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

   1 /*
   2  * Copyright (C) 2005 Junio C Hamano
   3  */
   4 #include "cache.h"
   5 #include "tempfile.h"
   6 #include "quote.h"
   7 #include "diff.h"
   8 #include "diffcore.h"
   9 #include "delta.h"
  10 #include "xdiff-interface.h"
  11 #include "color.h"
  12 #include "attr.h"
  13 #include "run-command.h"
  14 #include "utf8.h"
  15 #include "userdiff.h"
  16 #include "sigchain.h"
  17 #include "submodule-config.h"
  18 #include "submodule.h"
  19 #include "ll-merge.h"
  20 #include "string-list.h"
  21 #include "argv-array.h"
  22
  23 #ifdef NO_FAST_WORKING_DIRECTORY
  24 #define FAST_WORKING_DIRECTORY 0
  25 #else
  26 #define FAST_WORKING_DIRECTORY 1
  27 #endif
  28
  29 static int diff_detect_rename_default;
  30 static int diff_rename_limit_default = 400;
  31 static int diff_suppress_blank_empty;
  32 static int diff_use_color_default = -1;
  33 static int diff_context_default = 3;
  34 static const char *diff_word_regex_cfg;
  35 static const char *external_diff_cmd_cfg;
  36 static const char *diff_order_file_cfg;
  37 int diff_auto_refresh_index = 1;
  38 static int diff_mnemonic_prefix;
  39 static int diff_no_prefix;
  40 static int diff_stat_graph_width;
  41 static int diff_dirstat_permille_default = 30;
  42 static struct diff_options default_diff_options;
  43 static long diff_algorithm;
  44 static unsigned ws_error_highlight_default = WSEH_NEW;
  45
  46 static char diff_colors[][COLOR_MAXLEN] = {
  47         GIT_COLOR_RESET,
  48         GIT_COLOR_NORMAL,       /* CONTEXT */
  49         GIT_COLOR_BOLD,         /* METAINFO */
  50         GIT_COLOR_CYAN,         /* FRAGINFO */
  51         GIT_COLOR_RED,          /* OLD */
  52         GIT_COLOR_GREEN,        /* NEW */
  53         GIT_COLOR_YELLOW,       /* COMMIT */
  54         GIT_COLOR_BG_RED,       /* WHITESPACE */
  55         GIT_COLOR_NORMAL,       /* FUNCINFO */
  56 };
  57
  58 static int parse_diff_color_slot(const char *var)
  59 {
  60         if (!strcasecmp(var, "context") || !strcasecmp(var, "plain"))
  61                 return DIFF_CONTEXT;
  62         if (!strcasecmp(var, "meta"))
  63                 return DIFF_METAINFO;
  64         if (!strcasecmp(var, "frag"))
  65                 return DIFF_FRAGINFO;
  66         if (!strcasecmp(var, "old"))
  67                 return DIFF_FILE_OLD;
  68         if (!strcasecmp(var, "new"))
  69                 return DIFF_FILE_NEW;
  70         if (!strcasecmp(var, "commit"))
  71                 return DIFF_COMMIT;
  72         if (!strcasecmp(var, "whitespace"))
  73                 return DIFF_WHITESPACE;
  74         if (!strcasecmp(var, "func"))
  75                 return DIFF_FUNCINFO;
  76         return -1;
  77 }
  78
  79 static int parse_dirstat_params(struct diff_options *options, const char *params_string,
  80                                 struct strbuf *errmsg)
  81 {
  82         char *params_copy = xstrdup(params_string);
  83         struct string_list params = STRING_LIST_INIT_NODUP;
  84         int ret = 0;
  85         int i;
  86
  87         if (*params_copy)
  88                 string_list_split_in_place(&params, params_copy, ',', -1);
  89         for (i = 0; i < params.nr; i++) {
  90                 const char *p = params.items[i].string;
  91                 if (!strcmp(p, "changes")) {
  92                         DIFF_OPT_CLR(options, DIRSTAT_BY_LINE);
  93                         DIFF_OPT_CLR(options, DIRSTAT_BY_FILE);
  94                 } else if (!strcmp(p, "lines")) {
  95                         DIFF_OPT_SET(options, DIRSTAT_BY_LINE);
  96                         DIFF_OPT_CLR(options, DIRSTAT_BY_FILE);
  97                 } else if (!strcmp(p, "files")) {
  98                         DIFF_OPT_CLR(options, DIRSTAT_BY_LINE);
  99                         DIFF_OPT_SET(options, DIRSTAT_BY_FILE);
 100                 } else if (!strcmp(p, "noncumulative")) {
 101                         DIFF_OPT_CLR(options, DIRSTAT_CUMULATIVE);
 102                 } else if (!strcmp(p, "cumulative")) {
 103                         DIFF_OPT_SET(options, DIRSTAT_CUMULATIVE);
 104                 } else if (isdigit(*p)) {
 105                         char *end;
 106                         int permille = strtoul(p, &end, 10) * 10;
 107                         if (*end == '.' && isdigit(*++end)) {
 108                                 /* only use first digit */
 109                                 permille += *end - '0';
 110                                 /* .. and ignore any further digits */
 111                                 while (isdigit(*++end))
 112                                         ; /* nothing */
 113                         }
 114                         if (!*end)
 115                                 options->dirstat_permille = permille;
 116                         else {
 117                                 strbuf_addf(errmsg, _("  Failed to parse dirstat cut-off percentage '%s'\n"),
 118                                             p);
 119                                 ret++;
 120                         }
 121                 } else {
 122                         strbuf_addf(errmsg, _("  Unknown dirstat parameter '%s'\n"), p);
 123                         ret++;
 124                 }
 125
 126         }
 127         string_list_clear(&params, 0);
 128         free(params_copy);
 129         return ret;
 130 }
 131
 132 static int parse_submodule_params(struct diff_options *options, const char *value)
 133 {
 134         if (!strcmp(value, "log"))
 135                 DIFF_OPT_SET(options, SUBMODULE_LOG);
 136         else if (!strcmp(value, "short"))
 137                 DIFF_OPT_CLR(options, SUBMODULE_LOG);
 138         else
 139                 return -1;
 140         return 0;
 141 }
 142
 143 static int git_config_rename(const char *var, const char *value)
 144 {
 145         if (!value)
 146                 return DIFF_DETECT_RENAME;
 147         if (!strcasecmp(value, "copies") || !strcasecmp(value, "copy"))
 148                 return  DIFF_DETECT_COPY;
 149         return git_config_bool(var,value) ? DIFF_DETECT_RENAME : 0;
 150 }
 151
 152 long parse_algorithm_value(const char *value)
 153 {
 154         if (!value)
 155                 return -1;
 156         else if (!strcasecmp(value, "myers") || !strcasecmp(value, "default"))
 157                 return 0;
 158         else if (!strcasecmp(value, "minimal"))
 159                 return XDF_NEED_MINIMAL;
 160         else if (!strcasecmp(value, "patience"))
 161                 return XDF_PATIENCE_DIFF;
 162         else if (!strcasecmp(value, "histogram"))
 163                 return XDF_HISTOGRAM_DIFF;
 164         return -1;
 165 }
 166
 167 static int parse_one_token(const char **arg, const char *token)
 168 {
 169         const char *rest;
 170         if (skip_prefix(*arg, token, &rest) && (!*rest || *rest == ',')) {
 171                 *arg = rest;
 172                 return 1;
 173         }
 174         return 0;
 175 }
 176
 177 static int parse_ws_error_highlight(const char *arg)
 178 {
 179         const char *orig_arg = arg;
 180         unsigned val = 0;
 181
 182         while (*arg) {
 183                 if (parse_one_token(&arg, "none"))
 184                         val = 0;
 185                 else if (parse_one_token(&arg, "default"))
 186                         val = WSEH_NEW;
 187                 else if (parse_one_token(&arg, "all"))
 188                         val = WSEH_NEW | WSEH_OLD | WSEH_CONTEXT;
 189                 else if (parse_one_token(&arg, "new"))
 190                         val |= WSEH_NEW;
 191                 else if (parse_one_token(&arg, "old"))
 192                         val |= WSEH_OLD;
 193                 else if (parse_one_token(&arg, "context"))
 194                         val |= WSEH_CONTEXT;
 195                 else {
 196                         return -1 - (int)(arg - orig_arg);
 197                 }
 198                 if (*arg)
 199                         arg++;
 200         }
 201         return val;
 202 }
 203
 204 /*
 205  * These are to give UI layer defaults.
 206  * The core-level commands such as git-diff-files should
 207  * never be affected by the setting of diff.renames
 208  * the user happens to have in the configuration file.
 209  */
 210 int git_diff_ui_config(const char *var, const char *value, void *cb)
 211 {
 212         if (!strcmp(var, "diff.color") || !strcmp(var, "color.diff")) {
 213                 diff_use_color_default = git_config_colorbool(var, value);
 214                 return 0;
 215         }
 216         if (!strcmp(var, "diff.context")) {
 217                 diff_context_default = git_config_int(var, value);
 218                 if (diff_context_default < 0)
 219                         return -1;
 220                 return 0;
 221         }
 222         if (!strcmp(var, "diff.renames")) {
 223                 diff_detect_rename_default = git_config_rename(var, value);
 224                 return 0;
 225         }
 226         if (!strcmp(var, "diff.autorefreshindex")) {
 227                 diff_auto_refresh_index = git_config_bool(var, value);
 228                 return 0;
 229         }
 230         if (!strcmp(var, "diff.mnemonicprefix")) {
 231                 diff_mnemonic_prefix = git_config_bool(var, value);
 232                 return 0;
 233         }
 234         if (!strcmp(var, "diff.noprefix")) {
 235                 diff_no_prefix = git_config_bool(var, value);
 236                 return 0;
 237         }
 238         if (!strcmp(var, "diff.statgraphwidth")) {
 239                 diff_stat_graph_width = git_config_int(var, value);
 240                 return 0;
 241         }
 242         if (!strcmp(var, "diff.external"))
 243                 return git_config_string(&external_diff_cmd_cfg, var, value);
 244         if (!strcmp(var, "diff.wordregex"))
 245                 return git_config_string(&diff_word_regex_cfg, var, value);
 246         if (!strcmp(var, "diff.orderfile"))
 247                 return git_config_pathname(&diff_order_file_cfg, var, value);
 248
 249         if (!strcmp(var, "diff.ignoresubmodules"))
 250                 handle_ignore_submodules_arg(&default_diff_options, value);
 251
 252         if (!strcmp(var, "diff.submodule")) {
 253                 if (parse_submodule_params(&default_diff_options, value))
 254                         warning(_("Unknown value for 'diff.submodule' config variable: '%s'"),
 255                                 value);
 256                 return 0;
 257         }
 258
 259         if (!strcmp(var, "diff.algorithm")) {
 260                 diff_algorithm = parse_algorithm_value(value);
 261                 if (diff_algorithm < 0)
 262                         return -1;
 263                 return 0;
 264         }
 265
 266         if (!strcmp(var, "diff.wserrorhighlight")) {
 267                 int val = parse_ws_error_highlight(value);
 268                 if (val < 0)
 269                         return -1;
 270                 ws_error_highlight_default = val;
 271                 return 0;
 272         }
 273
 274         if (git_color_config(var, value, cb) < 0)
 275                 return -1;
 276
 277         return git_diff_basic_config(var, value, cb);
 278 }
 279
 280 int git_diff_basic_config(const char *var, const char *value, void *cb)
 281 {
 282         const char *name;
 283
 284         if (!strcmp(var, "diff.renamelimit")) {
 285                 diff_rename_limit_default = git_config_int(var, value);
 286                 return 0;
 287         }
 288
 289         if (userdiff_config(var, value) < 0)
 290                 return -1;
 291
 292         if (skip_prefix(var, "diff.color.", &name) ||
 293             skip_prefix(var, "color.diff.", &name)) {
 294                 int slot = parse_diff_color_slot(name);
 295                 if (slot < 0)
 296                         return 0;
 297                 if (!value)
 298                         return config_error_nonbool(var);
 299                 return color_parse(value, diff_colors[slot]);
 300         }
 301
 302         /* like GNU diff's --suppress-blank-empty option  */
 303         if (!strcmp(var, "diff.suppressblankempty") ||
 304                         /* for backwards compatibility */
 305                         !strcmp(var, "diff.suppress-blank-empty")) {
 306                 diff_suppress_blank_empty = git_config_bool(var, value);
 307                 return 0;
 308         }
 309
 310         if (!strcmp(var, "diff.dirstat")) {
 311                 struct strbuf errmsg = STRBUF_INIT;
 312                 default_diff_options.dirstat_permille = diff_dirstat_permille_default;
 313                 if (parse_dirstat_params(&default_diff_options, value, &errmsg))
 314                         warning(_("Found errors in 'diff.dirstat' config variable:\n%s"),
 315                                 errmsg.buf);
 316                 strbuf_release(&errmsg);
 317                 diff_dirstat_permille_default = default_diff_options.dirstat_permille;
 318                 return 0;
 319         }
 320
 321         if (starts_with(var, "submodule."))
