175                 options->submodule_format = DIFF_SUBMODULE_SHORT;
 176         else if (!strcmp(value, "diff"))
 177                 options->submodule_format = DIFF_SUBMODULE_INLINE_DIFF;
 178         /*
 179          * Please update $__git_diff_submodule_formats in
 180          * git-completion.bash when you add new formats.
 181          */
 182         else
 183                 return -1;
 184         return 0;
 185 }
 186
 187 int git_config_rename(const char *var, const char *value)
 188 {
 189         if (!value)
 190                 return DIFF_DETECT_RENAME;
 191         if (!strcasecmp(value, "copies") || !strcasecmp(value, "copy"))
 192                 return  DIFF_DETECT_COPY;
 193         return git_config_bool(var,value) ? DIFF_DETECT_RENAME : 0;
 194 }
 195
 196 long parse_algorithm_value(const char *value)
 197 {
 198         if (!value)
 199                 return -1;
 200         else if (!strcasecmp(value, "myers") || !strcasecmp(value, "default"))
 201                 return 0;
 202         else if (!strcasecmp(value, "minimal"))
 203                 return XDF_NEED_MINIMAL;
 204         else if (!strcasecmp(value, "patience"))
 205                 return XDF_PATIENCE_DIFF;
 206         else if (!strcasecmp(value, "histogram"))
 207                 return XDF_HISTOGRAM_DIFF;
 208         /*
 209          * Please update $__git_diff_algorithms in git-completion.bash
 210          * when you add new algorithms.
 211          */
 212         return -1;
 213 }
 214
 215 static int parse_one_token(const char **arg, const char *token)
 216 {
 217         const char *rest;
 218         if (skip_prefix(*arg, token, &rest) && (!*rest || *rest == ',')) {
 219                 *arg = rest;
 220                 return 1;
 221         }
 222         return 0;
 223 }
 224
 225 static int parse_ws_error_highlight(const char *arg)
 226 {
 227         const char *orig_arg = arg;
 228         unsigned val = 0;
 229
 230         while (*arg) {
 231                 if (parse_one_token(&arg, "none"))
 232                         val = 0;
 233                 else if (parse_one_token(&arg, "default"))
 234                         val = WSEH_NEW;
 235                 else if (parse_one_token(&arg, "all"))
 236                         val = WSEH_NEW | WSEH_OLD | WSEH_CONTEXT;
 237                 else if (parse_one_token(&arg, "new"))
 238                         val |= WSEH_NEW;
 239                 else if (parse_one_token(&arg, "old"))
 240                         val |= WSEH_OLD;
 241                 else if (parse_one_token(&arg, "context"))
 242                         val |= WSEH_CONTEXT;
 243                 else {
 244                         return -1 - (int)(arg - orig_arg);
 245                 }
 246                 if (*arg)
 247                         arg++;
 248         }
 249         return val;
 250 }
 251
 252 /*
 253  * These are to give UI layer defaults.
 254  * The core-level commands such as git-diff-files should
 255  * never be affected by the setting of diff.renames
 256  * the user happens to have in the configuration file.
 257  */
 258 void init_diff_ui_defaults(void)
 259 {
 260         diff_detect_rename_default = DIFF_DETECT_RENAME;
 261 }
 262
 263 int git_diff_heuristic_config(const char *var, const char *value, void *cb)
 264 {
 265         if (!strcmp(var, "diff.indentheuristic"))
 266                 diff_indent_heuristic = git_config_bool(var, value);
 267         return 0;
 268 }
 269
 270 static int parse_color_moved(const char *arg)
 271 {
 272         switch (git_parse_maybe_bool(arg)) {
 273         case 0:
 274                 return COLOR_MOVED_NO;
 275         case 1:
 276                 return COLOR_MOVED_DEFAULT;
 277         default:
 278                 break;
 279         }
 280
 281         if (!strcmp(arg, "no"))
 282                 return COLOR_MOVED_NO;
 283         else if (!strcmp(arg, "plain"))
 284                 return COLOR_MOVED_PLAIN;
 285         else if (!strcmp(arg, "blocks"))
 286                 return COLOR_MOVED_BLOCKS;
 287         else if (!strcmp(arg, "zebra"))
 288                 return COLOR_MOVED_ZEBRA;
 289         else if (!strcmp(arg, "default"))
 290                 return COLOR_MOVED_DEFAULT;
 291         else if (!strcmp(arg, "dimmed-zebra"))
 292                 return COLOR_MOVED_ZEBRA_DIM;
 293         else if (!strcmp(arg, "dimmed_zebra"))
 294                 return COLOR_MOVED_ZEBRA_DIM;
 295         else
 296                 return error(_("color moved setting must be one of 'no', 'default', 'blocks', 'zebra', 'dimmed-zebra', 'plain'"));
 297 }
 298
 299 static unsigned parse_color_moved_ws(const char *arg)
 300 {
 301         int ret = 0;
 302         struct string_list l = STRING_LIST_INIT_DUP;
 303         struct string_list_item *i;
 304
 305         string_list_split(&l, arg, ',', -1);
 306
 307         for_each_string_list_item(i, &l) {
 308                 struct strbuf sb = STRBUF_INIT;
 309                 strbuf_addstr(&sb, i->string);
 310                 strbuf_trim(&sb);
 311
 312                 if (!strcmp(sb.buf, "no"))
 313                         ret = 0;
 314                 else if (!strcmp(sb.buf, "ignore-space-change"))
 315                         ret |= XDF_IGNORE_WHITESPACE_CHANGE;
 316                 else if (!strcmp(sb.buf, "ignore-space-at-eol"))
 317                         ret |= XDF_IGNORE_WHITESPACE_AT_EOL;
 318                 else if (!strcmp(sb.buf, "ignore-all-space"))
 319                         ret |= XDF_IGNORE_WHITESPACE;
 320                 else if (!strcmp(sb.buf, "allow-indentation-change"))
 321                         ret |= COLOR_MOVED_WS_ALLOW_INDENTATION_CHANGE;
 322                 else {
 323                         ret |= COLOR_MOVED_WS_ERROR;
 324                         error(_("unknown color-moved-ws mode '%s', possible values are 'ignore-space-change', 'ignore-space-at-eol', 'ignore-all-space', 'allow-indentation-change'"), sb.buf);
 325                 }
 326
 327                 strbuf_release(&sb);
 328         }
 329
 330         if ((ret & COLOR_MOVED_WS_ALLOW_INDENTATION_CHANGE) &&
 331             (ret & XDF_WHITESPACE_FLAGS)) {
 332                 error(_("color-moved-ws: allow-indentation-change cannot be combined with other whitespace modes"));
 333                 ret |= COLOR_MOVED_WS_ERROR;
 334         }
 335
 336         string_list_clear(&l, 0);
 337
 338         return ret;
 339 }
 340
 341 int git_diff_ui_config(const char *var, const char *value, void *cb)
 342 {
 343         if (!strcmp(var, "diff.color") || !strcmp(var, "color.diff")) {
 344                 diff_use_color_default = git_config_colorbool(var, value);
 345                 return 0;
 346         }
 347         if (!strcmp(var, "diff.colormoved")) {
 348                 int cm = parse_color_moved(value);
 349                 if (cm < 0)
 350                         return -1;
 351                 diff_color_moved_default = cm;
 352                 return 0;
 353         }
 354         if (!strcmp(var, "diff.colormovedws")) {
 355                 unsigned cm = parse_color_moved_ws(value);
 356                 if (cm & COLOR_MOVED_WS_ERROR)
 357                         return -1;
 358                 diff_color_moved_ws_default = cm;
 359                 return 0;
 360         }
 361         if (!strcmp(var, "diff.context")) {
 362                 diff_context_default = git_config_int(var, value);
 363                 if (diff_context_default < 0)
 364                         return -1;
 365                 return 0;
 366         }
 367         if (!strcmp(var, "diff.interhunkcontext")) {
 368                 diff_interhunk_context_default = git_config_int(var, value);
 369                 if (diff_interhunk_context_default < 0)
 370                         return -1;
 371                 return 0;
 372         }
 373         if (!strcmp(var, "diff.renames")) {
 374                 diff_detect_rename_default = git_config_rename(var, value);
 375                 return 0;
 376         }
 377         if (!strcmp(var, "diff.autorefreshindex")) {
 378                 diff_auto_refresh_index = git_config_bool(var, value);
 379                 return 0;
 380         }
 381         if (!strcmp(var, "diff.mnemonicprefix")) {
 382                 diff_mnemonic_prefix = git_config_bool(var, value);
 383                 return 0;
 384         }
 385         if (!strcmp(var, "diff.noprefix")) {
 386                 diff_no_prefix = git_config_bool(var, value);
 387                 return 0;
 388         }
 389         if (!strcmp(var, "diff.statgraphwidth")) {
 390                 diff_stat_graph_width = git_config_int(var, value);
 391                 return 0;
 392         }
 393         if (!strcmp(var, "diff.external"))
 394                 return git_config_string(&external_diff_cmd_cfg, var, value);
 395         if (!strcmp(var, "diff.wordregex"))
 396                 return git_config_string(&diff_word_regex_cfg, var, value);
 397         if (!strcmp(var, "diff.orderfile"))
 398                 return git_config_pathname(&diff_order_file_cfg, var, value);
 399
 400         if (!strcmp(var, "diff.ignoresubmodules"))
 401                 handle_ignore_submodules_arg(&default_diff_options, value);
 402
 403         if (!strcmp(var, "diff.submodule")) {
 404                 if (parse_submodule_params(&default_diff_options, value))
 405                         warning(_("Unknown value for 'diff.submodule' config variable: '%s'"),
 406                                 value);
 407                 return 0;
 408         }
 409
 410         if (!strcmp(var, "diff.algorithm")) {
 411                 diff_algorithm = parse_algorithm_value(value);
 412                 if (diff_algorithm < 0)
 413                         return -1;
 414                 return 0;
 415         }
 416
 417         if (!strcmp(var, "diff.wserrorhighlight")) {
 418                 int val = parse_ws_error_highlight(value);
 419                 if (val < 0)
 420                         return -1;
 421                 ws_error_highlight_default = val;
 422                 return 0;
 423         }
 424
 425         if (git_color_config(var, value, cb) < 0)
 426                 return -1;
 427
 428         return git_diff_basic_config(var, value, cb);
 429 }
 430
 431 int git_diff_basic_config(const char *var, const char *value, void *cb)
 432 {
 433         const char *name;
 434
 435         if (!strcmp(var, "diff.renamelimit")) {
 436                 diff_rename_limit_default = git_config_int(var, value);
 437                 return 0;
 438         }
 439
 440         if (userdiff_config(var, value) < 0)
 441                 return -1;
 442
 443         if (skip_prefix(var, "diff.color.", &name) ||
 444             skip_prefix(var, "color.diff.", &name)) {
 445                 int slot = parse_diff_color_slot(name);
 446                 if (slot < 0)
 447                         return 0;
 448                 if (!value)
 449                         return config_error_nonbool(var);
 450                 return color_parse(value, diff_colors[slot]);
 451         }
 452
 453         /* like GNU diff's --suppress-blank-empty option  */
 454         if (!strcmp(var, "diff.suppressblankempty") ||
 455                         /* for backwards compatibility */
 456                         !strcmp(var, "diff.suppress-blank-empty")) {
 457                 diff_suppress_blank_empty = git_config_bool(var, value);
 458                 return 0;
 459         }
 460
 461         if (!strcmp(var, "diff.dirstat")) {
 462                 struct strbuf errmsg = STRBUF_INIT;
 463                 default_diff_options.dirstat_permille = diff_dirstat_permille_default;
 464                 if (parse_dirstat_params(&default_diff_options, value, &errmsg))
 465                         warning(_("Found errors in 'diff.dirstat' config variable:\n%s"),
 466                                 errmsg.buf);
 467                 strbuf_release(&errmsg);
 468                 diff_dirstat_permille_default = default_diff_options.dirstat_permille;
 469                 return 0;
 470         }
 471
 472         if (git_diff_heuristic_config(var, value, cb) < 0)
 473                 return -1;
 474
 475         return git_default_config(var, value, cb);
 476 }
 477
 478 static char *quote_two(const char *one, const char *two)
 479 {
 480         int need_one = quote_c_style(one, NULL, NULL, 1);
 481         int need_two = quote_c_style(two, NULL, NULL, 1);
 482         struct strbuf res = STRBUF_INIT;
 483
 484         if (need_one + need_two) {
 485                 strbuf_addch(&res, '"');
 486                 quote_c_style(one, &res, NULL, 1);
 487                 quote_c_style(two, &res, NULL, 1);
 488                 strbuf_addch(&res, '"');
 489         } else {
 490                 strbuf_addstr(&res, one);
 491                 strbuf_addstr(&res, two);
 492         }
 493         return strbuf_detach(&res, NULL);
 494 }
 495
 496 static const char *external_diff(void)
 497 {
 498         static const char *external_diff_cmd = NULL;
 499         static int done_preparing = 0;
 500
 501         if (done_preparing)
 502                 return external_diff_cmd;
 503         external_diff_cmd = xstrdup_or_null(getenv("GIT_EXTERNAL_DIFF"));
 504         if (!external_diff_cmd)
 505                 external_diff_cmd = external_diff_cmd_cfg;
 506         done_preparing = 1;
 507         return external_diff_cmd;
 508 }
 509
 510 /*
 511  * Keep track of files used for diffing. Sometimes such an entry
 512  * refers to a temporary file, sometimes to an existing file, and
 513  * sometimes to "/dev/null".
 514  */
 515 static struct diff_tempfile {
 516         /*
 517          * filename external diff should read from, or NULL if this
 518          * entry is currently not in use:
 519          */
 520         const char *name;
 521
 522         char hex[GIT_MAX_HEXSZ + 1];
 523         char mode[10];
 524
 525         /*
 526          * If this diff_tempfile instance refers to a temporary file,
 527          * this tempfile object is used to manage its lifetime.
 528          */
 529         struct tempfile *tempfile;
 530 } diff_temp[2];
 531
 532 struct emit_callback {
 533         int color_diff;
 534         unsigned ws_rule;
 535         int blank_at_eof_in_preimage;
 536         int blank_at_eof_in_postimage;
 537         int lno_in_preimage;
 538         int lno_in_postimage;
 539         const char **label_path;
 540         struct diff_words_data *diff_words;
 541         struct diff_options *opt;
 542         struct strbuf *header;
 543 };
 544
 545 static int count_lines(const char *data, int size)
 546 {
 547         int count, ch, completely_empty = 1, nl_just_seen = 0;
 548         count = 0;
 549         while (0 < size--) {
 550                 ch = *data++;
 551                 if (ch == '\n') {
 552                         count++;
 553                         nl_just_seen = 1;
 554                         completely_empty = 0;
 555                 }
 556                 else {
 557                         nl_just_seen = 0;
 558                         completely_empty = 0;
 559                 }
 560         }
 561         if (completely_empty)
 562                 return 0;
 563         if (!nl_just_seen)
 564                 count++; /* no trailing newline */
 565         return count;
 566 }
 567
 568 static int fill_mmfile(struct repository *r, mmfile_t *mf,
 569                        struct diff_filespec *one)
 570 {
 571         if (!DIFF_FILE_VALID(one)) {
 572                 mf->ptr = (char *)""; /* does not matter */
 573                 mf->size = 0;
 574                 return 0;
 575         }
 576         else if (diff_populate_filespec(r, one, 0))
 577                 return -1;
 578
 579         mf->ptr = one->data;
 580         mf->size = one->size;
 581         return 0;
 582 }
 583
 584 /* like fill_mmfile, but only for size, so we can avoid retrieving blob */
 585 static unsigned long diff_filespec_size(struct repository *r,
 586                                         struct diff_filespec *one)
 587 {
 588         if (!DIFF_FILE_VALID(one))
 589                 return 0;
 590         diff_populate_filespec(r, one, CHECK_SIZE_ONLY);
 591         return one->size;
 592 }
 593
 594 static int count_trailing_blank(mmfile_t *mf, unsigned ws_rule)
 595 {
 596         char *ptr = mf->ptr;
 597         long size = mf->size;
 598         int cnt = 0;
 599
 600         if (!size)
 601                 return cnt;
 602         ptr += size - 1; /* pointing at the very end */
 603         if (*ptr != '\n')
 604                 ; /* incomplete line */
 605         else
 606                 ptr--; /* skip the last LF */
 607         while (mf->ptr < ptr) {
 608                 char *prev_eol;
 609                 for (prev_eol = ptr; mf->ptr <= prev_eol; prev_eol--)
 610                         if (*prev_eol == '\n')
 611                                 break;
 612                 if (!ws_blank_line(prev_eol + 1, ptr - prev_eol, ws_rule))
 613                         break;
 614                 cnt++;
 615                 ptr = prev_eol - 1;
 616         }
 617         return cnt;
 618 }
 619
 620 static void check_blank_at_eof(mmfile_t *mf1, mmfile_t *mf2,
 621                                struct emit_callback *ecbdata)
 622 {
 623         int l1, l2, at;
 624         unsigned ws_rule = ecbdata->ws_rule;
 625         l1 = count_trailing_blank(mf1, ws_rule);
 626         l2 = count_trailing_blank(mf2, ws_rule);
 627         if (l2 <= l1) {
 628                 ecbdata->blank_at_eof_in_preimage = 0;
 629                 ecbdata->blank_at_eof_in_postimage = 0;
 630                 return;
 631         }
 632         at = count_lines(mf1->ptr, mf1->size);
 633         ecbdata->blank_at_eof_in_preimage = (at - l1) + 1;
 634
 635         at = count_lines(mf2->ptr, mf2->size);
 636         ecbdata->blank_at_eof_in_postimage = (at - l2) + 1;
 637 }
 638
 639 static void emit_line_0(struct diff_options *o,
 640                         const char *set_sign, const char *set, unsigned reverse, const char *reset,
 641                         int first, const char *line, int len)
 642 {
 643         int has_trailing_newline, has_trailing_carriage_return;
 644         int needs_reset = 0; /* at the end of the line */
 645         FILE *file = o->file;
 646
 647         fputs(diff_line_prefix(o), file);
 648
 649         has_trailing_newline = (len > 0 && line[len-1] == '\n');
 650         if (has_trailing_newline)
 651                 len--;
 652
 653         has_trailing_carriage_return = (len > 0 && line[len-1] == '\r');
 654         if (has_trailing_carriage_return)
 655                 len--;
 656
 657         if (!len && !first)
 658                 goto end_of_line;
 659
 660         if (reverse && want_color(o->use_color)) {
 661                 fputs(GIT_COLOR_REVERSE, file);
 662                 needs_reset = 1;
 663         }
 664
 665         if (set_sign) {
 666                 fputs(set_sign, file);
 667                 needs_reset = 1;
 668         }
 669
 670         if (first)
 671                 fputc(first, file);
 672
 673         if (!len)
 674                 goto end_of_line;
 675
 676         if (set) {
 677                 if (set_sign && set != set_sign)
 678                         fputs(reset, file);
 679                 fputs(set, file);
 680                 needs_reset = 1;
 681         }
 682         fwrite(line, len, 1, file);
 683         needs_reset = 1; /* 'line' may contain color codes. */
 684
 685 end_of_line:
 686         if (needs_reset)
 687                 fputs(reset, file);
 688         if (has_trailing_carriage_return)
 689                 fputc('\r', file);
 690         if (has_trailing_newline)
 691                 fputc('\n', file);
 692 }
 693
 694 static void emit_line(struct diff_options *o, const char *set, const char *reset,
 695                       const char *line, int len)
 696 {
 697         emit_line_0(o, set, NULL, 0, reset, 0, line, len);
 698 }
 699
 700 enum diff_symbol {
 701         DIFF_SYMBOL_BINARY_DIFF_HEADER,
 702         DIFF_SYMBOL_BINARY_DIFF_HEADER_DELTA,
 703         DIFF_SYMBOL_BINARY_DIFF_HEADER_LITERAL,
 704         DIFF_SYMBOL_BINARY_DIFF_BODY,
 705         DIFF_SYMBOL_BINARY_DIFF_FOOTER,
 706         DIFF_SYMBOL_STATS_SUMMARY_NO_FILES,
 707         DIFF_SYMBOL_STATS_SUMMARY_ABBREV,
 708         DIFF_SYMBOL_STATS_SUMMARY_INSERTS_DELETES,
 709         DIFF_SYMBOL_STATS_LINE,
 710         DIFF_SYMBOL_WORD_DIFF,
 711         DIFF_SYMBOL_STAT_SEP,
 712         DIFF_SYMBOL_SUMMARY,
 713         DIFF_SYMBOL_SUBMODULE_ADD,
 714         DIFF_SYMBOL_SUBMODULE_DEL,
 715         DIFF_SYMBOL_SUBMODULE_UNTRACKED,
 716         DIFF_SYMBOL_SUBMODULE_MODIFIED,
 717         DIFF_SYMBOL_SUBMODULE_HEADER,
 718         DIFF_SYMBOL_SUBMODULE_ERROR,
 719         DIFF_SYMBOL_SUBMODULE_PIPETHROUGH,
 720         DIFF_SYMBOL_REWRITE_DIFF,
 721         DIFF_SYMBOL_BINARY_FILES,
 722         DIFF_SYMBOL_HEADER,
 723         DIFF_SYMBOL_FILEPAIR_PLUS,
 724         DIFF_SYMBOL_FILEPAIR_MINUS,
 725         DIFF_SYMBOL_WORDS_PORCELAIN,
 726         DIFF_SYMBOL_WORDS,
 727         DIFF_SYMBOL_CONTEXT,
 728         DIFF_SYMBOL_CONTEXT_INCOMPLETE,
 729         DIFF_SYMBOL_PLUS,
 730         DIFF_SYMBOL_MINUS,
 731         DIFF_SYMBOL_NO_LF_EOF,
 732         DIFF_SYMBOL_CONTEXT_FRAGINFO,
 733         DIFF_SYMBOL_CONTEXT_MARKER,
 734         DIFF_SYMBOL_SEPARATOR
 735 };
 736 /*
 737  * Flags for content lines:
 738  * 0..12 are whitespace rules
 739  * 13-15 are WSEH_NEW | WSEH_OLD | WSEH_CONTEXT
 740  * 16 is marking if the line is blank at EOF
 741  */
 742 #define DIFF_SYMBOL_CONTENT_BLANK_LINE_EOF      (1<<16)
 743 #define DIFF_SYMBOL_MOVED_LINE                  (1<<17)
 744 #define DIFF_SYMBOL_MOVED_LINE_ALT              (1<<18)
 745 #define DIFF_SYMBOL_MOVED_LINE_UNINTERESTING    (1<<19)
 746 #define DIFF_SYMBOL_CONTENT_WS_MASK (WSEH_NEW | WSEH_OLD | WSEH_CONTEXT | WS_RULE_MASK)
 747
 748 /*
 749  * This struct is used when we need to buffer the output of the diff output.
 750  *
 751  * NEEDSWORK: Instead of storing a copy of the line, add an offset pointer
 752  * into the pre/post image file. This pointer could be a union with the
 753  * line pointer. By storing an offset into the file instead of the literal line,
 754  * we can decrease the memory footprint for the buffered output. At first we
 755  * may want to only have indirection for the content lines, but we could also
 756  * enhance the state for emitting prefabricated lines, e.g. the similarity
 757  * score line or hunk/file headers would only need to store a number or path
 758  * and then the output can be constructed later on depending on state.
 759  */
 760 struct emitted_diff_symbol {
 761         const char *line;
 762         int len;
 763         int flags;
 764         int indent_off;   /* Offset to first non-whitespace character */
 765         int indent_width; /* The visual width of the indentation */
 766         enum diff_symbol s;
 767 };
 768 #define EMITTED_DIFF_SYMBOL_INIT {NULL}
 769
 770 struct emitted_diff_symbols {
 771         struct emitted_diff_symbol *buf;
 772         int nr, alloc;
 773 };
 774 #define EMITTED_DIFF_SYMBOLS_INIT {NULL, 0, 0}
 775
 776 static void append_emitted_diff_symbol(struct diff_options *o,
 777                                        struct emitted_diff_symbol *e)
 778 {
 779         struct emitted_diff_symbol *f;
 780
 781         ALLOC_GROW(o->emitted_symbols->buf,
 782                    o->emitted_symbols->nr + 1,
 783                    o->emitted_symbols->alloc);
 784         f = &o->emitted_symbols->buf[o->emitted_symbols->nr++];
 785
 786         memcpy(f, e, sizeof(struct emitted_diff_symbol));
 787         f->line = e->line ? xmemdupz(e->line, e->len) : NULL;
 788 }
 789
 790 struct moved_entry {
 791         struct hashmap_entry ent;
 792         const struct emitted_diff_symbol *es;
 793         struct moved_entry *next_line;
 794 };
 795
 796 struct moved_block {
 797         struct moved_entry *match;
 798         int wsd; /* The whitespace delta of this block */
 799 };
 800
 801 static void moved_block_clear(struct moved_block *b)
 802 {
 803         memset(b, 0, sizeof(*b));
 804 }
 805
 806 #define INDENT_BLANKLINE INT_MIN
 807
 808 static void fill_es_indent_data(struct emitted_diff_symbol *es)
 809 {
 810         unsigned int off = 0, i;
 811         int width = 0, tab_width = es->flags & WS_TAB_WIDTH_MASK;
 812         const char *s = es->line;
 813         const int len = es->len;
 814
 815         /* skip any \v \f \r at start of indentation */
 816         while (s[off] == '\f' || s[off] == '\v' ||
 817                (s[off] == '\r' && off < len - 1))
 818                 off++;
 819
 820         /* calculate the visual width of indentation */
 821         while(1) {
 822                 if (s[off] == ' ') {
 823                         width++;
 824                         off++;
 825                 } else if (s[off] == '\t') {
 826                         width += tab_width - (width % tab_width);
 827                         while (s[++off] == '\t')
 828                                 width += tab_width;
 829                 } else {
 830                         break;
 831                 }
 832         }
 833
 834         /* check if this line is blank */
 835         for (i = off; i < len; i++)
 836                 if (!isspace(s[i]))
 837                     break;
 838
 839         if (i == len) {
 840                 es->indent_width = INDENT_BLANKLINE;
 841                 es->indent_off = len;
 842         } else {
 843                 es->indent_off = off;
 844                 es->indent_width = width;
 845         }
 846 }
 847
 848 static int compute_ws_delta(const struct emitted_diff_symbol *a,
 849                             const struct emitted_diff_symbol *b,
 850                             int *out)
 851 {
 852         int a_len = a->len,
 853             b_len = b->len,
 854             a_off = a->indent_off,
 855             a_width = a->indent_width,
 856             b_off = b->indent_off,
 857             b_width = b->indent_width;
 858         int delta;
 859
 860         if (a_width == INDENT_BLANKLINE && b_width == INDENT_BLANKLINE) {
 861                 *out = INDENT_BLANKLINE;
 862                 return 1;
 863         }
 864
 865         if (a->s == DIFF_SYMBOL_PLUS)
 866                 delta = a_width - b_width;
 867         else
 868                 delta = b_width - a_width;
 869
 870         if (a_len - a_off != b_len - b_off ||
 871             memcmp(a->line + a_off, b->line + b_off, a_len - a_off))
 872                 return 0;
 873
 874         *out = delta;
 875
 876         return 1;
 877 }
 878
 879 static int cmp_in_block_with_wsd(const struct diff_options *o,
 880                                  const struct moved_entry *cur,
 881                                  const struct moved_entry *match,
 882                                  struct moved_block *pmb,
 883                                  int n)
 884 {
 885         struct emitted_diff_symbol *l = &o->emitted_symbols->buf[n];
 886         int al = cur->es->len, bl = match->es->len, cl = l->len;
 887         const char *a = cur->es->line,
 888                    *b = match->es->line,
 889                    *c = l->line;
 890         int a_off = cur->es->indent_off,
 891             a_width = cur->es->indent_width,
 892             c_off = l->indent_off,
 893             c_width = l->indent_width;
 894         int delta;
 895
 896         /*
 897          * We need to check if 'cur' is equal to 'match'.  As those
 898          * are from the same (+/-) side, we do not need to adjust for
 899          * indent changes. However these were found using fuzzy
 900          * matching so we do have to check if they are equal. Here we
 901          * just check the lengths. We delay calling memcmp() to check
 902          * the contents until later as if the length comparison for a
 903          * and c fails we can avoid the call all together.
 904          */
 905         if (al != bl)
 906                 return 1;
 907
 908         /* If 'l' and 'cur' are both blank then they match. */
 909         if (a_width == INDENT_BLANKLINE && c_width == INDENT_BLANKLINE)
 910                 return 0;
 911
 912         /*
 913          * The indent changes of the block are known and stored in pmb->wsd;
 914          * however we need to check if the indent changes of the current line
 915          * match those of the current block and that the text of 'l' and 'cur'
 916          * after the indentation match.
 917          */
 918         if (cur->es->s == DIFF_SYMBOL_PLUS)
 919                 delta = a_width - c_width;
 920         else
 921                 delta = c_width - a_width;
 922
 923         /*
 924          * If the previous lines of this block were all blank then set its
 925          * whitespace delta.
 926          */
 927         if (pmb->wsd == INDENT_BLANKLINE)
 928                 pmb->wsd = delta;
 929
 930         return !(delta == pmb->wsd && al - a_off == cl - c_off &&
 931                  !memcmp(a, b, al) && !
 932                  memcmp(a + a_off, c + c_off, al - a_off));
 933 }
 934
 935 static int moved_entry_cmp(const void *hashmap_cmp_fn_data,
 936                            const struct hashmap_entry *eptr,
 937                            const struct hashmap_entry *entry_or_key,
 938                            const void *keydata)
 939 {
 940         const struct diff_options *diffopt = hashmap_cmp_fn_data;
 941         const struct moved_entry *a, *b;
 942         unsigned flags = diffopt->color_moved_ws_handling
 943                          & XDF_WHITESPACE_FLAGS;
 944
 945         a = container_of(eptr, const struct moved_entry, ent);
 946         b = container_of(entry_or_key, const struct moved_entry, ent);
 947
 948         if (diffopt->color_moved_ws_handling &
 949             COLOR_MOVED_WS_ALLOW_INDENTATION_CHANGE)
 950                 /*
 951                  * As there is not specific white space config given,
 952                  * we'd need to check for a new block, so ignore all
 953                  * white space. The setup of the white space
 954                  * configuration for the next block is done else where
 955                  */
 956                 flags |= XDF_IGNORE_WHITESPACE;
 957
 958         return !xdiff_compare_lines(a->es->line, a->es->len,
 959                                     b->es->line, b->es->len,
 960                                     flags);
 961 }
 962
 963 static struct moved_entry *prepare_entry(struct diff_options *o,
 964                                          int line_no)
 965 {
 966         struct moved_entry *ret = xmalloc(sizeof(*ret));
 967         struct emitted_diff_symbol *l = &o->emitted_symbols->buf[line_no];
 968         unsigned flags = o->color_moved_ws_handling & XDF_WHITESPACE_FLAGS;
 969         unsigned int hash = xdiff_hash_string(l->line, l->len, flags);
 970
 971         hashmap_entry_init(&ret->ent, hash);
 972         ret->es = l;
 973         ret->next_line = NULL;
 974
 975         return ret;
 976 }
 977
 978 static void add_lines_to_move_detection(struct diff_options *o,
 979                                         struct hashmap *add_lines,
 980                                         struct hashmap *del_lines)
 981 {
 982         struct moved_entry *prev_line = NULL;
 983
 984         int n;
 985         for (n = 0; n < o->emitted_symbols->nr; n++) {
 986                 struct hashmap *hm;
 987                 struct moved_entry *key;
 988
 989                 switch (o->emitted_symbols->buf[n].s) {
 990                 case DIFF_SYMBOL_PLUS:
 991                         hm = add_lines;
 992                         break;
 993                 case DIFF_SYMBOL_MINUS:
 994                         hm = del_lines;
 995                         break;
 996                 default:
 997                         prev_line = NULL;
 998                         continue;
 999                 }
1000
1001                 if (o->color_moved_ws_handling &
1002                     COLOR_MOVED_WS_ALLOW_INDENTATION_CHANGE)
1003                         fill_es_indent_data(&o->emitted_symbols->buf[n]);
1004                 key = prepare_entry(o, n);
1005                 if (prev_line && prev_line->es->s == o->emitted_symbols->buf[n].s)
1006                         prev_line->next_line = key;
1007
1008                 hashmap_add(hm, &key->ent);
1009                 prev_line = key;
1010         }
1011 }
1012
1013 static void pmb_advance_or_null(struct diff_options *o,
1014                                 struct moved_entry *match,
1015                                 struct hashmap *hm,
1016                                 struct moved_block *pmb,
1017                                 int pmb_nr)
1018 {
1019         int i;
1020         for (i = 0; i < pmb_nr; i++) {
1021                 struct moved_entry *prev = pmb[i].match;
1022                 struct moved_entry *cur = (prev && prev->next_line) ?
1023                                 prev->next_line : NULL;
1024                 if (cur && !hm->cmpfn(o, &cur->ent, &match->ent, NULL)) {
1025                         pmb[i].match = cur;
1026                 } else {
1027                         pmb[i].match = NULL;
1028                 }
1029         }
1030 }
1031
1032 static void pmb_advance_or_null_multi_match(struct diff_options *o,
1033                                             struct moved_entry *match,
1034                                             struct hashmap *hm,
1035                                             struct moved_block *pmb,
1036                                             int pmb_nr, int n)
1037 {
1038         int i;
1039         char *got_match = xcalloc(1, pmb_nr);
1040
1041         hashmap_for_each_entry_from(hm, match, struct moved_entry, ent) {
1042                 for (i = 0; i < pmb_nr; i++) {
1043                         struct moved_entry *prev = pmb[i].match;
1044                         struct moved_entry *cur = (prev && prev->next_line) ?
