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