 322                 return parse_submodule_config_option(var, value);
 323
 324         return git_default_config(var, value, cb);
 325 }
 326
 327 static char *quote_two(const char *one, const char *two)
 328 {
 329         int need_one = quote_c_style(one, NULL, NULL, 1);
 330         int need_two = quote_c_style(two, NULL, NULL, 1);
 331         struct strbuf res = STRBUF_INIT;
 332
 333         if (need_one + need_two) {
 334                 strbuf_addch(&res, '"');
 335                 quote_c_style(one, &res, NULL, 1);
 336                 quote_c_style(two, &res, NULL, 1);
 337                 strbuf_addch(&res, '"');
 338         } else {
 339                 strbuf_addstr(&res, one);
 340                 strbuf_addstr(&res, two);
 341         }
 342         return strbuf_detach(&res, NULL);
 343 }
 344
 345 static const char *external_diff(void)
 346 {
 347         static const char *external_diff_cmd = NULL;
 348         static int done_preparing = 0;
 349
 350         if (done_preparing)
 351                 return external_diff_cmd;
 352         external_diff_cmd = getenv("GIT_EXTERNAL_DIFF");
 353         if (!external_diff_cmd)
 354                 external_diff_cmd = external_diff_cmd_cfg;
 355         done_preparing = 1;
 356         return external_diff_cmd;
 357 }
 358
 359 /*
 360  * Keep track of files used for diffing. Sometimes such an entry
 361  * refers to a temporary file, sometimes to an existing file, and
 362  * sometimes to "/dev/null".
 363  */
 364 static struct diff_tempfile {
 365         /*
 366          * filename external diff should read from, or NULL if this
 367          * entry is currently not in use:
 368          */
 369         const char *name;
 370
 371         char hex[41];
 372         char mode[10];
 373
 374         /*
 375          * If this diff_tempfile instance refers to a temporary file,
 376          * this tempfile object is used to manage its lifetime.
 377          */
 378         struct tempfile tempfile;
 379 } diff_temp[2];
 380
 381 typedef unsigned long (*sane_truncate_fn)(char *line, unsigned long len);
 382
 383 struct emit_callback {
 384         int color_diff;
 385         unsigned ws_rule;
 386         int blank_at_eof_in_preimage;
 387         int blank_at_eof_in_postimage;
 388         int lno_in_preimage;
 389         int lno_in_postimage;
 390         sane_truncate_fn truncate;
 391         const char **label_path;
 392         struct diff_words_data *diff_words;
 393         struct diff_options *opt;
 394         int *found_changesp;
 395         struct strbuf *header;
 396 };
 397
 398 static int count_lines(const char *data, int size)
 399 {
 400         int count, ch, completely_empty = 1, nl_just_seen = 0;
 401         count = 0;
 402         while (0 < size--) {
 403                 ch = *data++;
 404                 if (ch == '\n') {
 405                         count++;
 406                         nl_just_seen = 1;
 407                         completely_empty = 0;
 408                 }
 409                 else {
 410                         nl_just_seen = 0;
 411                         completely_empty = 0;
 412                 }
 413         }
 414         if (completely_empty)
 415                 return 0;
 416         if (!nl_just_seen)
 417                 count++; /* no trailing newline */
 418         return count;
 419 }
 420
 421 static int fill_mmfile(mmfile_t *mf, struct diff_filespec *one)
 422 {
 423         if (!DIFF_FILE_VALID(one)) {
 424                 mf->ptr = (char *)""; /* does not matter */
 425                 mf->size = 0;
 426                 return 0;
 427         }
 428         else if (diff_populate_filespec(one, 0))
 429                 return -1;
 430
 431         mf->ptr = one->data;
 432         mf->size = one->size;
 433         return 0;
 434 }
 435
 436 /* like fill_mmfile, but only for size, so we can avoid retrieving blob */
 437 static unsigned long diff_filespec_size(struct diff_filespec *one)
 438 {
 439         if (!DIFF_FILE_VALID(one))
 440                 return 0;
 441         diff_populate_filespec(one, CHECK_SIZE_ONLY);
 442         return one->size;
 443 }
 444
 445 static int count_trailing_blank(mmfile_t *mf, unsigned ws_rule)
 446 {
 447         char *ptr = mf->ptr;
 448         long size = mf->size;
 449         int cnt = 0;
 450
 451         if (!size)
 452                 return cnt;
 453         ptr += size - 1; /* pointing at the very end */
 454         if (*ptr != '\n')
 455                 ; /* incomplete line */
 456         else
 457                 ptr--; /* skip the last LF */
 458         while (mf->ptr < ptr) {
 459                 char *prev_eol;
 460                 for (prev_eol = ptr; mf->ptr <= prev_eol; prev_eol--)
 461                         if (*prev_eol == '\n')
 462                                 break;
 463                 if (!ws_blank_line(prev_eol + 1, ptr - prev_eol, ws_rule))
 464                         break;
 465                 cnt++;
 466                 ptr = prev_eol - 1;
 467         }
 468         return cnt;
 469 }
 470
 471 static void check_blank_at_eof(mmfile_t *mf1, mmfile_t *mf2,
 472                                struct emit_callback *ecbdata)
 473 {
 474         int l1, l2, at;
 475         unsigned ws_rule = ecbdata->ws_rule;
 476         l1 = count_trailing_blank(mf1, ws_rule);
 477         l2 = count_trailing_blank(mf2, ws_rule);
 478         if (l2 <= l1) {
 479                 ecbdata->blank_at_eof_in_preimage = 0;
 480                 ecbdata->blank_at_eof_in_postimage = 0;
 481                 return;
 482         }
 483         at = count_lines(mf1->ptr, mf1->size);
 484         ecbdata->blank_at_eof_in_preimage = (at - l1) + 1;
 485
 486         at = count_lines(mf2->ptr, mf2->size);
 487         ecbdata->blank_at_eof_in_postimage = (at - l2) + 1;
 488 }
 489
 490 static void emit_line_0(struct diff_options *o, const char *set, const char *reset,
 491                         int first, const char *line, int len)
 492 {
 493         int has_trailing_newline, has_trailing_carriage_return;
 494         int nofirst;
 495         FILE *file = o->file;
 496
 497         fputs(diff_line_prefix(o), file);
 498
 499         if (len == 0) {
 500                 has_trailing_newline = (first == '\n');
 501                 has_trailing_carriage_return = (!has_trailing_newline &&
 502                                                 (first == '\r'));
 503                 nofirst = has_trailing_newline || has_trailing_carriage_return;
 504         } else {
 505                 has_trailing_newline = (len > 0 && line[len-1] == '\n');
 506                 if (has_trailing_newline)
 507                         len--;
 508                 has_trailing_carriage_return = (len > 0 && line[len-1] == '\r');
 509                 if (has_trailing_carriage_return)
 510                         len--;
 511                 nofirst = 0;
 512         }
 513
 514         if (len || !nofirst) {
 515                 fputs(set, file);
 516                 if (!nofirst)
 517                         fputc(first, file);
 518                 fwrite(line, len, 1, file);
 519                 fputs(reset, file);
 520         }
 521         if (has_trailing_carriage_return)
 522                 fputc('\r', file);
 523         if (has_trailing_newline)
 524                 fputc('\n', file);
 525 }
 526
 527 static void emit_line(struct diff_options *o, const char *set, const char *reset,
 528                       const char *line, int len)
 529 {
 530         emit_line_0(o, set, reset, line[0], line+1, len-1);
 531 }
 532
 533 static int new_blank_line_at_eof(struct emit_callback *ecbdata, const char *line, int len)
 534 {
 535         if (!((ecbdata->ws_rule & WS_BLANK_AT_EOF) &&
 536               ecbdata->blank_at_eof_in_preimage &&
 537               ecbdata->blank_at_eof_in_postimage &&
 538               ecbdata->blank_at_eof_in_preimage <= ecbdata->lno_in_preimage &&
 539               ecbdata->blank_at_eof_in_postimage <= ecbdata->lno_in_postimage))
 540                 return 0;
 541         return ws_blank_line(line, len, ecbdata->ws_rule);
 542 }
 543
 544 static void emit_line_checked(const char *reset,
 545                               struct emit_callback *ecbdata,
 546                               const char *line, int len,
 547                               enum color_diff color,
 548                               unsigned ws_error_highlight,
 549                               char sign)
 550 {
 551         const char *set = diff_get_color(ecbdata->color_diff, color);
 552         const char *ws = NULL;
 553
 554         if (ecbdata->opt->ws_error_highlight & ws_error_highlight) {
 555                 ws = diff_get_color(ecbdata->color_diff, DIFF_WHITESPACE);
 556                 if (!*ws)
 557                         ws = NULL;
 558         }
 559
 560         if (!ws)
 561                 emit_line_0(ecbdata->opt, set, reset, sign, line, len);
 562         else if (sign == '+' && new_blank_line_at_eof(ecbdata, line, len))
 563                 /* Blank line at EOF - paint '+' as well */
 564                 emit_line_0(ecbdata->opt, ws, reset, sign, line, len);
 565         else {
 566                 /* Emit just the prefix, then the rest. */
 567                 emit_line_0(ecbdata->opt, set, reset, sign, "", 0);
 568                 ws_check_emit(line, len, ecbdata->ws_rule,
 569                               ecbdata->opt->file, set, reset, ws);
 570         }
 571 }
 572
 573 static void emit_add_line(const char *reset,
 574                           struct emit_callback *ecbdata,
 575                           const char *line, int len)
 576 {
 577         emit_line_checked(reset, ecbdata, line, len,
 578                           DIFF_FILE_NEW, WSEH_NEW, '+');
 579 }
 580
 581 static void emit_del_line(const char *reset,
 582                           struct emit_callback *ecbdata,
 583                           const char *line, int len)
 584 {
 585         emit_line_checked(reset, ecbdata, line, len,
 586                           DIFF_FILE_OLD, WSEH_OLD, '-');
 587 }
 588
 589 static void emit_context_line(const char *reset,
 590                               struct emit_callback *ecbdata,
 591                               const char *line, int len)
 592 {
 593         emit_line_checked(reset, ecbdata, line, len,
 594                           DIFF_CONTEXT, WSEH_CONTEXT, ' ');
 595 }
 596
 597 static void emit_hunk_header(struct emit_callback *ecbdata,
 598                              const char *line, int len)
 599 {
 600         const char *context = diff_get_color(ecbdata->color_diff, DIFF_CONTEXT);
 601         const char *frag = diff_get_color(ecbdata->color_diff, DIFF_FRAGINFO);
 602         const char *func = diff_get_color(ecbdata->color_diff, DIFF_FUNCINFO);
 603         const char *reset = diff_get_color(ecbdata->color_diff, DIFF_RESET);
 604         static const char atat[2] = { '@', '@' };
 605         const char *cp, *ep;
 606         struct strbuf msgbuf = STRBUF_INIT;
 607         int org_len = len;
 608         int i = 1;
 609
 610         /*
 611          * As a hunk header must begin with "@@ -<old>, +<new> @@",
 612          * it always is at least 10 bytes long.