1045                                         prev->next_line : NULL;
1046                         if (!cur)
1047                                 continue;
1048                         if (!cmp_in_block_with_wsd(o, cur, match, &pmb[i], n))
1049                                 got_match[i] |= 1;
1050                 }
1051         }
1052
1053         for (i = 0; i < pmb_nr; i++) {
1054                 if (got_match[i]) {
1055                         /* Advance to the next line */
1056                         pmb[i].match = pmb[i].match->next_line;
1057                 } else {
1058                         moved_block_clear(&pmb[i]);
1059                 }
1060         }
1061
1062         free(got_match);
1063 }
1064
1065 static int shrink_potential_moved_blocks(struct moved_block *pmb,
1066                                          int pmb_nr)
1067 {
1068         int lp, rp;
1069
1070         /* Shrink the set of potential block to the remaining running */
1071         for (lp = 0, rp = pmb_nr - 1; lp <= rp;) {
1072                 while (lp < pmb_nr && pmb[lp].match)
1073                         lp++;
1074                 /* lp points at the first NULL now */
1075
1076                 while (rp > -1 && !pmb[rp].match)
1077                         rp--;
1078                 /* rp points at the last non-NULL */
1079
1080                 if (lp < pmb_nr && rp > -1 && lp < rp) {
1081                         pmb[lp] = pmb[rp];
1082                         memset(&pmb[rp], 0, sizeof(pmb[rp]));
1083                         rp--;
1084                         lp++;
1085                 }
1086         }
1087
1088         /* Remember the number of running sets */
1089         return rp + 1;
1090 }
1091
1092 /*
1093  * If o->color_moved is COLOR_MOVED_PLAIN, this function does nothing.
1094  *
1095  * Otherwise, if the last block has fewer alphanumeric characters than
1096  * COLOR_MOVED_MIN_ALNUM_COUNT, unset DIFF_SYMBOL_MOVED_LINE on all lines in
1097  * that block.
1098  *
1099  * The last block consists of the (n - block_length)'th line up to but not
1100  * including the nth line.
1101  *
1102  * Returns 0 if the last block is empty or is unset by this function, non zero
1103  * otherwise.
1104  *
1105  * NEEDSWORK: This uses the same heuristic as blame_entry_score() in blame.c.
1106  * Think of a way to unify them.
1107  */
1108 static int adjust_last_block(struct diff_options *o, int n, int block_length)
1109 {
1110         int i, alnum_count = 0;
1111         if (o->color_moved == COLOR_MOVED_PLAIN)
1112                 return block_length;
1113         for (i = 1; i < block_length + 1; i++) {
1114                 const char *c = o->emitted_symbols->buf[n - i].line;
1115                 for (; *c; c++) {
1116                         if (!isalnum(*c))
1117                                 continue;
1118                         alnum_count++;
1119                         if (alnum_count >= COLOR_MOVED_MIN_ALNUM_COUNT)
1120                                 return 1;
1121                 }
1122         }
1123         for (i = 1; i < block_length + 1; i++)
1124                 o->emitted_symbols->buf[n - i].flags &= ~DIFF_SYMBOL_MOVED_LINE;
1125         return 0;
1126 }
1127
1128 /* Find blocks of moved code, delegate actual coloring decision to helper */
1129 static void mark_color_as_moved(struct diff_options *o,
1130                                 struct hashmap *add_lines,
1131                                 struct hashmap *del_lines)
1132 {
1133         struct moved_block *pmb = NULL; /* potentially moved blocks */
1134         int pmb_nr = 0, pmb_alloc = 0;
1135         int n, flipped_block = 0, block_length = 0;
1136
1137
1138         for (n = 0; n < o->emitted_symbols->nr; n++) {
1139                 struct hashmap *hm = NULL;
1140                 struct moved_entry *key;
1141                 struct moved_entry *match = NULL;
1142                 struct emitted_diff_symbol *l = &o->emitted_symbols->buf[n];
1143                 enum diff_symbol last_symbol = 0;
1144
1145                 switch (l->s) {
1146                 case DIFF_SYMBOL_PLUS:
1147                         hm = del_lines;
1148                         key = prepare_entry(o, n);
1149                         match = hashmap_get_entry(hm, key, NULL,
1150                                                 struct moved_entry, ent);
1151                         free(key);
1152                         break;
1153                 case DIFF_SYMBOL_MINUS:
1154                         hm = add_lines;
1155                         key = prepare_entry(o, n);
1156                         match = hashmap_get_entry(hm, key, NULL,
1157                                                 struct moved_entry, ent);
1158                         free(key);
1159                         break;
1160                 default:
1161                         flipped_block = 0;
1162                 }
1163
1164                 if (!match) {
1165                         int i;
1166
1167                         adjust_last_block(o, n, block_length);
1168                         for(i = 0; i < pmb_nr; i++)
1169                                 moved_block_clear(&pmb[i]);
1170                         pmb_nr = 0;
1171                         block_length = 0;
1172                         flipped_block = 0;
1173                         last_symbol = l->s;
1174                         continue;
1175                 }
1176
1177                 if (o->color_moved == COLOR_MOVED_PLAIN) {
1178                         last_symbol = l->s;
1179                         l->flags |= DIFF_SYMBOL_MOVED_LINE;
1180                         continue;
1181                 }
1182
1183                 if (o->color_moved_ws_handling &
1184                     COLOR_MOVED_WS_ALLOW_INDENTATION_CHANGE)
1185                         pmb_advance_or_null_multi_match(o, match, hm, pmb, pmb_nr, n);
1186                 else
1187                         pmb_advance_or_null(o, match, hm, pmb, pmb_nr);
1188
1189                 pmb_nr = shrink_potential_moved_blocks(pmb, pmb_nr);
1190
1191                 if (pmb_nr == 0) {
1192                         /*
1193                          * The current line is the start of a new block.
1194                          * Setup the set of potential blocks.
1195                          */
1196                         hashmap_for_each_entry_from(hm, match,
1197                                                 struct moved_entry, ent) {
1198                                 ALLOC_GROW(pmb, pmb_nr + 1, pmb_alloc);
1199                                 if (o->color_moved_ws_handling &
1200                                     COLOR_MOVED_WS_ALLOW_INDENTATION_CHANGE) {
1201                                         if (compute_ws_delta(l, match->es,
1202                                                              &pmb[pmb_nr].wsd))
1203                                                 pmb[pmb_nr++].match = match;
1204                                 } else {
1205                                         pmb[pmb_nr].wsd = 0;
1206                                         pmb[pmb_nr++].match = match;
1207                                 }
1208                         }
1209
1210                         if (adjust_last_block(o, n, block_length) &&
1211                             pmb_nr && last_symbol != l->s)
1212                                 flipped_block = (flipped_block + 1) % 2;
1213                         else
1214                                 flipped_block = 0;
1215
1216                         block_length = 0;
1217                 }
1218
1219                 if (pmb_nr) {
1220                         block_length++;
1221                         l->flags |= DIFF_SYMBOL_MOVED_LINE;
1222                         if (flipped_block && o->color_moved != COLOR_MOVED_BLOCKS)
1223                                 l->flags |= DIFF_SYMBOL_MOVED_LINE_ALT;
1224                 }
1225                 last_symbol = l->s;
1226         }
1227         adjust_last_block(o, n, block_length);
1228
1229         for(n = 0; n < pmb_nr; n++)
1230                 moved_block_clear(&pmb[n]);
1231         free(pmb);
1232 }
1233
1234 #define DIFF_SYMBOL_MOVED_LINE_ZEBRA_MASK \
1235   (DIFF_SYMBOL_MOVED_LINE | DIFF_SYMBOL_MOVED_LINE_ALT)
1236 static void dim_moved_lines(struct diff_options *o)
1237 {
1238         int n;
1239         for (n = 0; n < o->emitted_symbols->nr; n++) {
1240                 struct emitted_diff_symbol *prev = (n != 0) ?
1241                                 &o->emitted_symbols->buf[n - 1] : NULL;
1242                 struct emitted_diff_symbol *l = &o->emitted_symbols->buf[n];
1243                 struct emitted_diff_symbol *next =
1244                                 (n < o->emitted_symbols->nr - 1) ?
1245                                 &o->emitted_symbols->buf[n + 1] : NULL;
1246
1247                 /* Not a plus or minus line? */
1248                 if (l->s != DIFF_SYMBOL_PLUS && l->s != DIFF_SYMBOL_MINUS)
1249                         continue;
1250
1251                 /* Not a moved line? */
1252                 if (!(l->flags & DIFF_SYMBOL_MOVED_LINE))
1253                         continue;
1254
1255                 /*
1256                  * If prev or next are not a plus or minus line,
1257                  * pretend they don't exist
1258                  */
1259                 if (prev && prev->s != DIFF_SYMBOL_PLUS &&
1260                             prev->s != DIFF_SYMBOL_MINUS)
1261                         prev = NULL;
1262                 if (next && next->s != DIFF_SYMBOL_PLUS &&
1263                             next->s != DIFF_SYMBOL_MINUS)
1264                         next = NULL;
1265
1266                 /* Inside a block? */
1267                 if ((prev &&
1268                     (prev->flags & DIFF_SYMBOL_MOVED_LINE_ZEBRA_MASK) ==
1269                     (l->flags & DIFF_SYMBOL_MOVED_LINE_ZEBRA_MASK)) &&
1270                     (next &&
1271                     (next->flags & DIFF_SYMBOL_MOVED_LINE_ZEBRA_MASK) ==
1272                     (l->flags & DIFF_SYMBOL_MOVED_LINE_ZEBRA_MASK))) {
1273                         l->flags |= DIFF_SYMBOL_MOVED_LINE_UNINTERESTING;
1274                         continue;
1275                 }
1276
1277                 /* Check if we are at an interesting bound: */
1278                 if (prev && (prev->flags & DIFF_SYMBOL_MOVED_LINE) &&
1279                     (prev->flags & DIFF_SYMBOL_MOVED_LINE_ALT) !=
1280                        (l->flags & DIFF_SYMBOL_MOVED_LINE_ALT))
1281                         continue;
1282                 if (next && (next->flags & DIFF_SYMBOL_MOVED_LINE) &&
1283                     (next->flags & DIFF_SYMBOL_MOVED_LINE_ALT) !=
1284                        (l->flags & DIFF_SYMBOL_MOVED_LINE_ALT))
1285                         continue;
1286
1287                 /*
1288                  * The boundary to prev and next are not interesting,
1289                  * so this line is not interesting as a whole
1290                  */
1291                 l->flags |= DIFF_SYMBOL_MOVED_LINE_UNINTERESTING;
1292         }
1293 }
1294
1295 static void emit_line_ws_markup(struct diff_options *o,
1296                                 const char *set_sign, const char *set,
1297                                 const char *reset,
1298                                 int sign_index, const char *line, int len,
1299                                 unsigned ws_rule, int blank_at_eof)
1300 {
1301         const char *ws = NULL;
1302         int sign = o->output_indicators[sign_index];
1303
1304         if (o->ws_error_highlight & ws_rule) {
1305                 ws = diff_get_color_opt(o, DIFF_WHITESPACE);
1306                 if (!*ws)
1307                         ws = NULL;
1308         }
1309
1310         if (!ws && !set_sign)
1311                 emit_line_0(o, set, NULL, 0, reset, sign, line, len);
1312         else if (!ws) {
1313                 emit_line_0(o, set_sign, set, !!set_sign, reset, sign, line, len);
1314         } else if (blank_at_eof)
1315                 /* Blank line at EOF - paint '+' as well */
1316                 emit_line_0(o, ws, NULL, 0, reset, sign, line, len);
1317         else {
1318                 /* Emit just the prefix, then the rest. */
1319                 emit_line_0(o, set_sign ? set_sign : set, NULL, !!set_sign, reset,
1320                             sign, "", 0);
1321                 ws_check_emit(line, len, ws_rule,
1322                               o->file, set, reset, ws);
1323         }
1324 }
1325
1326 static void emit_diff_symbol_from_struct(struct diff_options *o,
1327                                          struct emitted_diff_symbol *eds)
1328 {
1329         static const char *nneof = " No newline at end of file\n";
1330         const char *context, *reset, *set, *set_sign, *meta, *fraginfo;
1331         struct strbuf sb = STRBUF_INIT;
1332
1333         enum diff_symbol s = eds->s;
1334         const char *line = eds->line;
1335         int len = eds->len;
1336         unsigned flags = eds->flags;
1337
1338         switch (s) {
1339         case DIFF_SYMBOL_NO_LF_EOF:
1340                 context = diff_get_color_opt(o, DIFF_CONTEXT);
1341                 reset = diff_get_color_opt(o, DIFF_RESET);
1342                 putc('\n', o->file);
1343                 emit_line_0(o, context, NULL, 0, reset, '\\',
1344                             nneof, strlen(nneof));
1345                 break;
1346         case DIFF_SYMBOL_SUBMODULE_HEADER:
1347         case DIFF_SYMBOL_SUBMODULE_ERROR:
1348         case DIFF_SYMBOL_SUBMODULE_PIPETHROUGH:
1349         case DIFF_SYMBOL_STATS_SUMMARY_INSERTS_DELETES:
1350         case DIFF_SYMBOL_SUMMARY:
1351         case DIFF_SYMBOL_STATS_LINE:
1352         case DIFF_SYMBOL_BINARY_DIFF_BODY:
1353         case DIFF_SYMBOL_CONTEXT_FRAGINFO:
1354                 emit_line(o, "", "", line, len);
1355                 break;
1356         case DIFF_SYMBOL_CONTEXT_INCOMPLETE:
1357         case DIFF_SYMBOL_CONTEXT_MARKER:
1358                 context = diff_get_color_opt(o, DIFF_CONTEXT);
1359                 reset = diff_get_color_opt(o, DIFF_RESET);
1360                 emit_line(o, context, reset, line, len);
1361                 break;
1362         case DIFF_SYMBOL_SEPARATOR:
1363                 fprintf(o->file, "%s%c",
1364                         diff_line_prefix(o),
1365                         o->line_termination);
1366                 break;
1367         case DIFF_SYMBOL_CONTEXT:
1368                 set = diff_get_color_opt(o, DIFF_CONTEXT);
1369                 reset = diff_get_color_opt(o, DIFF_RESET);
1370                 set_sign = NULL;
1371                 if (o->flags.dual_color_diffed_diffs) {
1372                         char c = !len ? 0 : line[0];
1373
1374                         if (c == '+')
1375                                 set = diff_get_color_opt(o, DIFF_FILE_NEW);
1376                         else if (c == '@')
1377                                 set = diff_get_color_opt(o, DIFF_FRAGINFO);
1378                         else if (c == '-')
1379                                 set = diff_get_color_opt(o, DIFF_FILE_OLD);
1380                 }
1381                 emit_line_ws_markup(o, set_sign, set, reset,
1382                                     OUTPUT_INDICATOR_CONTEXT, line, len,
1383                                     flags & (DIFF_SYMBOL_CONTENT_WS_MASK), 0);
1384                 break;
1385         case DIFF_SYMBOL_PLUS:
1386                 switch (flags & (DIFF_SYMBOL_MOVED_LINE |
1387                                  DIFF_SYMBOL_MOVED_LINE_ALT |
1388                                  DIFF_SYMBOL_MOVED_LINE_UNINTERESTING)) {
1389                 case DIFF_SYMBOL_MOVED_LINE |
1390                      DIFF_SYMBOL_MOVED_LINE_ALT |
1391                      DIFF_SYMBOL_MOVED_LINE_UNINTERESTING:
1392                         set = diff_get_color_opt(o, DIFF_FILE_NEW_MOVED_ALT_DIM);
1393                         break;
1394                 case DIFF_SYMBOL_MOVED_LINE |
1395                      DIFF_SYMBOL_MOVED_LINE_ALT:
1396                         set = diff_get_color_opt(o, DIFF_FILE_NEW_MOVED_ALT);
1397                         break;
1398                 case DIFF_SYMBOL_MOVED_LINE |
1399                      DIFF_SYMBOL_MOVED_LINE_UNINTERESTING:
1400                         set = diff_get_color_opt(o, DIFF_FILE_NEW_MOVED_DIM);
1401                         break;
1402                 case DIFF_SYMBOL_MOVED_LINE:
1403                         set = diff_get_color_opt(o, DIFF_FILE_NEW_MOVED);
1404                         break;
1405                 default:
1406                         set = diff_get_color_opt(o, DIFF_FILE_NEW);
1407                 }
1408                 reset = diff_get_color_opt(o, DIFF_RESET);
1409                 if (!o->flags.dual_color_diffed_diffs)
1410                         set_sign = NULL;
1411                 else {
1412                         char c = !len ? 0 : line[0];
1413
1414                         set_sign = set;
1415                         if (c == '-')
1416                                 set = diff_get_color_opt(o, DIFF_FILE_OLD_BOLD);
1417                         else if (c == '@')
1418                                 set = diff_get_color_opt(o, DIFF_FRAGINFO);
1419                         else if (c == '+')
1420                                 set = diff_get_color_opt(o, DIFF_FILE_NEW_BOLD);
1421                         else
1422                                 set = diff_get_color_opt(o, DIFF_CONTEXT_BOLD);
1423                         flags &= ~DIFF_SYMBOL_CONTENT_WS_MASK;
1424                 }
1425                 emit_line_ws_markup(o, set_sign, set, reset,
1426                                     OUTPUT_INDICATOR_NEW, line, len,
1427                                     flags & DIFF_SYMBOL_CONTENT_WS_MASK,
1428                                     flags & DIFF_SYMBOL_CONTENT_BLANK_LINE_EOF);
1429                 break;
1430         case DIFF_SYMBOL_MINUS:
1431                 switch (flags & (DIFF_SYMBOL_MOVED_LINE |
1432                                  DIFF_SYMBOL_MOVED_LINE_ALT |
1433                                  DIFF_SYMBOL_MOVED_LINE_UNINTERESTING)) {
1434                 case DIFF_SYMBOL_MOVED_LINE |
1435                      DIFF_SYMBOL_MOVED_LINE_ALT |
1436                      DIFF_SYMBOL_MOVED_LINE_UNINTERESTING:
1437                         set = diff_get_color_opt(o, DIFF_FILE_OLD_MOVED_ALT_DIM);
1438                         break;
1439                 case DIFF_SYMBOL_MOVED_LINE |
1440                      DIFF_SYMBOL_MOVED_LINE_ALT:
1441                         set = diff_get_color_opt(o, DIFF_FILE_OLD_MOVED_ALT);
1442                         break;
1443                 case DIFF_SYMBOL_MOVED_LINE |
1444                      DIFF_SYMBOL_MOVED_LINE_UNINTERESTING:
1445                         set = diff_get_color_opt(o, DIFF_FILE_OLD_MOVED_DIM);
1446                         break;
1447                 case DIFF_SYMBOL_MOVED_LINE:
1448                         set = diff_get_color_opt(o, DIFF_FILE_OLD_MOVED);
1449                         break;
1450                 default:
1451                         set = diff_get_color_opt(o, DIFF_FILE_OLD);
1452                 }
1453                 reset = diff_get_color_opt(o, DIFF_RESET);
1454                 if (!o->flags.dual_color_diffed_diffs)
1455                         set_sign = NULL;
1456                 else {
1457                         char c = !len ? 0 : line[0];
1458
1459                         set_sign = set;
1460                         if (c == '+')
1461                                 set = diff_get_color_opt(o, DIFF_FILE_NEW_DIM);
1462                         else if (c == '@')
1463                                 set = diff_get_color_opt(o, DIFF_FRAGINFO);
1464                         else if (c == '-')
1465                                 set = diff_get_color_opt(o, DIFF_FILE_OLD_DIM);
1466                         else
1467                                 set = diff_get_color_opt(o, DIFF_CONTEXT_DIM);
1468                 }
1469                 emit_line_ws_markup(o, set_sign, set, reset,
1470                                     OUTPUT_INDICATOR_OLD, line, len,
1471                                     flags & DIFF_SYMBOL_CONTENT_WS_MASK, 0);
1472                 break;
1473         case DIFF_SYMBOL_WORDS_PORCELAIN:
1474                 context = diff_get_color_opt(o, DIFF_CONTEXT);
1475                 reset = diff_get_color_opt(o, DIFF_RESET);
1476                 emit_line(o, context, reset, line, len);
1477                 fputs("~\n", o->file);
1478                 break;
1479         case DIFF_SYMBOL_WORDS:
1480                 context = diff_get_color_opt(o, DIFF_CONTEXT);
1481                 reset = diff_get_color_opt(o, DIFF_RESET);
1482                 /*
1483                  * Skip the prefix character, if any.  With
1484                  * diff_suppress_blank_empty, there may be
1485                  * none.
1486                  */
1487                 if (line[0] != '\n') {
1488                         line++;
1489                         len--;
1490                 }
1491                 emit_line(o, context, reset, line, len);
1492                 break;
1493         case DIFF_SYMBOL_FILEPAIR_PLUS:
1494                 meta = diff_get_color_opt(o, DIFF_METAINFO);
1495                 reset = diff_get_color_opt(o, DIFF_RESET);
1496                 fprintf(o->file, "%s%s+++ %s%s%s\n", diff_line_prefix(o), meta,
1497                         line, reset,
1498                         strchr(line, ' ') ? "\t" : "");
1499                 break;
1500         case DIFF_SYMBOL_FILEPAIR_MINUS:
1501                 meta = diff_get_color_opt(o, DIFF_METAINFO);
1502                 reset = diff_get_color_opt(o, DIFF_RESET);
1503                 fprintf(o->file, "%s%s--- %s%s%s\n", diff_line_prefix(o), meta,
1504                         line, reset,
1505                         strchr(line, ' ') ? "\t" : "");
1506                 break;
1507         case DIFF_SYMBOL_BINARY_FILES:
1508         case DIFF_SYMBOL_HEADER:
1509                 fprintf(o->file, "%s", line);
1510                 break;
1511         case DIFF_SYMBOL_BINARY_DIFF_HEADER:
1512                 fprintf(o->file, "%sGIT binary patch\n", diff_line_prefix(o));
1513                 break;
1514         case DIFF_SYMBOL_BINARY_DIFF_HEADER_DELTA:
1515                 fprintf(o->file, "%sdelta %s\n", diff_line_prefix(o), line);
1516                 break;
1517         case DIFF_SYMBOL_BINARY_DIFF_HEADER_LITERAL:
1518                 fprintf(o->file, "%sliteral %s\n", diff_line_prefix(o), line);
1519                 break;
1520         case DIFF_SYMBOL_BINARY_DIFF_FOOTER:
1521                 fputs(diff_line_prefix(o), o->file);
1522                 fputc('\n', o->file);
1523                 break;
1524         case DIFF_SYMBOL_REWRITE_DIFF:
1525                 fraginfo = diff_get_color(o->use_color, DIFF_FRAGINFO);
1526                 reset = diff_get_color_opt(o, DIFF_RESET);
1527                 emit_line(o, fraginfo, reset, line, len);
1528                 break;
1529         case DIFF_SYMBOL_SUBMODULE_ADD:
1530                 set = diff_get_color_opt(o, DIFF_FILE_NEW);
1531                 reset = diff_get_color_opt(o, DIFF_RESET);
1532                 emit_line(o, set, reset, line, len);
1533                 break;
1534         case DIFF_SYMBOL_SUBMODULE_DEL:
1535                 set = diff_get_color_opt(o, DIFF_FILE_OLD);
1536                 reset = diff_get_color_opt(o, DIFF_RESET);
1537                 emit_line(o, set, reset, line, len);
1538                 break;
1539         case DIFF_SYMBOL_SUBMODULE_UNTRACKED:
1540                 fprintf(o->file, "%sSubmodule %s contains untracked content\n",
1541                         diff_line_prefix(o), line);
1542                 break;
1543         case DIFF_SYMBOL_SUBMODULE_MODIFIED:
1544                 fprintf(o->file, "%sSubmodule %s contains modified content\n",
1545                         diff_line_prefix(o), line);
1546                 break;
1547         case DIFF_SYMBOL_STATS_SUMMARY_NO_FILES:
1548                 emit_line(o, "", "", " 0 files changed\n",
1549                           strlen(" 0 files changed\n"));
1550                 break;
1551         case DIFF_SYMBOL_STATS_SUMMARY_ABBREV:
1552                 emit_line(o, "", "", " ...\n", strlen(" ...\n"));
1553                 break;
1554         case DIFF_SYMBOL_WORD_DIFF:
1555                 fprintf(o->file, "%.*s", len, line);
1556                 break;
1557         case DIFF_SYMBOL_STAT_SEP:
1558                 fputs(o->stat_sep, o->file);
1559                 break;
1560         default:
1561                 BUG("unknown diff symbol");
1562         }
1563         strbuf_release(&sb);
1564 }
1565
1566 static void emit_diff_symbol(struct diff_options *o, enum diff_symbol s,
1567                              const char *line, int len, unsigned flags)
1568 {
1569         struct emitted_diff_symbol e = {line, len, flags, 0, 0, s};
1570
1571         if (o->emitted_symbols)
1572                 append_emitted_diff_symbol(o, &e);
1573         else
1574                 emit_diff_symbol_from_struct(o, &e);
1575 }
1576
1577 void diff_emit_submodule_del(struct diff_options *o, const char *line)
1578 {
1579         emit_diff_symbol(o, DIFF_SYMBOL_SUBMODULE_DEL, line, strlen(line), 0);
1580 }
1581
1582 void diff_emit_submodule_add(struct diff_options *o, const char *line)
1583 {
1584         emit_diff_symbol(o, DIFF_SYMBOL_SUBMODULE_ADD, line, strlen(line), 0);
1585 }
1586
1587 void diff_emit_submodule_untracked(struct diff_options *o, const char *path)
1588 {
1589         emit_diff_symbol(o, DIFF_SYMBOL_SUBMODULE_UNTRACKED,
1590                          path, strlen(path), 0);
1591 }
1592
1593 void diff_emit_submodule_modified(struct diff_options *o, const char *path)
1594 {
1595         emit_diff_symbol(o, DIFF_SYMBOL_SUBMODULE_MODIFIED,
1596                          path, strlen(path), 0);
1597 }
1598
1599 void diff_emit_submodule_header(struct diff_options *o, const char *header)
1600 {
1601         emit_diff_symbol(o, DIFF_SYMBOL_SUBMODULE_HEADER,
1602                          header, strlen(header), 0);
1603 }
1604
1605 void diff_emit_submodule_error(struct diff_options *o, const char *err)
1606 {
1607         emit_diff_symbol(o, DIFF_SYMBOL_SUBMODULE_ERROR, err, strlen(err), 0);
1608 }
1609
1610 void diff_emit_submodule_pipethrough(struct diff_options *o,
1611                                      const char *line, int len)
1612 {
1613         emit_diff_symbol(o, DIFF_SYMBOL_SUBMODULE_PIPETHROUGH, line, len, 0);
1614 }
1615
1616 static int new_blank_line_at_eof(struct emit_callback *ecbdata, const char *line, int len)
1617 {
1618         if (!((ecbdata->ws_rule & WS_BLANK_AT_EOF) &&
1619               ecbdata->blank_at_eof_in_preimage &&
1620               ecbdata->blank_at_eof_in_postimage &&
1621               ecbdata->blank_at_eof_in_preimage <= ecbdata->lno_in_preimage &&
1622               ecbdata->blank_at_eof_in_postimage <= ecbdata->lno_in_postimage))
1623                 return 0;
1624         return ws_blank_line(line, len, ecbdata->ws_rule);
1625 }
1626
1627 static void emit_add_line(struct emit_callback *ecbdata,
1628                           const char *line, int len)
1629 {
1630         unsigned flags = WSEH_NEW | ecbdata->ws_rule;
1631         if (new_blank_line_at_eof(ecbdata, line, len))
1632                 flags |= DIFF_SYMBOL_CONTENT_BLANK_LINE_EOF;
1633
1634         emit_diff_symbol(ecbdata->opt, DIFF_SYMBOL_PLUS, line, len, flags);
1635 }
1636
1637 static void emit_del_line(struct emit_callback *ecbdata,
1638                           const char *line, int len)
1639 {
1640         unsigned flags = WSEH_OLD | ecbdata->ws_rule;
1641         emit_diff_symbol(ecbdata->opt, DIFF_SYMBOL_MINUS, line, len, flags);
1642 }
1643
1644 static void emit_context_line(struct emit_callback *ecbdata,
1645                               const char *line, int len)
1646 {
1647         unsigned flags = WSEH_CONTEXT | ecbdata->ws_rule;
1648         emit_diff_symbol(ecbdata->opt, DIFF_SYMBOL_CONTEXT, line, len, flags);
1649 }
1650
1651 static void emit_hunk_header(struct emit_callback *ecbdata,
1652                              const char *line, int len)
1653 {
1654         const char *context = diff_get_color(ecbdata->color_diff, DIFF_CONTEXT);
1655         const char *frag = diff_get_color(ecbdata->color_diff, DIFF_FRAGINFO);
1656         const char *func = diff_get_color(ecbdata->color_diff, DIFF_FUNCINFO);
1657         const char *reset = diff_get_color(ecbdata->color_diff, DIFF_RESET);
1658         const char *reverse = ecbdata->color_diff ? GIT_COLOR_REVERSE : "";
1659         static const char atat[2] = { '@', '@' };
1660         const char *cp, *ep;
1661         struct strbuf msgbuf = STRBUF_INIT;
1662         int org_len = len;
1663         int i = 1;
1664
1665         /*
1666          * As a hunk header must begin with "@@ -<old>, +<new> @@",
1667          * it always is at least 10 bytes long.
