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