 613          */
 614         if (len < 10 ||
 615             memcmp(line, atat, 2) ||
 616             !(ep = memmem(line + 2, len - 2, atat, 2))) {
 617                 emit_line(ecbdata->opt, context, reset, line, len);
 618                 return;
 619         }
 620         ep += 2; /* skip over @@ */
 621
 622         /* The hunk header in fraginfo color */
 623         strbuf_addstr(&msgbuf, frag);
 624         strbuf_add(&msgbuf, line, ep - line);
 625         strbuf_addstr(&msgbuf, reset);
 626
 627         /*
 628          * trailing "\r\n"
 629          */
 630         for ( ; i < 3; i++)
 631                 if (line[len - i] == '\r' || line[len - i] == '\n')
 632                         len--;
 633
 634         /* blank before the func header */
 635         for (cp = ep; ep - line < len; ep++)
 636                 if (*ep != ' ' && *ep != '\t')
 637                         break;
 638         if (ep != cp) {
 639                 strbuf_addstr(&msgbuf, context);
 640                 strbuf_add(&msgbuf, cp, ep - cp);
 641                 strbuf_addstr(&msgbuf, reset);
 642         }
 643
 644         if (ep < line + len) {
 645                 strbuf_addstr(&msgbuf, func);
 646                 strbuf_add(&msgbuf, ep, line + len - ep);
 647                 strbuf_addstr(&msgbuf, reset);
 648         }
 649
 650         strbuf_add(&msgbuf, line + len, org_len - len);
 651         emit_line(ecbdata->opt, "", "", msgbuf.buf, msgbuf.len);
 652         strbuf_release(&msgbuf);
 653 }
 654
 655 static struct diff_tempfile *claim_diff_tempfile(void) {
 656         int i;
 657         for (i = 0; i < ARRAY_SIZE(diff_temp); i++)
 658                 if (!diff_temp[i].name)
 659                         return diff_temp + i;
 660         die("BUG: diff is failing to clean up its tempfiles");
 661 }
 662
 663 static void remove_tempfile(void)
 664 {
 665         int i;
 666         for (i = 0; i < ARRAY_SIZE(diff_temp); i++) {
 667                 if (is_tempfile_active(&diff_temp[i].tempfile))
 668                         delete_tempfile(&diff_temp[i].tempfile);
 669                 diff_temp[i].name = NULL;
 670         }
 671 }
 672
 673 static void print_line_count(FILE *file, int count)
 674 {
 675         switch (count) {
 676         case 0:
 677                 fprintf(file, "0,0");
 678                 break;
 679         case 1:
 680                 fprintf(file, "1");
 681                 break;
 682         default:
 683                 fprintf(file, "1,%d", count);
 684                 break;
 685         }
 686 }
 687
 688 static void emit_rewrite_lines(struct emit_callback *ecb,
 689                                int prefix, const char *data, int size)
 690 {
 691         const char *endp = NULL;
 692         static const char *nneof = " No newline at end of file\n";
 693         const char *reset = diff_get_color(ecb->color_diff, DIFF_RESET);
 694
 695         while (0 < size) {
 696                 int len;
 697
 698                 endp = memchr(data, '\n', size);
 699                 len = endp ? (endp - data + 1) : size;
 700                 if (prefix != '+') {
 701                         ecb->lno_in_preimage++;
 702                         emit_del_line(reset, ecb, data, len);
 703                 } else {
 704                         ecb->lno_in_postimage++;
 705                         emit_add_line(reset, ecb, data, len);
 706                 }
 707                 size -= len;
 708                 data += len;
 709         }
 710         if (!endp) {
 711                 const char *context = diff_get_color(ecb->color_diff,
 712                                                      DIFF_CONTEXT);
 713                 putc('\n', ecb->opt->file);
 714                 emit_line_0(ecb->opt, context, reset, '\\',
 715                             nneof, strlen(nneof));
 716         }
 717 }
 718
 719 static void emit_rewrite_diff(const char *name_a,
 720                               const char *name_b,
 721                               struct diff_filespec *one,
 722                               struct diff_filespec *two,
 723                               struct userdiff_driver *textconv_one,
 724                               struct userdiff_driver *textconv_two,
 725                               struct diff_options *o)
 726 {
 727         int lc_a, lc_b;
 728         const char *name_a_tab, *name_b_tab;
 729         const char *metainfo = diff_get_color(o->use_color, DIFF_METAINFO);
 730         const char *fraginfo = diff_get_color(o->use_color, DIFF_FRAGINFO);
 731         const char *reset = diff_get_color(o->use_color, DIFF_RESET);
 732         static struct strbuf a_name = STRBUF_INIT, b_name = STRBUF_INIT;
 733         const char *a_prefix, *b_prefix;
 734         char *data_one, *data_two;
 735         size_t size_one, size_two;
 736         struct emit_callback ecbdata;
 737         const char *line_prefix = diff_line_prefix(o);
 738
 739         if (diff_mnemonic_prefix && DIFF_OPT_TST(o, REVERSE_DIFF)) {
 740                 a_prefix = o->b_prefix;
 741                 b_prefix = o->a_prefix;
 742         } else {
 743                 a_prefix = o->a_prefix;
 744                 b_prefix = o->b_prefix;
 745         }
 746
 747         name_a += (*name_a == '/');
 748         name_b += (*name_b == '/');
 749         name_a_tab = strchr(name_a, ' ') ? "\t" : "";
 750         name_b_tab = strchr(name_b, ' ') ? "\t" : "";
 751
 752         strbuf_reset(&a_name);
 753         strbuf_reset(&b_name);
 754         quote_two_c_style(&a_name, a_prefix, name_a, 0);
 755         quote_two_c_style(&b_name, b_prefix, name_b, 0);
 756
 757         size_one = fill_textconv(textconv_one, one, &data_one);
 758         size_two = fill_textconv(textconv_two, two, &data_two);
 759
 760         memset(&ecbdata, 0, sizeof(ecbdata));
 761         ecbdata.color_diff = want_color(o->use_color);
 762         ecbdata.found_changesp = &o->found_changes;
 763         ecbdata.ws_rule = whitespace_rule(name_b);
 764         ecbdata.opt = o;
 765         if (ecbdata.ws_rule & WS_BLANK_AT_EOF) {
 766                 mmfile_t mf1, mf2;
 767                 mf1.ptr = (char *)data_one;
 768                 mf2.ptr = (char *)data_two;
 769                 mf1.size = size_one;
 770                 mf2.size = size_two;
 771                 check_blank_at_eof(&mf1, &mf2, &ecbdata);
 772         }
 773         ecbdata.lno_in_preimage = 1;
 774         ecbdata.lno_in_postimage = 1;
 775
 776         lc_a = count_lines(data_one, size_one);
 777         lc_b = count_lines(data_two, size_two);
 778         fprintf(o->file,
 779                 "%s%s--- %s%s%s\n%s%s+++ %s%s%s\n%s%s@@ -",
 780                 line_prefix, metainfo, a_name.buf, name_a_tab, reset,
 781                 line_prefix, metainfo, b_name.buf, name_b_tab, reset,
 782                 line_prefix, fraginfo);
 783         if (!o->irreversible_delete)
 784                 print_line_count(o->file, lc_a);
 785         else
 786                 fprintf(o->file, "?,?");
 787         fprintf(o->file, " +");
 788         print_line_count(o->file, lc_b);
 789         fprintf(o->file, " @@%s\n", reset);
 790         if (lc_a && !o->irreversible_delete)
 791                 emit_rewrite_lines(&ecbdata, '-', data_one, size_one);
 792         if (lc_b)
 793                 emit_rewrite_lines(&ecbdata, '+', data_two, size_two);
 794         if (textconv_one)
 795                 free((char *)data_one);
 796         if (textconv_two)
 797                 free((char *)data_two);
 798 }
 799
 800 struct diff_words_buffer {
 801         mmfile_t text;
 802         long alloc;
 803         struct diff_words_orig {
 804                 const char *begin, *end;
 805         } *orig;
 806         int orig_nr, orig_alloc;
 807 };
 808
 809 static void diff_words_append(char *line, unsigned long len,
 810                 struct diff_words_buffer *buffer)
 811 {
 812         ALLOC_GROW(buffer->text.ptr, buffer->text.size + len, buffer->alloc);
 813         line++;
 814         len--;
 815         memcpy(buffer->text.ptr + buffer->text.size, line, len);
 816         buffer->text.size += len;
 817         buffer->text.ptr[buffer->text.size] = '\0';
 818 }
 819
 820 struct diff_words_style_elem {
 821         const char *prefix;
 822         const char *suffix;
 823         const char *color; /* NULL; filled in by the setup code if
 824                             * color is enabled */
 825 };
 826
 827 struct diff_words_style {
 828         enum diff_words_type type;
 829         struct diff_words_style_elem new, old, ctx;
 830         const char *newline;
 831 };
 832
 833 static struct diff_words_style diff_words_styles[] = {
 834         { DIFF_WORDS_PORCELAIN, {"+", "\n"}, {"-", "\n"}, {" ", "\n"}, "~\n" },
 835         { DIFF_WORDS_PLAIN, {"{+", "+}"}, {"[-", "-]"}, {"", ""}, "\n" },
 836         { DIFF_WORDS_COLOR, {"", ""}, {"", ""}, {"", ""}, "\n" }
 837 };
 838
 839 struct diff_words_data {
 840         struct diff_words_buffer minus, plus;
 841         const char *current_plus;
 842         int last_minus;
 843         struct diff_options *opt;
 844         regex_t *word_regex;
 845         enum diff_words_type type;
 846         struct diff_words_style *style;
 847 };
 848
 849 static int fn_out_diff_words_write_helper(FILE *fp,
 850                                           struct diff_words_style_elem *st_el,
 851                                           const char *newline,
 852                                           size_t count, const char *buf,
 853                                           const char *line_prefix)
 854 {
 855         int print = 0;
 856
 857         while (count) {
 858                 char *p = memchr(buf, '\n', count);
 859                 if (print)
 860                         fputs(line_prefix, fp);
 861                 if (p != buf) {
 862                         if (st_el->color && fputs(st_el->color, fp) < 0)
 863                                 return -1;
 864                         if (fputs(st_el->prefix, fp) < 0 ||
 865                             fwrite(buf, p ? p - buf : count, 1, fp) != 1 ||
 866                             fputs(st_el->suffix, fp) < 0)
 867                                 return -1;
 868                         if (st_el->color && *st_el->color
 869                             && fputs(GIT_COLOR_RESET, fp) < 0)
 870                                 return -1;
 871                 }
 872                 if (!p)
 873                         return 0;
 874                 if (fputs(newline, fp) < 0)
 875                         return -1;
 876                 count -= p + 1 - buf;
 877                 buf = p + 1;
 878                 print = 1;
 879         }
 880         return 0;
 881 }
 882
 883 /*
 884  * '--color-words' algorithm can be described as:
 885  *
 886  *   1. collect a the minus/plus lines of a diff hunk, divided into
 887  *      minus-lines and plus-lines;
 888  *
 889  *   2. break both minus-lines and plus-lines into words and
 890  *      place them into two mmfile_t with one word for each line;
 891  *
 892  *   3. use xdiff to run diff on the two mmfile_t to get the words level diff;
 893  *
 894  * And for the common parts of the both file, we output the plus side text.
 895  * diff_words->current_plus is used to trace the current position of the plus file
 896  * which printed. diff_words->last_minus is used to trace the last minus word
 897  * printed.
 898  *
 899  * For '--graph' to work with '--color-words', we need to output the graph prefix
 900  * on each line of color words output. Generally, there are two conditions on
 901  * which we should output the prefix.
 902  *
 903  *   1. diff_words->last_minus == 0 &&
 904  *      diff_words->current_plus == diff_words->plus.text.ptr
 905  *
 906  *      that is: the plus text must start as a new line, and if there is no minus
 907  *      word printed, a graph prefix must be printed.
 908  *
 909  *   2. diff_words->current_plus > diff_words->plus.text.ptr &&
 910  *      *(diff_words->current_plus - 1) == '\n'
 911  *
 912  *      that is: a graph prefix must be printed following a '\n'
 913  */
 914 static int color_words_output_graph_prefix(struct diff_words_data *diff_words)
 915 {
 916         if ((diff_words->last_minus == 0 &&
 917                 diff_words->current_plus == diff_words->plus.text.ptr) ||
 918                 (diff_words->current_plus > diff_words->plus.text.ptr &&
 919                 *(diff_words->current_plus - 1) == '\n')) {
 920                 return 1;
 921         } else {
 922                 return 0;
 923         }
 924 }
 925
 926 static void fn_out_diff_words_aux(void *priv, char *line, unsigned long len)
 927 {
 928         struct diff_words_data *diff_words = priv;
 929         struct diff_words_style *style = diff_words->style;
 930         int minus_first, minus_len, plus_first, plus_len;
 931         const char *minus_begin, *minus_end, *plus_begin, *plus_end;
 932         struct diff_options *opt = diff_words->opt;
 933         const char *line_prefix;
 934
 935         if (line[0] != '@' || parse_hunk_header(line, len,
 936                         &minus_first, &minus_len, &plus_first, &plus_len))
 937                 return;
 938
 939         assert(opt);
 940         line_prefix = diff_line_prefix(opt);
 941
 942         /* POSIX requires that first be decremented by one if len == 0... */
 943         if (minus_len) {
 944                 minus_begin = diff_words->minus.orig[minus_first].begin;
 945                 minus_end =
 946                         diff_words->minus.orig[minus_first + minus_len - 1].end;
 947         } else
 948                 minus_begin = minus_end =
 949                         diff_words->minus.orig[minus_first].end;
 950
 951         if (plus_len) {
 952                 plus_begin = diff_words->plus.orig[plus_first].begin;
 953                 plus_end = diff_words->plus.orig[plus_first + plus_len - 1].end;
 954         } else
 955                 plus_begin = plus_end = diff_words->plus.orig[plus_first].end;
 956
 957         if (color_words_output_graph_prefix(diff_words)) {
 958                 fputs(line_prefix, diff_words->opt->file);
 959         }
 960         if (diff_words->current_plus != plus_begin) {
 961                 fn_out_diff_words_write_helper(diff_words->opt->file,
 962                                 &style->ctx, style->newline,
 963                                 plus_begin - diff_words->current_plus,
 964                                 diff_words->current_plus, line_prefix);
 965                 if (*(plus_begin - 1) == '\n')
 966                         fputs(line_prefix, diff_words->opt->file);
 967         }
 968         if (minus_begin != minus_end) {
 969                 fn_out_diff_words_write_helper(diff_words->opt->file,
 970                                 &style->old, style->newline,
 971                                 minus_end - minus_begin, minus_begin,
 972                                 line_prefix);
 973         }
 974         if (plus_begin != plus_end) {
 975                 fn_out_diff_words_write_helper(diff_words->opt->file,
 976                                 &style->new, style->newline,
 977                                 plus_end - plus_begin, plus_begin,
 978                                 line_prefix);
 979         }
 980
 981         diff_words->current_plus = plus_end;
 982         diff_words->last_minus = minus_first;
 983 }
 984
 985 /* This function starts looking at *begin, and returns 0 iff a word was found. */
 986 static int find_word_boundaries(mmfile_t *buffer, regex_t *word_regex,
 987                 int *begin, int *end)
 988 {
 989         if (word_regex && *begin < buffer->size) {
 990                 regmatch_t match[1];
 991                 if (!regexec(word_regex, buffer->ptr + *begin, 1, match, 0)) {
 992                         char *p = memchr(buffer->ptr + *begin + match[0].rm_so,
 993                                         '\n', match[0].rm_eo - match[0].rm_so);
 994                         *end = p ? p - buffer->ptr : match[0].rm_eo + *begin;
 995                         *begin += match[0].rm_so;
 996                         return *begin >= *end;
 997                 }
 998                 return -1;
 999         }
1000
1001         /* find the next word */
1002         while (*begin < buffer->size && isspace(buffer->ptr[*begin]))
1003                 (*begin)++;
1004         if (*begin >= buffer->size)
1005                 return -1;
1006
1007         /* find the end of the word */
1008         *end = *begin + 1;
1009         while (*end < buffer->size && !isspace(buffer->ptr[*end]))
1010                 (*end)++;
1011
1012         return 0;
1013 }
1014
1015 /*
1016  * This function splits the words in buffer->text, stores the list with
1017  * newline separator into out, and saves the offsets of the original words
1018  * in buffer->orig.