1668          */
1669         if (len < 10 ||
1670             memcmp(line, atat, 2) ||
1671             !(ep = memmem(line + 2, len - 2, atat, 2))) {
1672                 emit_diff_symbol(ecbdata->opt,
1673                                  DIFF_SYMBOL_CONTEXT_MARKER, line, len, 0);
1674                 return;
1675         }
1676         ep += 2; /* skip over @@ */
1677
1678         /* The hunk header in fraginfo color */
1679         if (ecbdata->opt->flags.dual_color_diffed_diffs)
1680                 strbuf_addstr(&msgbuf, reverse);
1681         strbuf_addstr(&msgbuf, frag);
1682         if (ecbdata->opt->flags.suppress_hunk_header_line_count)
1683                 strbuf_add(&msgbuf, atat, sizeof(atat));
1684         else
1685                 strbuf_add(&msgbuf, line, ep - line);
1686         strbuf_addstr(&msgbuf, reset);
1687
1688         /*
1689          * trailing "\r\n"
1690          */
1691         for ( ; i < 3; i++)
1692                 if (line[len - i] == '\r' || line[len - i] == '\n')
1693                         len--;
1694
1695         /* blank before the func header */
1696         for (cp = ep; ep - line < len; ep++)
1697                 if (*ep != ' ' && *ep != '\t')
1698                         break;
1699         if (ep != cp) {
1700                 strbuf_addstr(&msgbuf, context);
1701                 strbuf_add(&msgbuf, cp, ep - cp);
1702                 strbuf_addstr(&msgbuf, reset);
1703         }
1704
1705         if (ep < line + len) {
1706                 strbuf_addstr(&msgbuf, func);
1707                 strbuf_add(&msgbuf, ep, line + len - ep);
1708                 strbuf_addstr(&msgbuf, reset);
1709         }
1710
1711         strbuf_add(&msgbuf, line + len, org_len - len);
1712         strbuf_complete_line(&msgbuf);
1713         emit_diff_symbol(ecbdata->opt,
1714                          DIFF_SYMBOL_CONTEXT_FRAGINFO, msgbuf.buf, msgbuf.len, 0);
1715         strbuf_release(&msgbuf);
1716 }
1717
1718 static struct diff_tempfile *claim_diff_tempfile(void)
1719 {
1720         int i;
1721         for (i = 0; i < ARRAY_SIZE(diff_temp); i++)
1722                 if (!diff_temp[i].name)
1723                         return diff_temp + i;
1724         BUG("diff is failing to clean up its tempfiles");
1725 }
1726
1727 static void remove_tempfile(void)
1728 {
1729         int i;
1730         for (i = 0; i < ARRAY_SIZE(diff_temp); i++) {
1731                 if (is_tempfile_active(diff_temp[i].tempfile))
1732                         delete_tempfile(&diff_temp[i].tempfile);
1733                 diff_temp[i].name = NULL;
1734         }
1735 }
1736
1737 static void add_line_count(struct strbuf *out, int count)
1738 {
1739         switch (count) {
1740         case 0:
1741                 strbuf_addstr(out, "0,0");
1742                 break;
1743         case 1:
1744                 strbuf_addstr(out, "1");
1745                 break;
1746         default:
1747                 strbuf_addf(out, "1,%d", count);
1748                 break;
1749         }
1750 }
1751
1752 static void emit_rewrite_lines(struct emit_callback *ecb,
1753                                int prefix, const char *data, int size)
1754 {
1755         const char *endp = NULL;
1756
1757         while (0 < size) {
1758                 int len;
1759
1760                 endp = memchr(data, '\n', size);
1761                 len = endp ? (endp - data + 1) : size;
1762                 if (prefix != '+') {
1763                         ecb->lno_in_preimage++;
1764                         emit_del_line(ecb, data, len);
1765                 } else {
1766                         ecb->lno_in_postimage++;
1767                         emit_add_line(ecb, data, len);
1768                 }
1769                 size -= len;
1770                 data += len;
1771         }
1772         if (!endp)
1773                 emit_diff_symbol(ecb->opt, DIFF_SYMBOL_NO_LF_EOF, NULL, 0, 0);
1774 }
1775
1776 static void emit_rewrite_diff(const char *name_a,
1777                               const char *name_b,
1778                               struct diff_filespec *one,
1779                               struct diff_filespec *two,
1780                               struct userdiff_driver *textconv_one,
1781                               struct userdiff_driver *textconv_two,
1782                               struct diff_options *o)
1783 {
1784         int lc_a, lc_b;
1785         static struct strbuf a_name = STRBUF_INIT, b_name = STRBUF_INIT;
1786         const char *a_prefix, *b_prefix;
1787         char *data_one, *data_two;
1788         size_t size_one, size_two;
1789         struct emit_callback ecbdata;
1790         struct strbuf out = STRBUF_INIT;
1791
1792         if (diff_mnemonic_prefix && o->flags.reverse_diff) {
1793                 a_prefix = o->b_prefix;
1794                 b_prefix = o->a_prefix;
1795         } else {
1796                 a_prefix = o->a_prefix;
1797                 b_prefix = o->b_prefix;
1798         }
1799
1800         name_a += (*name_a == '/');
1801         name_b += (*name_b == '/');
1802
1803         strbuf_reset(&a_name);
1804         strbuf_reset(&b_name);
1805         quote_two_c_style(&a_name, a_prefix, name_a, 0);
1806         quote_two_c_style(&b_name, b_prefix, name_b, 0);
1807
1808         size_one = fill_textconv(o->repo, textconv_one, one, &data_one);
1809         size_two = fill_textconv(o->repo, textconv_two, two, &data_two);
1810
1811         memset(&ecbdata, 0, sizeof(ecbdata));
1812         ecbdata.color_diff = want_color(o->use_color);
1813         ecbdata.ws_rule = whitespace_rule(o->repo->index, name_b);
1814         ecbdata.opt = o;
1815         if (ecbdata.ws_rule & WS_BLANK_AT_EOF) {
1816                 mmfile_t mf1, mf2;
1817                 mf1.ptr = (char *)data_one;
1818                 mf2.ptr = (char *)data_two;
1819                 mf1.size = size_one;
1820                 mf2.size = size_two;
1821                 check_blank_at_eof(&mf1, &mf2, &ecbdata);
1822         }
1823         ecbdata.lno_in_preimage = 1;
1824         ecbdata.lno_in_postimage = 1;
1825
1826         lc_a = count_lines(data_one, size_one);
1827         lc_b = count_lines(data_two, size_two);
1828
1829         emit_diff_symbol(o, DIFF_SYMBOL_FILEPAIR_MINUS,
1830                          a_name.buf, a_name.len, 0);
1831         emit_diff_symbol(o, DIFF_SYMBOL_FILEPAIR_PLUS,
1832                          b_name.buf, b_name.len, 0);
1833
1834         strbuf_addstr(&out, "@@ -");
1835         if (!o->irreversible_delete)
1836                 add_line_count(&out, lc_a);
1837         else
1838                 strbuf_addstr(&out, "?,?");
1839         strbuf_addstr(&out, " +");
1840         add_line_count(&out, lc_b);
1841         strbuf_addstr(&out, " @@\n");
1842         emit_diff_symbol(o, DIFF_SYMBOL_REWRITE_DIFF, out.buf, out.len, 0);
1843         strbuf_release(&out);
1844
1845         if (lc_a && !o->irreversible_delete)
1846                 emit_rewrite_lines(&ecbdata, '-', data_one, size_one);
1847         if (lc_b)
1848                 emit_rewrite_lines(&ecbdata, '+', data_two, size_two);
1849         if (textconv_one)
1850                 free((char *)data_one);
1851         if (textconv_two)
1852                 free((char *)data_two);
1853 }
1854
1855 struct diff_words_buffer {
1856         mmfile_t text;
1857         unsigned long alloc;
1858         struct diff_words_orig {
1859                 const char *begin, *end;
1860         } *orig;
1861         int orig_nr, orig_alloc;
1862 };
1863
1864 static void diff_words_append(char *line, unsigned long len,
1865                 struct diff_words_buffer *buffer)
1866 {
1867         ALLOC_GROW(buffer->text.ptr, buffer->text.size + len, buffer->alloc);
1868         line++;
1869         len--;
1870         memcpy(buffer->text.ptr + buffer->text.size, line, len);
1871         buffer->text.size += len;
1872         buffer->text.ptr[buffer->text.size] = '\0';
1873 }
1874
1875 struct diff_words_style_elem {
1876         const char *prefix;
1877         const char *suffix;
1878         const char *color; /* NULL; filled in by the setup code if
1879                             * color is enabled */
1880 };
1881
1882 struct diff_words_style {
1883         enum diff_words_type type;
1884         struct diff_words_style_elem new_word, old_word, ctx;
1885         const char *newline;
1886 };
1887
1888 static struct diff_words_style diff_words_styles[] = {
1889         { DIFF_WORDS_PORCELAIN, {"+", "\n"}, {"-", "\n"}, {" ", "\n"}, "~\n" },
1890         { DIFF_WORDS_PLAIN, {"{+", "+}"}, {"[-", "-]"}, {"", ""}, "\n" },
1891         { DIFF_WORDS_COLOR, {"", ""}, {"", ""}, {"", ""}, "\n" }
1892 };
1893
1894 struct diff_words_data {
1895         struct diff_words_buffer minus, plus;
1896         const char *current_plus;
1897         int last_minus;
1898         struct diff_options *opt;
1899         regex_t *word_regex;
1900         enum diff_words_type type;
1901         struct diff_words_style *style;
1902 };
1903
1904 static int fn_out_diff_words_write_helper(struct diff_options *o,
1905                                           struct diff_words_style_elem *st_el,
1906                                           const char *newline,
1907                                           size_t count, const char *buf)
1908 {
1909         int print = 0;
1910         struct strbuf sb = STRBUF_INIT;
1911
1912         while (count) {
1913                 char *p = memchr(buf, '\n', count);
1914                 if (print)
1915                         strbuf_addstr(&sb, diff_line_prefix(o));
1916
1917                 if (p != buf) {
1918                         const char *reset = st_el->color && *st_el->color ?
1919                                             GIT_COLOR_RESET : NULL;
1920                         if (st_el->color && *st_el->color)
1921                                 strbuf_addstr(&sb, st_el->color);
1922                         strbuf_addstr(&sb, st_el->prefix);
1923                         strbuf_add(&sb, buf, p ? p - buf : count);
1924                         strbuf_addstr(&sb, st_el->suffix);
1925                         if (reset)
1926                                 strbuf_addstr(&sb, reset);
1927                 }
1928                 if (!p)
1929                         goto out;
1930
1931                 strbuf_addstr(&sb, newline);
1932                 count -= p + 1 - buf;
1933                 buf = p + 1;
1934                 print = 1;
1935                 if (count) {
1936                         emit_diff_symbol(o, DIFF_SYMBOL_WORD_DIFF,
1937                                          sb.buf, sb.len, 0);
1938                         strbuf_reset(&sb);
1939                 }
1940         }
1941
1942 out:
1943         if (sb.len)
1944                 emit_diff_symbol(o, DIFF_SYMBOL_WORD_DIFF,
1945                                  sb.buf, sb.len, 0);
1946         strbuf_release(&sb);
1947         return 0;
1948 }
1949
1950 /*
1951  * '--color-words' algorithm can be described as:
1952  *
1953  *   1. collect the minus/plus lines of a diff hunk, divided into
1954  *      minus-lines and plus-lines;
1955  *
1956  *   2. break both minus-lines and plus-lines into words and
1957  *      place them into two mmfile_t with one word for each line;
1958  *
1959  *   3. use xdiff to run diff on the two mmfile_t to get the words level diff;
1960  *
1961  * And for the common parts of the both file, we output the plus side text.
1962  * diff_words->current_plus is used to trace the current position of the plus file
1963  * which printed. diff_words->last_minus is used to trace the last minus word
1964  * printed.
1965  *
1966  * For '--graph' to work with '--color-words', we need to output the graph prefix
1967  * on each line of color words output. Generally, there are two conditions on
1968  * which we should output the prefix.
1969  *
1970  *   1. diff_words->last_minus == 0 &&
1971  *      diff_words->current_plus == diff_words->plus.text.ptr
1972  *
1973  *      that is: the plus text must start as a new line, and if there is no minus
1974  *      word printed, a graph prefix must be printed.
1975  *
1976  *   2. diff_words->current_plus > diff_words->plus.text.ptr &&
1977  *      *(diff_words->current_plus - 1) == '\n'
1978  *
1979  *      that is: a graph prefix must be printed following a '\n'
1980  */
1981 static int color_words_output_graph_prefix(struct diff_words_data *diff_words)
1982 {
1983         if ((diff_words->last_minus == 0 &&
1984                 diff_words->current_plus == diff_words->plus.text.ptr) ||
1985                 (diff_words->current_plus > diff_words->plus.text.ptr &&
1986                 *(diff_words->current_plus - 1) == '\n')) {
1987                 return 1;
1988         } else {
1989                 return 0;
1990         }
1991 }
1992
1993 static void fn_out_diff_words_aux(void *priv,
1994                                   long minus_first, long minus_len,
1995                                   long plus_first, long plus_len,
1996                                   const char *func, long funclen)
1997 {
1998         struct diff_words_data *diff_words = priv;
1999         struct diff_words_style *style = diff_words->style;
2000         const char *minus_begin, *minus_end, *plus_begin, *plus_end;
2001         struct diff_options *opt = diff_words->opt;
2002         const char *line_prefix;
2003
2004         assert(opt);
2005         line_prefix = diff_line_prefix(opt);
2006
2007         /* POSIX requires that first be decremented by one if len == 0... */
2008         if (minus_len) {
2009                 minus_begin = diff_words->minus.orig[minus_first].begin;
2010                 minus_end =
2011                         diff_words->minus.orig[minus_first + minus_len - 1].end;
2012         } else
2013                 minus_begin = minus_end =
2014                         diff_words->minus.orig[minus_first].end;
2015
2016         if (plus_len) {
2017                 plus_begin = diff_words->plus.orig[plus_first].begin;
2018                 plus_end = diff_words->plus.orig[plus_first + plus_len - 1].end;
2019         } else
2020                 plus_begin = plus_end = diff_words->plus.orig[plus_first].end;
2021
2022         if (color_words_output_graph_prefix(diff_words)) {
2023                 fputs(line_prefix, diff_words->opt->file);
2024         }
2025         if (diff_words->current_plus != plus_begin) {
2026                 fn_out_diff_words_write_helper(diff_words->opt,
2027                                 &style->ctx, style->newline,
2028                                 plus_begin - diff_words->current_plus,
2029                                 diff_words->current_plus);
2030         }
2031         if (minus_begin != minus_end) {
2032                 fn_out_diff_words_write_helper(diff_words->opt,
2033                                 &style->old_word, style->newline,
2034                                 minus_end - minus_begin, minus_begin);
2035         }
2036         if (plus_begin != plus_end) {
2037                 fn_out_diff_words_write_helper(diff_words->opt,
2038                                 &style->new_word, style->newline,
2039                                 plus_end - plus_begin, plus_begin);
2040         }
2041
2042         diff_words->current_plus = plus_end;
2043         diff_words->last_minus = minus_first;
2044 }
2045
2046 /* This function starts looking at *begin, and returns 0 iff a word was found. */
2047 static int find_word_boundaries(mmfile_t *buffer, regex_t *word_regex,
2048                 int *begin, int *end)
2049 {
2050         if (word_regex && *begin < buffer->size) {
2051                 regmatch_t match[1];
2052                 if (!regexec_buf(word_regex, buffer->ptr + *begin,
2053                                  buffer->size - *begin, 1, match, 0)) {
2054                         char *p = memchr(buffer->ptr + *begin + match[0].rm_so,
2055                                         '\n', match[0].rm_eo - match[0].rm_so);
2056                         *end = p ? p - buffer->ptr : match[0].rm_eo + *begin;
2057                         *begin += match[0].rm_so;
2058                         return *begin >= *end;
2059                 }
2060                 return -1;
2061         }
2062
2063         /* find the next word */
2064         while (*begin < buffer->size && isspace(buffer->ptr[*begin]))
2065                 (*begin)++;
2066         if (*begin >= buffer->size)
2067                 return -1;
2068
2069         /* find the end of the word */
2070         *end = *begin + 1;
2071         while (*end < buffer->size && !isspace(buffer->ptr[*end]))
2072                 (*end)++;
2073
2074         return 0;
2075 }
2076
2077 /*
2078  * This function splits the words in buffer->text, stores the list with
2079  * newline separator into out, and saves the offsets of the original words
2080  * in buffer->orig.
2081  */
2082 static void diff_words_fill(struct diff_words_buffer *buffer, mmfile_t *out,
2083                 regex_t *word_regex)
2084 {
2085         int i, j;
2086         long alloc = 0;
2087
2088         out->size = 0;
2089         out->ptr = NULL;
2090
2091         /* fake an empty "0th" word */
2092         ALLOC_GROW(buffer->orig, 1, buffer->orig_alloc);
2093         buffer->orig[0].begin = buffer->orig[0].end = buffer->text.ptr;
2094         buffer->orig_nr = 1;
2095
2096         for (i = 0; i < buffer->text.size; i++) {
2097                 if (find_word_boundaries(&buffer->text, word_regex, &i, &j))
2098                         return;
2099
2100                 /* store original boundaries */
2101                 ALLOC_GROW(buffer->orig, buffer->orig_nr + 1,
2102                                 buffer->orig_alloc);
2103                 buffer->orig[buffer->orig_nr].begin = buffer->text.ptr + i;
2104                 buffer->orig[buffer->orig_nr].end = buffer->text.ptr + j;
2105                 buffer->orig_nr++;
2106
2107                 /* store one word */
2108                 ALLOC_GROW(out->ptr, out->size + j - i + 1, alloc);
2109                 memcpy(out->ptr + out->size, buffer->text.ptr + i, j - i);
2110                 out->ptr[out->size + j - i] = '\n';
2111                 out->size += j - i + 1;
2112
2113                 i = j - 1;
2114         }
2115 }
2116
2117 /* this executes the word diff on the accumulated buffers */
2118 static void diff_words_show(struct diff_words_data *diff_words)
2119 {
2120         xpparam_t xpp;
2121         xdemitconf_t xecfg;
2122         mmfile_t minus, plus;
2123         struct diff_words_style *style = diff_words->style;
2124
2125         struct diff_options *opt = diff_words->opt;
2126         const char *line_prefix;
2127
2128         assert(opt);
2129         line_prefix = diff_line_prefix(opt);
2130
2131         /* special case: only removal */
2132         if (!diff_words->plus.text.size) {
2133                 emit_diff_symbol(diff_words->opt, DIFF_SYMBOL_WORD_DIFF,
2134                                  line_prefix, strlen(line_prefix), 0);
2135                 fn_out_diff_words_write_helper(diff_words->opt,
2136                         &style->old_word, style->newline,
2137                         diff_words->minus.text.size,
2138                         diff_words->minus.text.ptr);
2139                 diff_words->minus.text.size = 0;
2140                 return;
2141         }
2142
2143         diff_words->current_plus = diff_words->plus.text.ptr;
2144         diff_words->last_minus = 0;
2145
2146         memset(&xpp, 0, sizeof(xpp));
2147         memset(&xecfg, 0, sizeof(xecfg));
2148         diff_words_fill(&diff_words->minus, &minus, diff_words->word_regex);
2149         diff_words_fill(&diff_words->plus, &plus, diff_words->word_regex);
2150         xpp.flags = 0;
2151         /* as only the hunk header will be parsed, we need a 0-context */
2152         xecfg.ctxlen = 0;
2153         if (xdi_diff_outf(&minus, &plus, fn_out_diff_words_aux, NULL,
2154                           diff_words, &xpp, &xecfg))
2155                 die("unable to generate word diff");
2156         free(minus.ptr);
2157         free(plus.ptr);
2158         if (diff_words->current_plus != diff_words->plus.text.ptr +
2159                         diff_words->plus.text.size) {
2160                 if (color_words_output_graph_prefix(diff_words))
2161                         emit_diff_symbol(diff_words->opt, DIFF_SYMBOL_WORD_DIFF,
2162                                          line_prefix, strlen(line_prefix), 0);
2163                 fn_out_diff_words_write_helper(diff_words->opt,
2164                         &style->ctx, style->newline,
2165                         diff_words->plus.text.ptr + diff_words->plus.text.size
2166                         - diff_words->current_plus, diff_words->current_plus);
2167         }
2168         diff_words->minus.text.size = diff_words->plus.text.size = 0;
2169 }
2170
2171 /* In "color-words" mode, show word-diff of words accumulated in the buffer */
2172 static void diff_words_flush(struct emit_callback *ecbdata)
2173 {
2174         struct diff_options *wo = ecbdata->diff_words->opt;
2175
2176         if (ecbdata->diff_words->minus.text.size ||
2177             ecbdata->diff_words->plus.text.size)
2178                 diff_words_show(ecbdata->diff_words);
2179
2180         if (wo->emitted_symbols) {
2181                 struct diff_options *o = ecbdata->opt;
2182                 struct emitted_diff_symbols *wol = wo->emitted_symbols;
2183                 int i;
2184
2185                 /*
2186                  * NEEDSWORK:
2187                  * Instead of appending each, concat all words to a line?
2188                  */
2189                 for (i = 0; i < wol->nr; i++)
2190                         append_emitted_diff_symbol(o, &wol->buf[i]);
2191
2192                 for (i = 0; i < wol->nr; i++)
2193                         free((void *)wol->buf[i].line);
2194
2195                 wol->nr = 0;
2196         }
2197 }
2198
2199 static void diff_filespec_load_driver(struct diff_filespec *one,
2200                                       struct index_state *istate)
2201 {
2202         /* Use already-loaded driver */
2203         if (one->driver)
2204                 return;
2205
2206         if (S_ISREG(one->mode))
2207                 one->driver = userdiff_find_by_path(istate, one->path);
2208
2209         /* Fallback to default settings */
2210         if (!one->driver)
2211                 one->driver = userdiff_find_by_name("default");
2212 }
2213
2214 static const char *userdiff_word_regex(struct diff_filespec *one,
2215                                        struct index_state *istate)
2216 {
2217         diff_filespec_load_driver(one, istate);
2218         return one->driver->word_regex;
2219 }
2220
2221 static void init_diff_words_data(struct emit_callback *ecbdata,
2222                                  struct diff_options *orig_opts,
2223                                  struct diff_filespec *one,
2224                                  struct diff_filespec *two)
2225 {
2226         int i;
2227         struct diff_options *o = xmalloc(sizeof(struct diff_options));
2228         memcpy(o, orig_opts, sizeof(struct diff_options));
2229
2230         ecbdata->diff_words =
2231                 xcalloc(1, sizeof(struct diff_words_data));
2232         ecbdata->diff_words->type = o->word_diff;
2233         ecbdata->diff_words->opt = o;
2234
2235         if (orig_opts->emitted_symbols)
2236                 o->emitted_symbols =
2237                         xcalloc(1, sizeof(struct emitted_diff_symbols));
2238
2239         if (!o->word_regex)
2240                 o->word_regex = userdiff_word_regex(one, o->repo->index);
2241         if (!o->word_regex)
2242                 o->word_regex = userdiff_word_regex(two, o->repo->index);
2243         if (!o->word_regex)
2244                 o->word_regex = diff_word_regex_cfg;
2245         if (o->word_regex) {
2246                 ecbdata->diff_words->word_regex = (regex_t *)
2247                         xmalloc(sizeof(regex_t));
2248                 if (regcomp(ecbdata->diff_words->word_regex,
2249                             o->word_regex,
2250                             REG_EXTENDED | REG_NEWLINE))
2251                         die("invalid regular expression: %s",
2252                             o->word_regex);
2253         }
2254         for (i = 0; i < ARRAY_SIZE(diff_words_styles); i++) {
2255                 if (o->word_diff == diff_words_styles[i].type) {
2256                         ecbdata->diff_words->style =
2257                                 &diff_words_styles[i];
2258                         break;
2259                 }
2260         }
2261         if (want_color(o->use_color)) {
2262                 struct diff_words_style *st = ecbdata->diff_words->style;
2263                 st->old_word.color = diff_get_color_opt(o, DIFF_FILE_OLD);
2264                 st->new_word.color = diff_get_color_opt(o, DIFF_FILE_NEW);
2265                 st->ctx.color = diff_get_color_opt(o, DIFF_CONTEXT);
2266         }
2267 }
2268
2269 static void free_diff_words_data(struct emit_callback *ecbdata)
2270 {
2271         if (ecbdata->diff_words) {
2272                 diff_words_flush(ecbdata);
2273                 free (ecbdata->diff_words->opt->emitted_symbols);
2274                 free (ecbdata->diff_words->opt);
2275                 free (ecbdata->diff_words->minus.text.ptr);
2276                 free (ecbdata->diff_words->minus.orig);
2277                 free (ecbdata->diff_words->plus.text.ptr);
2278                 free (ecbdata->diff_words->plus.orig);
2279                 if (ecbdata->diff_words->word_regex) {
2280                         regfree(ecbdata->diff_words->word_regex);
2281                         free(ecbdata->diff_words->word_regex);
2282                 }
2283                 FREE_AND_NULL(ecbdata->diff_words);
2284         }
2285 }
2286
2287 const char *diff_get_color(int diff_use_color, enum color_diff ix)
2288 {
2289         if (want_color(diff_use_color))
2290                 return diff_colors[ix];
2291         return "";
2292 }
2293
2294 const char *diff_line_prefix(struct diff_options *opt)
2295 {
2296         struct strbuf *msgbuf;
2297         if (!opt->output_prefix)
2298                 return "";
2299
2300         msgbuf = opt->output_prefix(opt, opt->output_prefix_data);
2301         return msgbuf->buf;
2302 }
2303
2304 static unsigned long sane_truncate_line(char *line, unsigned long len)
2305 {
2306         const char *cp;
2307         unsigned long allot;
2308         size_t l = len;
2309
2310         cp = line;
2311         allot = l;
2312         while (0 < l) {
2313                 (void) utf8_width(&cp, &l);
2314                 if (!cp)
2315                         break; /* truncated in the middle? */
2316         }
2317         return allot - l;
2318 }
2319
2320 static void find_lno(const char *line, struct emit_callback *ecbdata)
2321 {
2322         const char *p;
2323         ecbdata->lno_in_preimage = 0;
2324         ecbdata->lno_in_postimage = 0;
2325         p = strchr(line, '-');
2326         if (!p)
2327                 return; /* cannot happen */
2328         ecbdata->lno_in_preimage = strtol(p + 1, NULL, 10);
2329         p = strchr(p, '+');
2330         if (!p)
2331                 return; /* cannot happen */
2332         ecbdata->lno_in_postimage = strtol(p + 1, NULL, 10);
2333 }
2334
2335 static void fn_out_consume(void *priv, char *line, unsigned long len)
2336 {
2337         struct emit_callback *ecbdata = priv;
2338         struct diff_options *o = ecbdata->opt;
2339
2340         o->found_changes = 1;
2341
2342         if (ecbdata->header) {
2343                 emit_diff_symbol(o, DIFF_SYMBOL_HEADER,
2344                                  ecbdata->header->buf, ecbdata->header->len, 0);
2345                 strbuf_reset(ecbdata->header);
2346                 ecbdata->header = NULL;
2347         }
2348
2349         if (ecbdata->label_path[0]) {
2350                 emit_diff_symbol(o, DIFF_SYMBOL_FILEPAIR_MINUS,
2351                                  ecbdata->label_path[0],
2352                                  strlen(ecbdata->label_path[0]), 0);
2353                 emit_diff_symbol(o, DIFF_SYMBOL_FILEPAIR_PLUS,
2354                                  ecbdata->label_path[1],
2355                                  strlen(ecbdata->label_path[1]), 0);
2356                 ecbdata->label_path[0] = ecbdata->label_path[1] = NULL;
2357         }
2358
2359         if (diff_suppress_blank_empty
2360             && len == 2 && line[0] == ' ' && line[1] == '\n') {
2361                 line[0] = '\n';
2362                 len = 1;
2363         }
2364
2365         if (line[0] == '@') {
2366                 if (ecbdata->diff_words)
2367                         diff_words_flush(ecbdata);
2368                 len = sane_truncate_line(line, len);
2369                 find_lno(line, ecbdata);
2370                 emit_hunk_header(ecbdata, line, len);
2371                 return;
2372         }
2373
2374         if (ecbdata->diff_words) {
2375                 enum diff_symbol s =
2376                         ecbdata->diff_words->type == DIFF_WORDS_PORCELAIN ?
2377                         DIFF_SYMBOL_WORDS_PORCELAIN : DIFF_SYMBOL_WORDS;
2378                 if (line[0] == '-') {
2379                         diff_words_append(line, len,
2380                                           &ecbdata->diff_words->minus);
2381                         return;
2382                 } else if (line[0] == '+') {
2383                         diff_words_append(line, len,
2384                                           &ecbdata->diff_words->plus);
2385                         return;
2386                 } else if (starts_with(line, "\\ ")) {
2387                         /*
2388                          * Eat the "no newline at eof" marker as if we
2389                          * saw a "+" or "-" line with nothing on it,
2390                          * and return without diff_words_flush() to
2391                          * defer processing. If this is the end of
2392                          * preimage, more "+" lines may come after it.
2393                          */
2394                         return;
2395                 }
2396                 diff_words_flush(ecbdata);
2397                 emit_diff_symbol(o, s, line, len, 0);
2398                 return;
2399         }
2400
2401         switch (line[0]) {
2402         case '+':
2403                 ecbdata->lno_in_postimage++;
2404                 emit_add_line(ecbdata, line + 1, len - 1);
2405                 break;
2406         case '-':
2407                 ecbdata->lno_in_preimage++;
2408                 emit_del_line(ecbdata, line + 1, len - 1);
2409                 break;
2410         case ' ':
2411                 ecbdata->lno_in_postimage++;
2412                 ecbdata->lno_in_preimage++;
2413                 emit_context_line(ecbdata, line + 1, len - 1);
2414                 break;
2415         default:
2416                 /* incomplete line at the end */
2417                 ecbdata->lno_in_preimage++;
2418                 emit_diff_symbol(o, DIFF_SYMBOL_CONTEXT_INCOMPLETE,
2419                                  line, len, 0);
2420                 break;
2421         }
2422 }
2423
2424 static void pprint_rename(struct strbuf *name, const char *a, const char *b)
2425 {
2426         const char *old_name = a;
2427         const char *new_name = b;
2428         int pfx_length, sfx_length;
2429         int pfx_adjust_for_slash;
2430         int len_a = strlen(a);
2431         int len_b = strlen(b);
2432         int a_midlen, b_midlen;
2433         int qlen_a = quote_c_style(a, NULL, NULL, 0);
2434         int qlen_b = quote_c_style(b, NULL, NULL, 0);
2435
2436         if (qlen_a || qlen_b) {
2437                 quote_c_style(a, name, NULL, 0);
2438                 strbuf_addstr(name, " => ");
2439                 quote_c_style(b, name, NULL, 0);
2440                 return;
2441         }
2442
2443         /* Find common prefix */
2444         pfx_length = 0;
2445         while (*old_name && *new_name && *old_name == *new_name) {
2446                 if (*old_name == '/')
2447                         pfx_length = old_name - a + 1;
2448                 old_name++;
2449                 new_name++;
2450         }
2451
2452         /* Find common suffix */
2453         old_name = a + len_a;
2454         new_name = b + len_b;
2455         sfx_length = 0;
2456         /*
2457          * If there is a common prefix, it must end in a slash.  In
2458          * that case we let this loop run 1 into the prefix to see the
2459          * same slash.
2460          *
2461          * If there is no common prefix, we cannot do this as it would
2462          * underrun the input strings.
2463          */
2464         pfx_adjust_for_slash = (pfx_length ? 1 : 0);
2465         while (a + pfx_length - pfx_adjust_for_slash <= old_name &&
2466                b + pfx_length - pfx_adjust_for_slash <= new_name &&
2467                *old_name == *new_name) {
2468                 if (*old_name == '/')
2469                         sfx_length = len_a - (old_name - a);
2470                 old_name--;
2471                 new_name--;
2472         }
2473
2474         /*
2475          * pfx{mid-a => mid-b}sfx
2476          * {pfx-a => pfx-b}sfx
2477          * pfx{sfx-a => sfx-b}
2478          * name-a => name-b
2479          */
2480         a_midlen = len_a - pfx_length - sfx_length;
2481         b_midlen = len_b - pfx_length - sfx_length;
2482         if (a_midlen < 0)
2483                 a_midlen = 0;
2484         if (b_midlen < 0)
2485                 b_midlen = 0;
2486
2487         strbuf_grow(name, pfx_length + a_midlen + b_midlen + sfx_length + 7);
2488         if (pfx_length + sfx_length) {
2489                 strbuf_add(name, a, pfx_length);
2490                 strbuf_addch(name, '{');
2491         }
2492         strbuf_add(name, a + pfx_length, a_midlen);
2493         strbuf_addstr(name, " => ");
2494         strbuf_add(name, b + pfx_length, b_midlen);
2495         if (pfx_length + sfx_length) {
2496                 strbuf_addch(name, '}');
2497                 strbuf_add(name, a + len_a - sfx_length, sfx_length);
2498         }
2499 }
2500
2501 struct diffstat_t {
2502         int nr;
2503         int alloc;
2504         struct diffstat_file {
2505                 char *from_name;
2506                 char *name;
2507                 char *print_name;
2508                 const char *comments;
2509                 unsigned is_unmerged:1;
2510                 unsigned is_binary:1;
2511                 unsigned is_renamed:1;
2512                 unsigned is_interesting:1;
2513                 uintmax_t added, deleted;
2514         } **files;
2515 };
2516
2517 static struct diffstat_file *diffstat_add(struct diffstat_t *diffstat,
2518                                           const char *name_a,
2519                                           const char *name_b)
2520 {
2521         struct diffstat_file *x;
2522         x = xcalloc(1, sizeof(*x));
2523         ALLOC_GROW(diffstat->files, diffstat->nr + 1, diffstat->alloc);
2524         diffstat->files[diffstat->nr++] = x;
2525         if (name_b) {
2526                 x->from_name = xstrdup(name_a);
2527                 x->name = xstrdup(name_b);
2528                 x->is_renamed = 1;
2529         }
2530         else {
2531                 x->from_name = NULL;
2532                 x->name = xstrdup(name_a);
2533         }
2534         return x;
2535 }
2536
2537 static void diffstat_consume(void *priv, char *line, unsigned long len)
2538 {
2539         struct diffstat_t *diffstat = priv;
2540         struct diffstat_file *x = diffstat->files[diffstat->nr - 1];
2541
2542         if (line[0] == '+')
2543                 x->added++;
2544         else if (line[0] == '-')
2545                 x->deleted++;
2546 }
2547
2548 const char mime_boundary_leader[] = "------------";
2549
2550 static int scale_linear(int it, int width, int max_change)
2551 {
2552         if (!it)
2553                 return 0;
2554         /*
2555          * make sure that at least one '-' or '+' is printed if
2556          * there is any change to this path. The easiest way is to
2557          * scale linearly as if the alloted width is one column shorter
2558          * than it is, and then add 1 to the result.
2559          */
2560         return 1 + (it * (width - 1) / max_change);
2561 }
2562
2563 static void show_graph(struct strbuf *out, char ch, int cnt,
2564                        const char *set, const char *reset)
2565 {
2566         if (cnt <= 0)
2567                 return;
2568         strbuf_addstr(out, set);
2569         strbuf_addchars(out, ch, cnt);
2570         strbuf_addstr(out, reset);
2571 }
2572
2573 static void fill_print_name(struct diffstat_file *file)
2574 {
2575         struct strbuf pname = STRBUF_INIT;
2576
2577         if (file->print_name)
2578                 return;
2579
2580         if (file->is_renamed)
2581                 pprint_rename(&pname, file->from_name, file->name);
2582         else
2583                 quote_c_style(file->name, &pname, NULL, 0);
2584
2585         if (file->comments)
2586                 strbuf_addf(&pname, " (%s)", file->comments);
2587
2588         file->print_name = strbuf_detach(&pname, NULL);
2589 }
2590
2591 static void print_stat_summary_inserts_deletes(struct diff_options *options,
2592                 int files, int insertions, int deletions)
2593 {
2594         struct strbuf sb = STRBUF_INIT;
2595
2596         if (!files) {
2597                 assert(insertions == 0 && deletions == 0);
2598                 emit_diff_symbol(options, DIFF_SYMBOL_STATS_SUMMARY_NO_FILES,
2599                                  NULL, 0, 0);
2600                 return;
2601         }
2602
2603         strbuf_addf(&sb,
2604                     (files == 1) ? " %d file changed" : " %d files changed",
2605                     files);
2606
2607         /*
2608          * For binary diff, the caller may want to print "x files
2609          * changed" with insertions == 0 && deletions == 0.
2610          *
2611          * Not omitting "0 insertions(+), 0 deletions(-)" in this case
2612          * is probably less confusing (i.e skip over "2 files changed
2613          * but nothing about added/removed lines? Is this a bug in Git?").