1019  */
1020 static void diff_words_fill(struct diff_words_buffer *buffer, mmfile_t *out,
1021                 regex_t *word_regex)
1022 {
1023         int i, j;
1024         long alloc = 0;
1025
1026         out->size = 0;
1027         out->ptr = NULL;
1028
1029         /* fake an empty "0th" word */
1030         ALLOC_GROW(buffer->orig, 1, buffer->orig_alloc);
1031         buffer->orig[0].begin = buffer->orig[0].end = buffer->text.ptr;
1032         buffer->orig_nr = 1;
1033
1034         for (i = 0; i < buffer->text.size; i++) {
1035                 if (find_word_boundaries(&buffer->text, word_regex, &i, &j))
1036                         return;
1037
1038                 /* store original boundaries */
1039                 ALLOC_GROW(buffer->orig, buffer->orig_nr + 1,
1040                                 buffer->orig_alloc);
1041                 buffer->orig[buffer->orig_nr].begin = buffer->text.ptr + i;
1042                 buffer->orig[buffer->orig_nr].end = buffer->text.ptr + j;
1043                 buffer->orig_nr++;
1044
1045                 /* store one word */
1046                 ALLOC_GROW(out->ptr, out->size + j - i + 1, alloc);
1047                 memcpy(out->ptr + out->size, buffer->text.ptr + i, j - i);
1048                 out->ptr[out->size + j - i] = '\n';
1049                 out->size += j - i + 1;
1050
1051                 i = j - 1;
1052         }
1053 }
1054
1055 /* this executes the word diff on the accumulated buffers */
1056 static void diff_words_show(struct diff_words_data *diff_words)
1057 {
1058         xpparam_t xpp;
1059         xdemitconf_t xecfg;
1060         mmfile_t minus, plus;
1061         struct diff_words_style *style = diff_words->style;
1062
1063         struct diff_options *opt = diff_words->opt;
1064         const char *line_prefix;
1065
1066         assert(opt);
1067         line_prefix = diff_line_prefix(opt);
1068
1069         /* special case: only removal */
1070         if (!diff_words->plus.text.size) {
1071                 fputs(line_prefix, diff_words->opt->file);
1072                 fn_out_diff_words_write_helper(diff_words->opt->file,
1073                         &style->old, style->newline,
1074                         diff_words->minus.text.size,
1075                         diff_words->minus.text.ptr, line_prefix);
1076                 diff_words->minus.text.size = 0;
1077                 return;
1078         }
1079
1080         diff_words->current_plus = diff_words->plus.text.ptr;
1081         diff_words->last_minus = 0;
1082
1083         memset(&xpp, 0, sizeof(xpp));
1084         memset(&xecfg, 0, sizeof(xecfg));
1085         diff_words_fill(&diff_words->minus, &minus, diff_words->word_regex);
1086         diff_words_fill(&diff_words->plus, &plus, diff_words->word_regex);
1087         xpp.flags = 0;
1088         /* as only the hunk header will be parsed, we need a 0-context */
1089         xecfg.ctxlen = 0;
1090         if (xdi_diff_outf(&minus, &plus, fn_out_diff_words_aux, diff_words,
1091                           &xpp, &xecfg))
1092                 die("unable to generate word diff");
1093         free(minus.ptr);
1094         free(plus.ptr);
1095         if (diff_words->current_plus != diff_words->plus.text.ptr +
1096                         diff_words->plus.text.size) {
1097                 if (color_words_output_graph_prefix(diff_words))
1098                         fputs(line_prefix, diff_words->opt->file);
1099                 fn_out_diff_words_write_helper(diff_words->opt->file,
1100                         &style->ctx, style->newline,
1101                         diff_words->plus.text.ptr + diff_words->plus.text.size
1102                         - diff_words->current_plus, diff_words->current_plus,
1103                         line_prefix);
1104         }
1105         diff_words->minus.text.size = diff_words->plus.text.size = 0;
1106 }
1107
1108 /* In "color-words" mode, show word-diff of words accumulated in the buffer */
1109 static void diff_words_flush(struct emit_callback *ecbdata)
1110 {
1111         if (ecbdata->diff_words->minus.text.size ||
1112             ecbdata->diff_words->plus.text.size)
1113                 diff_words_show(ecbdata->diff_words);
1114 }
1115
1116 static void diff_filespec_load_driver(struct diff_filespec *one)
1117 {
1118         /* Use already-loaded driver */
1119         if (one->driver)
1120                 return;
1121
1122         if (S_ISREG(one->mode))
1123                 one->driver = userdiff_find_by_path(one->path);
1124
1125         /* Fallback to default settings */
1126         if (!one->driver)
1127                 one->driver = userdiff_find_by_name("default");
1128 }
1129
1130 static const char *userdiff_word_regex(struct diff_filespec *one)
1131 {
1132         diff_filespec_load_driver(one);
1133         return one->driver->word_regex;
1134 }
1135
1136 static void init_diff_words_data(struct emit_callback *ecbdata,
1137                                  struct diff_options *orig_opts,
1138                                  struct diff_filespec *one,
1139                                  struct diff_filespec *two)
1140 {
1141         int i;
1142         struct diff_options *o = xmalloc(sizeof(struct diff_options));
1143         memcpy(o, orig_opts, sizeof(struct diff_options));
1144
1145         ecbdata->diff_words =
1146                 xcalloc(1, sizeof(struct diff_words_data));
1147         ecbdata->diff_words->type = o->word_diff;
1148         ecbdata->diff_words->opt = o;
1149         if (!o->word_regex)
1150                 o->word_regex = userdiff_word_regex(one);
1151         if (!o->word_regex)
1152                 o->word_regex = userdiff_word_regex(two);
1153         if (!o->word_regex)
1154                 o->word_regex = diff_word_regex_cfg;
1155         if (o->word_regex) {
1156                 ecbdata->diff_words->word_regex = (regex_t *)
1157                         xmalloc(sizeof(regex_t));
1158                 if (regcomp(ecbdata->diff_words->word_regex,
1159                             o->word_regex,
1160                             REG_EXTENDED | REG_NEWLINE))
1161                         die ("Invalid regular expression: %s",
1162                              o->word_regex);
1163         }
1164         for (i = 0; i < ARRAY_SIZE(diff_words_styles); i++) {
1165                 if (o->word_diff == diff_words_styles[i].type) {
1166                         ecbdata->diff_words->style =
1167                                 &diff_words_styles[i];
1168                         break;
1169                 }
1170         }
1171         if (want_color(o->use_color)) {
1172                 struct diff_words_style *st = ecbdata->diff_words->style;
1173                 st->old.color = diff_get_color_opt(o, DIFF_FILE_OLD);
1174                 st->new.color = diff_get_color_opt(o, DIFF_FILE_NEW);
1175                 st->ctx.color = diff_get_color_opt(o, DIFF_CONTEXT);
1176         }
1177 }
1178
1179 static void free_diff_words_data(struct emit_callback *ecbdata)
1180 {
1181         if (ecbdata->diff_words) {
1182                 diff_words_flush(ecbdata);
1183                 free (ecbdata->diff_words->opt);
1184                 free (ecbdata->diff_words->minus.text.ptr);
1185                 free (ecbdata->diff_words->minus.orig);
1186                 free (ecbdata->diff_words->plus.text.ptr);
1187                 free (ecbdata->diff_words->plus.orig);
1188                 if (ecbdata->diff_words->word_regex) {
1189                         regfree(ecbdata->diff_words->word_regex);
1190                         free(ecbdata->diff_words->word_regex);
1191                 }
1192                 free(ecbdata->diff_words);
1193                 ecbdata->diff_words = NULL;
1194         }
1195 }
1196
1197 const char *diff_get_color(int diff_use_color, enum color_diff ix)
1198 {
1199         if (want_color(diff_use_color))
1200                 return diff_colors[ix];
1201         return "";
1202 }
1203
1204 const char *diff_line_prefix(struct diff_options *opt)
1205 {
1206         struct strbuf *msgbuf;
1207         if (!opt->output_prefix)
1208                 return "";
1209
1210         msgbuf = opt->output_prefix(opt, opt->output_prefix_data);
1211         return msgbuf->buf;
1212 }
1213
1214 static unsigned long sane_truncate_line(struct emit_callback *ecb, char *line, unsigned long len)
1215 {
1216         const char *cp;
1217         unsigned long allot;
1218         size_t l = len;
1219
1220         if (ecb->truncate)
1221                 return ecb->truncate(line, len);
1222         cp = line;
1223         allot = l;
1224         while (0 < l) {
1225                 (void) utf8_width(&cp, &l);
1226                 if (!cp)
1227                         break; /* truncated in the middle? */
1228         }
1229         return allot - l;
1230 }
1231
1232 static void find_lno(const char *line, struct emit_callback *ecbdata)
1233 {
1234         const char *p;
1235         ecbdata->lno_in_preimage = 0;
1236         ecbdata->lno_in_postimage = 0;
1237         p = strchr(line, '-');
1238         if (!p)
1239                 return; /* cannot happen */
1240         ecbdata->lno_in_preimage = strtol(p + 1, NULL, 10);
1241         p = strchr(p, '+');
1242         if (!p)
1243                 return; /* cannot happen */
1244         ecbdata->lno_in_postimage = strtol(p + 1, NULL, 10);
1245 }
1246
1247 static void fn_out_consume(void *priv, char *line, unsigned long len)
1248 {
1249         struct emit_callback *ecbdata = priv;
1250         const char *meta = diff_get_color(ecbdata->color_diff, DIFF_METAINFO);
1251         const char *context = diff_get_color(ecbdata->color_diff, DIFF_CONTEXT);
1252         const char *reset = diff_get_color(ecbdata->color_diff, DIFF_RESET);
1253         struct diff_options *o = ecbdata->opt;
1254         const char *line_prefix = diff_line_prefix(o);
1255
1256         if (ecbdata->header) {
1257                 fprintf(ecbdata->opt->file, "%s", ecbdata->header->buf);
1258                 strbuf_reset(ecbdata->header);
1259                 ecbdata->header = NULL;
1260         }
1261         *(ecbdata->found_changesp) = 1;
1262
1263         if (ecbdata->label_path[0]) {
1264                 const char *name_a_tab, *name_b_tab;
1265
1266                 name_a_tab = strchr(ecbdata->label_path[0], ' ') ? "\t" : "";
1267                 name_b_tab = strchr(ecbdata->label_path[1], ' ') ? "\t" : "";
1268
1269                 fprintf(ecbdata->opt->file, "%s%s--- %s%s%s\n",
1270                         line_prefix, meta, ecbdata->label_path[0], reset, name_a_tab);
1271                 fprintf(ecbdata->opt->file, "%s%s+++ %s%s%s\n",
1272                         line_prefix, meta, ecbdata->label_path[1], reset, name_b_tab);
1273                 ecbdata->label_path[0] = ecbdata->label_path[1] = NULL;
1274         }
1275
1276         if (diff_suppress_blank_empty
1277             && len == 2 && line[0] == ' ' && line[1] == '\n') {
1278                 line[0] = '\n';
1279                 len = 1;
1280         }
1281
1282         if (line[0] == '@') {
1283                 if (ecbdata->diff_words)
1284                         diff_words_flush(ecbdata);
1285                 len = sane_truncate_line(ecbdata, line, len);
1286                 find_lno(line, ecbdata);
1287                 emit_hunk_header(ecbdata, line, len);
1288                 if (line[len-1] != '\n')
1289                         putc('\n', ecbdata->opt->file);
1290                 return;
1291         }
1292
1293         if (len < 1) {
1294                 emit_line(ecbdata->opt, reset, reset, line, len);
1295                 if (ecbdata->diff_words
1296                     && ecbdata->diff_words->type == DIFF_WORDS_PORCELAIN)
1297                         fputs("~\n", ecbdata->opt->file);
1298                 return;
1299         }
1300
1301         if (ecbdata->diff_words) {
1302                 if (line[0] == '-') {
1303                         diff_words_append(line, len,
1304                                           &ecbdata->diff_words->minus);
1305                         return;
1306                 } else if (line[0] == '+') {
1307                         diff_words_append(line, len,
1308                                           &ecbdata->diff_words->plus);
1309                         return;
1310                 } else if (starts_with(line, "\\ ")) {
1311                         /*
1312                          * Eat the "no newline at eof" marker as if we
1313                          * saw a "+" or "-" line with nothing on it,
1314                          * and return without diff_words_flush() to
1315                          * defer processing. If this is the end of
1316                          * preimage, more "+" lines may come after it.
1317                          */
1318                         return;
1319                 }
1320                 diff_words_flush(ecbdata);
1321                 if (ecbdata->diff_words->type == DIFF_WORDS_PORCELAIN) {
1322                         emit_line(ecbdata->opt, context, reset, line, len);
1323                         fputs("~\n", ecbdata->opt->file);
1324                 } else {
1325                         /*
1326                          * Skip the prefix character, if any.  With
1327                          * diff_suppress_blank_empty, there may be
1328                          * none.
1329                          */
1330                         if (line[0] != '\n') {
1331                               line++;
1332                               len--;
1333                         }
1334                         emit_line(ecbdata->opt, context, reset, line, len);
1335                 }
1336                 return;
1337         }
1338
1339         switch (line[0]) {
1340         case '+':
1341                 ecbdata->lno_in_postimage++;
1342                 emit_add_line(reset, ecbdata, line + 1, len - 1);
1343                 break;
1344         case '-':
1345                 ecbdata->lno_in_preimage++;
1346                 emit_del_line(reset, ecbdata, line + 1, len - 1);
1347                 break;
1348         case ' ':
1349                 ecbdata->lno_in_postimage++;
1350                 ecbdata->lno_in_preimage++;
1351                 emit_context_line(reset, ecbdata, line + 1, len - 1);
1352                 break;
1353         default:
1354                 /* incomplete line at the end */
1355                 ecbdata->lno_in_preimage++;
1356                 emit_line(ecbdata->opt,
1357                           diff_get_color(ecbdata->color_diff, DIFF_CONTEXT),
1358                           reset, line, len);
1359                 break;
1360         }
1361 }
1362
1363 static char *pprint_rename(const char *a, const char *b)
1364 {
1365         const char *old = a;
1366         const char *new = b;
1367         struct strbuf name = STRBUF_INIT;
1368         int pfx_length, sfx_length;
1369         int pfx_adjust_for_slash;
1370         int len_a = strlen(a);
1371         int len_b = strlen(b);
1372         int a_midlen, b_midlen;
1373         int qlen_a = quote_c_style(a, NULL, NULL, 0);
1374         int qlen_b = quote_c_style(b, NULL, NULL, 0);
1375
1376         if (qlen_a || qlen_b) {
1377                 quote_c_style(a, &name, NULL, 0);
1378                 strbuf_addstr(&name, " => ");
1379                 quote_c_style(b, &name, NULL, 0);
1380                 return strbuf_detach(&name, NULL);
1381         }
1382
1383         /* Find common prefix */
1384         pfx_length = 0;
1385         while (*old && *new && *old == *new) {
1386                 if (*old == '/')
1387                         pfx_length = old - a + 1;
1388                 old++;
1389                 new++;
1390         }
1391
1392         /* Find common suffix */
1393         old = a + len_a;
1394         new = b + len_b;
1395         sfx_length = 0;
1396         /*
1397          * If there is a common prefix, it must end in a slash.  In
1398          * that case we let this loop run 1 into the prefix to see the
1399          * same slash.
1400          *
1401          * If there is no common prefix, we cannot do this as it would
1402          * underrun the input strings.
1403          */
1404         pfx_adjust_for_slash = (pfx_length ? 1 : 0);
1405         while (a + pfx_length - pfx_adjust_for_slash <= old &&
1406                b + pfx_length - pfx_adjust_for_slash <= new &&
1407                *old == *new) {
1408                 if (*old == '/')
1409                         sfx_length = len_a - (old - a);
1410                 old--;
1411                 new--;
1412         }
1413
1414         /*
1415          * pfx{mid-a => mid-b}sfx
1416          * {pfx-a => pfx-b}sfx
1417          * pfx{sfx-a => sfx-b}
1418          * name-a => name-b
1419          */
1420         a_midlen = len_a - pfx_length - sfx_length;
1421         b_midlen = len_b - pfx_length - sfx_length;
1422         if (a_midlen < 0)
1423                 a_midlen = 0;
1424         if (b_midlen < 0)
1425                 b_midlen = 0;
1426
1427         strbuf_grow(&name, pfx_length + a_midlen + b_midlen + sfx_length + 7);
1428         if (pfx_length + sfx_length) {
1429                 strbuf_add(&name, a, pfx_length);
1430                 strbuf_addch(&name, '{');
1431         }
1432         strbuf_add(&name, a + pfx_length, a_midlen);
1433         strbuf_addstr(&name, " => ");
1434         strbuf_add(&name, b + pfx_length, b_midlen);
1435         if (pfx_length + sfx_length) {
1436                 strbuf_addch(&name, '}');
1437                 strbuf_add(&name, a + len_a - sfx_length, sfx_length);
1438         }
1439         return strbuf_detach(&name, NULL);
1440 }
1441
1442 struct diffstat_t {
1443         int nr;
1444         int alloc;
1445         struct diffstat_file {
1446                 char *from_name;
1447                 char *name;
1448                 char *print_name;
1449                 unsigned is_unmerged:1;
1450                 unsigned is_binary:1;
1451                 unsigned is_renamed:1;
1452                 unsigned is_interesting:1;
1453                 uintmax_t added, deleted;
1454         } **files;
1455 };
1456
1457 static struct diffstat_file *diffstat_add(struct diffstat_t *diffstat,
1458                                           const char *name_a,
1459                                           const char *name_b)
1460 {
1461         struct diffstat_file *x;
1462         x = xcalloc(1, sizeof(*x));
1463         ALLOC_GROW(diffstat->files, diffstat->nr + 1, diffstat->alloc);
1464         diffstat->files[diffstat->nr++] = x;
1465         if (name_b) {
1466                 x->from_name = xstrdup(name_a);
1467                 x->name = xstrdup(name_b);
1468                 x->is_renamed = 1;
1469         }
1470         else {
1471                 x->from_name = NULL;
1472                 x->name = xstrdup(name_a);
1473         }
1474         return x;
1475 }
1476
1477 static void diffstat_consume(void *priv, char *line, unsigned long len)
1478 {
1479         struct diffstat_t *diffstat = priv;
1480         struct diffstat_file *x = diffstat->files[diffstat->nr - 1];
1481
1482         if (line[0] == '+')
1483                 x->added++;
1484         else if (line[0] == '-')
1485                 x->deleted++;
1486 }
1487
1488 const char mime_boundary_leader[] = "------------";
1489
1490 static int scale_linear(int it, int width, int max_change)
1491 {
1492         if (!it)
1493                 return 0;
1494         /*
1495          * make sure that at least one '-' or '+' is printed if
1496          * there is any change to this path. The easiest way is to
1497          * scale linearly as if the alloted width is one column shorter
1498          * than it is, and then add 1 to the result.
1499          */
1500         return 1 + (it * (width - 1) / max_change);
1501 }
1502
1503 static void show_name(FILE *file,
1504                       const char *prefix, const char *name, int len)
1505 {
1506         fprintf(file, " %s%-*s |", prefix, len, name);
1507 }
1508
1509 static void show_graph(FILE *file, char ch, int cnt, const char *set, const char *reset)
1510 {
1511         if (cnt <= 0)
1512                 return;
1513         fprintf(file, "%s", set);
1514         while (cnt--)
1515                 putc(ch, file);
1516         fprintf(file, "%s", reset);
1517 }
1518
1519 static void fill_print_name(struct diffstat_file *file)
1520 {
1521         char *pname;
1522
1523         if (file->print_name)
1524                 return;
1525
1526         if (!file->is_renamed) {
1527                 struct strbuf buf = STRBUF_INIT;
1528                 if (quote_c_style(file->name, &buf, NULL, 0)) {
1529                         pname = strbuf_detach(&buf, NULL);
1530                 } else {
1531                         pname = file->name;
1532                         strbuf_release(&buf);
1533                 }
1534         } else {
1535                 pname = pprint_rename(file->from_name, file->name);
1536         }
1537         file->print_name = pname;
1538 }
1539
1540 int print_stat_summary(FILE *fp, int files, int insertions, int deletions)
1541 {
1542         struct strbuf sb = STRBUF_INIT;
1543         int ret;
1544
1545         if (!files) {
1546                 assert(insertions == 0 && deletions == 0);
1547                 return fprintf(fp, "%s\n", " 0 files changed");
1548         }
1549
1550         strbuf_addf(&sb,
1551                     (files == 1) ? " %d file changed" : " %d files changed",
1552                     files);
1553
1554         /*
1555          * For binary diff, the caller may want to print "x files
1556          * changed" with insertions == 0 && deletions == 0.
1557          *
1558          * Not omitting "0 insertions(+), 0 deletions(-)" in this case
1559          * is probably less confusing (i.e skip over "2 files changed
1560          * but nothing about added/removed lines? Is this a bug in Git?").
1561          */
1562         if (insertions || deletions == 0) {
1563                 strbuf_addf(&sb,
1564                             (insertions == 1) ? ", %d insertion(+)" : ", %d insertions(+)",
1565                             insertions);
1566         }
1567
1568         if (deletions || insertions == 0) {
1569                 strbuf_addf(&sb,
1570                             (deletions == 1) ? ", %d deletion(-)" : ", %d deletions(-)",
1571                             deletions);
1572         }
1573         strbuf_addch(&sb, '\n');
1574         ret = fputs(sb.buf, fp);
1575         strbuf_release(&sb);
1576         return ret;
1577 }
1578
1579 static void show_stats(struct diffstat_t *data, struct diff_options *options)
1580 {
1581         int i, len, add, del, adds = 0, dels = 0;
1582         uintmax_t max_change = 0, max_len = 0;
1583         int total_files = data->nr, count;
1584         int width, name_width, graph_width, number_width = 0, bin_width = 0;
1585         const char *reset, *add_c, *del_c;
1586         const char *line_prefix = "";
1587         int extra_shown = 0;
1588
1589         if (data->nr == 0)
1590                 return;
1591
1592         line_prefix = diff_line_prefix(options);
1593         count = options->stat_count ? options->stat_count : data->nr;
1594
1595         reset = diff_get_color_opt(options, DIFF_RESET);
1596         add_c = diff_get_color_opt(options, DIFF_FILE_NEW);
1597         del_c = diff_get_color_opt(options, DIFF_FILE_OLD);
1598
1599         /*
1600          * Find the longest filename and max number of changes
1601          */
1602         for (i = 0; (i < count) && (i < data->nr); i++) {
1603                 struct diffstat_file *file = data->files[i];
1604                 uintmax_t change = file->added + file->deleted;
1605
1606                 if (!file->is_interesting && (change == 0)) {
1607                         count++; /* not shown == room for one more */
1608                         continue;
1609                 }
1610                 fill_print_name(file);
1611                 len = strlen(file->print_name);
1612                 if (max_len < len)
1613                         max_len = len;
1614
1615                 if (file->is_unmerged) {
1616                         /* "Unmerged" is 8 characters */
1617                         bin_width = bin_width < 8 ? 8 : bin_width;
1618                         continue;
1619                 }
1620                 if (file->is_binary) {
1621                         /* "Bin XXX -> YYY bytes" */
1622                         int w = 14 + decimal_width(file->added)
1623                                 + decimal_width(file->deleted);
1624                         bin_width = bin_width < w ? w : bin_width;
1625                         /* Display change counts aligned with "Bin" */
1626                         number_width = 3;
1627                         continue;
1628                 }
1629
1630                 if (max_change < change)
1631                         max_change = change;
1632         }
1633         count = i; /* where we can stop scanning in data->files[] */
1634
1635         /*
1636          * We have width = stat_width or term_columns() columns total.
1637          * We want a maximum of min(max_len, stat_name_width) for the name part.
1638          * We want a maximum of min(max_change, stat_graph_width) for the +- part.
1639          * We also need 1 for " " and 4 + decimal_width(max_change)
1640          * for " | NNNN " and one the empty column at the end, altogether
1641          * 6 + decimal_width(max_change).
1642          *
1643          * If there's not enough space, we will use the smaller of
1644          * stat_name_width (if set) and 5/8*width for the filename,
1645          * and the rest for constant elements + graph part, but no more
1646          * than stat_graph_width for the graph part.
1647          * (5/8 gives 50 for filename and 30 for the constant parts + graph
1648          * for the standard terminal size).
1649          *
1650          * In other words: stat_width limits the maximum width, and
1651          * stat_name_width fixes the maximum width of the filename,
1652          * and is also used to divide available columns if there
1653          * aren't enough.
1654          *
1655          * Binary files are displayed with "Bin XXX -> YYY bytes"
1656          * instead of the change count and graph. This part is treated
1657          * similarly to the graph part, except that it is not
1658          * "scaled". If total width is too small to accommodate the
1659          * guaranteed minimum width of the filename part and the
1660          * separators and this message, this message will "overflow"
1661          * making the line longer than the maximum width.
1662          */
1663
1664         if (options->stat_width == -1)
1665                 width = term_columns() - options->output_prefix_length;
1666         else
1667                 width = options->stat_width ? options->stat_width : 80;
1668         number_width = decimal_width(max_change) > number_width ?
1669                 decimal_width(max_change) : number_width;
1670
1671         if (options->stat_graph_width == -1)
1672                 options->stat_graph_width = diff_stat_graph_width;
1673
1674         /*
1675          * Guarantee 3/8*16==6 for the graph part
1676          * and 5/8*16==10 for the filename part
1677          */
1678         if (width < 16 + 6 + number_width)
1679                 width = 16 + 6 + number_width;
1680
1681         /*
1682          * First assign sizes that are wanted, ignoring available width.
1683          * strlen("Bin XXX -> YYY bytes") == bin_width, and the part
1684          * starting from "XXX" should fit in graph_width.
1685          */
1686         graph_width = max_change + 4 > bin_width ? max_change : bin_width - 4;
1687         if (options->stat_graph_width &&
1688             options->stat_graph_width < graph_width)
1689                 graph_width = options->stat_graph_width;
1690
1691         name_width = (options->stat_name_width > 0 &&
1692                       options->stat_name_width < max_len) ?
1693                 options->stat_name_width : max_len;
1694
1695         /*
1696          * Adjust adjustable widths not to exceed maximum width
1697          */
1698         if (name_width + number_width + 6 + graph_width > width) {
1699                 if (graph_width > width * 3/8 - number_width - 6) {
1700                         graph_width = width * 3/8 - number_width - 6;
1701                         if (graph_width < 6)
1702                                 graph_width = 6;
1703                 }
1704
1705                 if (options->stat_graph_width &&
1706                     graph_width > options->stat_graph_width)
1707                         graph_width = options->stat_graph_width;
1708                 if (name_width > width - number_width - 6 - graph_width)
1709                         name_width = width - number_width - 6 - graph_width;
1710                 else
1711                         graph_width = width - number_width - 6 - name_width;
1712         }
1713
1714         /*
1715          * From here name_width is the width of the name area,
1716          * and graph_width is the width of the graph area.
1717          * max_change is used to scale graph properly.