2614          */
2615         if (insertions || deletions == 0) {
2616                 strbuf_addf(&sb,
2617                             (insertions == 1) ? ", %d insertion(+)" : ", %d insertions(+)",
2618                             insertions);
2619         }
2620
2621         if (deletions || insertions == 0) {
2622                 strbuf_addf(&sb,
2623                             (deletions == 1) ? ", %d deletion(-)" : ", %d deletions(-)",
2624                             deletions);
2625         }
2626         strbuf_addch(&sb, '\n');
2627         emit_diff_symbol(options, DIFF_SYMBOL_STATS_SUMMARY_INSERTS_DELETES,
2628                          sb.buf, sb.len, 0);
2629         strbuf_release(&sb);
2630 }
2631
2632 void print_stat_summary(FILE *fp, int files,
2633                         int insertions, int deletions)
2634 {
2635         struct diff_options o;
2636         memset(&o, 0, sizeof(o));
2637         o.file = fp;
2638
2639         print_stat_summary_inserts_deletes(&o, files, insertions, deletions);
2640 }
2641
2642 static void show_stats(struct diffstat_t *data, struct diff_options *options)
2643 {
2644         int i, len, add, del, adds = 0, dels = 0;
2645         uintmax_t max_change = 0, max_len = 0;
2646         int total_files = data->nr, count;
2647         int width, name_width, graph_width, number_width = 0, bin_width = 0;
2648         const char *reset, *add_c, *del_c;
2649         int extra_shown = 0;
2650         const char *line_prefix = diff_line_prefix(options);
2651         struct strbuf out = STRBUF_INIT;
2652
2653         if (data->nr == 0)
2654                 return;
2655
2656         count = options->stat_count ? options->stat_count : data->nr;
2657
2658         reset = diff_get_color_opt(options, DIFF_RESET);
2659         add_c = diff_get_color_opt(options, DIFF_FILE_NEW);
2660         del_c = diff_get_color_opt(options, DIFF_FILE_OLD);
2661
2662         /*
2663          * Find the longest filename and max number of changes
2664          */
2665         for (i = 0; (i < count) && (i < data->nr); i++) {
2666                 struct diffstat_file *file = data->files[i];
2667                 uintmax_t change = file->added + file->deleted;
2668
2669                 if (!file->is_interesting && (change == 0)) {
2670                         count++; /* not shown == room for one more */
2671                         continue;
2672                 }
2673                 fill_print_name(file);
2674                 len = strlen(file->print_name);
2675                 if (max_len < len)
2676                         max_len = len;
2677
2678                 if (file->is_unmerged) {
2679                         /* "Unmerged" is 8 characters */
2680                         bin_width = bin_width < 8 ? 8 : bin_width;
2681                         continue;
2682                 }
2683                 if (file->is_binary) {
2684                         /* "Bin XXX -> YYY bytes" */
2685                         int w = 14 + decimal_width(file->added)
2686                                 + decimal_width(file->deleted);
2687                         bin_width = bin_width < w ? w : bin_width;
2688                         /* Display change counts aligned with "Bin" */
2689                         number_width = 3;
2690                         continue;
2691                 }
2692
2693                 if (max_change < change)
2694                         max_change = change;
2695         }
2696         count = i; /* where we can stop scanning in data->files[] */
2697
2698         /*
2699          * We have width = stat_width or term_columns() columns total.
2700          * We want a maximum of min(max_len, stat_name_width) for the name part.
2701          * We want a maximum of min(max_change, stat_graph_width) for the +- part.
2702          * We also need 1 for " " and 4 + decimal_width(max_change)
2703          * for " | NNNN " and one the empty column at the end, altogether
2704          * 6 + decimal_width(max_change).
2705          *
2706          * If there's not enough space, we will use the smaller of
2707          * stat_name_width (if set) and 5/8*width for the filename,
2708          * and the rest for constant elements + graph part, but no more
2709          * than stat_graph_width for the graph part.
2710          * (5/8 gives 50 for filename and 30 for the constant parts + graph
2711          * for the standard terminal size).
2712          *
2713          * In other words: stat_width limits the maximum width, and
2714          * stat_name_width fixes the maximum width of the filename,
2715          * and is also used to divide available columns if there
2716          * aren't enough.
2717          *
2718          * Binary files are displayed with "Bin XXX -> YYY bytes"
2719          * instead of the change count and graph. This part is treated
2720          * similarly to the graph part, except that it is not
2721          * "scaled". If total width is too small to accommodate the
2722          * guaranteed minimum width of the filename part and the
2723          * separators and this message, this message will "overflow"
2724          * making the line longer than the maximum width.
2725          */
2726
2727         if (options->stat_width == -1)
2728                 width = term_columns() - strlen(line_prefix);
2729         else
2730                 width = options->stat_width ? options->stat_width : 80;
2731         number_width = decimal_width(max_change) > number_width ?
2732                 decimal_width(max_change) : number_width;
2733
2734         if (options->stat_graph_width == -1)
2735                 options->stat_graph_width = diff_stat_graph_width;
2736
2737         /*
2738          * Guarantee 3/8*16==6 for the graph part
2739          * and 5/8*16==10 for the filename part
2740          */
2741         if (width < 16 + 6 + number_width)
2742                 width = 16 + 6 + number_width;
2743
2744         /*
2745          * First assign sizes that are wanted, ignoring available width.
2746          * strlen("Bin XXX -> YYY bytes") == bin_width, and the part
2747          * starting from "XXX" should fit in graph_width.
2748          */
2749         graph_width = max_change + 4 > bin_width ? max_change : bin_width - 4;
2750         if (options->stat_graph_width &&
2751             options->stat_graph_width < graph_width)
2752                 graph_width = options->stat_graph_width;
2753
2754         name_width = (options->stat_name_width > 0 &&
2755                       options->stat_name_width < max_len) ?
2756                 options->stat_name_width : max_len;
2757
2758         /*
2759          * Adjust adjustable widths not to exceed maximum width
2760          */
2761         if (name_width + number_width + 6 + graph_width > width) {
2762                 if (graph_width > width * 3/8 - number_width - 6) {
2763                         graph_width = width * 3/8 - number_width - 6;
2764                         if (graph_width < 6)
2765                                 graph_width = 6;
2766                 }
2767
2768                 if (options->stat_graph_width &&
2769                     graph_width > options->stat_graph_width)
2770                         graph_width = options->stat_graph_width;
2771                 if (name_width > width - number_width - 6 - graph_width)
2772                         name_width = width - number_width - 6 - graph_width;
2773                 else
2774                         graph_width = width - number_width - 6 - name_width;
2775         }
2776
2777         /*
2778          * From here name_width is the width of the name area,
2779          * and graph_width is the width of the graph area.
2780          * max_change is used to scale graph properly.
2781          */
2782         for (i = 0; i < count; i++) {
2783                 const char *prefix = "";
2784                 struct diffstat_file *file = data->files[i];
2785                 char *name = file->print_name;
2786                 uintmax_t added = file->added;
2787                 uintmax_t deleted = file->deleted;
2788                 int name_len;
2789
2790                 if (!file->is_interesting && (added + deleted == 0))
2791                         continue;
2792
2793                 /*
2794                  * "scale" the filename
2795                  */
2796                 len = name_width;
2797                 name_len = strlen(name);
2798                 if (name_width < name_len) {
2799                         char *slash;
2800                         prefix = "...";
2801                         len -= 3;
2802                         name += name_len - len;
2803                         slash = strchr(name, '/');
2804                         if (slash)
2805                                 name = slash;
2806                 }
2807
2808                 if (file->is_binary) {
2809                         strbuf_addf(&out, " %s%-*s |", prefix, len, name);
2810                         strbuf_addf(&out, " %*s", number_width, "Bin");
2811                         if (!added && !deleted) {
2812                                 strbuf_addch(&out, '\n');
2813                                 emit_diff_symbol(options, DIFF_SYMBOL_STATS_LINE,
2814                                                  out.buf, out.len, 0);
2815                                 strbuf_reset(&out);
2816                                 continue;
2817                         }
2818                         strbuf_addf(&out, " %s%"PRIuMAX"%s",
2819                                 del_c, deleted, reset);
2820                         strbuf_addstr(&out, " -> ");
2821                         strbuf_addf(&out, "%s%"PRIuMAX"%s",
2822                                 add_c, added, reset);
2823                         strbuf_addstr(&out, " bytes\n");
2824                         emit_diff_symbol(options, DIFF_SYMBOL_STATS_LINE,
2825                                          out.buf, out.len, 0);
2826                         strbuf_reset(&out);
2827                         continue;
2828                 }
2829                 else if (file->is_unmerged) {
2830                         strbuf_addf(&out, " %s%-*s |", prefix, len, name);
2831                         strbuf_addstr(&out, " Unmerged\n");
2832                         emit_diff_symbol(options, DIFF_SYMBOL_STATS_LINE,
2833                                          out.buf, out.len, 0);
2834                         strbuf_reset(&out);
2835                         continue;
2836                 }
2837
2838                 /*
2839                  * scale the add/delete
2840                  */
2841                 add = added;
2842                 del = deleted;
2843
2844                 if (graph_width <= max_change) {
2845                         int total = scale_linear(add + del, graph_width, max_change);
2846                         if (total < 2 && add && del)
2847                                 /* width >= 2 due to the sanity check */
2848                                 total = 2;
2849                         if (add < del) {
2850                                 add = scale_linear(add, graph_width, max_change);
2851                                 del = total - add;
2852                         } else {
2853                                 del = scale_linear(del, graph_width, max_change);
2854                                 add = total - del;
2855                         }
2856                 }
2857                 strbuf_addf(&out, " %s%-*s |", prefix, len, name);
2858                 strbuf_addf(&out, " %*"PRIuMAX"%s",
2859                         number_width, added + deleted,
2860                         added + deleted ? " " : "");
2861                 show_graph(&out, '+', add, add_c, reset);
2862                 show_graph(&out, '-', del, del_c, reset);
2863                 strbuf_addch(&out, '\n');
2864                 emit_diff_symbol(options, DIFF_SYMBOL_STATS_LINE,
2865                                  out.buf, out.len, 0);
2866                 strbuf_reset(&out);
2867         }
2868
2869         for (i = 0; i < data->nr; i++) {
2870                 struct diffstat_file *file = data->files[i];
2871                 uintmax_t added = file->added;
2872                 uintmax_t deleted = file->deleted;
2873
2874                 if (file->is_unmerged ||
2875                     (!file->is_interesting && (added + deleted == 0))) {
2876                         total_files--;
2877                         continue;
2878                 }
2879
2880                 if (!file->is_binary) {
2881                         adds += added;
2882                         dels += deleted;
2883                 }
2884                 if (i < count)
2885                         continue;
2886                 if (!extra_shown)
2887                         emit_diff_symbol(options,
2888                                          DIFF_SYMBOL_STATS_SUMMARY_ABBREV,
2889                                          NULL, 0, 0);
2890                 extra_shown = 1;
2891         }
2892
2893         print_stat_summary_inserts_deletes(options, total_files, adds, dels);
2894         strbuf_release(&out);
2895 }
2896
2897 static void show_shortstats(struct diffstat_t *data, struct diff_options *options)
2898 {
2899         int i, adds = 0, dels = 0, total_files = data->nr;
2900
2901         if (data->nr == 0)
2902                 return;
2903
2904         for (i = 0; i < data->nr; i++) {
2905                 int added = data->files[i]->added;
2906                 int deleted = data->files[i]->deleted;
2907
2908                 if (data->files[i]->is_unmerged ||
2909                     (!data->files[i]->is_interesting && (added + deleted == 0))) {
2910                         total_files--;
2911                 } else if (!data->files[i]->is_binary) { /* don't count bytes */
2912                         adds += added;
2913                         dels += deleted;
2914                 }
2915         }
2916         print_stat_summary_inserts_deletes(options, total_files, adds, dels);
2917 }
2918
2919 static void show_numstat(struct diffstat_t *data, struct diff_options *options)
2920 {
2921         int i;
2922
2923         if (data->nr == 0)
2924                 return;
2925
2926         for (i = 0; i < data->nr; i++) {
2927                 struct diffstat_file *file = data->files[i];
2928
2929                 fprintf(options->file, "%s", diff_line_prefix(options));
2930
2931                 if (file->is_binary)
2932                         fprintf(options->file, "-\t-\t");
2933                 else
2934                         fprintf(options->file,
2935                                 "%"PRIuMAX"\t%"PRIuMAX"\t",
2936                                 file->added, file->deleted);
2937                 if (options->line_termination) {
2938                         fill_print_name(file);
2939                         if (!file->is_renamed)
2940                                 write_name_quoted(file->name, options->file,
2941                                                   options->line_termination);
2942                         else {
2943                                 fputs(file->print_name, options->file);
2944                                 putc(options->line_termination, options->file);
2945                         }
2946                 } else {
2947                         if (file->is_renamed) {
2948                                 putc('\0', options->file);
2949                                 write_name_quoted(file->from_name, options->file, '\0');
2950                         }
2951                         write_name_quoted(file->name, options->file, '\0');
2952                 }
2953         }
2954 }
2955
2956 struct dirstat_file {
2957         const char *name;
2958         unsigned long changed;
2959 };
2960
2961 struct dirstat_dir {
2962         struct dirstat_file *files;
2963         int alloc, nr, permille, cumulative;
2964 };
2965
2966 static long gather_dirstat(struct diff_options *opt, struct dirstat_dir *dir,
2967                 unsigned long changed, const char *base, int baselen)
2968 {
2969         unsigned long sum_changes = 0;
2970         unsigned int sources = 0;
2971         const char *line_prefix = diff_line_prefix(opt);
2972
2973         while (dir->nr) {
2974                 struct dirstat_file *f = dir->files;
2975                 int namelen = strlen(f->name);
2976                 unsigned long changes;
2977                 char *slash;
2978
2979                 if (namelen < baselen)
2980                         break;
2981                 if (memcmp(f->name, base, baselen))
2982                         break;
2983                 slash = strchr(f->name + baselen, '/');
2984                 if (slash) {
2985                         int newbaselen = slash + 1 - f->name;
2986                         changes = gather_dirstat(opt, dir, changed, f->name, newbaselen);
2987                         sources++;
2988                 } else {
2989                         changes = f->changed;
2990                         dir->files++;
2991                         dir->nr--;
2992                         sources += 2;
2993                 }
2994                 sum_changes += changes;
2995         }
2996
2997         /*
2998          * We don't report dirstat's for
2999          *  - the top level
3000          *  - or cases where everything came from a single directory
3001          *    under this directory (sources == 1).
3002          */
3003         if (baselen && sources != 1) {
3004                 if (sum_changes) {
3005                         int permille = sum_changes * 1000 / changed;
3006                         if (permille >= dir->permille) {
3007                                 fprintf(opt->file, "%s%4d.%01d%% %.*s\n", line_prefix,
3008                                         permille / 10, permille % 10, baselen, base);
3009                                 if (!dir->cumulative)
3010                                         return 0;
3011                         }
3012                 }
3013         }
3014         return sum_changes;
3015 }
3016
3017 static int dirstat_compare(const void *_a, const void *_b)
3018 {
3019         const struct dirstat_file *a = _a;
3020         const struct dirstat_file *b = _b;
3021         return strcmp(a->name, b->name);
3022 }
3023
3024 static void show_dirstat(struct diff_options *options)
3025 {
3026         int i;
3027         unsigned long changed;
3028         struct dirstat_dir dir;
3029         struct diff_queue_struct *q = &diff_queued_diff;
3030
3031         dir.files = NULL;
3032         dir.alloc = 0;
3033         dir.nr = 0;
3034         dir.permille = options->dirstat_permille;
3035         dir.cumulative = options->flags.dirstat_cumulative;
3036
3037         changed = 0;
3038         for (i = 0; i < q->nr; i++) {
3039                 struct diff_filepair *p = q->queue[i];
3040                 const char *name;
3041                 unsigned long copied, added, damage;
3042
3043                 name = p->two->path ? p->two->path : p->one->path;
3044
3045                 if (p->one->oid_valid && p->two->oid_valid &&
3046                     oideq(&p->one->oid, &p->two->oid)) {
3047                         /*
3048                          * The SHA1 has not changed, so pre-/post-content is
3049                          * identical. We can therefore skip looking at the
3050                          * file contents altogether.
3051                          */
3052                         damage = 0;
3053                         goto found_damage;
3054                 }
3055
3056                 if (options->flags.dirstat_by_file) {
3057                         /*
3058                          * In --dirstat-by-file mode, we don't really need to
3059                          * look at the actual file contents at all.
3060                          * The fact that the SHA1 changed is enough for us to
3061                          * add this file to the list of results
3062                          * (with each file contributing equal damage).
3063                          */
3064                         damage = 1;
3065                         goto found_damage;
3066                 }
3067
3068                 if (DIFF_FILE_VALID(p->one) && DIFF_FILE_VALID(p->two)) {
3069                         diff_populate_filespec(options->repo, p->one, 0);
3070                         diff_populate_filespec(options->repo, p->two, 0);
3071                         diffcore_count_changes(options->repo,
3072                                                p->one, p->two, NULL, NULL,
3073                                                &copied, &added);
3074                         diff_free_filespec_data(p->one);
3075                         diff_free_filespec_data(p->two);
3076                 } else if (DIFF_FILE_VALID(p->one)) {
3077                         diff_populate_filespec(options->repo, p->one, CHECK_SIZE_ONLY);
3078                         copied = added = 0;
3079                         diff_free_filespec_data(p->one);
3080                 } else if (DIFF_FILE_VALID(p->two)) {
3081                         diff_populate_filespec(options->repo, p->two, CHECK_SIZE_ONLY);
3082                         copied = 0;
3083                         added = p->two->size;
3084                         diff_free_filespec_data(p->two);
3085                 } else
3086                         continue;
3087
3088                 /*
3089                  * Original minus copied is the removed material,
3090                  * added is the new material.  They are both damages
3091                  * made to the preimage.
3092                  * If the resulting damage is zero, we know that
3093                  * diffcore_count_changes() considers the two entries to
3094                  * be identical, but since the oid changed, we
3095                  * know that there must have been _some_ kind of change,
3096                  * so we force all entries to have damage > 0.
3097                  */
3098                 damage = (p->one->size - copied) + added;
3099                 if (!damage)
3100                         damage = 1;
3101
3102 found_damage:
3103                 ALLOC_GROW(dir.files, dir.nr + 1, dir.alloc);
3104                 dir.files[dir.nr].name = name;
3105                 dir.files[dir.nr].changed = damage;
3106                 changed += damage;
3107                 dir.nr++;
3108         }
3109
3110         /* This can happen even with many files, if everything was renames */
3111         if (!changed)
3112                 return;
3113
3114         /* Show all directories with more than x% of the changes */
3115         QSORT(dir.files, dir.nr, dirstat_compare);
3116         gather_dirstat(options, &dir, changed, "", 0);
3117 }
3118
3119 static void show_dirstat_by_line(struct diffstat_t *data, struct diff_options *options)
3120 {
3121         int i;
3122         unsigned long changed;
3123         struct dirstat_dir dir;
3124
3125         if (data->nr == 0)
3126                 return;
3127
3128         dir.files = NULL;
3129         dir.alloc = 0;
3130         dir.nr = 0;
3131         dir.permille = options->dirstat_permille;
3132         dir.cumulative = options->flags.dirstat_cumulative;
3133
3134         changed = 0;
3135         for (i = 0; i < data->nr; i++) {
3136                 struct diffstat_file *file = data->files[i];
3137                 unsigned long damage = file->added + file->deleted;
3138                 if (file->is_binary)
3139                         /*
3140                          * binary files counts bytes, not lines. Must find some
3141                          * way to normalize binary bytes vs. textual lines.
3142                          * The following heuristic assumes that there are 64
3143                          * bytes per "line".
3144                          * This is stupid and ugly, but very cheap...
3145                          */
3146                         damage = DIV_ROUND_UP(damage, 64);
3147                 ALLOC_GROW(dir.files, dir.nr + 1, dir.alloc);
3148                 dir.files[dir.nr].name = file->name;
3149                 dir.files[dir.nr].changed = damage;
3150                 changed += damage;
3151                 dir.nr++;
3152         }
3153
3154         /* This can happen even with many files, if everything was renames */
3155         if (!changed)
3156                 return;
3157
3158         /* Show all directories with more than x% of the changes */
3159         QSORT(dir.files, dir.nr, dirstat_compare);
3160         gather_dirstat(options, &dir, changed, "", 0);
3161 }
3162
3163 static void free_diffstat_info(struct diffstat_t *diffstat)
3164 {
3165         int i;
3166         for (i = 0; i < diffstat->nr; i++) {
3167                 struct diffstat_file *f = diffstat->files[i];
3168                 free(f->print_name);
3169                 free(f->name);
3170                 free(f->from_name);
3171                 free(f);
3172         }
3173         free(diffstat->files);
3174 }
3175
3176 struct checkdiff_t {
3177         const char *filename;
3178         int lineno;
3179         int conflict_marker_size;
3180         struct diff_options *o;
3181         unsigned ws_rule;
3182         unsigned status;
3183 };
3184
3185 static int is_conflict_marker(const char *line, int marker_size, unsigned long len)
3186 {
3187         char firstchar;
3188         int cnt;
3189
3190         if (len < marker_size + 1)
3191                 return 0;
3192         firstchar = line[0];
3193         switch (firstchar) {
3194         case '=': case '>': case '<': case '|':
3195                 break;
3196         default:
3197                 return 0;
3198         }
3199         for (cnt = 1; cnt < marker_size; cnt++)
3200                 if (line[cnt] != firstchar)
3201                         return 0;
3202         /* line[1] thru line[marker_size-1] are same as firstchar */
3203         if (len < marker_size + 1 || !isspace(line[marker_size]))
3204                 return 0;
3205         return 1;
3206 }
3207
3208 static void checkdiff_consume_hunk(void *priv,
3209                                    long ob, long on, long nb, long nn,
3210                                    const char *func, long funclen)
3211
3212 {
3213         struct checkdiff_t *data = priv;
3214         data->lineno = nb - 1;
3215 }
3216
3217 static void checkdiff_consume(void *priv, char *line, unsigned long len)
3218 {
3219         struct checkdiff_t *data = priv;
3220         int marker_size = data->conflict_marker_size;
3221         const char *ws = diff_get_color(data->o->use_color, DIFF_WHITESPACE);
3222         const char *reset = diff_get_color(data->o->use_color, DIFF_RESET);
3223         const char *set = diff_get_color(data->o->use_color, DIFF_FILE_NEW);
3224         char *err;
3225         const char *line_prefix;
3226
3227         assert(data->o);
3228         line_prefix = diff_line_prefix(data->o);
3229
3230         if (line[0] == '+') {
3231                 unsigned bad;
3232                 data->lineno++;
3233                 if (is_conflict_marker(line + 1, marker_size, len - 1)) {
3234                         data->status |= 1;
3235                         fprintf(data->o->file,
3236                                 "%s%s:%d: leftover conflict marker\n",
3237                                 line_prefix, data->filename, data->lineno);
3238                 }
3239                 bad = ws_check(line + 1, len - 1, data->ws_rule);
3240                 if (!bad)
3241                         return;
3242                 data->status |= bad;
3243                 err = whitespace_error_string(bad);
3244                 fprintf(data->o->file, "%s%s:%d: %s.\n",
3245                         line_prefix, data->filename, data->lineno, err);
3246                 free(err);
3247                 emit_line(data->o, set, reset, line, 1);
3248                 ws_check_emit(line + 1, len - 1, data->ws_rule,
3249                               data->o->file, set, reset, ws);
3250         } else if (line[0] == ' ') {
3251                 data->lineno++;
3252         }
3253 }
3254
3255 static unsigned char *deflate_it(char *data,
3256                                  unsigned long size,
3257                                  unsigned long *result_size)
3258 {
3259         int bound;
3260         unsigned char *deflated;
3261         git_zstream stream;
3262
3263         git_deflate_init(&stream, zlib_compression_level);
3264         bound = git_deflate_bound(&stream, size);
3265         deflated = xmalloc(bound);
3266         stream.next_out = deflated;
3267         stream.avail_out = bound;
3268
3269         stream.next_in = (unsigned char *)data;
3270         stream.avail_in = size;
3271         while (git_deflate(&stream, Z_FINISH) == Z_OK)
3272                 ; /* nothing */
3273         git_deflate_end(&stream);
3274         *result_size = stream.total_out;
3275         return deflated;
3276 }
3277
3278 static void emit_binary_diff_body(struct diff_options *o,
3279                                   mmfile_t *one, mmfile_t *two)
3280 {
3281         void *cp;
3282         void *delta;
3283         void *deflated;
3284         void *data;
3285         unsigned long orig_size;
3286         unsigned long delta_size;
3287         unsigned long deflate_size;
3288         unsigned long data_size;
3289
3290         /* We could do deflated delta, or we could do just deflated two,
3291          * whichever is smaller.
3292          */
3293         delta = NULL;
3294         deflated = deflate_it(two->ptr, two->size, &deflate_size);
3295         if (one->size && two->size) {
3296                 delta = diff_delta(one->ptr, one->size,
3297                                    two->ptr, two->size,
3298                                    &delta_size, deflate_size);
3299                 if (delta) {
3300                         void *to_free = delta;
3301                         orig_size = delta_size;
3302                         delta = deflate_it(delta, delta_size, &delta_size);
3303                         free(to_free);
3304                 }
3305         }
3306
3307         if (delta && delta_size < deflate_size) {
3308                 char *s = xstrfmt("%"PRIuMAX , (uintmax_t)orig_size);
3309                 emit_diff_symbol(o, DIFF_SYMBOL_BINARY_DIFF_HEADER_DELTA,
3310                                  s, strlen(s), 0);
3311                 free(s);
3312                 free(deflated);
3313                 data = delta;
3314                 data_size = delta_size;
3315         } else {
3316                 char *s = xstrfmt("%lu", two->size);
3317                 emit_diff_symbol(o, DIFF_SYMBOL_BINARY_DIFF_HEADER_LITERAL,
3318                                  s, strlen(s), 0);
3319                 free(s);
3320                 free(delta);
3321                 data = deflated;
3322                 data_size = deflate_size;
3323         }
3324
3325         /* emit data encoded in base85 */
3326         cp = data;
3327         while (data_size) {
3328                 int len;
3329                 int bytes = (52 < data_size) ? 52 : data_size;
3330                 char line[71];
3331                 data_size -= bytes;
3332                 if (bytes <= 26)
3333                         line[0] = bytes + 'A' - 1;
3334                 else
3335                         line[0] = bytes - 26 + 'a' - 1;
3336                 encode_85(line + 1, cp, bytes);
3337                 cp = (char *) cp + bytes;
3338
3339                 len = strlen(line);
3340                 line[len++] = '\n';
3341                 line[len] = '\0';
3342
3343                 emit_diff_symbol(o, DIFF_SYMBOL_BINARY_DIFF_BODY,
3344                                  line, len, 0);
3345         }
3346         emit_diff_symbol(o, DIFF_SYMBOL_BINARY_DIFF_FOOTER, NULL, 0, 0);
3347         free(data);
3348 }
3349
3350 static void emit_binary_diff(struct diff_options *o,
3351                              mmfile_t *one, mmfile_t *two)
3352 {
3353         emit_diff_symbol(o, DIFF_SYMBOL_BINARY_DIFF_HEADER, NULL, 0, 0);
3354         emit_binary_diff_body(o, one, two);
3355         emit_binary_diff_body(o, two, one);
3356 }
3357
3358 int diff_filespec_is_binary(struct repository *r,
3359                             struct diff_filespec *one)
3360 {
3361         if (one->is_binary == -1) {
3362                 diff_filespec_load_driver(one, r->index);
3363                 if (one->driver->binary != -1)
3364                         one->is_binary = one->driver->binary;
3365                 else {
3366                         if (!one->data && DIFF_FILE_VALID(one))
3367                                 diff_populate_filespec(r, one, CHECK_BINARY);
3368                         if (one->is_binary == -1 && one->data)
3369                                 one->is_binary = buffer_is_binary(one->data,
3370                                                 one->size);
3371                         if (one->is_binary == -1)
3372                                 one->is_binary = 0;
3373                 }
3374         }
3375         return one->is_binary;
3376 }
3377
3378 static const struct userdiff_funcname *
3379 diff_funcname_pattern(struct diff_options *o, struct diff_filespec *one)
3380 {
3381         diff_filespec_load_driver(one, o->repo->index);
3382         return one->driver->funcname.pattern ? &one->driver->funcname : NULL;
3383 }
3384
3385 void diff_set_mnemonic_prefix(struct diff_options *options, const char *a, const char *b)
3386 {
3387         if (!options->a_prefix)
3388                 options->a_prefix = a;
3389         if (!options->b_prefix)
3390                 options->b_prefix = b;
3391 }
3392
3393 struct userdiff_driver *get_textconv(struct repository *r,
3394                                      struct diff_filespec *one)
3395 {
3396         if (!DIFF_FILE_VALID(one))
3397                 return NULL;
3398
3399         diff_filespec_load_driver(one, r->index);
3400         return userdiff_get_textconv(r, one->driver);
3401 }
3402
3403 static void builtin_diff(const char *name_a,
3404                          const char *name_b,
3405                          struct diff_filespec *one,
3406                          struct diff_filespec *two,
3407                          const char *xfrm_msg,
3408                          int must_show_header,
3409                          struct diff_options *o,
3410                          int complete_rewrite)
3411 {
3412         mmfile_t mf1, mf2;
3413         const char *lbl[2];
3414         char *a_one, *b_two;
3415         const char *meta = diff_get_color_opt(o, DIFF_METAINFO);
3416         const char *reset = diff_get_color_opt(o, DIFF_RESET);
3417         const char *a_prefix, *b_prefix;
3418         struct userdiff_driver *textconv_one = NULL;
3419         struct userdiff_driver *textconv_two = NULL;
3420         struct strbuf header = STRBUF_INIT;
3421         const char *line_prefix = diff_line_prefix(o);
3422
3423         diff_set_mnemonic_prefix(o, "a/", "b/");
3424         if (o->flags.reverse_diff) {
3425                 a_prefix = o->b_prefix;
3426                 b_prefix = o->a_prefix;
3427         } else {
3428                 a_prefix = o->a_prefix;
3429                 b_prefix = o->b_prefix;
3430         }
3431
3432         if (o->submodule_format == DIFF_SUBMODULE_LOG &&
3433             (!one->mode || S_ISGITLINK(one->mode)) &&
3434             (!two->mode || S_ISGITLINK(two->mode))) {
3435                 show_submodule_summary(o, one->path ? one->path : two->path,
3436                                 &one->oid, &two->oid,
3437                                 two->dirty_submodule);
3438                 return;
3439         } else if (o->submodule_format == DIFF_SUBMODULE_INLINE_DIFF &&
3440                    (!one->mode || S_ISGITLINK(one->mode)) &&
3441                    (!two->mode || S_ISGITLINK(two->mode))) {
3442                 show_submodule_inline_diff(o, one->path ? one->path : two->path,
3443                                 &one->oid, &two->oid,
3444                                 two->dirty_submodule);
3445                 return;
3446         }
3447
3448         if (o->flags.allow_textconv) {
3449                 textconv_one = get_textconv(o->repo, one);
3450                 textconv_two = get_textconv(o->repo, two);
3451         }
3452
3453         /* Never use a non-valid filename anywhere if at all possible */
3454         name_a = DIFF_FILE_VALID(one) ? name_a : name_b;
3455         name_b = DIFF_FILE_VALID(two) ? name_b : name_a;
3456
3457         a_one = quote_two(a_prefix, name_a + (*name_a == '/'));
3458         b_two = quote_two(b_prefix, name_b + (*name_b == '/'));
3459         lbl[0] = DIFF_FILE_VALID(one) ? a_one : "/dev/null";
3460         lbl[1] = DIFF_FILE_VALID(two) ? b_two : "/dev/null";
3461         strbuf_addf(&header, "%s%sdiff --git %s %s%s\n", line_prefix, meta, a_one, b_two, reset);
3462         if (lbl[0][0] == '/') {
3463                 /* /dev/null */
3464                 strbuf_addf(&header, "%s%snew file mode %06o%s\n", line_prefix, meta, two->mode, reset);
3465                 if (xfrm_msg)
3466                         strbuf_addstr(&header, xfrm_msg);
3467                 must_show_header = 1;
3468         }
3469         else if (lbl[1][0] == '/') {
3470                 strbuf_addf(&header, "%s%sdeleted file mode %06o%s\n", line_prefix, meta, one->mode, reset);
3471                 if (xfrm_msg)
3472                         strbuf_addstr(&header, xfrm_msg);
3473                 must_show_header = 1;
3474         }
3475         else {
3476                 if (one->mode != two->mode) {
3477                         strbuf_addf(&header, "%s%sold mode %06o%s\n", line_prefix, meta, one->mode, reset);
3478                         strbuf_addf(&header, "%s%snew mode %06o%s\n", line_prefix, meta, two->mode, reset);
3479                         must_show_header = 1;
3480                 }
3481                 if (xfrm_msg)
3482                         strbuf_addstr(&header, xfrm_msg);
3483
3484                 /*
3485                  * we do not run diff between different kind
3486                  * of objects.