1718          */
1719         for (i = 0; i < count; i++) {
1720                 const char *prefix = "";
1721                 struct diffstat_file *file = data->files[i];
1722                 char *name = file->print_name;
1723                 uintmax_t added = file->added;
1724                 uintmax_t deleted = file->deleted;
1725                 int name_len;
1726
1727                 if (!file->is_interesting && (added + deleted == 0))
1728                         continue;
1729
1730                 /*
1731                  * "scale" the filename
1732                  */
1733                 len = name_width;
1734                 name_len = strlen(name);
1735                 if (name_width < name_len) {
1736                         char *slash;
1737                         prefix = "...";
1738                         len -= 3;
1739                         name += name_len - len;
1740                         slash = strchr(name, '/');
1741                         if (slash)
1742                                 name = slash;
1743                 }
1744
1745                 if (file->is_binary) {
1746                         fprintf(options->file, "%s", line_prefix);
1747                         show_name(options->file, prefix, name, len);
1748                         fprintf(options->file, " %*s", number_width, "Bin");
1749                         if (!added && !deleted) {
1750                                 putc('\n', options->file);
1751                                 continue;
1752                         }
1753                         fprintf(options->file, " %s%"PRIuMAX"%s",
1754                                 del_c, deleted, reset);
1755                         fprintf(options->file, " -> ");
1756                         fprintf(options->file, "%s%"PRIuMAX"%s",
1757                                 add_c, added, reset);
1758                         fprintf(options->file, " bytes");
1759                         fprintf(options->file, "\n");
1760                         continue;
1761                 }
1762                 else if (file->is_unmerged) {
1763                         fprintf(options->file, "%s", line_prefix);
1764                         show_name(options->file, prefix, name, len);
1765                         fprintf(options->file, " Unmerged\n");
1766                         continue;
1767                 }
1768
1769                 /*
1770                  * scale the add/delete
1771                  */
1772                 add = added;
1773                 del = deleted;
1774
1775                 if (graph_width <= max_change) {
1776                         int total = scale_linear(add + del, graph_width, max_change);
1777                         if (total < 2 && add && del)
1778                                 /* width >= 2 due to the sanity check */
1779                                 total = 2;
1780                         if (add < del) {
1781                                 add = scale_linear(add, graph_width, max_change);
1782                                 del = total - add;
1783                         } else {
1784                                 del = scale_linear(del, graph_width, max_change);
1785                                 add = total - del;
1786                         }
1787                 }
1788                 fprintf(options->file, "%s", line_prefix);
1789                 show_name(options->file, prefix, name, len);
1790                 fprintf(options->file, " %*"PRIuMAX"%s",
1791                         number_width, added + deleted,
1792                         added + deleted ? " " : "");
1793                 show_graph(options->file, '+', add, add_c, reset);
1794                 show_graph(options->file, '-', del, del_c, reset);
1795                 fprintf(options->file, "\n");
1796         }
1797
1798         for (i = 0; i < data->nr; i++) {
1799                 struct diffstat_file *file = data->files[i];
1800                 uintmax_t added = file->added;
1801                 uintmax_t deleted = file->deleted;
1802
1803                 if (file->is_unmerged ||
1804                     (!file->is_interesting && (added + deleted == 0))) {
1805                         total_files--;
1806                         continue;
1807                 }
1808
1809                 if (!file->is_binary) {
1810                         adds += added;
1811                         dels += deleted;
1812                 }
1813                 if (i < count)
1814                         continue;
1815                 if (!extra_shown)
1816                         fprintf(options->file, "%s ...\n", line_prefix);
1817                 extra_shown = 1;
1818         }
1819         fprintf(options->file, "%s", line_prefix);
1820         print_stat_summary(options->file, total_files, adds, dels);
1821 }
1822
1823 static void show_shortstats(struct diffstat_t *data, struct diff_options *options)
1824 {
1825         int i, adds = 0, dels = 0, total_files = data->nr;
1826
1827         if (data->nr == 0)
1828                 return;
1829
1830         for (i = 0; i < data->nr; i++) {
1831                 int added = data->files[i]->added;
1832                 int deleted= data->files[i]->deleted;
1833
1834                 if (data->files[i]->is_unmerged ||
1835                     (!data->files[i]->is_interesting && (added + deleted == 0))) {
1836                         total_files--;
1837                 } else if (!data->files[i]->is_binary) { /* don't count bytes */
1838                         adds += added;
1839                         dels += deleted;
1840                 }
1841         }
1842         fprintf(options->file, "%s", diff_line_prefix(options));
1843         print_stat_summary(options->file, total_files, adds, dels);
1844 }
1845
1846 static void show_numstat(struct diffstat_t *data, struct diff_options *options)
1847 {
1848         int i;
1849
1850         if (data->nr == 0)
1851                 return;
1852
1853         for (i = 0; i < data->nr; i++) {
1854                 struct diffstat_file *file = data->files[i];
1855
1856                 fprintf(options->file, "%s", diff_line_prefix(options));
1857
1858                 if (file->is_binary)
1859                         fprintf(options->file, "-\t-\t");
1860                 else
1861                         fprintf(options->file,
1862                                 "%"PRIuMAX"\t%"PRIuMAX"\t",
1863                                 file->added, file->deleted);
1864                 if (options->line_termination) {
1865                         fill_print_name(file);
1866                         if (!file->is_renamed)
1867                                 write_name_quoted(file->name, options->file,
1868                                                   options->line_termination);
1869                         else {
1870                                 fputs(file->print_name, options->file);
1871                                 putc(options->line_termination, options->file);
1872                         }
1873                 } else {
1874                         if (file->is_renamed) {
1875                                 putc('\0', options->file);
1876                                 write_name_quoted(file->from_name, options->file, '\0');
1877                         }
1878                         write_name_quoted(file->name, options->file, '\0');
1879                 }
1880         }
1881 }
1882
1883 struct dirstat_file {
1884         const char *name;
1885         unsigned long changed;
1886 };
1887
1888 struct dirstat_dir {
1889         struct dirstat_file *files;
1890         int alloc, nr, permille, cumulative;
1891 };
1892
1893 static long gather_dirstat(struct diff_options *opt, struct dirstat_dir *dir,
1894                 unsigned long changed, const char *base, int baselen)
1895 {
1896         unsigned long this_dir = 0;
1897         unsigned int sources = 0;
1898         const char *line_prefix = diff_line_prefix(opt);
1899
1900         while (dir->nr) {
1901                 struct dirstat_file *f = dir->files;
1902                 int namelen = strlen(f->name);
1903                 unsigned long this;
1904                 char *slash;
1905
1906                 if (namelen < baselen)
1907                         break;
1908                 if (memcmp(f->name, base, baselen))
1909                         break;
1910                 slash = strchr(f->name + baselen, '/');
1911                 if (slash) {
1912                         int newbaselen = slash + 1 - f->name;
1913                         this = gather_dirstat(opt, dir, changed, f->name, newbaselen);
1914                         sources++;
1915                 } else {
1916                         this = f->changed;
1917                         dir->files++;
1918                         dir->nr--;
1919                         sources += 2;
1920                 }
1921                 this_dir += this;
1922         }
1923
1924         /*
1925          * We don't report dirstat's for
1926          *  - the top level
1927          *  - or cases where everything came from a single directory
1928          *    under this directory (sources == 1).
1929          */
1930         if (baselen && sources != 1) {
1931                 if (this_dir) {
1932                         int permille = this_dir * 1000 / changed;
1933                         if (permille >= dir->permille) {
1934                                 fprintf(opt->file, "%s%4d.%01d%% %.*s\n", line_prefix,
1935                                         permille / 10, permille % 10, baselen, base);
1936                                 if (!dir->cumulative)
1937                                         return 0;
1938                         }
1939                 }
1940         }
1941         return this_dir;
1942 }
1943
1944 static int dirstat_compare(const void *_a, const void *_b)
1945 {
1946         const struct dirstat_file *a = _a;
1947         const struct dirstat_file *b = _b;
1948         return strcmp(a->name, b->name);
1949 }
1950
1951 static void show_dirstat(struct diff_options *options)
1952 {
1953         int i;
1954         unsigned long changed;
1955         struct dirstat_dir dir;
1956         struct diff_queue_struct *q = &diff_queued_diff;
1957
1958         dir.files = NULL;
1959         dir.alloc = 0;
1960         dir.nr = 0;
1961         dir.permille = options->dirstat_permille;
1962         dir.cumulative = DIFF_OPT_TST(options, DIRSTAT_CUMULATIVE);
1963
1964         changed = 0;
1965         for (i = 0; i < q->nr; i++) {
1966                 struct diff_filepair *p = q->queue[i];
1967                 const char *name;
1968                 unsigned long copied, added, damage;
1969                 int content_changed;
1970
1971                 name = p->two->path ? p->two->path : p->one->path;
1972
1973                 if (p->one->sha1_valid && p->two->sha1_valid)
1974                         content_changed = hashcmp(p->one->sha1, p->two->sha1);
1975                 else
1976                         content_changed = 1;
1977
1978                 if (!content_changed) {
1979                         /*
1980                          * The SHA1 has not changed, so pre-/post-content is
1981                          * identical. We can therefore skip looking at the
1982                          * file contents altogether.
1983                          */
1984                         damage = 0;
1985                         goto found_damage;
1986                 }
1987
1988                 if (DIFF_OPT_TST(options, DIRSTAT_BY_FILE)) {
1989                         /*
1990                          * In --dirstat-by-file mode, we don't really need to
1991                          * look at the actual file contents at all.
1992                          * The fact that the SHA1 changed is enough for us to
1993                          * add this file to the list of results
1994                          * (with each file contributing equal damage).
1995                          */
1996                         damage = 1;
1997                         goto found_damage;
1998                 }
1999
2000                 if (DIFF_FILE_VALID(p->one) && DIFF_FILE_VALID(p->two)) {
2001                         diff_populate_filespec(p->one, 0);
2002                         diff_populate_filespec(p->two, 0);
2003                         diffcore_count_changes(p->one, p->two, NULL, NULL, 0,
2004                                                &copied, &added);
2005                         diff_free_filespec_data(p->one);
2006                         diff_free_filespec_data(p->two);
2007                 } else if (DIFF_FILE_VALID(p->one)) {
2008                         diff_populate_filespec(p->one, CHECK_SIZE_ONLY);
2009                         copied = added = 0;
2010                         diff_free_filespec_data(p->one);
2011                 } else if (DIFF_FILE_VALID(p->two)) {
2012                         diff_populate_filespec(p->two, CHECK_SIZE_ONLY);
2013                         copied = 0;
2014                         added = p->two->size;
2015                         diff_free_filespec_data(p->two);
2016                 } else
2017                         continue;
2018
2019                 /*
2020                  * Original minus copied is the removed material,
2021                  * added is the new material.  They are both damages
2022                  * made to the preimage.
2023                  * If the resulting damage is zero, we know that
2024                  * diffcore_count_changes() considers the two entries to
2025                  * be identical, but since content_changed is true, we
2026                  * know that there must have been _some_ kind of change,
2027                  * so we force all entries to have damage > 0.
2028                  */
2029                 damage = (p->one->size - copied) + added;
2030                 if (!damage)
2031                         damage = 1;
2032
2033 found_damage:
2034                 ALLOC_GROW(dir.files, dir.nr + 1, dir.alloc);
2035                 dir.files[dir.nr].name = name;
2036                 dir.files[dir.nr].changed = damage;
2037                 changed += damage;
2038                 dir.nr++;
2039         }
2040
2041         /* This can happen even with many files, if everything was renames */
2042         if (!changed)
2043                 return;
2044
2045         /* Show all directories with more than x% of the changes */
2046         qsort(dir.files, dir.nr, sizeof(dir.files[0]), dirstat_compare);
2047         gather_dirstat(options, &dir, changed, "", 0);
2048 }
2049
2050 static void show_dirstat_by_line(struct diffstat_t *data, struct diff_options *options)
2051 {
2052         int i;
2053         unsigned long changed;
2054         struct dirstat_dir dir;
2055
2056         if (data->nr == 0)
2057                 return;
2058
2059         dir.files = NULL;
2060         dir.alloc = 0;
2061         dir.nr = 0;
2062         dir.permille = options->dirstat_permille;
2063         dir.cumulative = DIFF_OPT_TST(options, DIRSTAT_CUMULATIVE);
2064
2065         changed = 0;
2066         for (i = 0; i < data->nr; i++) {
2067                 struct diffstat_file *file = data->files[i];
2068                 unsigned long damage = file->added + file->deleted;
2069                 if (file->is_binary)
2070                         /*
2071                          * binary files counts bytes, not lines. Must find some
2072                          * way to normalize binary bytes vs. textual lines.
2073                          * The following heuristic assumes that there are 64
2074                          * bytes per "line".
2075                          * This is stupid and ugly, but very cheap...
2076                          */
2077                         damage = (damage + 63) / 64;
2078                 ALLOC_GROW(dir.files, dir.nr + 1, dir.alloc);
2079                 dir.files[dir.nr].name = file->name;
2080                 dir.files[dir.nr].changed = damage;
2081                 changed += damage;
2082                 dir.nr++;
2083         }
2084
2085         /* This can happen even with many files, if everything was renames */
2086         if (!changed)
2087                 return;
2088
2089         /* Show all directories with more than x% of the changes */
2090         qsort(dir.files, dir.nr, sizeof(dir.files[0]), dirstat_compare);
2091         gather_dirstat(options, &dir, changed, "", 0);
2092 }
2093
2094 static void free_diffstat_info(struct diffstat_t *diffstat)
2095 {
2096         int i;
2097         for (i = 0; i < diffstat->nr; i++) {
2098                 struct diffstat_file *f = diffstat->files[i];
2099                 if (f->name != f->print_name)
2100                         free(f->print_name);
2101                 free(f->name);
2102                 free(f->from_name);
2103                 free(f);
2104         }
2105         free(diffstat->files);
2106 }
2107
2108 struct checkdiff_t {
2109         const char *filename;
2110         int lineno;
2111         int conflict_marker_size;
2112         struct diff_options *o;
2113         unsigned ws_rule;
2114         unsigned status;
2115 };
2116
2117 static int is_conflict_marker(const char *line, int marker_size, unsigned long len)
2118 {
2119         char firstchar;
2120         int cnt;
2121
2122         if (len < marker_size + 1)
2123                 return 0;
2124         firstchar = line[0];
2125         switch (firstchar) {
2126         case '=': case '>': case '<': case '|':
2127                 break;
2128         default:
2129                 return 0;
2130         }
2131         for (cnt = 1; cnt < marker_size; cnt++)
2132                 if (line[cnt] != firstchar)
2133                         return 0;
2134         /* line[1] thru line[marker_size-1] are same as firstchar */
2135         if (len < marker_size + 1 || !isspace(line[marker_size]))
2136                 return 0;
2137         return 1;
2138 }
2139
2140 static void checkdiff_consume(void *priv, char *line, unsigned long len)
2141 {
2142         struct checkdiff_t *data = priv;
2143         int marker_size = data->conflict_marker_size;
2144         const char *ws = diff_get_color(data->o->use_color, DIFF_WHITESPACE);
2145         const char *reset = diff_get_color(data->o->use_color, DIFF_RESET);
2146         const char *set = diff_get_color(data->o->use_color, DIFF_FILE_NEW);
2147         char *err;
2148         const char *line_prefix;
2149
2150         assert(data->o);
2151         line_prefix = diff_line_prefix(data->o);
2152
2153         if (line[0] == '+') {
2154                 unsigned bad;
2155                 data->lineno++;
2156                 if (is_conflict_marker(line + 1, marker_size, len - 1)) {
2157                         data->status |= 1;
2158                         fprintf(data->o->file,
2159                                 "%s%s:%d: leftover conflict marker\n",
2160                                 line_prefix, data->filename, data->lineno);
2161                 }
2162                 bad = ws_check(line + 1, len - 1, data->ws_rule);
2163                 if (!bad)
2164                         return;
2165                 data->status |= bad;
2166                 err = whitespace_error_string(bad);
2167                 fprintf(data->o->file, "%s%s:%d: %s.\n",
2168                         line_prefix, data->filename, data->lineno, err);
2169                 free(err);
2170                 emit_line(data->o, set, reset, line, 1);
2171                 ws_check_emit(line + 1, len - 1, data->ws_rule,
2172                               data->o->file, set, reset, ws);
2173         } else if (line[0] == ' ') {
2174                 data->lineno++;
2175         } else if (line[0] == '@') {
2176                 char *plus = strchr(line, '+');
2177                 if (plus)
2178                         data->lineno = strtol(plus, NULL, 10) - 1;
2179                 else
2180                         die("invalid diff");
2181         }
2182 }
2183
2184 static unsigned char *deflate_it(char *data,
2185                                  unsigned long size,
2186                                  unsigned long *result_size)
2187 {
2188         int bound;
2189         unsigned char *deflated;
2190         git_zstream stream;
2191
2192         git_deflate_init(&stream, zlib_compression_level);
2193         bound = git_deflate_bound(&stream, size);
2194         deflated = xmalloc(bound);
2195         stream.next_out = deflated;
2196         stream.avail_out = bound;
2197
2198         stream.next_in = (unsigned char *)data;
2199         stream.avail_in = size;
2200         while (git_deflate(&stream, Z_FINISH) == Z_OK)
2201                 ; /* nothing */
2202         git_deflate_end(&stream);
2203         *result_size = stream.total_out;
2204         return deflated;
2205 }
2206
2207 static void emit_binary_diff_body(FILE *file, mmfile_t *one, mmfile_t *two,
2208                                   const char *prefix)
2209 {
2210         void *cp;
2211         void *delta;
2212         void *deflated;
2213         void *data;
2214         unsigned long orig_size;
2215         unsigned long delta_size;
2216         unsigned long deflate_size;
2217         unsigned long data_size;
2218
2219         /* We could do deflated delta, or we could do just deflated two,
2220          * whichever is smaller.
2221          */
2222         delta = NULL;
2223         deflated = deflate_it(two->ptr, two->size, &deflate_size);
2224         if (one->size && two->size) {
2225                 delta = diff_delta(one->ptr, one->size,
2226                                    two->ptr, two->size,
2227                                    &delta_size, deflate_size);
2228                 if (delta) {
2229                         void *to_free = delta;
2230                         orig_size = delta_size;
2231                         delta = deflate_it(delta, delta_size, &delta_size);
2232                         free(to_free);
2233                 }
2234         }
2235
2236         if (delta && delta_size < deflate_size) {
2237                 fprintf(file, "%sdelta %lu\n", prefix, orig_size);
2238                 free(deflated);
2239                 data = delta;
2240                 data_size = delta_size;
2241         }
2242         else {
2243                 fprintf(file, "%sliteral %lu\n", prefix, two->size);
2244                 free(delta);
2245                 data = deflated;
2246                 data_size = deflate_size;
2247         }
2248
2249         /* emit data encoded in base85 */
2250         cp = data;
2251         while (data_size) {
2252                 int bytes = (52 < data_size) ? 52 : data_size;
2253                 char line[70];
2254                 data_size -= bytes;
2255                 if (bytes <= 26)
2256                         line[0] = bytes + 'A' - 1;
2257                 else
2258                         line[0] = bytes - 26 + 'a' - 1;
2259                 encode_85(line + 1, cp, bytes);
2260                 cp = (char *) cp + bytes;
2261                 fprintf(file, "%s", prefix);
2262                 fputs(line, file);
2263                 fputc('\n', file);
2264         }
2265         fprintf(file, "%s\n", prefix);
2266         free(data);
2267 }
2268
2269 static void emit_binary_diff(FILE *file, mmfile_t *one, mmfile_t *two,
2270                              const char *prefix)
2271 {
2272         fprintf(file, "%sGIT binary patch\n", prefix);
2273         emit_binary_diff_body(file, one, two, prefix);
2274         emit_binary_diff_body(file, two, one, prefix);
2275 }
2276
2277 int diff_filespec_is_binary(struct diff_filespec *one)
2278 {
2279         if (one->is_binary == -1) {
2280                 diff_filespec_load_driver(one);
2281                 if (one->driver->binary != -1)
2282                         one->is_binary = one->driver->binary;
2283                 else {
2284                         if (!one->data && DIFF_FILE_VALID(one))
2285                                 diff_populate_filespec(one, CHECK_BINARY);
2286                         if (one->is_binary == -1 && one->data)
2287                                 one->is_binary = buffer_is_binary(one->data,
2288                                                 one->size);
2289                         if (one->is_binary == -1)
2290                                 one->is_binary = 0;
2291                 }
2292         }
2293         return one->is_binary;
2294 }
2295
2296 static const struct userdiff_funcname *diff_funcname_pattern(struct diff_filespec *one)
2297 {
2298         diff_filespec_load_driver(one);
2299         return one->driver->funcname.pattern ? &one->driver->funcname : NULL;
2300 }
2301
2302 void diff_set_mnemonic_prefix(struct diff_options *options, const char *a, const char *b)
2303 {
2304         if (!options->a_prefix)
2305                 options->a_prefix = a;
2306         if (!options->b_prefix)
2307                 options->b_prefix = b;
2308 }
2309
2310 struct userdiff_driver *get_textconv(struct diff_filespec *one)
2311 {
2312         if (!DIFF_FILE_VALID(one))
2313                 return NULL;
2314
2315         diff_filespec_load_driver(one);
2316         return userdiff_get_textconv(one->driver);
2317 }
2318
2319 static void builtin_diff(const char *name_a,
2320                          const char *name_b,
2321                          struct diff_filespec *one,
2322                          struct diff_filespec *two,
2323                          const char *xfrm_msg,
2324                          int must_show_header,
2325                          struct diff_options *o,
2326                          int complete_rewrite)
2327 {
2328         mmfile_t mf1, mf2;
2329         const char *lbl[2];
2330         char *a_one, *b_two;
2331         const char *meta = diff_get_color_opt(o, DIFF_METAINFO);
2332         const char *reset = diff_get_color_opt(o, DIFF_RESET);
2333         const char *a_prefix, *b_prefix;
2334         struct userdiff_driver *textconv_one = NULL;
2335         struct userdiff_driver *textconv_two = NULL;
2336         struct strbuf header = STRBUF_INIT;
2337         const char *line_prefix = diff_line_prefix(o);
2338
2339         if (DIFF_OPT_TST(o, SUBMODULE_LOG) &&
2340                         (!one->mode || S_ISGITLINK(one->mode)) &&
2341                         (!two->mode || S_ISGITLINK(two->mode))) {
2342                 const char *del = diff_get_color_opt(o, DIFF_FILE_OLD);
2343                 const char *add = diff_get_color_opt(o, DIFF_FILE_NEW);
2344                 show_submodule_summary(o->file, one->path ? one->path : two->path,
2345                                 line_prefix,
2346                                 one->sha1, two->sha1, two->dirty_submodule,
2347                                 meta, del, add, reset);
2348                 return;
2349         }
2350
2351         if (DIFF_OPT_TST(o, ALLOW_TEXTCONV)) {
2352                 textconv_one = get_textconv(one);
2353                 textconv_two = get_textconv(two);
2354         }
2355
2356         diff_set_mnemonic_prefix(o, "a/", "b/");
2357         if (DIFF_OPT_TST(o, REVERSE_DIFF)) {
2358                 a_prefix = o->b_prefix;
2359                 b_prefix = o->a_prefix;
2360         } else {
2361                 a_prefix = o->a_prefix;
2362                 b_prefix = o->b_prefix;
2363         }
2364
2365         /* Never use a non-valid filename anywhere if at all possible */
2366         name_a = DIFF_FILE_VALID(one) ? name_a : name_b;
2367         name_b = DIFF_FILE_VALID(two) ? name_b : name_a;
2368
2369         a_one = quote_two(a_prefix, name_a + (*name_a == '/'));
2370         b_two = quote_two(b_prefix, name_b + (*name_b == '/'));
2371         lbl[0] = DIFF_FILE_VALID(one) ? a_one : "/dev/null";
2372         lbl[1] = DIFF_FILE_VALID(two) ? b_two : "/dev/null";
2373         strbuf_addf(&header, "%s%sdiff --git %s %s%s\n", line_prefix, meta, a_one, b_two, reset);
2374         if (lbl[0][0] == '/') {
2375                 /* /dev/null */
2376                 strbuf_addf(&header, "%s%snew file mode %06o%s\n", line_prefix, meta, two->mode, reset);
2377                 if (xfrm_msg)
2378                         strbuf_addstr(&header, xfrm_msg);
2379                 must_show_header = 1;
2380         }
2381         else if (lbl[1][0] == '/') {
2382                 strbuf_addf(&header, "%s%sdeleted file mode %06o%s\n", line_prefix, meta, one->mode, reset);
2383                 if (xfrm_msg)
2384                         strbuf_addstr(&header, xfrm_msg);
2385                 must_show_header = 1;
2386         }
2387         else {
2388                 if (one->mode != two->mode) {
2389                         strbuf_addf(&header, "%s%sold mode %06o%s\n", line_prefix, meta, one->mode, reset);
2390                         strbuf_addf(&header, "%s%snew mode %06o%s\n", line_prefix, meta, two->mode, reset);
2391                         must_show_header = 1;
2392                 }
2393                 if (xfrm_msg)
2394                         strbuf_addstr(&header, xfrm_msg);
2395
2396                 /*
2397                  * we do not run diff between different kind
2398                  * of objects.