3487                  */
3488                 if ((one->mode ^ two->mode) & S_IFMT)
3489                         goto free_ab_and_return;
3490                 if (complete_rewrite &&
3491                     (textconv_one || !diff_filespec_is_binary(o->repo, one)) &&
3492                     (textconv_two || !diff_filespec_is_binary(o->repo, two))) {
3493                         emit_diff_symbol(o, DIFF_SYMBOL_HEADER,
3494                                          header.buf, header.len, 0);
3495                         strbuf_reset(&header);
3496                         emit_rewrite_diff(name_a, name_b, one, two,
3497                                           textconv_one, textconv_two, o);
3498                         o->found_changes = 1;
3499                         goto free_ab_and_return;
3500                 }
3501         }
3502
3503         if (o->irreversible_delete && lbl[1][0] == '/') {
3504                 emit_diff_symbol(o, DIFF_SYMBOL_HEADER, header.buf,
3505                                  header.len, 0);
3506                 strbuf_reset(&header);
3507                 goto free_ab_and_return;
3508         } else if (!o->flags.text &&
3509                    ( (!textconv_one && diff_filespec_is_binary(o->repo, one)) ||
3510                      (!textconv_two && diff_filespec_is_binary(o->repo, two)) )) {
3511                 struct strbuf sb = STRBUF_INIT;
3512                 if (!one->data && !two->data &&
3513                     S_ISREG(one->mode) && S_ISREG(two->mode) &&
3514                     !o->flags.binary) {
3515                         if (oideq(&one->oid, &two->oid)) {
3516                                 if (must_show_header)
3517                                         emit_diff_symbol(o, DIFF_SYMBOL_HEADER,
3518                                                          header.buf, header.len,
3519                                                          0);
3520                                 goto free_ab_and_return;
3521                         }
3522                         emit_diff_symbol(o, DIFF_SYMBOL_HEADER,
3523                                          header.buf, header.len, 0);
3524                         strbuf_addf(&sb, "%sBinary files %s and %s differ\n",
3525                                     diff_line_prefix(o), lbl[0], lbl[1]);
3526                         emit_diff_symbol(o, DIFF_SYMBOL_BINARY_FILES,
3527                                          sb.buf, sb.len, 0);
3528                         strbuf_release(&sb);
3529                         goto free_ab_and_return;
3530                 }
3531                 if (fill_mmfile(o->repo, &mf1, one) < 0 ||
3532                     fill_mmfile(o->repo, &mf2, two) < 0)
3533                         die("unable to read files to diff");
3534                 /* Quite common confusing case */
3535                 if (mf1.size == mf2.size &&
3536                     !memcmp(mf1.ptr, mf2.ptr, mf1.size)) {
3537                         if (must_show_header)
3538                                 emit_diff_symbol(o, DIFF_SYMBOL_HEADER,
3539                                                  header.buf, header.len, 0);
3540                         goto free_ab_and_return;
3541                 }
3542                 emit_diff_symbol(o, DIFF_SYMBOL_HEADER, header.buf, header.len, 0);
3543                 strbuf_reset(&header);
3544                 if (o->flags.binary)
3545                         emit_binary_diff(o, &mf1, &mf2);
3546                 else {
3547                         strbuf_addf(&sb, "%sBinary files %s and %s differ\n",
3548                                     diff_line_prefix(o), lbl[0], lbl[1]);
3549                         emit_diff_symbol(o, DIFF_SYMBOL_BINARY_FILES,
3550                                          sb.buf, sb.len, 0);
3551                         strbuf_release(&sb);
3552                 }
3553                 o->found_changes = 1;
3554         } else {
3555                 /* Crazy xdl interfaces.. */
3556                 const char *diffopts;
3557                 const char *v;
3558                 xpparam_t xpp;
3559                 xdemitconf_t xecfg;
3560                 struct emit_callback ecbdata;
3561                 const struct userdiff_funcname *pe;
3562
3563                 if (must_show_header) {
3564                         emit_diff_symbol(o, DIFF_SYMBOL_HEADER,
3565                                          header.buf, header.len, 0);
3566                         strbuf_reset(&header);
3567                 }
3568
3569                 mf1.size = fill_textconv(o->repo, textconv_one, one, &mf1.ptr);
3570                 mf2.size = fill_textconv(o->repo, textconv_two, two, &mf2.ptr);
3571
3572                 pe = diff_funcname_pattern(o, one);
3573                 if (!pe)
3574                         pe = diff_funcname_pattern(o, two);
3575
3576                 memset(&xpp, 0, sizeof(xpp));
3577                 memset(&xecfg, 0, sizeof(xecfg));
3578                 memset(&ecbdata, 0, sizeof(ecbdata));
3579                 if (o->flags.suppress_diff_headers)
3580                         lbl[0] = NULL;
3581                 ecbdata.label_path = lbl;
3582                 ecbdata.color_diff = want_color(o->use_color);
3583                 ecbdata.ws_rule = whitespace_rule(o->repo->index, name_b);
3584                 if (ecbdata.ws_rule & WS_BLANK_AT_EOF)
3585                         check_blank_at_eof(&mf1, &mf2, &ecbdata);
3586                 ecbdata.opt = o;
3587                 if (header.len && !o->flags.suppress_diff_headers)
3588                         ecbdata.header = &header;
3589                 xpp.flags = o->xdl_opts;
3590                 xpp.anchors = o->anchors;
3591                 xpp.anchors_nr = o->anchors_nr;
3592                 xecfg.ctxlen = o->context;
3593                 xecfg.interhunkctxlen = o->interhunkcontext;
3594                 xecfg.flags = XDL_EMIT_FUNCNAMES;
3595                 if (o->flags.funccontext)
3596                         xecfg.flags |= XDL_EMIT_FUNCCONTEXT;
3597                 if (pe)
3598                         xdiff_set_find_func(&xecfg, pe->pattern, pe->cflags);
3599
3600                 diffopts = getenv("GIT_DIFF_OPTS");
3601                 if (!diffopts)
3602                         ;
3603                 else if (skip_prefix(diffopts, "--unified=", &v))
3604                         xecfg.ctxlen = strtoul(v, NULL, 10);
3605                 else if (skip_prefix(diffopts, "-u", &v))
3606                         xecfg.ctxlen = strtoul(v, NULL, 10);
3607
3608                 if (o->word_diff)
3609                         init_diff_words_data(&ecbdata, o, one, two);
3610                 if (xdi_diff_outf(&mf1, &mf2, NULL, fn_out_consume,
3611                                   &ecbdata, &xpp, &xecfg))
3612                         die("unable to generate diff for %s", one->path);
3613                 if (o->word_diff)
3614                         free_diff_words_data(&ecbdata);
3615                 if (textconv_one)
3616                         free(mf1.ptr);
3617                 if (textconv_two)
3618                         free(mf2.ptr);
3619                 xdiff_clear_find_func(&xecfg);
3620         }
3621
3622  free_ab_and_return:
3623         strbuf_release(&header);
3624         diff_free_filespec_data(one);
3625         diff_free_filespec_data(two);
3626         free(a_one);
3627         free(b_two);
3628         return;
3629 }
3630
3631 static char *get_compact_summary(const struct diff_filepair *p, int is_renamed)
3632 {
3633         if (!is_renamed) {
3634                 if (p->status == DIFF_STATUS_ADDED) {
3635                         if (S_ISLNK(p->two->mode))
3636                                 return "new +l";
3637                         else if ((p->two->mode & 0777) == 0755)
3638                                 return "new +x";
3639                         else
3640                                 return "new";
3641                 } else if (p->status == DIFF_STATUS_DELETED)
3642                         return "gone";
3643         }
3644         if (S_ISLNK(p->one->mode) && !S_ISLNK(p->two->mode))
3645                 return "mode -l";
3646         else if (!S_ISLNK(p->one->mode) && S_ISLNK(p->two->mode))
3647                 return "mode +l";
3648         else if ((p->one->mode & 0777) == 0644 &&
3649                  (p->two->mode & 0777) == 0755)
3650                 return "mode +x";
3651         else if ((p->one->mode & 0777) == 0755 &&
3652                  (p->two->mode & 0777) == 0644)
3653                 return "mode -x";
3654         return NULL;
3655 }
3656
3657 static void builtin_diffstat(const char *name_a, const char *name_b,
3658                              struct diff_filespec *one,
3659                              struct diff_filespec *two,
3660                              struct diffstat_t *diffstat,
3661                              struct diff_options *o,
3662                              struct diff_filepair *p)
3663 {
3664         mmfile_t mf1, mf2;
3665         struct diffstat_file *data;
3666         int same_contents;
3667         int complete_rewrite = 0;
3668
3669         if (!DIFF_PAIR_UNMERGED(p)) {
3670                 if (p->status == DIFF_STATUS_MODIFIED && p->score)
3671                         complete_rewrite = 1;
3672         }
3673
3674         data = diffstat_add(diffstat, name_a, name_b);
3675         data->is_interesting = p->status != DIFF_STATUS_UNKNOWN;
3676         if (o->flags.stat_with_summary)
3677                 data->comments = get_compact_summary(p, data->is_renamed);
3678
3679         if (!one || !two) {
3680                 data->is_unmerged = 1;
3681                 return;
3682         }
3683
3684         same_contents = oideq(&one->oid, &two->oid);
3685
3686         if (diff_filespec_is_binary(o->repo, one) ||
3687             diff_filespec_is_binary(o->repo, two)) {
3688                 data->is_binary = 1;
3689                 if (same_contents) {
3690                         data->added = 0;
3691                         data->deleted = 0;
3692                 } else {
3693                         data->added = diff_filespec_size(o->repo, two);
3694                         data->deleted = diff_filespec_size(o->repo, one);
3695                 }
3696         }
3697
3698         else if (complete_rewrite) {
3699                 diff_populate_filespec(o->repo, one, 0);
3700                 diff_populate_filespec(o->repo, two, 0);
3701                 data->deleted = count_lines(one->data, one->size);
3702                 data->added = count_lines(two->data, two->size);
3703         }
3704
3705         else if (!same_contents) {
3706                 /* Crazy xdl interfaces.. */
3707                 xpparam_t xpp;
3708                 xdemitconf_t xecfg;
3709
3710                 if (fill_mmfile(o->repo, &mf1, one) < 0 ||
3711                     fill_mmfile(o->repo, &mf2, two) < 0)
3712                         die("unable to read files to diff");
3713
3714                 memset(&xpp, 0, sizeof(xpp));
3715                 memset(&xecfg, 0, sizeof(xecfg));
3716                 xpp.flags = o->xdl_opts;
3717                 xpp.anchors = o->anchors;
3718                 xpp.anchors_nr = o->anchors_nr;
3719                 xecfg.ctxlen = o->context;
3720                 xecfg.interhunkctxlen = o->interhunkcontext;
3721                 if (xdi_diff_outf(&mf1, &mf2, discard_hunk_line,
3722                                   diffstat_consume, diffstat, &xpp, &xecfg))
3723                         die("unable to generate diffstat for %s", one->path);
3724         }
3725
3726         diff_free_filespec_data(one);
3727         diff_free_filespec_data(two);
3728 }
3729
3730 static void builtin_checkdiff(const char *name_a, const char *name_b,
3731                               const char *attr_path,
3732                               struct diff_filespec *one,
3733                               struct diff_filespec *two,
3734                               struct diff_options *o)
3735 {
3736         mmfile_t mf1, mf2;
3737         struct checkdiff_t data;
3738
3739         if (!two)
3740                 return;
3741
3742         memset(&data, 0, sizeof(data));
3743         data.filename = name_b ? name_b : name_a;
3744         data.lineno = 0;
3745         data.o = o;
3746         data.ws_rule = whitespace_rule(o->repo->index, attr_path);
3747         data.conflict_marker_size = ll_merge_marker_size(o->repo->index, attr_path);
3748
3749         if (fill_mmfile(o->repo, &mf1, one) < 0 ||
3750             fill_mmfile(o->repo, &mf2, two) < 0)
3751                 die("unable to read files to diff");
3752
3753         /*
3754          * All the other codepaths check both sides, but not checking
3755          * the "old" side here is deliberate.  We are checking the newly
3756          * introduced changes, and as long as the "new" side is text, we
3757          * can and should check what it introduces.
3758          */
3759         if (diff_filespec_is_binary(o->repo, two))
3760                 goto free_and_return;
3761         else {
3762                 /* Crazy xdl interfaces.. */
3763                 xpparam_t xpp;
3764                 xdemitconf_t xecfg;
3765
3766                 memset(&xpp, 0, sizeof(xpp));
3767                 memset(&xecfg, 0, sizeof(xecfg));
3768                 xecfg.ctxlen = 1; /* at least one context line */
3769                 xpp.flags = 0;
3770                 if (xdi_diff_outf(&mf1, &mf2, checkdiff_consume_hunk,
3771                                   checkdiff_consume, &data,
3772                                   &xpp, &xecfg))
3773                         die("unable to generate checkdiff for %s", one->path);
3774
3775                 if (data.ws_rule & WS_BLANK_AT_EOF) {
3776                         struct emit_callback ecbdata;
3777                         int blank_at_eof;
3778
3779                         ecbdata.ws_rule = data.ws_rule;
3780                         check_blank_at_eof(&mf1, &mf2, &ecbdata);
3781                         blank_at_eof = ecbdata.blank_at_eof_in_postimage;
3782
3783                         if (blank_at_eof) {
3784                                 static char *err;
3785                                 if (!err)
3786                                         err = whitespace_error_string(WS_BLANK_AT_EOF);
3787                                 fprintf(o->file, "%s:%d: %s.\n",
3788                                         data.filename, blank_at_eof, err);
3789                                 data.status = 1; /* report errors */
3790                         }
3791                 }
3792         }
3793  free_and_return:
3794         diff_free_filespec_data(one);
3795         diff_free_filespec_data(two);
3796         if (data.status)
3797                 o->flags.check_failed = 1;
3798 }
3799
3800 struct diff_filespec *alloc_filespec(const char *path)
3801 {
3802         struct diff_filespec *spec;
3803
3804         FLEXPTR_ALLOC_STR(spec, path, path);
3805         spec->count = 1;
3806         spec->is_binary = -1;
3807         return spec;
3808 }
3809
3810 void free_filespec(struct diff_filespec *spec)
3811 {
3812         if (!--spec->count) {
3813                 diff_free_filespec_data(spec);
3814                 free(spec);
3815         }
3816 }
3817
3818 void fill_filespec(struct diff_filespec *spec, const struct object_id *oid,
3819                    int oid_valid, unsigned short mode)
3820 {
3821         if (mode) {
3822                 spec->mode = canon_mode(mode);
3823                 oidcpy(&spec->oid, oid);
3824                 spec->oid_valid = oid_valid;
3825         }
3826 }
3827
3828 /*
3829  * Given a name and sha1 pair, if the index tells us the file in
3830  * the work tree has that object contents, return true, so that
3831  * prepare_temp_file() does not have to inflate and extract.
3832  */
3833 static int reuse_worktree_file(struct index_state *istate,
3834                                const char *name,
3835                                const struct object_id *oid,
3836                                int want_file)
3837 {
3838         const struct cache_entry *ce;
3839         struct stat st;
3840         int pos, len;
3841
3842         /*
3843          * We do not read the cache ourselves here, because the
3844          * benchmark with my previous version that always reads cache
3845          * shows that it makes things worse for diff-tree comparing
3846          * two linux-2.6 kernel trees in an already checked out work
3847          * tree.  This is because most diff-tree comparisons deal with
3848          * only a small number of files, while reading the cache is
3849          * expensive for a large project, and its cost outweighs the
3850          * savings we get by not inflating the object to a temporary
3851          * file.  Practically, this code only helps when we are used
3852          * by diff-cache --cached, which does read the cache before
3853          * calling us.
3854          */
3855         if (!istate->cache)
3856                 return 0;
3857
3858         /* We want to avoid the working directory if our caller
3859          * doesn't need the data in a normal file, this system
3860          * is rather slow with its stat/open/mmap/close syscalls,
3861          * and the object is contained in a pack file.  The pack
3862          * is probably already open and will be faster to obtain
3863          * the data through than the working directory.  Loose
3864          * objects however would tend to be slower as they need
3865          * to be individually opened and inflated.
3866          */
3867         if (!FAST_WORKING_DIRECTORY && !want_file && has_object_pack(oid))
3868                 return 0;
3869
3870         /*
3871          * Similarly, if we'd have to convert the file contents anyway, that
3872          * makes the optimization not worthwhile.
3873          */
3874         if (!want_file && would_convert_to_git(istate, name))
3875                 return 0;
3876
3877         len = strlen(name);
3878         pos = index_name_pos(istate, name, len);
3879         if (pos < 0)
3880                 return 0;
3881         ce = istate->cache[pos];
3882
3883         /*
3884          * This is not the sha1 we are looking for, or
3885          * unreusable because it is not a regular file.
3886          */
3887         if (!oideq(oid, &ce->oid) || !S_ISREG(ce->ce_mode))
3888                 return 0;
3889
3890         /*
3891          * If ce is marked as "assume unchanged", there is no
3892          * guarantee that work tree matches what we are looking for.
3893          */
3894         if ((ce->ce_flags & CE_VALID) || ce_skip_worktree(ce))
3895                 return 0;
3896
3897         /*
3898          * If ce matches the file in the work tree, we can reuse it.
3899          */
3900         if (ce_uptodate(ce) ||
3901             (!lstat(name, &st) && !ie_match_stat(istate, ce, &st, 0)))
3902                 return 1;
3903
3904         return 0;
3905 }
3906
3907 static int diff_populate_gitlink(struct diff_filespec *s, int size_only)
3908 {
3909         struct strbuf buf = STRBUF_INIT;
3910         char *dirty = "";
3911
3912         /* Are we looking at the work tree? */
3913         if (s->dirty_submodule)
3914                 dirty = "-dirty";
3915
3916         strbuf_addf(&buf, "Subproject commit %s%s\n",
3917                     oid_to_hex(&s->oid), dirty);
3918         s->size = buf.len;
3919         if (size_only) {
3920                 s->data = NULL;
3921                 strbuf_release(&buf);
3922         } else {
3923                 s->data = strbuf_detach(&buf, NULL);
3924                 s->should_free = 1;
3925         }
3926         return 0;
3927 }
3928
3929 /*
3930  * While doing rename detection and pickaxe operation, we may need to
3931  * grab the data for the blob (or file) for our own in-core comparison.
3932  * diff_filespec has data and size fields for this purpose.
3933  */
3934 int diff_populate_filespec(struct repository *r,
3935                            struct diff_filespec *s,
3936                            unsigned int flags)
3937 {
3938         int size_only = flags & CHECK_SIZE_ONLY;
3939         int err = 0;
3940         int conv_flags = global_conv_flags_eol;
3941         /*
3942          * demote FAIL to WARN to allow inspecting the situation
3943          * instead of refusing.
3944          */
3945         if (conv_flags & CONV_EOL_RNDTRP_DIE)
3946                 conv_flags = CONV_EOL_RNDTRP_WARN;
3947
3948         if (!DIFF_FILE_VALID(s))
3949                 die("internal error: asking to populate invalid file.");
3950         if (S_ISDIR(s->mode))
3951                 return -1;
3952
3953         if (s->data)
3954                 return 0;
3955
3956         if (size_only && 0 < s->size)
3957                 return 0;
3958
3959         if (S_ISGITLINK(s->mode))
3960                 return diff_populate_gitlink(s, size_only);
3961
3962         if (!s->oid_valid ||
3963             reuse_worktree_file(r->index, s->path, &s->oid, 0)) {
3964                 struct strbuf buf = STRBUF_INIT;
3965                 struct stat st;
3966                 int fd;
3967
3968                 if (lstat(s->path, &st) < 0) {
3969                 err_empty:
3970                         err = -1;
3971                 empty:
3972                         s->data = (char *)"";
3973                         s->size = 0;
3974                         return err;
3975                 }
3976                 s->size = xsize_t(st.st_size);
3977                 if (!s->size)
3978                         goto empty;
3979                 if (S_ISLNK(st.st_mode)) {
3980                         struct strbuf sb = STRBUF_INIT;
3981
3982                         if (strbuf_readlink(&sb, s->path, s->size))
3983                                 goto err_empty;
3984                         s->size = sb.len;
3985                         s->data = strbuf_detach(&sb, NULL);
3986                         s->should_free = 1;
3987                         return 0;
3988                 }
3989
3990                 /*
3991                  * Even if the caller would be happy with getting
3992                  * only the size, we cannot return early at this
3993                  * point if the path requires us to run the content
3994                  * conversion.
3995                  */
3996                 if (size_only && !would_convert_to_git(r->index, s->path))
3997                         return 0;
3998
3999                 /*
4000                  * Note: this check uses xsize_t(st.st_size) that may
4001                  * not be the true size of the blob after it goes
4002                  * through convert_to_git().  This may not strictly be
4003                  * correct, but the whole point of big_file_threshold
4004                  * and is_binary check being that we want to avoid
4005                  * opening the file and inspecting the contents, this
4006                  * is probably fine.
4007                  */
4008                 if ((flags & CHECK_BINARY) &&
4009                     s->size > big_file_threshold && s->is_binary == -1) {
4010                         s->is_binary = 1;
4011                         return 0;
4012                 }
4013                 fd = open(s->path, O_RDONLY);
4014                 if (fd < 0)
4015                         goto err_empty;
4016                 s->data = xmmap(NULL, s->size, PROT_READ, MAP_PRIVATE, fd, 0);
4017                 close(fd);
4018                 s->should_munmap = 1;
4019
4020                 /*
4021                  * Convert from working tree format to canonical git format
4022                  */
4023                 if (convert_to_git(r->index, s->path, s->data, s->size, &buf, conv_flags)) {
4024                         size_t size = 0;
4025                         munmap(s->data, s->size);
4026                         s->should_munmap = 0;
4027                         s->data = strbuf_detach(&buf, &size);
4028                         s->size = size;
4029                         s->should_free = 1;
4030                 }
4031         }
4032         else {
4033                 enum object_type type;
4034                 if (size_only || (flags & CHECK_BINARY)) {
4035                         type = oid_object_info(r, &s->oid, &s->size);
4036                         if (type < 0)
4037                                 die("unable to read %s",
4038                                     oid_to_hex(&s->oid));
4039                         if (size_only)
4040                                 return 0;
4041                         if (s->size > big_file_threshold && s->is_binary == -1) {
4042                                 s->is_binary = 1;
4043                                 return 0;
4044                         }
4045                 }
4046                 s->data = read_object_file(&s->oid, &type, &s->size);
4047                 if (!s->data)
4048                         die("unable to read %s", oid_to_hex(&s->oid));
4049                 s->should_free = 1;
4050         }
4051         return 0;
4052 }
4053
4054 void diff_free_filespec_blob(struct diff_filespec *s)
4055 {
4056         if (s->should_free)
4057                 free(s->data);
4058         else if (s->should_munmap)
4059                 munmap(s->data, s->size);
4060
4061         if (s->should_free || s->should_munmap) {
4062                 s->should_free = s->should_munmap = 0;
4063                 s->data = NULL;
4064         }
4065 }
4066
4067 void diff_free_filespec_data(struct diff_filespec *s)
4068 {
4069         diff_free_filespec_blob(s);
4070         FREE_AND_NULL(s->cnt_data);
4071 }
4072
4073 static void prep_temp_blob(struct index_state *istate,
4074                            const char *path, struct diff_tempfile *temp,
4075                            void *blob,
4076                            unsigned long size,
4077                            const struct object_id *oid,
4078                            int mode)
4079 {
4080         struct strbuf buf = STRBUF_INIT;
4081         struct strbuf tempfile = STRBUF_INIT;
4082         char *path_dup = xstrdup(path);
4083         const char *base = basename(path_dup);
4084
4085         /* Generate "XXXXXX_basename.ext" */
4086         strbuf_addstr(&tempfile, "XXXXXX_");
4087         strbuf_addstr(&tempfile, base);
4088
4089         temp->tempfile = mks_tempfile_ts(tempfile.buf, strlen(base) + 1);
4090         if (!temp->tempfile)
4091                 die_errno("unable to create temp-file");
4092         if (convert_to_working_tree(istate, path,
4093                         (const char *)blob, (size_t)size, &buf)) {
4094                 blob = buf.buf;
4095                 size = buf.len;
4096         }
4097         if (write_in_full(temp->tempfile->fd, blob, size) < 0 ||
4098             close_tempfile_gently(temp->tempfile))
4099                 die_errno("unable to write temp-file");
4100         temp->name = get_tempfile_path(temp->tempfile);
4101         oid_to_hex_r(temp->hex, oid);
4102         xsnprintf(temp->mode, sizeof(temp->mode), "%06o", mode);
4103         strbuf_release(&buf);
4104         strbuf_release(&tempfile);
4105         free(path_dup);
4106 }
4107
4108 static struct diff_tempfile *prepare_temp_file(struct repository *r,
4109                                                const char *name,
4110                                                struct diff_filespec *one)
4111 {
4112         struct diff_tempfile *temp = claim_diff_tempfile();
4113
4114         if (!DIFF_FILE_VALID(one)) {
4115         not_a_valid_file:
4116                 /* A '-' entry produces this for file-2, and
4117                  * a '+' entry produces this for file-1.
4118                  */
4119                 temp->name = "/dev/null";
4120                 xsnprintf(temp->hex, sizeof(temp->hex), ".");
4121                 xsnprintf(temp->mode, sizeof(temp->mode), ".");
4122                 return temp;
4123         }
4124
4125         if (!S_ISGITLINK(one->mode) &&
4126             (!one->oid_valid ||
4127              reuse_worktree_file(r->index, name, &one->oid, 1))) {
4128                 struct stat st;
4129                 if (lstat(name, &st) < 0) {
4130                         if (errno == ENOENT)
4131                                 goto not_a_valid_file;
4132                         die_errno("stat(%s)", name);
4133                 }
4134                 if (S_ISLNK(st.st_mode)) {
4135                         struct strbuf sb = STRBUF_INIT;
4136                         if (strbuf_readlink(&sb, name, st.st_size) < 0)
4137                                 die_errno("readlink(%s)", name);
4138                         prep_temp_blob(r->index, name, temp, sb.buf, sb.len,
4139                                        (one->oid_valid ?
4140                                         &one->oid : &null_oid),
4141                                        (one->oid_valid ?
4142                                         one->mode : S_IFLNK));
4143                         strbuf_release(&sb);
4144                 }
4145                 else {
4146                         /* we can borrow from the file in the work tree */
4147                         temp->name = name;
4148                         if (!one->oid_valid)
4149                                 oid_to_hex_r(temp->hex, &null_oid);
4150                         else
4151                                 oid_to_hex_r(temp->hex, &one->oid);
4152                         /* Even though we may sometimes borrow the
4153                          * contents from the work tree, we always want
4154                          * one->mode.  mode is trustworthy even when
4155                          * !(one->oid_valid), as long as
4156                          * DIFF_FILE_VALID(one).
4157                          */
4158                         xsnprintf(temp->mode, sizeof(temp->mode), "%06o", one->mode);
4159                 }
4160                 return temp;
4161         }
4162         else {
4163                 if (diff_populate_filespec(r, one, 0))
4164                         die("cannot read data blob for %s", one->path);
4165                 prep_temp_blob(r->index, name, temp,
4166                                one->data, one->size,
4167                                &one->oid, one->mode);
4168         }
4169         return temp;
4170 }
4171
4172 static void add_external_diff_name(struct repository *r,
4173                                    struct argv_array *argv,
4174                                    const char *name,
4175                                    struct diff_filespec *df)
4176 {
4177         struct diff_tempfile *temp = prepare_temp_file(r, name, df);
4178         argv_array_push(argv, temp->name);
4179         argv_array_push(argv, temp->hex);
4180         argv_array_push(argv, temp->mode);
4181 }
4182
4183 /* An external diff command takes:
4184  *
4185  * diff-cmd name infile1 infile1-sha1 infile1-mode \
4186  *               infile2 infile2-sha1 infile2-mode [ rename-to ]
4187  *
4188  */
4189 static void run_external_diff(const char *pgm,
4190                               const char *name,
4191                               const char *other,
4192                               struct diff_filespec *one,
4193                               struct diff_filespec *two,
4194                               const char *xfrm_msg,
4195                               struct diff_options *o)
4196 {
4197         struct argv_array argv = ARGV_ARRAY_INIT;
4198         struct argv_array env = ARGV_ARRAY_INIT;
4199         struct diff_queue_struct *q = &diff_queued_diff;
4200
4201         argv_array_push(&argv, pgm);
4202         argv_array_push(&argv, name);
4203
4204         if (one && two) {
4205                 add_external_diff_name(o->repo, &argv, name, one);
4206                 if (!other)
4207                         add_external_diff_name(o->repo, &argv, name, two);
4208                 else {
4209                         add_external_diff_name(o->repo, &argv, other, two);
4210                         argv_array_push(&argv, other);
4211                         argv_array_push(&argv, xfrm_msg);
4212                 }
4213         }
4214
4215         argv_array_pushf(&env, "GIT_DIFF_PATH_COUNTER=%d", ++o->diff_path_counter);
4216         argv_array_pushf(&env, "GIT_DIFF_PATH_TOTAL=%d", q->nr);
4217
4218         diff_free_filespec_data(one);
4219         diff_free_filespec_data(two);
4220         if (run_command_v_opt_cd_env(argv.argv, RUN_USING_SHELL, NULL, env.argv))
4221                 die(_("external diff died, stopping at %s"), name);
4222
4223         remove_tempfile();
4224         argv_array_clear(&argv);
4225         argv_array_clear(&env);
4226 }
4227
4228 static int similarity_index(struct diff_filepair *p)
4229 {
4230         return p->score * 100 / MAX_SCORE;
4231 }
4232
4233 static const char *diff_abbrev_oid(const struct object_id *oid, int abbrev)
4234 {
4235         if (startup_info->have_repository)
4236                 return find_unique_abbrev(oid, abbrev);
4237         else {
4238                 char *hex = oid_to_hex(oid);
4239                 if (abbrev < 0)
4240                         abbrev = FALLBACK_DEFAULT_ABBREV;
4241                 if (abbrev > the_hash_algo->hexsz)
4242                         BUG("oid abbreviation out of range: %d", abbrev);
4243                 if (abbrev)
4244                         hex[abbrev] = '\0';
4245                 return hex;
4246         }
4247 }
4248
4249 static void fill_metainfo(struct strbuf *msg,
4250                           const char *name,
4251                           const char *other,
4252                           struct diff_filespec *one,
4253                           struct diff_filespec *two,
4254                           struct diff_options *o,
4255                           struct diff_filepair *p,
4256                           int *must_show_header,
4257                           int use_color)
4258 {
4259         const char *set = diff_get_color(use_color, DIFF_METAINFO);
4260         const char *reset = diff_get_color(use_color, DIFF_RESET);
4261         const char *line_prefix = diff_line_prefix(o);
4262
4263         *must_show_header = 1;
4264         strbuf_init(msg, PATH_MAX * 2 + 300);
4265         switch (p->status) {
4266         case DIFF_STATUS_COPIED:
4267                 strbuf_addf(msg, "%s%ssimilarity index %d%%",
4268                             line_prefix, set, similarity_index(p));
4269                 strbuf_addf(msg, "%s\n%s%scopy from ",
4270                             reset,  line_prefix, set);
4271                 quote_c_style(name, msg, NULL, 0);
4272                 strbuf_addf(msg, "%s\n%s%scopy to ", reset, line_prefix, set);
4273                 quote_c_style(other, msg, NULL, 0);
4274                 strbuf_addf(msg, "%s\n", reset);
4275                 break;
4276         case DIFF_STATUS_RENAMED:
4277                 strbuf_addf(msg, "%s%ssimilarity index %d%%",
4278                             line_prefix, set, similarity_index(p));
4279                 strbuf_addf(msg, "%s\n%s%srename from ",
4280                             reset, line_prefix, set);
4281                 quote_c_style(name, msg, NULL, 0);
4282                 strbuf_addf(msg, "%s\n%s%srename to ",
4283                             reset, line_prefix, set);
4284                 quote_c_style(other, msg, NULL, 0);
4285                 strbuf_addf(msg, "%s\n", reset);
4286                 break;
4287         case DIFF_STATUS_MODIFIED:
4288                 if (p->score) {
4289                         strbuf_addf(msg, "%s%sdissimilarity index %d%%%s\n",
4290                                     line_prefix,
4291                                     set, similarity_index(p), reset);
4292                         break;
4293                 }
4294                 /* fallthru */
4295         default:
4296                 *must_show_header = 0;
4297         }
4298         if (one && two && !oideq(&one->oid, &two->oid)) {
4299                 const unsigned hexsz = the_hash_algo->hexsz;
4300                 int abbrev = o->flags.full_index ? hexsz : DEFAULT_ABBREV;
4301
4302                 if (o->flags.binary) {
4303                         mmfile_t mf;
4304                         if ((!fill_mmfile(o->repo, &mf, one) &&
4305                              diff_filespec_is_binary(o->repo, one)) ||
4306                             (!fill_mmfile(o->repo, &mf, two) &&
4307                              diff_filespec_is_binary(o->repo, two)))
4308                                 abbrev = hexsz;
4309                 }
4310                 strbuf_addf(msg, "%s%sindex %s..%s", line_prefix, set,
4311                             diff_abbrev_oid(&one->oid, abbrev),
4312                             diff_abbrev_oid(&two->oid, abbrev));
4313                 if (one->mode == two->mode)
4314                         strbuf_addf(msg, " %06o", one->mode);
4315                 strbuf_addf(msg, "%s\n", reset);
4316         }
4317 }
4318
4319 static void run_diff_cmd(const char *pgm,
4320                          const char *name,
4321                          const char *other,
4322                          const char *attr_path,
4323                          struct diff_filespec *one,
4324                          struct diff_filespec *two,
4325                          struct strbuf *msg,
4326                          struct diff_options *o,
4327                          struct diff_filepair *p)
4328 {
4329         const char *xfrm_msg = NULL;
4330         int complete_rewrite = (p->status == DIFF_STATUS_MODIFIED) && p->score;
4331         int must_show_header = 0;
4332
4333
4334         if (o->flags.allow_external) {
4335                 struct userdiff_driver *drv;
4336
4337                 drv = userdiff_find_by_path(o->repo->index, attr_path);
4338                 if (drv && drv->external)
4339                         pgm = drv->external;
4340         }
4341
4342         if (msg) {
4343                 /*
4344                  * don't use colors when the header is intended for an
4345                  * external diff driver
4346                  */
4347                 fill_metainfo(msg, name, other, one, two, o, p,
4348                               &must_show_header,
4349                               want_color(o->use_color) && !pgm);
4350                 xfrm_msg = msg->len ? msg->buf : NULL;
4351         }
4352
4353         if (pgm) {
4354                 run_external_diff(pgm, name, other, one, two, xfrm_msg, o);
4355                 return;
4356         }
4357         if (one && two)
4358                 builtin_diff(name, other ? other : name,
4359                              one, two, xfrm_msg, must_show_header,
4360                              o, complete_rewrite);
4361         else
4362                 fprintf(o->file, "* Unmerged path %s\n", name);
4363 }
4364
4365 static void diff_fill_oid_info(struct diff_filespec *one, struct index_state *istate)
4366 {
4367         if (DIFF_FILE_VALID(one)) {
4368                 if (!one->oid_valid) {
4369                         struct stat st;
4370                         if (one->is_stdin) {
4371                                 oidclr(&one->oid);
4372                                 return;
4373                         }
4374                         if (lstat(one->path, &st) < 0)
4375                                 die_errno("stat '%s'", one->path);
4376                         if (index_path(istate, &one->oid, one->path, &st, 0))
4377                                 die("cannot hash %s", one->path);
4378                 }
4379         }
4380         else
4381                 oidclr(&one->oid);
4382 }
4383
4384 static void strip_prefix(int prefix_length, const char **namep, const char **otherp)
4385 {
4386         /* Strip the prefix but do not molest /dev/null and absolute paths */
4387         if (*namep && !is_absolute_path(*namep)) {
4388                 *namep += prefix_length;
4389                 if (**namep == '/')
4390                         ++*namep;
4391         }
4392         if (*otherp && !is_absolute_path(*otherp)) {
4393                 *otherp += prefix_length;
4394                 if (**otherp == '/')
4395                         ++*otherp;
4396         }
4397 }
4398
4399 static void run_diff(struct diff_filepair *p, struct diff_options *o)
4400 {
4401         const char *pgm = external_diff();
4402         struct strbuf msg;
4403         struct diff_filespec *one = p->one;
4404         struct diff_filespec *two = p->two;
4405         const char *name;
4406         const char *other;
4407         const char *attr_path;
4408
4409         name  = one->path;
4410         other = (strcmp(name, two->path) ? two->path : NULL);
4411         attr_path = name;
4412         if (o->prefix_length)
4413                 strip_prefix(o->prefix_length, &name, &other);
4414
4415         if (!o->flags.allow_external)
4416                 pgm = NULL;
4417
4418         if (DIFF_PAIR_UNMERGED(p)) {
4419                 run_diff_cmd(pgm, name, NULL, attr_path,
4420                              NULL, NULL, NULL, o, p);
4421                 return;
4422         }
4423
4424         diff_fill_oid_info(one, o->repo->index);
4425         diff_fill_oid_info(two, o->repo->index);
4426
4427         if (!pgm &&
4428             DIFF_FILE_VALID(one) && DIFF_FILE_VALID(two) &&
4429             (S_IFMT & one->mode) != (S_IFMT & two->mode)) {
4430                 /*
4431                  * a filepair that changes between file and symlink
4432                  * needs to be split into deletion and creation.
4433                  */
4434                 struct diff_filespec *null = alloc_filespec(two->path);
4435                 run_diff_cmd(NULL, name, other, attr_path,
4436                              one, null, &msg,
4437                              o, p);
4438                 free(null);
4439                 strbuf_release(&msg);
4440
4441                 null = alloc_filespec(one->path);
4442                 run_diff_cmd(NULL, name, other, attr_path,
4443                              null, two, &msg, o, p);
4444                 free(null);
4445         }
4446         else
4447                 run_diff_cmd(pgm, name, other, attr_path,
4448                              one, two, &msg, o, p);
4449
4450         strbuf_release(&msg);
4451 }
4452
4453 static void run_diffstat(struct diff_filepair *p, struct diff_options *o,
4454                          struct diffstat_t *diffstat)
4455 {
4456         const char *name;
4457         const char *other;
4458
4459         if (DIFF_PAIR_UNMERGED(p)) {
4460                 /* unmerged */
4461                 builtin_diffstat(p->one->path, NULL, NULL, NULL,
4462                                  diffstat, o, p);
4463                 return;
4464         }
4465
4466         name = p->one->path;
4467         other = (strcmp(name, p->two->path) ? p->two->path : NULL);
4468
4469         if (o->prefix_length)
4470                 strip_prefix(o->prefix_length, &name, &other);
4471
4472         diff_fill_oid_info(p->one, o->repo->index);
4473         diff_fill_oid_info(p->two, o->repo->index);
4474
4475         builtin_diffstat(name, other, p->one, p->two,
4476                          diffstat, o, p);
4477 }
4478
4479 static void run_checkdiff(struct diff_filepair *p, struct diff_options *o)
4480 {
4481         const char *name;
4482         const char *other;
4483         const char *attr_path;
4484
4485         if (DIFF_PAIR_UNMERGED(p)) {
4486                 /* unmerged */
4487                 return;
4488         }
4489
4490         name = p->one->path;
4491         other = (strcmp(name, p->two->path) ? p->two->path : NULL);
4492         attr_path = other ? other : name;
4493
4494         if (o->prefix_length)
4495                 strip_prefix(o->prefix_length, &name, &other);
4496
4497         diff_fill_oid_info(p->one, o->repo->index);
4498         diff_fill_oid_info(p->two, o->repo->index);
4499
4500         builtin_checkdiff(name, other, attr_path, p->one, p->two, o);
4501 }
4502
4503 static void prep_parse_options(struct diff_options *options);
4504
4505 void repo_diff_setup(struct repository *r, struct diff_options *options)
4506 {
4507         memcpy(options, &default_diff_options, sizeof(*options));
4508
4509         options->file = stdout;
4510         options->repo = r;
4511
4512         options->output_indicators[OUTPUT_INDICATOR_NEW] = '+';
4513         options->output_indicators[OUTPUT_INDICATOR_OLD] = '-';
4514         options->output_indicators[OUTPUT_INDICATOR_CONTEXT] = ' ';
4515         options->abbrev = DEFAULT_ABBREV;
4516         options->line_termination = '\n';
4517         options->break_opt = -1;
4518         options->rename_limit = -1;
4519         options->dirstat_permille = diff_dirstat_permille_default;
4520         options->context = diff_context_default;
4521         options->interhunkcontext = diff_interhunk_context_default;
4522         options->ws_error_highlight = ws_error_highlight_default;
4523         options->flags.rename_empty = 1;
4524         options->objfind = NULL;
4525
4526         /* pathchange left =NULL by default */
4527         options->change = diff_change;
4528         options->add_remove = diff_addremove;
4529         options->use_color = diff_use_color_default;
4530         options->detect_rename = diff_detect_rename_default;
4531         options->xdl_opts |= diff_algorithm;
4532         if (diff_indent_heuristic)
4533                 DIFF_XDL_SET(options, INDENT_HEURISTIC);
4534
4535         options->orderfile = diff_order_file_cfg;
4536
4537         if (diff_no_prefix) {
4538                 options->a_prefix = options->b_prefix = "";
4539         } else if (!diff_mnemonic_prefix) {
4540                 options->a_prefix = "a/";
4541                 options->b_prefix = "b/";
4542         }
4543
4544         options->color_moved = diff_color_moved_default;
4545         options->color_moved_ws_handling = diff_color_moved_ws_default;
4546
4547         prep_parse_options(options);
4548 }
4549
4550 void diff_setup_done(struct diff_options *options)
4551 {
4552         unsigned check_mask = DIFF_FORMAT_NAME |
4553                               DIFF_FORMAT_NAME_STATUS |
4554                               DIFF_FORMAT_CHECKDIFF |
4555                               DIFF_FORMAT_NO_OUTPUT;
4556         /*
4557          * This must be signed because we're comparing against a potentially
4558          * negative value.
4559          */
4560         const int hexsz = the_hash_algo->hexsz;
4561
4562         if (options->set_default)
4563                 options->set_default(options);
4564
4565         if (HAS_MULTI_BITS(options->output_format & check_mask))
4566                 die(_("--name-only, --name-status, --check and -s are mutually exclusive"));
4567
4568         if (HAS_MULTI_BITS(options->pickaxe_opts & DIFF_PICKAXE_KINDS_MASK))
4569                 die(_("-G, -S and --find-object are mutually exclusive"));
4570
4571         /*
4572          * Most of the time we can say "there are changes"
4573          * only by checking if there are changed paths, but
4574          * --ignore-whitespace* options force us to look
4575          * inside contents.
4576          */
4577
4578         if ((options->xdl_opts & XDF_WHITESPACE_FLAGS))
4579                 options->flags.diff_from_contents = 1;
4580         else
4581                 options->flags.diff_from_contents = 0;
4582
4583         if (options->flags.find_copies_harder)
4584                 options->detect_rename = DIFF_DETECT_COPY;
4585
4586         if (!options->flags.relative_name)
4587                 options->prefix = NULL;
4588         if (options->prefix)
4589                 options->prefix_length = strlen(options->prefix);
4590         else
4591                 options->prefix_length = 0;
4592
4593         if (options->output_format & (DIFF_FORMAT_NAME |
4594                                       DIFF_FORMAT_NAME_STATUS |
4595                                       DIFF_FORMAT_CHECKDIFF |
4596                                       DIFF_FORMAT_NO_OUTPUT))
4597                 options->output_format &= ~(DIFF_FORMAT_RAW |
4598                                             DIFF_FORMAT_NUMSTAT |
4599                                             DIFF_FORMAT_DIFFSTAT |
4600                                             DIFF_FORMAT_SHORTSTAT |
4601                                             DIFF_FORMAT_DIRSTAT |
4602                                             DIFF_FORMAT_SUMMARY |
4603                                             DIFF_FORMAT_PATCH);
4604
4605         /*
4606          * These cases always need recursive; we do not drop caller-supplied
4607          * recursive bits for other formats here.
4608          */
4609         if (options->output_format & (DIFF_FORMAT_PATCH |
4610                                       DIFF_FORMAT_NUMSTAT |
4611                                       DIFF_FORMAT_DIFFSTAT |
4612                                       DIFF_FORMAT_SHORTSTAT |
4613                                       DIFF_FORMAT_DIRSTAT |
4614                                       DIFF_FORMAT_SUMMARY |
4615                                       DIFF_FORMAT_CHECKDIFF))
4616                 options->flags.recursive = 1;
4617         /*
4618          * Also pickaxe would not work very well if you do not say recursive
4619          */
4620         if (options->pickaxe_opts & DIFF_PICKAXE_KINDS_MASK)
4621                 options->flags.recursive = 1;
4622         /*
4623          * When patches are generated, submodules diffed against the work tree
4624          * must be checked for dirtiness too so it can be shown in the output
4625          */
4626         if (options->output_format & DIFF_FORMAT_PATCH)
4627                 options->flags.dirty_submodules = 1;
4628
4629         if (options->detect_rename && options->rename_limit < 0)
4630                 options->rename_limit = diff_rename_limit_default;
4631         if (hexsz < options->abbrev)
4632                 options->abbrev = hexsz; /* full */
4633
4634         /*
4635          * It does not make sense to show the first hit we happened
4636          * to have found.  It does not make sense not to return with
4637          * exit code in such a case either.
4638          */
4639         if (options->flags.quick) {
4640                 options->output_format = DIFF_FORMAT_NO_OUTPUT;
4641                 options->flags.exit_with_status = 1;
4642         }
4643
4644         options->diff_path_counter = 0;
4645
4646         if (options->flags.follow_renames && options->pathspec.nr != 1)
4647                 die(_("--follow requires exactly one pathspec"));
4648
4649         if (!options->use_color || external_diff())
4650                 options->color_moved = 0;
4651
4652         FREE_AND_NULL(options->parseopts);
4653 }
4654
4655 int parse_long_opt(const char *opt, const char **argv,
4656                    const char **optarg)
4657 {
4658         const char *arg = argv[0];
4659         if (!skip_prefix(arg, "--", &arg))
4660                 return 0;
4661         if (!skip_prefix(arg, opt, &arg))
4662                 return 0;
4663         if (*arg == '=') { /* stuck form: --option=value */
4664                 *optarg = arg + 1;
4665                 return 1;
4666         }
4667         if (*arg != '\0')
4668                 return 0;
4669         /* separate form: --option value */
4670         if (!argv[1])
4671                 die("Option '--%s' requires a value", opt);
4672         *optarg = argv[1];
4673         return 2;
4674 }
4675
4676 static int diff_opt_stat(const struct option *opt, const char *value, int unset)
4677 {
4678         struct diff_options *options = opt->value;
4679         int width = options->stat_width;
4680         int name_width = options->stat_name_width;
4681         int graph_width = options->stat_graph_width;
4682         int count = options->stat_count;
4683         char *end;
4684
4685         BUG_ON_OPT_NEG(unset);
4686
4687         if (!strcmp(opt->long_name, "stat")) {
4688                 if (value) {
4689                         width = strtoul(value, &end, 10);
4690                         if (*end == ',')
4691                                 name_width = strtoul(end+1, &end, 10);
4692                         if (*end == ',')
4693                                 count = strtoul(end+1, &end, 10);
4694                         if (*end)
4695                                 return error(_("invalid --stat value: %s"), value);
4696                 }
4697         } else if (!strcmp(opt->long_name, "stat-width")) {
4698                 width = strtoul(value, &end, 10);
4699                 if (*end)
4700                         return error(_("%s expects a numerical value"),
4701                                      opt->long_name);
4702         } else if (!strcmp(opt->long_name, "stat-name-width")) {
4703                 name_width = strtoul(value, &end, 10);
4704                 if (*end)
4705                         return error(_("%s expects a numerical value"),
4706                                      opt->long_name);
4707         } else if (!strcmp(opt->long_name, "stat-graph-width")) {
4708                 graph_width = strtoul(value, &end, 10);
4709                 if (*end)
4710                         return error(_("%s expects a numerical value"),
4711                                      opt->long_name);
4712         } else if (!strcmp(opt->long_name, "stat-count")) {
4713                 count = strtoul(value, &end, 10);
4714                 if (*end)
4715                         return error(_("%s expects a numerical value"),
4716                                      opt->long_name);
4717         } else
4718                 BUG("%s should not get here", opt->long_name);
4719
4720         options->output_format |= DIFF_FORMAT_DIFFSTAT;
4721         options->stat_name_width = name_width;
4722         options->stat_graph_width = graph_width;
4723         options->stat_width = width;
4724         options->stat_count = count;
4725         return 0;
4726 }
4727
4728 static int parse_dirstat_opt(struct diff_options *options, const char *params)
4729 {
4730         struct strbuf errmsg = STRBUF_INIT;
4731         if (parse_dirstat_params(options, params, &errmsg))
4732                 die(_("Failed to parse --dirstat/-X option parameter:\n%s"),
4733                     errmsg.buf);
4734         strbuf_release(&errmsg);
4735         /*
4736          * The caller knows a dirstat-related option is given from the command
4737          * line; allow it to say "return this_function();"
4738          */
4739         options->output_format |= DIFF_FORMAT_DIRSTAT;
4740         return 1;
4741 }
4742
4743 static const char diff_status_letters[] = {
4744         DIFF_STATUS_ADDED,
4745         DIFF_STATUS_COPIED,
4746         DIFF_STATUS_DELETED,
4747         DIFF_STATUS_MODIFIED,
4748         DIFF_STATUS_RENAMED,
4749         DIFF_STATUS_TYPE_CHANGED,
4750         DIFF_STATUS_UNKNOWN,
4751         DIFF_STATUS_UNMERGED,
4752         DIFF_STATUS_FILTER_AON,
4753         DIFF_STATUS_FILTER_BROKEN,
4754         '\0',
4755 };
4756
4757 static unsigned int filter_bit['Z' + 1];
4758
4759 static void prepare_filter_bits(void)
4760 {
4761         int i;
4762
4763         if (!filter_bit[DIFF_STATUS_ADDED]) {
4764                 for (i = 0; diff_status_letters[i]; i++)
4765                         filter_bit[(int) diff_status_letters[i]] = (1 << i);
4766         }
4767 }
4768
4769 static unsigned filter_bit_tst(char status, const struct diff_options *opt)
4770 {
4771         return opt->filter & filter_bit[(int) status];
4772 }
4773
4774 unsigned diff_filter_bit(char status)
4775 {
4776         prepare_filter_bits();
4777         return filter_bit[(int) status];
4778 }
4779
4780 static int diff_opt_diff_filter(const struct option *option,
4781                                 const char *optarg, int unset)
4782 {
4783         struct diff_options *opt = option->value;
4784         int i, optch;
4785
4786         BUG_ON_OPT_NEG(unset);
4787         prepare_filter_bits();
4788
4789         /*
4790          * If there is a negation e.g. 'd' in the input, and we haven't
4791          * initialized the filter field with another --diff-filter, start
4792          * from full set of bits, except for AON.
4793          */
4794         if (!opt->filter) {
4795                 for (i = 0; (optch = optarg[i]) != '\0'; i++) {
4796                         if (optch < 'a' || 'z' < optch)
4797                                 continue;
4798                         opt->filter = (1 << (ARRAY_SIZE(diff_status_letters) - 1)) - 1;
4799                         opt->filter &= ~filter_bit[DIFF_STATUS_FILTER_AON];
4800                         break;
4801                 }
4802         }
4803
4804         for (i = 0; (optch = optarg[i]) != '\0'; i++) {
4805                 unsigned int bit;
4806                 int negate;
4807
4808                 if ('a' <= optch && optch <= 'z') {
4809                         negate = 1;
4810                         optch = toupper(optch);
4811                 } else {
4812                         negate = 0;
4813                 }
4814
4815                 bit = (0 <= optch && optch <= 'Z') ? filter_bit[optch] : 0;
4816                 if (!bit)
4817                         return error(_("unknown change class '%c' in --diff-filter=%s"),
4818                                      optarg[i], optarg);
4819                 if (negate)
4820                         opt->filter &= ~bit;
4821                 else
4822                         opt->filter |= bit;
4823         }
4824         return 0;
4825 }
4826
4827 static void enable_patch_output(int *fmt)
4828 {
4829         *fmt &= ~DIFF_FORMAT_NO_OUTPUT;
4830         *fmt |= DIFF_FORMAT_PATCH;
4831 }
4832
4833 static int diff_opt_ws_error_highlight(const struct option *option,
4834                                        const char *arg, int unset)
4835 {
4836         struct diff_options *opt = option->value;
4837         int val = parse_ws_error_highlight(arg);
4838
4839         BUG_ON_OPT_NEG(unset);
4840         if (val < 0)
4841                 return error(_("unknown value after ws-error-highlight=%.*s"),
4842                              -1 - val, arg);
4843         opt->ws_error_highlight = val;
4844         return 0;
4845 }
4846
4847 static int diff_opt_find_object(const struct option *option,
4848                                 const char *arg, int unset)
4849 {
4850         struct diff_options *opt = option->value;
4851         struct object_id oid;
4852
4853         BUG_ON_OPT_NEG(unset);
4854         if (get_oid(arg, &oid))
4855                 return error(_("unable to resolve '%s'"), arg);
4856
4857         if (!opt->objfind)
4858                 opt->objfind = xcalloc(1, sizeof(*opt->objfind));
4859
4860         opt->pickaxe_opts |= DIFF_PICKAXE_KIND_OBJFIND;
4861         opt->flags.recursive = 1;
4862         opt->flags.tree_in_recursive = 1;
4863         oidset_insert(opt->objfind, &oid);
4864         return 0;
4865 }
4866
4867 static int diff_opt_anchored(const struct option *opt,
4868                              const char *arg, int unset)
4869 {
4870         struct diff_options *options = opt->value;
4871
4872         BUG_ON_OPT_NEG(unset);
4873         options->xdl_opts = DIFF_WITH_ALG(options, PATIENCE_DIFF);
4874         ALLOC_GROW(options->anchors, options->anchors_nr + 1,
4875                    options->anchors_alloc);
4876         options->anchors[options->anchors_nr++] = xstrdup(arg);
4877         return 0;
4878 }
4879
4880 static int diff_opt_binary(const struct option *opt,
4881                            const char *arg, int unset)
4882 {
4883         struct diff_options *options = opt->value;
4884
4885         BUG_ON_OPT_NEG(unset);
4886         BUG_ON_OPT_ARG(arg);
4887         enable_patch_output(&options->output_format);
4888         options->flags.binary = 1;
4889         return 0;
4890 }
4891
4892 static int diff_opt_break_rewrites(const struct option *opt,
4893                                    const char *arg, int unset)
4894 {
4895         int *break_opt = opt->value;
4896         int opt1, opt2;
4897
4898         BUG_ON_OPT_NEG(unset);
4899         if (!arg)
4900                 arg = "";
4901         opt1 = parse_rename_score(&arg);
4902         if (*arg == 0)
4903                 opt2 = 0;
4904         else if (*arg != '/')
4905                 return error(_("%s expects <n>/<m> form"), opt->long_name);
4906         else {
4907                 arg++;
4908                 opt2 = parse_rename_score(&arg);
4909         }
4910         if (*arg != 0)
4911                 return error(_("%s expects <n>/<m> form"), opt->long_name);
4912         *break_opt = opt1 | (opt2 << 16);
4913         return 0;
4914 }
4915
4916 static int diff_opt_char(const struct option *opt,
4917                          const char *arg, int unset)
4918 {
4919         char *value = opt->value;
4920
4921         BUG_ON_OPT_NEG(unset);
4922         if (arg[1])
4923                 return error(_("%s expects a character, got '%s'"),
4924                              opt->long_name, arg);
4925         *value = arg[0];
4926         return 0;
4927 }
4928
4929 static int diff_opt_color_moved(const struct option *opt,
4930                                 const char *arg, int unset)
4931 {
4932         struct diff_options *options = opt->value;
4933
4934         if (unset) {
4935                 options->color_moved = COLOR_MOVED_NO;
4936         } else if (!arg) {
4937                 if (diff_color_moved_default)
4938                         options->color_moved = diff_color_moved_default;
4939                 if (options->color_moved == COLOR_MOVED_NO)
4940                         options->color_moved = COLOR_MOVED_DEFAULT;
4941         } else {
4942                 int cm = parse_color_moved(arg);
4943                 if (cm < 0)
4944                         return error(_("bad --color-moved argument: %s"), arg);
4945                 options->color_moved = cm;
4946         }
4947         return 0;
4948 }
4949
4950 static int diff_opt_color_moved_ws(const struct option *opt,
4951                                    const char *arg, int unset)
4952 {
4953         struct diff_options *options = opt->value;
4954         unsigned cm;
4955
4956         if (unset) {
4957                 options->color_moved_ws_handling = 0;
4958                 return 0;
4959         }
4960
4961         cm = parse_color_moved_ws(arg);
4962         if (cm & COLOR_MOVED_WS_ERROR)
4963                 return error(_("invalid mode '%s' in --color-moved-ws"), arg);
4964         options->color_moved_ws_handling = cm;
4965         return 0;
4966 }
4967
4968 static int diff_opt_color_words(const struct option *opt,
4969                                 const char *arg, int unset)
4970 {
4971         struct diff_options *options = opt->value;
4972
4973         BUG_ON_OPT_NEG(unset);
4974         options->use_color = 1;
4975         options->word_diff = DIFF_WORDS_COLOR;
4976         options->word_regex = arg;
4977         return 0;
4978 }
4979
4980 static int diff_opt_compact_summary(const struct option *opt,
4981                                     const char *arg, int unset)
4982 {
4983         struct diff_options *options = opt->value;
4984
4985         BUG_ON_OPT_ARG(arg);
4986         if (unset) {
4987                 options->flags.stat_with_summary = 0;
4988         } else {
4989                 options->flags.stat_with_summary = 1;
4990                 options->output_format |= DIFF_FORMAT_DIFFSTAT;
4991         }
4992         return 0;
4993 }
4994
4995 static int diff_opt_diff_algorithm(const struct option *opt,
4996                                    const char *arg, int unset)
4997 {
4998         struct diff_options *options = opt->value;
4999         long value = parse_algorithm_value(arg);
5000
5001         BUG_ON_OPT_NEG(unset);
5002         if (value < 0)
5003                 return error(_("option diff-algorithm accepts \"myers\", "
5004                                "\"minimal\", \"patience\" and \"histogram\""));
5005
5006         /* clear out previous settings */
5007         DIFF_XDL_CLR(options, NEED_MINIMAL);
5008         options->xdl_opts &= ~XDF_DIFF_ALGORITHM_MASK;
5009         options->xdl_opts |= value;
5010         return 0;
5011 }
5012
5013 static int diff_opt_dirstat(const struct option *opt,
5014                             const char *arg, int unset)
5015 {
5016         struct diff_options *options = opt->value;
5017
5018         BUG_ON_OPT_NEG(unset);
5019         if (!strcmp(opt->long_name, "cumulative")) {
5020                 if (arg)
5021                         BUG("how come --cumulative take a value?");
5022                 arg = "cumulative";
5023         } else if (!strcmp(opt->long_name, "dirstat-by-file"))
5024                 parse_dirstat_opt(options, "files");
5025         parse_dirstat_opt(options, arg ? arg : "");
5026         return 0;
5027 }
5028
5029 static int diff_opt_find_copies(const struct option *opt,
5030                                 const char *arg, int unset)
5031 {
5032         struct diff_options *options = opt->value;
5033
5034         BUG_ON_OPT_NEG(unset);
5035         if (!arg)
5036                 arg = "";
5037         options->rename_score = parse_rename_score(&arg);
5038         if (*arg != 0)
5039                 return error(_("invalid argument to %s"), opt->long_name);
5040
5041         if (options->detect_rename == DIFF_DETECT_COPY)
5042                 options->flags.find_copies_harder = 1;
5043         else
5044                 options->detect_rename = DIFF_DETECT_COPY;
5045
5046         return 0;
5047 }
5048
5049 static int diff_opt_find_renames(const struct option *opt,
5050                                  const char *arg, int unset)
5051 {
5052         struct diff_options *options = opt->value;
5053
5054         BUG_ON_OPT_NEG(unset);
5055         if (!arg)
5056                 arg = "";
5057         options->rename_score = parse_rename_score(&arg);
5058         if (*arg != 0)
5059                 return error(_("invalid argument to %s"), opt->long_name);
5060
5061         options->detect_rename = DIFF_DETECT_RENAME;
5062         return 0;
5063 }
5064
5065 static int diff_opt_follow(const struct option *opt,
5066                            const char *arg, int unset)
5067 {
5068         struct diff_options *options = opt->value;
5069
5070         BUG_ON_OPT_ARG(arg);
5071         if (unset) {
5072                 options->flags.follow_renames = 0;
5073                 options->flags.default_follow_renames = 0;
5074         } else {
5075                 options->flags.follow_renames = 1;
5076         }
5077         return 0;
5078 }
5079
5080 static int diff_opt_ignore_submodules(const struct option *opt,
5081                                       const char *arg, int unset)
5082 {
5083         struct diff_options *options = opt->value;
5084
5085         BUG_ON_OPT_NEG(unset);
5086         if (!arg)
5087                 arg = "all";
5088         options->flags.override_submodule_config = 1;
5089         handle_ignore_submodules_arg(options, arg);
5090         return 0;
5091 }
5092
5093 static int diff_opt_line_prefix(const struct option *opt,
5094                                 const char *optarg, int unset)
5095 {
5096         struct diff_options *options = opt->value;
5097
5098         BUG_ON_OPT_NEG(unset);
5099         options->line_prefix = optarg;
5100         options->line_prefix_length = strlen(options->line_prefix);
5101         graph_setup_line_prefix(options);
5102         return 0;
5103 }
5104
5105 static int diff_opt_no_prefix(const struct option *opt,
5106                               const char *optarg, int unset)
5107 {
5108         struct diff_options *options = opt->value;
5109
5110         BUG_ON_OPT_NEG(unset);
5111         BUG_ON_OPT_ARG(optarg);
5112         options->a_prefix = "";
5113         options->b_prefix = "";
5114         return 0;
5115 }
5116
5117 static enum parse_opt_result diff_opt_output(struct parse_opt_ctx_t *ctx,
5118                                              const struct option *opt,
5119                                              const char *arg, int unset)
5120 {
5121         struct diff_options *options = opt->value;
5122         char *path;
5123
5124         BUG_ON_OPT_NEG(unset);
5125         path = prefix_filename(ctx->prefix, arg);
5126         options->file = xfopen(path, "w");
5127         options->close_file = 1;
5128         if (options->use_color != GIT_COLOR_ALWAYS)
5129                 options->use_color = GIT_COLOR_NEVER;
5130         free(path);
5131         return 0;
5132 }
5133
5134 static int diff_opt_patience(const struct option *opt,
5135                              const char *arg, int unset)
5136 {
5137         struct diff_options *options = opt->value;
5138         int i;
5139
5140         BUG_ON_OPT_NEG(unset);
5141         BUG_ON_OPT_ARG(arg);
5142         options->xdl_opts = DIFF_WITH_ALG(options, PATIENCE_DIFF);
5143         /*
5144          * Both --patience and --anchored use PATIENCE_DIFF
5145          * internally, so remove any anchors previously
5146          * specified.
5147          */
5148         for (i = 0; i < options->anchors_nr; i++)
5149                 free(options->anchors[i]);
5150         options->anchors_nr = 0;
5151         return 0;
5152 }
5153
5154 static int diff_opt_pickaxe_regex(const struct option *opt,
5155                                   const char *arg, int unset)
5156 {
5157         struct diff_options *options = opt->value;
5158
5159         BUG_ON_OPT_NEG(unset);
5160         options->pickaxe = arg;
5161         options->pickaxe_opts |= DIFF_PICKAXE_KIND_G;
5162         return 0;
5163 }
5164
5165 static int diff_opt_pickaxe_string(const struct option *opt,
5166                                    const char *arg, int unset)
5167 {
5168         struct diff_options *options = opt->value;
5169
5170         BUG_ON_OPT_NEG(unset);
5171         options->pickaxe = arg;
5172         options->pickaxe_opts |= DIFF_PICKAXE_KIND_S;
5173         return 0;
5174 }
5175
5176 static int diff_opt_relative(const struct option *opt,
5177                              const char *arg, int unset)
5178 {
5179         struct diff_options *options = opt->value;
5180
5181         BUG_ON_OPT_NEG(unset);
5182         options->flags.relative_name = 1;
5183         if (arg)
5184                 options->prefix = arg;
5185         return 0;
5186 }
5187
5188 static int diff_opt_submodule(const struct option *opt,
5189                               const char *arg, int unset)
5190 {
5191         struct diff_options *options = opt->value;
5192
5193         BUG_ON_OPT_NEG(unset);
5194         if (!arg)
5195                 arg = "log";
5196         if (parse_submodule_params(options, arg))
5197                 return error(_("failed to parse --submodule option parameter: '%s'"),
5198                              arg);
5199         return 0;
5200 }
5201
5202 static int diff_opt_textconv(const struct option *opt,
5203                              const char *arg, int unset)
5204 {
5205         struct diff_options *options = opt->value;
5206
5207         BUG_ON_OPT_ARG(arg);
5208         if (unset) {
5209                 options->flags.allow_textconv = 0;
5210         } else {
5211                 options->flags.allow_textconv = 1;
5212                 options->flags.textconv_set_via_cmdline = 1;
5213         }
5214         return 0;
5215 }
5216
5217 static int diff_opt_unified(const struct option *opt,
5218                             const char *arg, int unset)
5219 {
5220         struct diff_options *options = opt->value;
5221         char *s;
5222
5223         BUG_ON_OPT_NEG(unset);
5224
5225         if (arg) {
5226                 options->context = strtol(arg, &s, 10);
5227                 if (*s)
5228                         return error(_("%s expects a numerical value"), "--unified");
5229         }
5230         enable_patch_output(&options->output_format);
5231
5232         return 0;
5233 }
5234
5235 static int diff_opt_word_diff(const struct option *opt,
5236                               const char *arg, int unset)
5237 {
5238         struct diff_options *options = opt->value;
5239
5240         BUG_ON_OPT_NEG(unset);
5241         if (arg) {
5242                 if (!strcmp(arg, "plain"))
5243                         options->word_diff = DIFF_WORDS_PLAIN;
5244                 else if (!strcmp(arg, "color")) {
5245                         options->use_color = 1;
5246                         options->word_diff = DIFF_WORDS_COLOR;
5247                 }
5248                 else if (!strcmp(arg, "porcelain"))
5249                         options->word_diff = DIFF_WORDS_PORCELAIN;
5250                 else if (!strcmp(arg, "none"))
5251                         options->word_diff = DIFF_WORDS_NONE;
5252                 else
5253                         return error(_("bad --word-diff argument: %s"), arg);
5254         } else {
5255                 if (options->word_diff == DIFF_WORDS_NONE)
5256                         options->word_diff = DIFF_WORDS_PLAIN;
5257         }
5258         return 0;
5259 }
5260
5261 static int diff_opt_word_diff_regex(const struct option *opt,
5262                                     const char *arg, int unset)
5263 {
5264         struct diff_options *options = opt->value;
5265
5266         BUG_ON_OPT_NEG(unset);
5267         if (options->word_diff == DIFF_WORDS_NONE)
5268                 options->word_diff = DIFF_WORDS_PLAIN;
5269         options->word_regex = arg;
5270         return 0;
5271 }
5272
5273 static void prep_parse_options(struct diff_options *options)
5274 {
5275         struct option parseopts[] = {
5276                 OPT_GROUP(N_("Diff output format options")),
5277                 OPT_BITOP('p', "patch", &options->output_format,
5278                           N_("generate patch"),
5279                           DIFF_FORMAT_PATCH, DIFF_FORMAT_NO_OUTPUT),
5280                 OPT_BIT_F('s', "no-patch", &options->output_format,
5281                           N_("suppress diff output"),
5282                           DIFF_FORMAT_NO_OUTPUT, PARSE_OPT_NONEG),
5283                 OPT_BITOP('u', NULL, &options->output_format,
5284                           N_("generate patch"),
5285                           DIFF_FORMAT_PATCH, DIFF_FORMAT_NO_OUTPUT),
5286                 OPT_CALLBACK_F('U', "unified", options, N_("<n>"),
5287                                N_("generate diffs with <n> lines context"),
5288                                PARSE_OPT_NONEG | PARSE_OPT_OPTARG, diff_opt_unified),
5289                 OPT_BOOL('W', "function-context", &options->flags.funccontext,
5290                          N_("generate diffs with <n> lines context")),
5291                 OPT_BIT_F(0, "raw", &options->output_format,
5292                           N_("generate the diff in raw format"),
5293                           DIFF_FORMAT_RAW, PARSE_OPT_NONEG),
5294                 OPT_BITOP(0, "patch-with-raw", &options->output_format,
5295                           N_("synonym for '-p --raw'"),
5296                           DIFF_FORMAT_PATCH | DIFF_FORMAT_RAW,
5297                           DIFF_FORMAT_NO_OUTPUT),
5298                 OPT_BITOP(0, "patch-with-stat", &options->output_format,
5299                           N_("synonym for '-p --stat'"),
5300                           DIFF_FORMAT_PATCH | DIFF_FORMAT_DIFFSTAT,
5301                           DIFF_FORMAT_NO_OUTPUT),
5302                 OPT_BIT_F(0, "numstat", &options->output_format,
5303                           N_("machine friendly --stat"),
5304                           DIFF_FORMAT_NUMSTAT, PARSE_OPT_NONEG),
5305                 OPT_BIT_F(0, "shortstat", &options->output_format,
5306                           N_("output only the last line of --stat"),
5307                           DIFF_FORMAT_SHORTSTAT, PARSE_OPT_NONEG),
5308                 OPT_CALLBACK_F('X', "dirstat", options, N_("<param1,param2>..."),
5309                                N_("output the distribution of relative amount of changes for each sub-directory"),
5310                                PARSE_OPT_NONEG | PARSE_OPT_OPTARG,
5311                                diff_opt_dirstat),
5312                 OPT_CALLBACK_F(0, "cumulative", options, NULL,
5313                                N_("synonym for --dirstat=cumulative"),
5314                                PARSE_OPT_NONEG | PARSE_OPT_NOARG,
5315                                diff_opt_dirstat),
5316                 OPT_CALLBACK_F(0, "dirstat-by-file", options, N_("<param1,param2>..."),
5317                                N_("synonym for --dirstat=files,param1,param2..."),
5318                                PARSE_OPT_NONEG | PARSE_OPT_OPTARG,
5319                                diff_opt_dirstat),
5320                 OPT_BIT_F(0, "check", &options->output_format,
5321                           N_("warn if changes introduce conflict markers or whitespace errors"),
5322                           DIFF_FORMAT_CHECKDIFF, PARSE_OPT_NONEG),
5323                 OPT_BIT_F(0, "summary", &options->output_format,
5324                           N_("condensed summary such as creations, renames and mode changes"),
5325                           DIFF_FORMAT_SUMMARY, PARSE_OPT_NONEG),
5326                 OPT_BIT_F(0, "name-only", &options->output_format,
5327                           N_("show only names of changed files"),
5328                           DIFF_FORMAT_NAME, PARSE_OPT_NONEG),
5329                 OPT_BIT_F(0, "name-status", &options->output_format,
5330                           N_("show only names and status of changed files"),
5331                           DIFF_FORMAT_NAME_STATUS, PARSE_OPT_NONEG),
5332                 OPT_CALLBACK_F(0, "stat", options, N_("<width>[,<name-width>[,<count>]]"),
5333                                N_("generate diffstat"),
5334                                PARSE_OPT_NONEG | PARSE_OPT_OPTARG, diff_opt_stat),
5335                 OPT_CALLBACK_F(0, "stat-width", options, N_("<width>"),
5336                                N_("generate diffstat with a given width"),
5337                                PARSE_OPT_NONEG, diff_opt_stat),
5338                 OPT_CALLBACK_F(0, "stat-name-width", options, N_("<width>"),
5339                                N_("generate diffstat with a given name width"),
5340                                PARSE_OPT_NONEG, diff_opt_stat),
5341                 OPT_CALLBACK_F(0, "stat-graph-width", options, N_("<width>"),
5342                                N_("generate diffstat with a given graph width"),
5343                                PARSE_OPT_NONEG, diff_opt_stat),
5344                 OPT_CALLBACK_F(0, "stat-count", options, N_("<count>"),
5345                                N_("generate diffstat with limited lines"),
5346                                PARSE_OPT_NONEG, diff_opt_stat),
5347                 OPT_CALLBACK_F(0, "compact-summary", options, NULL,
5348                                N_("generate compact summary in diffstat"),
5349                                PARSE_OPT_NOARG, diff_opt_compact_summary),
5350                 OPT_CALLBACK_F(0, "binary", options, NULL,
5351                                N_("output a binary diff that can be applied"),
5352                                PARSE_OPT_NONEG | PARSE_OPT_NOARG, diff_opt_binary),
5353                 OPT_BOOL(0, "full-index", &options->flags.full_index,
5354                          N_("show full pre- and post-image object names on the \"index\" lines")),
5355                 OPT_COLOR_FLAG(0, "color", &options->use_color,
5356                                N_("show colored diff")),
5357                 OPT_CALLBACK_F(0, "ws-error-highlight", options, N_("<kind>"),
5358                                N_("highlight whitespace errors in the 'context', 'old' or 'new' lines in the diff"),
5359                                PARSE_OPT_NONEG, diff_opt_ws_error_highlight),
5360                 OPT_SET_INT('z', NULL, &options->line_termination,
5361                             N_("do not munge pathnames and use NULs as output field terminators in --raw or --numstat"),
5362                             0),
5363                 OPT__ABBREV(&options->abbrev),
5364                 OPT_STRING_F(0, "src-prefix", &options->a_prefix, N_("<prefix>"),
5365                              N_("show the given source prefix instead of \"a/\""),
5366                              PARSE_OPT_NONEG),
5367                 OPT_STRING_F(0, "dst-prefix", &options->b_prefix, N_("<prefix>"),
5368                              N_("show the given destination prefix instead of \"b/\""),
5369                              PARSE_OPT_NONEG),
5370                 OPT_CALLBACK_F(0, "line-prefix", options, N_("<prefix>"),
5371                                N_("prepend an additional prefix to every line of output"),
5372                                PARSE_OPT_NONEG, diff_opt_line_prefix),
5373                 OPT_CALLBACK_F(0, "no-prefix", options, NULL,
5374                                N_("do not show any source or destination prefix"),
5375                                PARSE_OPT_NONEG | PARSE_OPT_NOARG, diff_opt_no_prefix),
5376                 OPT_INTEGER_F(0, "inter-hunk-context", &options->interhunkcontext,
5377                               N_("show context between diff hunks up to the specified number of lines"),
5378                               PARSE_OPT_NONEG),
5379                 OPT_CALLBACK_F(0, "output-indicator-new",
5380                                &options->output_indicators[OUTPUT_INDICATOR_NEW],
5381                                N_("<char>"),
5382                                N_("specify the character to indicate a new line instead of '+'"),
5383                                PARSE_OPT_NONEG, diff_opt_char),
5384                 OPT_CALLBACK_F(0, "output-indicator-old",
5385                                &options->output_indicators[OUTPUT_INDICATOR_OLD],
5386                                N_("<char>"),
5387                                N_("specify the character to indicate an old line instead of '-'"),
5388                                PARSE_OPT_NONEG, diff_opt_char),
5389                 OPT_CALLBACK_F(0, "output-indicator-context",
5390                                &options->output_indicators[OUTPUT_INDICATOR_CONTEXT],
5391                                N_("<char>"),
5392                                N_("specify the character to indicate a context instead of ' '"),
5393                                PARSE_OPT_NONEG, diff_opt_char),
5394
5395                 OPT_GROUP(N_("Diff rename options")),
5396                 OPT_CALLBACK_F('B', "break-rewrites", &options->break_opt, N_("<n>[/<m>]"),
5397                                N_("break complete rewrite changes into pairs of delete and create"),
5398                                PARSE_OPT_NONEG | PARSE_OPT_OPTARG,
5399                                diff_opt_break_rewrites),
5400                 OPT_CALLBACK_F('M', "find-renames", options, N_("<n>"),
5401                                N_("detect renames"),
5402                                PARSE_OPT_NONEG | PARSE_OPT_OPTARG,
5403                                diff_opt_find_renames),
5404                 OPT_SET_INT_F('D', "irreversible-delete", &options->irreversible_delete,
5405                               N_("omit the preimage for deletes"),
5406                               1, PARSE_OPT_NONEG),
5407                 OPT_CALLBACK_F('C', "find-copies", options, N_("<n>"),
5408                                N_("detect copies"),
5409                                PARSE_OPT_NONEG | PARSE_OPT_OPTARG,
5410                                diff_opt_find_copies),
5411                 OPT_BOOL(0, "find-copies-harder", &options->flags.find_copies_harder,
5412                          N_("use unmodified files as source to find copies")),
5413                 OPT_SET_INT_F(0, "no-renames", &options->detect_rename,
5414                               N_("disable rename detection"),
5415                               0, PARSE_OPT_NONEG),
5416                 OPT_BOOL(0, "rename-empty", &options->flags.rename_empty,
5417                          N_("use empty blobs as rename source")),
5418                 OPT_CALLBACK_F(0, "follow", options, NULL,
5419                                N_("continue listing the history of a file beyond renames"),
5420                                PARSE_OPT_NOARG, diff_opt_follow),
5421                 OPT_INTEGER('l', NULL, &options->rename_limit,
5422                             N_("prevent rename/copy detection if the number of rename/copy targets exceeds given limit")),
5423
5424                 OPT_GROUP(N_("Diff algorithm options")),
5425                 OPT_BIT(0, "minimal", &options->xdl_opts,
5426                         N_("produce the smallest possible diff"),
5427                         XDF_NEED_MINIMAL),
5428                 OPT_BIT_F('w', "ignore-all-space", &options->xdl_opts,
5429                           N_("ignore whitespace when comparing lines"),
5430                           XDF_IGNORE_WHITESPACE, PARSE_OPT_NONEG),
5431                 OPT_BIT_F('b', "ignore-space-change", &options->xdl_opts,
5432                           N_("ignore changes in amount of whitespace"),
5433                           XDF_IGNORE_WHITESPACE_CHANGE, PARSE_OPT_NONEG),
5434                 OPT_BIT_F(0, "ignore-space-at-eol", &options->xdl_opts,
5435                           N_("ignore changes in whitespace at EOL"),
5436                           XDF_IGNORE_WHITESPACE_AT_EOL, PARSE_OPT_NONEG),
5437                 OPT_BIT_F(0, "ignore-cr-at-eol", &options->xdl_opts,
5438                           N_("ignore carrier-return at the end of line"),
5439                           XDF_IGNORE_CR_AT_EOL, PARSE_OPT_NONEG),
5440                 OPT_BIT_F(0, "ignore-blank-lines", &options->xdl_opts,
5441                           N_("ignore changes whose lines are all blank"),
5442                           XDF_IGNORE_BLANK_LINES, PARSE_OPT_NONEG),
5443                 OPT_BIT(0, "indent-heuristic", &options->xdl_opts,
5444                         N_("heuristic to shift diff hunk boundaries for easy reading"),
5445                         XDF_INDENT_HEURISTIC),
5446                 OPT_CALLBACK_F(0, "patience", options, NULL,
5447                                N_("generate diff using the \"patience diff\" algorithm"),
5448                                PARSE_OPT_NONEG | PARSE_OPT_NOARG,
5449                                diff_opt_patience),
5450                 OPT_BITOP(0, "histogram", &options->xdl_opts,
5451                           N_("generate diff using the \"histogram diff\" algorithm"),
5452                           XDF_HISTOGRAM_DIFF, XDF_DIFF_ALGORITHM_MASK),
5453                 OPT_CALLBACK_F(0, "diff-algorithm", options, N_("<algorithm>"),
5454                                N_("choose a diff algorithm"),
5455                                PARSE_OPT_NONEG, diff_opt_diff_algorithm),
5456                 OPT_CALLBACK_F(0, "anchored", options, N_("<text>"),
5457                                N_("generate diff using the \"anchored diff\" algorithm"),
5458                                PARSE_OPT_NONEG, diff_opt_anchored),
5459                 OPT_CALLBACK_F(0, "word-diff", options, N_("<mode>"),
5460                                N_("show word diff, using <mode> to delimit changed words"),
5461                                PARSE_OPT_NONEG | PARSE_OPT_OPTARG, diff_opt_word_diff),
5462                 OPT_CALLBACK_F(0, "word-diff-regex", options, N_("<regex>"),
5463                                N_("use <regex> to decide what a word is"),
5464                                PARSE_OPT_NONEG, diff_opt_word_diff_regex),
5465                 OPT_CALLBACK_F(0, "color-words", options, N_("<regex>"),
5466                                N_("equivalent to --word-diff=color --word-diff-regex=<regex>"),
5467                                PARSE_OPT_NONEG | PARSE_OPT_OPTARG, diff_opt_color_words),
5468                 OPT_CALLBACK_F(0, "color-moved", options, N_("<mode>"),
5469                                N_("moved lines of code are colored differently"),
5470                                PARSE_OPT_OPTARG, diff_opt_color_moved),
5471                 OPT_CALLBACK_F(0, "color-moved-ws", options, N_("<mode>"),
5472                                N_("how white spaces are ignored in --color-moved"),
5473                                0, diff_opt_color_moved_ws),
5474
5475                 OPT_GROUP(N_("Other diff options")),
5476                 OPT_CALLBACK_F(0, "relative", options, N_("<prefix>"),
5477                                N_("when run from subdir, exclude changes outside and show relative paths"),
5478                                PARSE_OPT_NONEG | PARSE_OPT_OPTARG,
5479                                diff_opt_relative),
5480                 OPT_BOOL('a', "text", &options->flags.text,
5481                          N_("treat all files as text")),
5482                 OPT_BOOL('R', NULL, &options->flags.reverse_diff,
5483                          N_("swap two inputs, reverse the diff")),
5484                 OPT_BOOL(0, "exit-code", &options->flags.exit_with_status,
5485                          N_("exit with 1 if there were differences, 0 otherwise")),
5486                 OPT_BOOL(0, "quiet", &options->flags.quick,
5487                          N_("disable all output of the program")),
5488                 OPT_BOOL(0, "ext-diff", &options->flags.allow_external,
5489                          N_("allow an external diff helper to be executed")),
5490                 OPT_CALLBACK_F(0, "textconv", options, NULL,
5491                                N_("run external text conversion filters when comparing binary files"),
5492                                PARSE_OPT_NOARG, diff_opt_textconv),
5493                 OPT_CALLBACK_F(0, "ignore-submodules", options, N_("<when>"),
5494                                N_("ignore changes to submodules in the diff generation"),
5495                                PARSE_OPT_NONEG | PARSE_OPT_OPTARG,
5496                                diff_opt_ignore_submodules),
5497                 OPT_CALLBACK_F(0, "submodule", options, N_("<format>"),
5498                                N_("specify how differences in submodules are shown"),
5499                                PARSE_OPT_NONEG | PARSE_OPT_OPTARG,
5500                                diff_opt_submodule),
5501                 OPT_SET_INT_F(0, "ita-invisible-in-index", &options->ita_invisible_in_index,
5502                               N_("hide 'git add -N' entries from the index"),
5503                               1, PARSE_OPT_NONEG),
5504                 OPT_SET_INT_F(0, "ita-visible-in-index", &options->ita_invisible_in_index,
5505                               N_("treat 'git add -N' entries as real in the index"),
5506                               0, PARSE_OPT_NONEG),
5507                 OPT_CALLBACK_F('S', NULL, options, N_("<string>"),
5508                                N_("look for differences that change the number of occurrences of the specified string"),
5509                                0, diff_opt_pickaxe_string),
5510                 OPT_CALLBACK_F('G', NULL, options, N_("<regex>"),
5511                                N_("look for differences that change the number of occurrences of the specified regex"),
5512                                0, diff_opt_pickaxe_regex),
5513                 OPT_BIT_F(0, "pickaxe-all", &options->pickaxe_opts,
5514                           N_("show all changes in the changeset with -S or -G"),
5515                           DIFF_PICKAXE_ALL, PARSE_OPT_NONEG),
5516                 OPT_BIT_F(0, "pickaxe-regex", &options->pickaxe_opts,
5517                           N_("treat <string> in -S as extended POSIX regular expression"),
5518                           DIFF_PICKAXE_REGEX, PARSE_OPT_NONEG),
5519                 OPT_FILENAME('O', NULL, &options->orderfile,
5520                              N_("control the order in which files appear in the output")),
5521                 OPT_CALLBACK_F(0, "find-object", options, N_("<object-id>"),
5522                                N_("look for differences that change the number of occurrences of the specified object"),
5523                                PARSE_OPT_NONEG, diff_opt_find_object),
5524                 OPT_CALLBACK_F(0, "diff-filter", options, N_("[(A|C|D|M|R|T|U|X|B)...[*]]"),
5525                                N_("select files by diff type"),
5526                                PARSE_OPT_NONEG, diff_opt_diff_filter),
5527                 { OPTION_CALLBACK, 0, "output", options, N_("<file>"),
5528                   N_("Output to a specific file"),
5529                   PARSE_OPT_NONEG, NULL, 0, diff_opt_output },
5530
5531                 OPT_END()
5532         };
5533
5534         ALLOC_ARRAY(options->parseopts, ARRAY_SIZE(parseopts));
5535         memcpy(options->parseopts, parseopts, sizeof(parseopts));
5536 }
5537
5538 int diff_opt_parse(struct diff_options *options,
5539                    const char **av, int ac, const char *prefix)
5540 {
5541         if (!prefix)
5542                 prefix = "";
5543
5544         ac = parse_options(ac, av, prefix, options->parseopts, NULL,
5545                            PARSE_OPT_KEEP_DASHDASH |
5546                            PARSE_OPT_KEEP_UNKNOWN |
5547                            PARSE_OPT_NO_INTERNAL_HELP |
5548                            PARSE_OPT_ONE_SHOT |
5549                            PARSE_OPT_STOP_AT_NON_OPTION);
5550
5551         return ac;
5552 }
5553
5554 int parse_rename_score(const char **cp_p)
5555 {
5556         unsigned long num, scale;
5557         int ch, dot;
5558         const char *cp = *cp_p;
5559
5560         num = 0;
5561         scale = 1;
5562         dot = 0;
5563         for (;;) {
5564                 ch = *cp;
5565                 if ( !dot && ch == '.' ) {
5566                         scale = 1;
5567                         dot = 1;
5568                 } else if ( ch == '%' ) {
5569                         scale = dot ? scale*100 : 100;
5570                         cp++;   /* % is always at the end */
5571                         break;
5572                 } else if ( ch >= '0' && ch <= '9' ) {
5573                         if ( scale < 100000 ) {
5574                                 scale *= 10;
5575                                 num = (num*10) + (ch-'0');
5576                         }
5577                 } else {
5578                         break;
5579                 }
5580                 cp++;
5581         }
5582         *cp_p = cp;
5583
5584         /* user says num divided by scale and we say internally that
5585          * is MAX_SCORE * num / scale.
5586          */
5587         return (int)((num >= scale) ? MAX_SCORE : (MAX_SCORE * num / scale));
5588 }
5589
5590 struct diff_queue_struct diff_queued_diff;
5591
5592 void diff_q(struct diff_queue_struct *queue, struct diff_filepair *dp)
5593 {
5594         ALLOC_GROW(queue->queue, queue->nr + 1, queue->alloc);
5595         queue->queue[queue->nr++] = dp;
5596 }
5597
5598 struct diff_filepair *diff_queue(struct diff_queue_struct *queue,
5599                                  struct diff_filespec *one,
5600                                  struct diff_filespec *two)
5601 {
5602         struct diff_filepair *dp = xcalloc(1, sizeof(*dp));
5603         dp->one = one;
5604         dp->two = two;
5605         if (queue)
5606                 diff_q(queue, dp);
5607         return dp;
5608 }
5609
5610 void diff_free_filepair(struct diff_filepair *p)
5611 {
5612         free_filespec(p->one);
5613         free_filespec(p->two);
5614         free(p);
5615 }
5616
5617 const char *diff_aligned_abbrev(const struct object_id *oid, int len)
5618 {
5619         int abblen;
5620         const char *abbrev;
5621
5622         /* Do we want all 40 hex characters? */
5623         if (len == the_hash_algo->hexsz)
5624                 return oid_to_hex(oid);
5625
5626         /* An abbreviated value is fine, possibly followed by an ellipsis. */
5627         abbrev = diff_abbrev_oid(oid, len);
5628
5629         if (!print_sha1_ellipsis())
5630                 return abbrev;
5631
5632         abblen = strlen(abbrev);
5633
5634         /*
5635          * In well-behaved cases, where the abbreviated result is the
5636          * same as the requested length, append three dots after the
5637          * abbreviation (hence the whole logic is limited to the case
5638          * where abblen < 37); when the actual abbreviated result is a
5639          * bit longer than the requested length, we reduce the number
5640          * of dots so that they match the well-behaved ones.  However,
5641          * if the actual abbreviation is longer than the requested
5642          * length by more than three, we give up on aligning, and add
5643          * three dots anyway, to indicate that the output is not the
5644          * full object name.  Yes, this may be suboptimal, but this
5645          * appears only in "diff --raw --abbrev" output and it is not
5646          * worth the effort to change it now.  Note that this would
5647          * likely to work fine when the automatic sizing of default
5648          * abbreviation length is used--we would be fed -1 in "len" in
5649          * that case, and will end up always appending three-dots, but
5650          * the automatic sizing is supposed to give abblen that ensures
5651          * uniqueness across all objects (statistically speaking).
5652          */
5653         if (abblen < the_hash_algo->hexsz - 3) {
5654                 static char hex[GIT_MAX_HEXSZ + 1];
5655                 if (len < abblen && abblen <= len + 2)
5656                         xsnprintf(hex, sizeof(hex), "%s%.*s", abbrev, len+3-abblen, "..");
5657                 else
5658                         xsnprintf(hex, sizeof(hex), "%s...", abbrev);
5659                 return hex;
5660         }
5661
5662         return oid_to_hex(oid);
5663 }
5664
5665 static void diff_flush_raw(struct diff_filepair *p, struct diff_options *opt)
5666 {
5667         int line_termination = opt->line_termination;
5668         int inter_name_termination = line_termination ? '\t' : '\0';
5669
5670         fprintf(opt->file, "%s", diff_line_prefix(opt));
5671         if (!(opt->output_format & DIFF_FORMAT_NAME_STATUS)) {
5672                 fprintf(opt->file, ":%06o %06o %s ", p->one->mode, p->two->mode,
5673                         diff_aligned_abbrev(&p->one->oid, opt->abbrev));
5674                 fprintf(opt->file, "%s ",
5675                         diff_aligned_abbrev(&p->two->oid, opt->abbrev));
5676         }
5677         if (p->score) {
5678                 fprintf(opt->file, "%c%03d%c", p->status, similarity_index(p),
5679                         inter_name_termination);
5680         } else {
5681                 fprintf(opt->file, "%c%c", p->status, inter_name_termination);
5682         }
5683
5684         if (p->status == DIFF_STATUS_COPIED ||
5685             p->status == DIFF_STATUS_RENAMED) {
5686                 const char *name_a, *name_b;
5687                 name_a = p->one->path;
5688                 name_b = p->two->path;
5689                 strip_prefix(opt->prefix_length, &name_a, &name_b);
5690                 write_name_quoted(name_a, opt->file, inter_name_termination);
5691                 write_name_quoted(name_b, opt->file, line_termination);
5692         } else {
5693                 const char *name_a, *name_b;
5694                 name_a = p->one->mode ? p->one->path : p->two->path;
5695                 name_b = NULL;
5696                 strip_prefix(opt->prefix_length, &name_a, &name_b);
5697                 write_name_quoted(name_a, opt->file, line_termination);
5698         }
5699 }
5700
5701 int diff_unmodified_pair(struct diff_filepair *p)
5702 {
5703         /* This function is written stricter than necessary to support
5704          * the currently implemented transformers, but the idea is to
5705          * let transformers to produce diff_filepairs any way they want,
5706          * and filter and clean them up here before producing the output.
5707          */
5708         struct diff_filespec *one = p->one, *two = p->two;
5709
5710         if (DIFF_PAIR_UNMERGED(p))
5711                 return 0; /* unmerged is interesting */
5712
5713         /* deletion, addition, mode or type change
5714          * and rename are all interesting.
5715          */
5716         if (DIFF_FILE_VALID(one) != DIFF_FILE_VALID(two) ||
5717             DIFF_PAIR_MODE_CHANGED(p) ||
5718             strcmp(one->path, two->path))
5719                 return 0;
5720
5721         /* both are valid and point at the same path.  that is, we are
5722          * dealing with a change.
5723          */
5724         if (one->oid_valid && two->oid_valid &&
5725             oideq(&one->oid, &two->oid) &&
5726             !one->dirty_submodule && !two->dirty_submodule)
5727                 return 1; /* no change */
5728         if (!one->oid_valid && !two->oid_valid)
5729                 return 1; /* both look at the same file on the filesystem. */
5730         return 0;
5731 }
5732
5733 static void diff_flush_patch(struct diff_filepair *p, struct diff_options *o)
5734 {
5735         if (diff_unmodified_pair(p))
5736                 return;
5737
5738         if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
5739             (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
5740                 return; /* no tree diffs in patch format */
5741
5742         run_diff(p, o);
5743 }
5744
5745 static void diff_flush_stat(struct diff_filepair *p, struct diff_options *o,
5746                             struct diffstat_t *diffstat)
5747 {
5748         if (diff_unmodified_pair(p))
5749                 return;
5750
5751         if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
5752             (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
5753                 return; /* no useful stat for tree diffs */
5754
5755         run_diffstat(p, o, diffstat);
5756 }
5757
5758 static void diff_flush_checkdiff(struct diff_filepair *p,
5759                 struct diff_options *o)
5760 {
5761         if (diff_unmodified_pair(p))
5762                 return;
5763
5764         if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
5765             (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
5766                 return; /* nothing to check in tree diffs */
5767
5768         run_checkdiff(p, o);
5769 }
5770
5771 int diff_queue_is_empty(void)
5772 {
5773         struct diff_queue_struct *q = &diff_queued_diff;
5774         int i;
5775         for (i = 0; i < q->nr; i++)
5776                 if (!diff_unmodified_pair(q->queue[i]))
5777                         return 0;
5778         return 1;
5779 }
5780
5781 #if DIFF_DEBUG
5782 void diff_debug_filespec(struct diff_filespec *s, int x, const char *one)
5783 {
5784         fprintf(stderr, "queue[%d] %s (%s) %s %06o %s\n",
5785                 x, one ? one : "",
5786                 s->path,
5787                 DIFF_FILE_VALID(s) ? "valid" : "invalid",
5788                 s->mode,
5789                 s->oid_valid ? oid_to_hex(&s->oid) : "");
5790         fprintf(stderr, "queue[%d] %s size %lu\n",
5791                 x, one ? one : "",
5792                 s->size);
5793 }
5794
5795 void diff_debug_filepair(const struct diff_filepair *p, int i)
5796 {
5797         diff_debug_filespec(p->one, i, "one");
5798         diff_debug_filespec(p->two, i, "two");
5799         fprintf(stderr, "score %d, status %c rename_used %d broken %d\n",
5800                 p->score, p->status ? p->status : '?',
5801                 p->one->rename_used, p->broken_pair);
5802 }
5803
5804 void diff_debug_queue(const char *msg, struct diff_queue_struct *q)
5805 {
5806         int i;
5807         if (msg)
5808                 fprintf(stderr, "%s\n", msg);
5809         fprintf(stderr, "q->nr = %d\n", q->nr);
5810         for (i = 0; i < q->nr; i++) {
5811                 struct diff_filepair *p = q->queue[i];
5812                 diff_debug_filepair(p, i);
5813         }
5814 }
5815 #endif
5816
5817 static void diff_resolve_rename_copy(void)
5818 {
5819         int i;
5820         struct diff_filepair *p;
5821         struct diff_queue_struct *q = &diff_queued_diff;
5822
5823         diff_debug_queue("resolve-rename-copy", q);
5824
5825         for (i = 0; i < q->nr; i++) {
5826                 p = q->queue[i];
5827                 p->status = 0; /* undecided */
5828                 if (DIFF_PAIR_UNMERGED(p))
5829                         p->status = DIFF_STATUS_UNMERGED;
5830                 else if (!DIFF_FILE_VALID(p->one))
5831                         p->status = DIFF_STATUS_ADDED;
5832                 else if (!DIFF_FILE_VALID(p->two))
5833                         p->status = DIFF_STATUS_DELETED;
5834                 else if (DIFF_PAIR_TYPE_CHANGED(p))
5835                         p->status = DIFF_STATUS_TYPE_CHANGED;
5836
5837                 /* from this point on, we are dealing with a pair
5838                  * whose both sides are valid and of the same type, i.e.
5839                  * either in-place edit or rename/copy edit.
5840                  */
5841                 else if (DIFF_PAIR_RENAME(p)) {
5842                         /*
5843                          * A rename might have re-connected a broken
5844                          * pair up, causing the pathnames to be the
5845                          * same again. If so, that's not a rename at
5846                          * all, just a modification..
5847                          *
5848                          * Otherwise, see if this source was used for
5849                          * multiple renames, in which case we decrement
5850                          * the count, and call it a copy.
5851                          */
5852                         if (!strcmp(p->one->path, p->two->path))
5853                                 p->status = DIFF_STATUS_MODIFIED;
5854                         else if (--p->one->rename_used > 0)
5855                                 p->status = DIFF_STATUS_COPIED;
5856                         else
5857                                 p->status = DIFF_STATUS_RENAMED;
5858                 }
5859                 else if (!oideq(&p->one->oid, &p->two->oid) ||
5860                          p->one->mode != p->two->mode ||
5861                          p->one->dirty_submodule ||
5862                          p->two->dirty_submodule ||
5863                          is_null_oid(&p->one->oid))
5864                         p->status = DIFF_STATUS_MODIFIED;
5865                 else {
5866                         /* This is a "no-change" entry and should not
5867                          * happen anymore, but prepare for broken callers.
5868                          */
5869                         error("feeding unmodified %s to diffcore",
5870                               p->one->path);
5871                         p->status = DIFF_STATUS_UNKNOWN;
5872                 }
5873         }
5874         diff_debug_queue("resolve-rename-copy done", q);
5875 }
5876
5877 static int check_pair_status(struct diff_filepair *p)
5878 {
5879         switch (p->status) {
5880         case DIFF_STATUS_UNKNOWN:
5881                 return 0;
5882         case 0:
5883                 die("internal error in diff-resolve-rename-copy");
5884         default:
5885                 return 1;
5886         }
5887 }
5888
5889 static void flush_one_pair(struct diff_filepair *p, struct diff_options *opt)
5890 {
5891         int fmt = opt->output_format;
5892
5893         if (fmt & DIFF_FORMAT_CHECKDIFF)
5894                 diff_flush_checkdiff(p, opt);
5895         else if (fmt & (DIFF_FORMAT_RAW | DIFF_FORMAT_NAME_STATUS))
5896                 diff_flush_raw(p, opt);
5897         else if (fmt & DIFF_FORMAT_NAME) {
5898                 const char *name_a, *name_b;
5899                 name_a = p->two->path;
5900                 name_b = NULL;
5901                 strip_prefix(opt->prefix_length, &name_a, &name_b);
5902                 fprintf(opt->file, "%s", diff_line_prefix(opt));
5903                 write_name_quoted(name_a, opt->file, opt->line_termination);
5904         }
5905 }
5906
5907 static void show_file_mode_name(struct diff_options *opt, const char *newdelete, struct diff_filespec *fs)
5908 {
5909         struct strbuf sb = STRBUF_INIT;
5910         if (fs->mode)
5911                 strbuf_addf(&sb, " %s mode %06o ", newdelete, fs->mode);
5912         else
5913                 strbuf_addf(&sb, " %s ", newdelete);
5914
5915         quote_c_style(fs->path, &sb, NULL, 0);
5916         strbuf_addch(&sb, '\n');
5917         emit_diff_symbol(opt, DIFF_SYMBOL_SUMMARY,
5918                          sb.buf, sb.len, 0);
5919         strbuf_release(&sb);
5920 }
5921
5922 static void show_mode_change(struct diff_options *opt, struct diff_filepair *p,
5923                 int show_name)
5924 {
5925         if (p->one->mode && p->two->mode && p->one->mode != p->two->mode) {
5926                 struct strbuf sb = STRBUF_INIT;
5927                 strbuf_addf(&sb, " mode change %06o => %06o",
5928                             p->one->mode, p->two->mode);
5929                 if (show_name) {
5930                         strbuf_addch(&sb, ' ');
5931                         quote_c_style(p->two->path, &sb, NULL, 0);
5932                 }
5933                 strbuf_addch(&sb, '\n');
5934                 emit_diff_symbol(opt, DIFF_SYMBOL_SUMMARY,
5935                                  sb.buf, sb.len, 0);
5936                 strbuf_release(&sb);
5937         }
5938 }
5939
5940 static void show_rename_copy(struct diff_options *opt, const char *renamecopy,
5941                 struct diff_filepair *p)
5942 {
5943         struct strbuf sb = STRBUF_INIT;
5944         struct strbuf names = STRBUF_INIT;
5945
5946         pprint_rename(&names, p->one->path, p->two->path);
5947         strbuf_addf(&sb, " %s %s (%d%%)\n",
5948                     renamecopy, names.buf, similarity_index(p));
5949         strbuf_release(&names);
5950         emit_diff_symbol(opt, DIFF_SYMBOL_SUMMARY,
5951                                  sb.buf, sb.len, 0);
5952         show_mode_change(opt, p, 0);
5953         strbuf_release(&sb);
5954 }
5955
5956 static void diff_summary(struct diff_options *opt, struct diff_filepair *p)
5957 {
5958         switch(p->status) {
5959         case DIFF_STATUS_DELETED:
5960                 show_file_mode_name(opt, "delete", p->one);
5961                 break;
5962         case DIFF_STATUS_ADDED:
5963                 show_file_mode_name(opt, "create", p->two);
5964                 break;
5965         case DIFF_STATUS_COPIED:
5966                 show_rename_copy(opt, "copy", p);
5967                 break;
5968         case DIFF_STATUS_RENAMED:
5969                 show_rename_copy(opt, "rename", p);
5970                 break;
5971         default:
5972                 if (p->score) {
5973                         struct strbuf sb = STRBUF_INIT;
5974                         strbuf_addstr(&sb, " rewrite ");
5975                         quote_c_style(p->two->path, &sb, NULL, 0);
5976                         strbuf_addf(&sb, " (%d%%)\n", similarity_index(p));
5977                         emit_diff_symbol(opt, DIFF_SYMBOL_SUMMARY,
5978                                          sb.buf, sb.len, 0);
5979                         strbuf_release(&sb);
5980                 }
5981                 show_mode_change(opt, p, !p->score);
5982                 break;
5983         }
5984 }
5985
5986 struct patch_id_t {
5987         git_SHA_CTX *ctx;
5988         int patchlen;
5989 };
5990
5991 static int remove_space(char *line, int len)
5992 {
5993         int i;
5994         char *dst = line;
5995         unsigned char c;
5996
5997         for (i = 0; i < len; i++)
5998                 if (!isspace((c = line[i])))
5999                         *dst++ = c;
6000
6001         return dst - line;
6002 }
6003
6004 void flush_one_hunk(struct object_id *result, git_SHA_CTX *ctx)
6005 {
6006         unsigned char hash[GIT_MAX_RAWSZ];
6007         unsigned short carry = 0;
6008         int i;
6009
6010         git_SHA1_Final(hash, ctx);
6011         git_SHA1_Init(ctx);
6012         /* 20-byte sum, with carry */
6013         for (i = 0; i < GIT_SHA1_RAWSZ; ++i) {
6014                 carry += result->hash[i] + hash[i];
6015                 result->hash[i] = carry;
6016                 carry >>= 8;
6017         }
6018 }
6019
6020 static void patch_id_consume(void *priv, char *line, unsigned long len)
6021 {
6022         struct patch_id_t *data = priv;
6023         int new_len;
6024
6025         new_len = remove_space(line, len);
6026
6027         git_SHA1_Update(data->ctx, line, new_len);
6028         data->patchlen += new_len;
6029 }
6030
6031 static void patch_id_add_string(git_SHA_CTX *ctx, const char *str)
6032 {
6033         git_SHA1_Update(ctx, str, strlen(str));
6034 }
6035
6036 static void patch_id_add_mode(git_SHA_CTX *ctx, unsigned mode)
6037 {
6038         /* large enough for 2^32 in octal */
6039         char buf[12];
6040         int len = xsnprintf(buf, sizeof(buf), "%06o", mode);
6041         git_SHA1_Update(ctx, buf, len);
6042 }
6043
6044 /* returns 0 upon success, and writes result into oid */
6045 static int diff_get_patch_id(struct diff_options *options, struct object_id *oid, int diff_header_only, int stable)
6046 {
6047         struct diff_queue_struct *q = &diff_queued_diff;
6048         int i;
6049         git_SHA_CTX ctx;
6050         struct patch_id_t data;
6051
6052         git_SHA1_Init(&ctx);
6053         memset(&data, 0, sizeof(struct patch_id_t));
6054         data.ctx = &ctx;
6055         oidclr(oid);
6056
6057         for (i = 0; i < q->nr; i++) {
6058                 xpparam_t xpp;
6059                 xdemitconf_t xecfg;
6060                 mmfile_t mf1, mf2;
6061                 struct diff_filepair *p = q->queue[i];
6062                 int len1, len2;
6063
6064                 memset(&xpp, 0, sizeof(xpp));
6065                 memset(&xecfg, 0, sizeof(xecfg));
6066                 if (p->status == 0)
6067                         return error("internal diff status error");
6068                 if (p->status == DIFF_STATUS_UNKNOWN)
6069                         continue;
6070                 if (diff_unmodified_pair(p))
6071                         continue;
6072                 if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
6073                     (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
6074                         continue;
6075                 if (DIFF_PAIR_UNMERGED(p))
6076                         continue;
6077
6078                 diff_fill_oid_info(p->one, options->repo->index);
6079                 diff_fill_oid_info(p->two, options->repo->index);
6080
6081                 len1 = remove_space(p->one->path, strlen(p->one->path));
6082                 len2 = remove_space(p->two->path, strlen(p->two->path));
6083                 patch_id_add_string(&ctx, "diff--git");
6084                 patch_id_add_string(&ctx, "a/");
6085                 git_SHA1_Update(&ctx, p->one->path, len1);
6086                 patch_id_add_string(&ctx, "b/");
6087                 git_SHA1_Update(&ctx, p->two->path, len2);
6088
6089                 if (p->one->mode == 0) {
6090                         patch_id_add_string(&ctx, "newfilemode");
6091                         patch_id_add_mode(&ctx, p->two->mode);
6092                         patch_id_add_string(&ctx, "---/dev/null");
6093                         patch_id_add_string(&ctx, "+++b/");
6094                         git_SHA1_Update(&ctx, p->two->path, len2);
6095                 } else if (p->two->mode == 0) {
6096                         patch_id_add_string(&ctx, "deletedfilemode");
6097                         patch_id_add_mode(&ctx, p->one->mode);
6098                         patch_id_add_string(&ctx, "---a/");
6099                         git_SHA1_Update(&ctx, p->one->path, len1);
6100                         patch_id_add_string(&ctx, "+++/dev/null");
6101                 } else {
6102                         patch_id_add_string(&ctx, "---a/");
6103                         git_SHA1_Update(&ctx, p->one->path, len1);
6104                         patch_id_add_string(&ctx, "+++b/");
6105                         git_SHA1_Update(&ctx, p->two->path, len2);
6106                 }
6107
6108                 if (diff_header_only)
6109                         continue;
6110
6111                 if (fill_mmfile(options->repo, &mf1, p->one) < 0 ||
6112                     fill_mmfile(options->repo, &mf2, p->two) < 0)
6113                         return error("unable to read files to diff");
6114
6115                 if (diff_filespec_is_binary(options->repo, p->one) ||
6116                     diff_filespec_is_binary(options->repo, p->two)) {
6117                         git_SHA1_Update(&ctx, oid_to_hex(&p->one->oid),
6118                                         GIT_SHA1_HEXSZ);
6119                         git_SHA1_Update(&ctx, oid_to_hex(&p->two->oid),
6120                                         GIT_SHA1_HEXSZ);
6121                         continue;
6122                 }
6123
6124                 xpp.flags = 0;
6125                 xecfg.ctxlen = 3;
6126                 xecfg.flags = 0;
6127                 if (xdi_diff_outf(&mf1, &mf2, discard_hunk_line,
6128                                   patch_id_consume, &data, &xpp, &xecfg))
6129                         return error("unable to generate patch-id diff for %s",
6130                                      p->one->path);
6131
6132                 if (stable)
6133                         flush_one_hunk(oid, &ctx);
6134         }
6135
6136         if (!stable)
6137                 git_SHA1_Final(oid->hash, &ctx);
6138
6139         return 0;
6140 }
6141
6142 int diff_flush_patch_id(struct diff_options *options, struct object_id *oid, int diff_header_only, int stable)
6143 {
6144         struct diff_queue_struct *q = &diff_queued_diff;
6145         int i;
6146         int result = diff_get_patch_id(options, oid, diff_header_only, stable);
6147
6148         for (i = 0; i < q->nr; i++)
6149                 diff_free_filepair(q->queue[i]);
6150
6151         free(q->queue);
6152         DIFF_QUEUE_CLEAR(q);
6153
6154         return result;
6155 }
6156
6157 static int is_summary_empty(const struct diff_queue_struct *q)
6158 {
6159         int i;
6160
6161         for (i = 0; i < q->nr; i++) {
6162                 const struct diff_filepair *p = q->queue[i];
6163
6164                 switch (p->status) {
6165                 case DIFF_STATUS_DELETED:
6166                 case DIFF_STATUS_ADDED:
6167                 case DIFF_STATUS_COPIED:
6168                 case DIFF_STATUS_RENAMED:
6169                         return 0;
6170                 default:
6171                         if (p->score)
6172                                 return 0;
6173                         if (p->one->mode && p->two->mode &&
6174                             p->one->mode != p->two->mode)
6175                                 return 0;
6176                         break;
6177                 }
6178         }
6179         return 1;
6180 }
6181
6182 static const char rename_limit_warning[] =
6183 N_("inexact rename detection was skipped due to too many files.");
6184
6185 static const char degrade_cc_to_c_warning[] =
6186 N_("only found copies from modified paths due to too many files.");
6187
6188 static const char rename_limit_advice[] =
6189 N_("you may want to set your %s variable to at least "
6190    "%d and retry the command.");
6191
6192 void diff_warn_rename_limit(const char *varname, int needed, int degraded_cc)
6193 {
6194         fflush(stdout);
6195         if (degraded_cc)
6196                 warning(_(degrade_cc_to_c_warning));
6197         else if (needed)
6198                 warning(_(rename_limit_warning));
6199         else
6200                 return;
6201         if (0 < needed)
6202                 warning(_(rename_limit_advice), varname, needed);
6203 }
6204
6205 static void diff_flush_patch_all_file_pairs(struct diff_options *o)
6206 {
6207         int i;
6208         static struct emitted_diff_symbols esm = EMITTED_DIFF_SYMBOLS_INIT;
6209         struct diff_queue_struct *q = &diff_queued_diff;
6210
6211         if (WSEH_NEW & WS_RULE_MASK)
6212                 BUG("WS rules bit mask overlaps with diff symbol flags");
6213
6214         if (o->color_moved)
6215                 o->emitted_symbols = &esm;
6216
6217         for (i = 0; i < q->nr; i++) {
6218                 struct diff_filepair *p = q->queue[i];
6219                 if (check_pair_status(p))
6220                         diff_flush_patch(p, o);
6221         }
6222
6223         if (o->emitted_symbols) {
6224                 if (o->color_moved) {
6225                         struct hashmap add_lines, del_lines;
6226
6227                         if (o->color_moved_ws_handling &
6228                             COLOR_MOVED_WS_ALLOW_INDENTATION_CHANGE)
6229                                 o->color_moved_ws_handling |= XDF_IGNORE_WHITESPACE;
6230
6231                         hashmap_init(&del_lines, moved_entry_cmp, o, 0);
6232                         hashmap_init(&add_lines, moved_entry_cmp, o, 0);
6233
6234                         add_lines_to_move_detection(o, &add_lines, &del_lines);
6235                         mark_color_as_moved(o, &add_lines, &del_lines);
6236                         if (o->color_moved == COLOR_MOVED_ZEBRA_DIM)
6237                                 dim_moved_lines(o);
6238
6239                         hashmap_free_entries(&add_lines, struct moved_entry,
6240                                                 ent);
6241                         hashmap_free_entries(&del_lines, struct moved_entry,
6242                                                 ent);
6243                 }
6244
6245                 for (i = 0; i < esm.nr; i++)
6246                         emit_diff_symbol_from_struct(o, &esm.buf[i]);
6247
6248                 for (i = 0; i < esm.nr; i++)
6249                         free((void *)esm.buf[i].line);
6250                 esm.nr = 0;
6251
6252                 o->emitted_symbols = NULL;
6253         }
6254 }
6255
6256 void diff_flush(struct diff_options *options)
6257 {
6258         struct diff_queue_struct *q = &diff_queued_diff;
6259         int i, output_format = options->output_format;
6260         int separator = 0;
6261         int dirstat_by_line = 0;
6262
6263         /*
6264          * Order: raw, stat, summary, patch
6265          * or:    name/name-status/checkdiff (other bits clear)
6266          */
6267         if (!q->nr)
6268                 goto free_queue;
6269
6270         if (output_format & (DIFF_FORMAT_RAW |
6271                              DIFF_FORMAT_NAME |
6272                              DIFF_FORMAT_NAME_STATUS |
6273                              DIFF_FORMAT_CHECKDIFF)) {
6274                 for (i = 0; i < q->nr; i++) {
6275                         struct diff_filepair *p = q->queue[i];
6276                         if (check_pair_status(p))
6277                                 flush_one_pair(p, options);
6278                 }
6279                 separator++;
6280         }
6281
6282         if (output_format & DIFF_FORMAT_DIRSTAT && options->flags.dirstat_by_line)
6283                 dirstat_by_line = 1;
6284
6285         if (output_format & (DIFF_FORMAT_DIFFSTAT|DIFF_FORMAT_SHORTSTAT|DIFF_FORMAT_NUMSTAT) ||
6286             dirstat_by_line) {
6287                 struct diffstat_t diffstat;
6288
6289                 memset(&diffstat, 0, sizeof(struct diffstat_t));
6290                 for (i = 0; i < q->nr; i++) {
6291                         struct diff_filepair *p = q->queue[i];
6292                         if (check_pair_status(p))
6293                                 diff_flush_stat(p, options, &diffstat);
6294                 }
6295                 if (output_format & DIFF_FORMAT_NUMSTAT)
6296                         show_numstat(&diffstat, options);
6297                 if (output_format & DIFF_FORMAT_DIFFSTAT)
6298                         show_stats(&diffstat, options);
6299                 if (output_format & DIFF_FORMAT_SHORTSTAT)
6300                         show_shortstats(&diffstat, options);
6301                 if (output_format & DIFF_FORMAT_DIRSTAT && dirstat_by_line)
6302                         show_dirstat_by_line(&diffstat, options);
6303                 free_diffstat_info(&diffstat);
6304                 separator++;
6305         }
6306         if ((output_format & DIFF_FORMAT_DIRSTAT) && !dirstat_by_line)
6307                 show_dirstat(options);
6308
6309         if (output_format & DIFF_FORMAT_SUMMARY && !is_summary_empty(q)) {
6310                 for (i = 0; i < q->nr; i++) {
6311                         diff_summary(options, q->queue[i]);
6312                 }
6313                 separator++;
6314         }
6315
6316         if (output_format & DIFF_FORMAT_NO_OUTPUT &&
6317             options->flags.exit_with_status &&
6318             options->flags.diff_from_contents) {
6319                 /*
6320                  * run diff_flush_patch for the exit status. setting
6321                  * options->file to /dev/null should be safe, because we
6322                  * aren't supposed to produce any output anyway.
6323                  */
6324                 if (options->close_file)
6325                         fclose(options->file);
6326                 options->file = xfopen("/dev/null", "w");
6327                 options->close_file = 1;
6328                 options->color_moved = 0;
6329                 for (i = 0; i < q->nr; i++) {
6330                         struct diff_filepair *p = q->queue[i];
6331                         if (check_pair_status(p))
6332                                 diff_flush_patch(p, options);
6333                         if (options->found_changes)
6334                                 break;
6335                 }
6336         }
6337
6338         if (output_format & DIFF_FORMAT_PATCH) {
6339                 if (separator) {
6340                         emit_diff_symbol(options, DIFF_SYMBOL_SEPARATOR, NULL, 0, 0);
6341                         if (options->stat_sep)
6342                                 /* attach patch instead of inline */
6343                                 emit_diff_symbol(options, DIFF_SYMBOL_STAT_SEP,
6344                                                  NULL, 0, 0);
6345                 }
6346
6347                 diff_flush_patch_all_file_pairs(options);
6348         }
6349
6350         if (output_format & DIFF_FORMAT_CALLBACK)
6351                 options->format_callback(q, options, options->format_callback_data);
6352
6353         for (i = 0; i < q->nr; i++)
6354                 diff_free_filepair(q->queue[i]);
6355 free_queue:
6356         free(q->queue);
6357         DIFF_QUEUE_CLEAR(q);
6358         if (options->close_file)
6359                 fclose(options->file);
6360
6361         /*
6362          * Report the content-level differences with HAS_CHANGES;
6363          * diff_addremove/diff_change does not set the bit when
6364          * DIFF_FROM_CONTENTS is in effect (e.g. with -w).
6365          */
6366         if (options->flags.diff_from_contents) {
6367                 if (options->found_changes)
6368                         options->flags.has_changes = 1;
6369                 else
6370                         options->flags.has_changes = 0;
6371         }
6372 }
6373
6374 static int match_filter(const struct diff_options *options, const struct diff_filepair *p)
6375 {
6376         return (((p->status == DIFF_STATUS_MODIFIED) &&
6377                  ((p->score &&
6378                    filter_bit_tst(DIFF_STATUS_FILTER_BROKEN, options)) ||
6379                   (!p->score &&
6380                    filter_bit_tst(DIFF_STATUS_MODIFIED, options)))) ||
6381                 ((p->status != DIFF_STATUS_MODIFIED) &&
6382                  filter_bit_tst(p->status, options)));
6383 }
6384
6385 static void diffcore_apply_filter(struct diff_options *options)
6386 {
6387         int i;
6388         struct diff_queue_struct *q = &diff_queued_diff;
6389         struct diff_queue_struct outq;
6390
6391         DIFF_QUEUE_CLEAR(&outq);
6392
6393         if (!options->filter)
6394                 return;
6395
6396         if (filter_bit_tst(DIFF_STATUS_FILTER_AON, options)) {
6397                 int found;
6398                 for (i = found = 0; !found && i < q->nr; i++) {
6399                         if (match_filter(options, q->queue[i]))
6400                                 found++;
6401                 }
6402                 if (found)
6403                         return;
6404
6405                 /* otherwise we will clear the whole queue
6406                  * by copying the empty outq at the end of this
6407                  * function, but first clear the current entries
6408                  * in the queue.
6409                  */
6410                 for (i = 0; i < q->nr; i++)
6411                         diff_free_filepair(q->queue[i]);
6412         }
6413         else {
6414                 /* Only the matching ones */
6415                 for (i = 0; i < q->nr; i++) {
6416                         struct diff_filepair *p = q->queue[i];
6417                         if (match_filter(options, p))
6418                                 diff_q(&outq, p);
6419                         else
6420                                 diff_free_filepair(p);
6421                 }
6422         }
6423         free(q->queue);
6424         *q = outq;
6425 }
6426
6427 /* Check whether two filespecs with the same mode and size are identical */
6428 static int diff_filespec_is_identical(struct repository *r,
6429                                       struct diff_filespec *one,
6430                                       struct diff_filespec *two)
6431 {
6432         if (S_ISGITLINK(one->mode))
6433                 return 0;
6434         if (diff_populate_filespec(r, one, 0))
6435                 return 0;
6436         if (diff_populate_filespec(r, two, 0))
6437                 return 0;
6438         return !memcmp(one->data, two->data, one->size);
6439 }
6440
6441 static int diff_filespec_check_stat_unmatch(struct repository *r,
6442                                             struct diff_filepair *p)
6443 {
6444         if (p->done_skip_stat_unmatch)
6445                 return p->skip_stat_unmatch_result;
6446
6447         p->done_skip_stat_unmatch = 1;
6448         p->skip_stat_unmatch_result = 0;
6449         /*
6450          * 1. Entries that come from stat info dirtiness
6451          *    always have both sides (iow, not create/delete),
6452          *    one side of the object name is unknown, with
6453          *    the same mode and size.  Keep the ones that
6454          *    do not match these criteria.  They have real
6455          *    differences.
6456          *
6457          * 2. At this point, the file is known to be modified,
6458          *    with the same mode and size, and the object
6459          *    name of one side is unknown.  Need to inspect
6460          *    the identical contents.
6461          */
6462         if (!DIFF_FILE_VALID(p->one) || /* (1) */
6463             !DIFF_FILE_VALID(p->two) ||
6464             (p->one->oid_valid && p->two->oid_valid) ||
6465             (p->one->mode != p->two->mode) ||
6466             diff_populate_filespec(r, p->one, CHECK_SIZE_ONLY) ||
6467             diff_populate_filespec(r, p->two, CHECK_SIZE_ONLY) ||
6468             (p->one->size != p->two->size) ||
6469             !diff_filespec_is_identical(r, p->one, p->two)) /* (2) */
6470                 p->skip_stat_unmatch_result = 1;
6471         return p->skip_stat_unmatch_result;
6472 }
6473
6474 static void diffcore_skip_stat_unmatch(struct diff_options *diffopt)
6475 {
6476         int i;
6477         struct diff_queue_struct *q = &diff_queued_diff;
6478         struct diff_queue_struct outq;
6479         DIFF_QUEUE_CLEAR(&outq);
6480
6481         for (i = 0; i < q->nr; i++) {
6482                 struct diff_filepair *p = q->queue[i];
6483
6484                 if (diff_filespec_check_stat_unmatch(diffopt->repo, p))
6485                         diff_q(&outq, p);
6486                 else {
6487                         /*
6488                          * The caller can subtract 1 from skip_stat_unmatch
6489                          * to determine how many paths were dirty only
6490                          * due to stat info mismatch.
6491                          */
6492                         if (!diffopt->flags.no_index)
6493                                 diffopt->skip_stat_unmatch++;
6494                         diff_free_filepair(p);
6495                 }
6496         }
6497         free(q->queue);
6498         *q = outq;
6499 }
6500
6501 static int diffnamecmp(const void *a_, const void *b_)
6502 {
6503         const struct diff_filepair *a = *((const struct diff_filepair **)a_);
6504         const struct diff_filepair *b = *((const struct diff_filepair **)b_);
6505         const char *name_a, *name_b;
6506
6507         name_a = a->one ? a->one->path : a->two->path;
6508         name_b = b->one ? b->one->path : b->two->path;
6509         return strcmp(name_a, name_b);
6510 }
6511
6512 void diffcore_fix_diff_index(void)
6513 {
6514         struct diff_queue_struct *q = &diff_queued_diff;
6515         QSORT(q->queue, q->nr, diffnamecmp);
6516 }
6517
6518 static void add_if_missing(struct repository *r,
6519                            struct oid_array *to_fetch,
6520                            const struct diff_filespec *filespec)
6521 {
6522         if (filespec && filespec->oid_valid &&
6523             oid_object_info_extended(r, &filespec->oid, NULL,
6524                                      OBJECT_INFO_FOR_PREFETCH))
6525                 oid_array_append(to_fetch, &filespec->oid);
6526 }
6527
6528 void diffcore_std(struct diff_options *options)
6529 {
6530         if (options->repo == the_repository &&
6531             repository_format_partial_clone) {
6532                 /*
6533                  * Prefetch the diff pairs that are about to be flushed.
6534                  */
6535                 int i;
6536                 struct diff_queue_struct *q = &diff_queued_diff;
6537                 struct oid_array to_fetch = OID_ARRAY_INIT;
6538
6539                 for (i = 0; i < q->nr; i++) {
6540                         struct diff_filepair *p = q->queue[i];
6541                         add_if_missing(options->repo, &to_fetch, p->one);
6542                         add_if_missing(options->repo, &to_fetch, p->two);
6543                 }
6544                 if (to_fetch.nr)
6545                         /*
6546                          * NEEDSWORK: Consider deduplicating the OIDs sent.
6547                          */
6548                         fetch_objects(repository_format_partial_clone,
6549                                       to_fetch.oid, to_fetch.nr);
6550                 oid_array_clear(&to_fetch);
6551         }
6552
6553         /* NOTE please keep the following in sync with diff_tree_combined() */
6554         if (options->skip_stat_unmatch)
6555                 diffcore_skip_stat_unmatch(options);
6556         if (!options->found_follow) {
6557                 /* See try_to_follow_renames() in tree-diff.c */
6558                 if (options->break_opt != -1)
6559                         diffcore_break(options->repo,
6560                                        options->break_opt);
6561                 if (options->detect_rename)
6562                         diffcore_rename(options);
6563                 if (options->break_opt != -1)
6564                         diffcore_merge_broken();
6565         }
6566         if (options->pickaxe_opts & DIFF_PICKAXE_KINDS_MASK)
6567                 diffcore_pickaxe(options);
6568         if (options->orderfile)
6569                 diffcore_order(options->orderfile);
6570         if (!options->found_follow)
6571                 /* See try_to_follow_renames() in tree-diff.c */
6572                 diff_resolve_rename_copy();
6573         diffcore_apply_filter(options);
6574
6575         if (diff_queued_diff.nr && !options->flags.diff_from_contents)
6576                 options->flags.has_changes = 1;
6577         else
6578                 options->flags.has_changes = 0;
6579
6580         options->found_follow = 0;
6581 }
6582
6583 int diff_result_code(struct diff_options *opt, int status)
6584 {
6585         int result = 0;
6586
6587         diff_warn_rename_limit("diff.renameLimit",
6588                                opt->needed_rename_limit,
6589                                opt->degraded_cc_to_c);
6590         if (!opt->flags.exit_with_status &&
6591             !(opt->output_format & DIFF_FORMAT_CHECKDIFF))
6592                 return status;
6593         if (opt->flags.exit_with_status &&
6594             opt->flags.has_changes)
6595                 result |= 01;
6596         if ((opt->output_format & DIFF_FORMAT_CHECKDIFF) &&
6597             opt->flags.check_failed)
6598                 result |= 02;
6599         return result;
6600 }
6601
6602 int diff_can_quit_early(struct diff_options *opt)
6603 {
6604         return (opt->flags.quick &&
6605                 !opt->filter &&
6606                 opt->flags.has_changes);
6607 }
6608
6609 /*
6610  * Shall changes to this submodule be ignored?
6611  *
6612  * Submodule changes can be configured to be ignored separately for each path,
6613  * but that configuration can be overridden from the command line.
6614  */
6615 static int is_submodule_ignored(const char *path, struct diff_options *options)
6616 {
6617         int ignored = 0;
6618         struct diff_flags orig_flags = options->flags;
6619         if (!options->flags.override_submodule_config)
6620                 set_diffopt_flags_from_submodule_config(options, path);
6621         if (options->flags.ignore_submodules)
6622                 ignored = 1;
6623         options->flags = orig_flags;
6624         return ignored;
6625 }
6626
6627 void diff_addremove(struct diff_options *options,
6628                     int addremove, unsigned mode,
6629                     const struct object_id *oid,
6630                     int oid_valid,
6631                     const char *concatpath, unsigned dirty_submodule)
6632 {
6633         struct diff_filespec *one, *two;
6634
6635         if (S_ISGITLINK(mode) && is_submodule_ignored(concatpath, options))
6636                 return;
6637
6638         /* This may look odd, but it is a preparation for
6639          * feeding "there are unchanged files which should
6640          * not produce diffs, but when you are doing copy
6641          * detection you would need them, so here they are"
6642          * entries to the diff-core.  They will be prefixed
6643          * with something like '=' or '*' (I haven't decided
6644          * which but should not make any difference).
6645          * Feeding the same new and old to diff_change()
6646          * also has the same effect.
6647          * Before the final output happens, they are pruned after
6648          * merged into rename/copy pairs as appropriate.
6649          */
6650         if (options->flags.reverse_diff)
6651                 addremove = (addremove == '+' ? '-' :
6652                              addremove == '-' ? '+' : addremove);
6653
6654         if (options->prefix &&
6655             strncmp(concatpath, options->prefix, options->prefix_length))
6656                 return;
6657
6658         one = alloc_filespec(concatpath);
6659         two = alloc_filespec(concatpath);
6660
6661         if (addremove != '+')
6662                 fill_filespec(one, oid, oid_valid, mode);
6663         if (addremove != '-') {
6664                 fill_filespec(two, oid, oid_valid, mode);
6665                 two->dirty_submodule = dirty_submodule;
6666         }
6667
6668         diff_queue(&diff_queued_diff, one, two);
6669         if (!options->flags.diff_from_contents)
6670                 options->flags.has_changes = 1;
6671 }
6672
6673 void diff_change(struct diff_options *options,
6674                  unsigned old_mode, unsigned new_mode,
6675                  const struct object_id *old_oid,
6676                  const struct object_id *new_oid,
6677                  int old_oid_valid, int new_oid_valid,
6678                  const char *concatpath,
6679                  unsigned old_dirty_submodule, unsigned new_dirty_submodule)
6680 {
6681         struct diff_filespec *one, *two;
6682         struct diff_filepair *p;
6683
6684         if (S_ISGITLINK(old_mode) && S_ISGITLINK(new_mode) &&
6685             is_submodule_ignored(concatpath, options))
6686                 return;
6687
6688         if (options->flags.reverse_diff) {
6689                 SWAP(old_mode, new_mode);
6690                 SWAP(old_oid, new_oid);
6691                 SWAP(old_oid_valid, new_oid_valid);
6692                 SWAP(old_dirty_submodule, new_dirty_submodule);
6693         }
6694
6695         if (options->prefix &&
6696             strncmp(concatpath, options->prefix, options->prefix_length))
6697                 return;
6698
6699         one = alloc_filespec(concatpath);
6700         two = alloc_filespec(concatpath);
6701         fill_filespec(one, old_oid, old_oid_valid, old_mode);
6702         fill_filespec(two, new_oid, new_oid_valid, new_mode);
6703         one->dirty_submodule = old_dirty_submodule;
6704         two->dirty_submodule = new_dirty_submodule;
6705         p = diff_queue(&diff_queued_diff, one, two);
6706
6707         if (options->flags.diff_from_contents)
6708                 return;
6709
6710         if (options->flags.quick && options->skip_stat_unmatch &&
6711             !diff_filespec_check_stat_unmatch(options->repo, p))
6712                 return;
6713
6714         options->flags.has_changes = 1;
6715 }
6716
6717 struct diff_filepair *diff_unmerge(struct diff_options *options, const char *path)
6718 {
6719         struct diff_filepair *pair;
6720         struct diff_filespec *one, *two;
6721
6722         if (options->prefix &&
6723             strncmp(path, options->prefix, options->prefix_length))
6724                 return NULL;
6725
6726         one = alloc_filespec(path);
6727         two = alloc_filespec(path);
6728         pair = diff_queue(&diff_queued_diff, one, two);
6729         pair->is_unmerged = 1;
6730         return pair;
6731 }
6732
6733 static char *run_textconv(struct repository *r,
6734                           const char *pgm,
6735                           struct diff_filespec *spec,
6736                           size_t *outsize)
6737 {
6738         struct diff_tempfile *temp;
6739         const char *argv[3];
6740         const char **arg = argv;
6741         struct child_process child = CHILD_PROCESS_INIT;
6742         struct strbuf buf = STRBUF_INIT;
6743         int err = 0;
6744
6745         temp = prepare_temp_file(r, spec->path, spec);
6746         *arg++ = pgm;
6747         *arg++ = temp->name;
6748         *arg = NULL;
6749
6750         child.use_shell = 1;
6751         child.argv = argv;
6752         child.out = -1;
6753         if (start_command(&child)) {
6754                 remove_tempfile();
6755                 return NULL;
6756         }
6757
6758         if (strbuf_read(&buf, child.out, 0) < 0)
6759                 err = error("error reading from textconv command '%s'", pgm);
6760         close(child.out);
6761
6762         if (finish_command(&child) || err) {
6763                 strbuf_release(&buf);
6764                 remove_tempfile();
6765                 return NULL;
6766         }
6767         remove_tempfile();
6768
6769         return strbuf_detach(&buf, outsize);
6770 }
6771
6772 size_t fill_textconv(struct repository *r,
6773                      struct userdiff_driver *driver,
6774                      struct diff_filespec *df,
6775                      char **outbuf)
6776 {
6777         size_t size;
6778
6779         if (!driver) {
6780                 if (!DIFF_FILE_VALID(df)) {
6781                         *outbuf = "";
6782                         return 0;
6783                 }
6784                 if (diff_populate_filespec(r, df, 0))
6785                         die("unable to read files to diff");
6786                 *outbuf = df->data;
6787                 return df->size;
6788         }
6789
6790         if (!driver->textconv)
6791                 BUG("fill_textconv called with non-textconv driver");
6792
6793         if (driver->textconv_cache && df->oid_valid) {
6794                 *outbuf = notes_cache_get(driver->textconv_cache,
6795                                           &df->oid,
6796                                           &size);
6797                 if (*outbuf)
6798                         return size;
6799         }
6800
6801         *outbuf = run_textconv(r, driver->textconv, df, &size);
6802         if (!*outbuf)
6803                 die("unable to read files to diff");
6804
6805         if (driver->textconv_cache && df->oid_valid) {
6806                 /* ignore errors, as we might be in a readonly repository */
6807                 notes_cache_put(driver->textconv_cache, &df->oid, *outbuf,
6808                                 size);
6809                 /*
6810                  * we could save up changes and flush them all at the end,
6811                  * but we would need an extra call after all diffing is done.
6812                  * Since generating a cache entry is the slow path anyway,
6813                  * this extra overhead probably isn't a big deal.
6814                  */
6815                 notes_cache_write(driver->textconv_cache);
6816         }
6817
6818         return size;
6819 }
6820
6821 int textconv_object(struct repository *r,
6822                     const char *path,
6823                     unsigned mode,
6824                     const struct object_id *oid,
6825                     int oid_valid,
6826                     char **buf,
6827                     unsigned long *buf_size)
6828 {
6829         struct diff_filespec *df;
6830         struct userdiff_driver *textconv;
6831
6832         df = alloc_filespec(path);
6833         fill_filespec(df, oid, oid_valid, mode);
6834         textconv = get_textconv(r, df);
6835         if (!textconv) {
6836                 free_filespec(df);
6837                 return 0;
6838         }
6839
6840         *buf_size = fill_textconv(r, textconv, df, buf);
6841         free_filespec(df);
6842         return 1;
6843 }
6844
6845 void setup_diff_pager(struct diff_options *opt)
6846 {
6847         /*
6848          * If the user asked for our exit code, then either they want --quiet
6849          * or --exit-code. We should definitely not bother with a pager in the
6850          * former case, as we will generate no output. Since we still properly
6851          * report our exit code even when a pager is run, we _could_ run a
6852          * pager with --exit-code. But since we have not done so historically,
6853          * and because it is easy to find people oneline advising "git diff
6854          * --exit-code" in hooks and other scripts, we do not do so.
6855          */
6856         if (!opt->flags.exit_with_status &&
6857             check_pager_config("diff") != 0)
6858                 setup_pager();
6859 }