2399                  */
2400                 if ((one->mode ^ two->mode) & S_IFMT)
2401                         goto free_ab_and_return;
2402                 if (complete_rewrite &&
2403                     (textconv_one || !diff_filespec_is_binary(one)) &&
2404                     (textconv_two || !diff_filespec_is_binary(two))) {
2405                         fprintf(o->file, "%s", header.buf);
2406                         strbuf_reset(&header);
2407                         emit_rewrite_diff(name_a, name_b, one, two,
2408                                                 textconv_one, textconv_two, o);
2409                         o->found_changes = 1;
2410                         goto free_ab_and_return;
2411                 }
2412         }
2413
2414         if (o->irreversible_delete && lbl[1][0] == '/') {
2415                 fprintf(o->file, "%s", header.buf);
2416                 strbuf_reset(&header);
2417                 goto free_ab_and_return;
2418         } else if (!DIFF_OPT_TST(o, TEXT) &&
2419             ( (!textconv_one && diff_filespec_is_binary(one)) ||
2420               (!textconv_two && diff_filespec_is_binary(two)) )) {
2421                 if (!one->data && !two->data &&
2422                     S_ISREG(one->mode) && S_ISREG(two->mode) &&
2423                     !DIFF_OPT_TST(o, BINARY)) {
2424                         if (!hashcmp(one->sha1, two->sha1)) {
2425                                 if (must_show_header)
2426                                         fprintf(o->file, "%s", header.buf);
2427                                 goto free_ab_and_return;
2428                         }
2429                         fprintf(o->file, "%s", header.buf);
2430                         fprintf(o->file, "%sBinary files %s and %s differ\n",
2431                                 line_prefix, lbl[0], lbl[1]);
2432                         goto free_ab_and_return;
2433                 }
2434                 if (fill_mmfile(&mf1, one) < 0 || fill_mmfile(&mf2, two) < 0)
2435                         die("unable to read files to diff");
2436                 /* Quite common confusing case */
2437                 if (mf1.size == mf2.size &&
2438                     !memcmp(mf1.ptr, mf2.ptr, mf1.size)) {
2439                         if (must_show_header)
2440                                 fprintf(o->file, "%s", header.buf);
2441                         goto free_ab_and_return;
2442                 }
2443                 fprintf(o->file, "%s", header.buf);
2444                 strbuf_reset(&header);
2445                 if (DIFF_OPT_TST(o, BINARY))
2446                         emit_binary_diff(o->file, &mf1, &mf2, line_prefix);
2447                 else
2448                         fprintf(o->file, "%sBinary files %s and %s differ\n",
2449                                 line_prefix, lbl[0], lbl[1]);
2450                 o->found_changes = 1;
2451         } else {
2452                 /* Crazy xdl interfaces.. */
2453                 const char *diffopts = getenv("GIT_DIFF_OPTS");
2454                 const char *v;
2455                 xpparam_t xpp;
2456                 xdemitconf_t xecfg;
2457                 struct emit_callback ecbdata;
2458                 const struct userdiff_funcname *pe;
2459
2460                 if (must_show_header) {
2461                         fprintf(o->file, "%s", header.buf);
2462                         strbuf_reset(&header);
2463                 }
2464
2465                 mf1.size = fill_textconv(textconv_one, one, &mf1.ptr);
2466                 mf2.size = fill_textconv(textconv_two, two, &mf2.ptr);
2467
2468                 pe = diff_funcname_pattern(one);
2469                 if (!pe)
2470                         pe = diff_funcname_pattern(two);
2471
2472                 memset(&xpp, 0, sizeof(xpp));
2473                 memset(&xecfg, 0, sizeof(xecfg));
2474                 memset(&ecbdata, 0, sizeof(ecbdata));
2475                 ecbdata.label_path = lbl;
2476                 ecbdata.color_diff = want_color(o->use_color);
2477                 ecbdata.found_changesp = &o->found_changes;
2478                 ecbdata.ws_rule = whitespace_rule(name_b);
2479                 if (ecbdata.ws_rule & WS_BLANK_AT_EOF)
2480                         check_blank_at_eof(&mf1, &mf2, &ecbdata);
2481                 ecbdata.opt = o;
2482                 ecbdata.header = header.len ? &header : NULL;
2483                 xpp.flags = o->xdl_opts;
2484                 xecfg.ctxlen = o->context;
2485                 xecfg.interhunkctxlen = o->interhunkcontext;
2486                 xecfg.flags = XDL_EMIT_FUNCNAMES;
2487                 if (DIFF_OPT_TST(o, FUNCCONTEXT))
2488                         xecfg.flags |= XDL_EMIT_FUNCCONTEXT;
2489                 if (pe)
2490                         xdiff_set_find_func(&xecfg, pe->pattern, pe->cflags);
2491                 if (!diffopts)
2492                         ;
2493                 else if (skip_prefix(diffopts, "--unified=", &v))
2494                         xecfg.ctxlen = strtoul(v, NULL, 10);
2495                 else if (skip_prefix(diffopts, "-u", &v))
2496                         xecfg.ctxlen = strtoul(v, NULL, 10);
2497                 if (o->word_diff)
2498                         init_diff_words_data(&ecbdata, o, one, two);
2499                 if (xdi_diff_outf(&mf1, &mf2, fn_out_consume, &ecbdata,
2500                                   &xpp, &xecfg))
2501                         die("unable to generate diff for %s", one->path);
2502                 if (o->word_diff)
2503                         free_diff_words_data(&ecbdata);
2504                 if (textconv_one)
2505                         free(mf1.ptr);
2506                 if (textconv_two)
2507                         free(mf2.ptr);
2508                 xdiff_clear_find_func(&xecfg);
2509         }
2510
2511  free_ab_and_return:
2512         strbuf_release(&header);
2513         diff_free_filespec_data(one);
2514         diff_free_filespec_data(two);
2515         free(a_one);
2516         free(b_two);
2517         return;
2518 }
2519
2520 static void builtin_diffstat(const char *name_a, const char *name_b,
2521                              struct diff_filespec *one,
2522                              struct diff_filespec *two,
2523                              struct diffstat_t *diffstat,
2524                              struct diff_options *o,
2525                              struct diff_filepair *p)
2526 {
2527         mmfile_t mf1, mf2;
2528         struct diffstat_file *data;
2529         int same_contents;
2530         int complete_rewrite = 0;
2531
2532         if (!DIFF_PAIR_UNMERGED(p)) {
2533                 if (p->status == DIFF_STATUS_MODIFIED && p->score)
2534                         complete_rewrite = 1;
2535         }
2536
2537         data = diffstat_add(diffstat, name_a, name_b);
2538         data->is_interesting = p->status != DIFF_STATUS_UNKNOWN;
2539
2540         if (!one || !two) {
2541                 data->is_unmerged = 1;
2542                 return;
2543         }
2544
2545         same_contents = !hashcmp(one->sha1, two->sha1);
2546
2547         if (diff_filespec_is_binary(one) || diff_filespec_is_binary(two)) {
2548                 data->is_binary = 1;
2549                 if (same_contents) {
2550                         data->added = 0;
2551                         data->deleted = 0;
2552                 } else {
2553                         data->added = diff_filespec_size(two);
2554                         data->deleted = diff_filespec_size(one);
2555                 }
2556         }
2557
2558         else if (complete_rewrite) {
2559                 diff_populate_filespec(one, 0);
2560                 diff_populate_filespec(two, 0);
2561                 data->deleted = count_lines(one->data, one->size);
2562                 data->added = count_lines(two->data, two->size);
2563         }
2564
2565         else if (!same_contents) {
2566                 /* Crazy xdl interfaces.. */
2567                 xpparam_t xpp;
2568                 xdemitconf_t xecfg;
2569
2570                 if (fill_mmfile(&mf1, one) < 0 || fill_mmfile(&mf2, two) < 0)
2571                         die("unable to read files to diff");
2572
2573                 memset(&xpp, 0, sizeof(xpp));
2574                 memset(&xecfg, 0, sizeof(xecfg));
2575                 xpp.flags = o->xdl_opts;
2576                 xecfg.ctxlen = o->context;
2577                 xecfg.interhunkctxlen = o->interhunkcontext;
2578                 if (xdi_diff_outf(&mf1, &mf2, diffstat_consume, diffstat,
2579                                   &xpp, &xecfg))
2580                         die("unable to generate diffstat for %s", one->path);
2581         }
2582
2583         diff_free_filespec_data(one);
2584         diff_free_filespec_data(two);
2585 }
2586
2587 static void builtin_checkdiff(const char *name_a, const char *name_b,
2588                               const char *attr_path,
2589                               struct diff_filespec *one,
2590                               struct diff_filespec *two,
2591                               struct diff_options *o)
2592 {
2593         mmfile_t mf1, mf2;
2594         struct checkdiff_t data;
2595
2596         if (!two)
2597                 return;
2598
2599         memset(&data, 0, sizeof(data));
2600         data.filename = name_b ? name_b : name_a;
2601         data.lineno = 0;
2602         data.o = o;
2603         data.ws_rule = whitespace_rule(attr_path);
2604         data.conflict_marker_size = ll_merge_marker_size(attr_path);
2605
2606         if (fill_mmfile(&mf1, one) < 0 || fill_mmfile(&mf2, two) < 0)
2607                 die("unable to read files to diff");
2608
2609         /*
2610          * All the other codepaths check both sides, but not checking
2611          * the "old" side here is deliberate.  We are checking the newly
2612          * introduced changes, and as long as the "new" side is text, we
2613          * can and should check what it introduces.
2614          */
2615         if (diff_filespec_is_binary(two))
2616                 goto free_and_return;
2617         else {
2618                 /* Crazy xdl interfaces.. */
2619                 xpparam_t xpp;
2620                 xdemitconf_t xecfg;
2621
2622                 memset(&xpp, 0, sizeof(xpp));
2623                 memset(&xecfg, 0, sizeof(xecfg));
2624                 xecfg.ctxlen = 1; /* at least one context line */
2625                 xpp.flags = 0;
2626                 if (xdi_diff_outf(&mf1, &mf2, checkdiff_consume, &data,
2627                                   &xpp, &xecfg))
2628                         die("unable to generate checkdiff for %s", one->path);
2629
2630                 if (data.ws_rule & WS_BLANK_AT_EOF) {
2631                         struct emit_callback ecbdata;
2632                         int blank_at_eof;
2633
2634                         ecbdata.ws_rule = data.ws_rule;
2635                         check_blank_at_eof(&mf1, &mf2, &ecbdata);
2636                         blank_at_eof = ecbdata.blank_at_eof_in_postimage;
2637
2638                         if (blank_at_eof) {
2639                                 static char *err;
2640                                 if (!err)
2641                                         err = whitespace_error_string(WS_BLANK_AT_EOF);
2642                                 fprintf(o->file, "%s:%d: %s.\n",
2643                                         data.filename, blank_at_eof, err);
2644                                 data.status = 1; /* report errors */
2645                         }
2646                 }
2647         }
2648  free_and_return:
2649         diff_free_filespec_data(one);
2650         diff_free_filespec_data(two);
2651         if (data.status)
2652                 DIFF_OPT_SET(o, CHECK_FAILED);
2653 }
2654
2655 struct diff_filespec *alloc_filespec(const char *path)
2656 {
2657         int namelen = strlen(path);
2658         struct diff_filespec *spec = xmalloc(sizeof(*spec) + namelen + 1);
2659
2660         memset(spec, 0, sizeof(*spec));
2661         spec->path = (char *)(spec + 1);
2662         memcpy(spec->path, path, namelen+1);
2663         spec->count = 1;
2664         spec->is_binary = -1;
2665         return spec;
2666 }
2667
2668 void free_filespec(struct diff_filespec *spec)
2669 {
2670         if (!--spec->count) {
2671                 diff_free_filespec_data(spec);
2672                 free(spec);
2673         }
2674 }
2675
2676 void fill_filespec(struct diff_filespec *spec, const unsigned char *sha1,
2677                    int sha1_valid, unsigned short mode)
2678 {
2679         if (mode) {
2680                 spec->mode = canon_mode(mode);
2681                 hashcpy(spec->sha1, sha1);
2682                 spec->sha1_valid = sha1_valid;
2683         }
2684 }
2685
2686 /*
2687  * Given a name and sha1 pair, if the index tells us the file in
2688  * the work tree has that object contents, return true, so that
2689  * prepare_temp_file() does not have to inflate and extract.
2690  */
2691 static int reuse_worktree_file(const char *name, const unsigned char *sha1, int want_file)
2692 {
2693         const struct cache_entry *ce;
2694         struct stat st;
2695         int pos, len;
2696
2697         /*
2698          * We do not read the cache ourselves here, because the
2699          * benchmark with my previous version that always reads cache
2700          * shows that it makes things worse for diff-tree comparing
2701          * two linux-2.6 kernel trees in an already checked out work
2702          * tree.  This is because most diff-tree comparisons deal with
2703          * only a small number of files, while reading the cache is
2704          * expensive for a large project, and its cost outweighs the
2705          * savings we get by not inflating the object to a temporary
2706          * file.  Practically, this code only helps when we are used
2707          * by diff-cache --cached, which does read the cache before
2708          * calling us.
2709          */
2710         if (!active_cache)
2711                 return 0;
2712
2713         /* We want to avoid the working directory if our caller
2714          * doesn't need the data in a normal file, this system
2715          * is rather slow with its stat/open/mmap/close syscalls,
2716          * and the object is contained in a pack file.  The pack
2717          * is probably already open and will be faster to obtain
2718          * the data through than the working directory.  Loose
2719          * objects however would tend to be slower as they need
2720          * to be individually opened and inflated.
2721          */
2722         if (!FAST_WORKING_DIRECTORY && !want_file && has_sha1_pack(sha1))
2723                 return 0;
2724
2725         len = strlen(name);
2726         pos = cache_name_pos(name, len);
2727         if (pos < 0)
2728                 return 0;
2729         ce = active_cache[pos];
2730
2731         /*
2732          * This is not the sha1 we are looking for, or
2733          * unreusable because it is not a regular file.
2734          */
2735         if (hashcmp(sha1, ce->sha1) || !S_ISREG(ce->ce_mode))
2736                 return 0;
2737
2738         /*
2739          * If ce is marked as "assume unchanged", there is no
2740          * guarantee that work tree matches what we are looking for.
2741          */
2742         if ((ce->ce_flags & CE_VALID) || ce_skip_worktree(ce))
2743                 return 0;
2744
2745         /*
2746          * If ce matches the file in the work tree, we can reuse it.
2747          */
2748         if (ce_uptodate(ce) ||
2749             (!lstat(name, &st) && !ce_match_stat(ce, &st, 0)))
2750                 return 1;
2751
2752         return 0;
2753 }
2754
2755 static int diff_populate_gitlink(struct diff_filespec *s, int size_only)
2756 {
2757         int len;
2758         char *data = xmalloc(100), *dirty = "";
2759
2760         /* Are we looking at the work tree? */
2761         if (s->dirty_submodule)
2762                 dirty = "-dirty";
2763
2764         len = snprintf(data, 100,
2765                        "Subproject commit %s%s\n", sha1_to_hex(s->sha1), dirty);
2766         s->data = data;
2767         s->size = len;
2768         s->should_free = 1;
2769         if (size_only) {
2770                 s->data = NULL;
2771                 free(data);
2772         }
2773         return 0;
2774 }
2775
2776 /*
2777  * While doing rename detection and pickaxe operation, we may need to
2778  * grab the data for the blob (or file) for our own in-core comparison.
2779  * diff_filespec has data and size fields for this purpose.
2780  */
2781 int diff_populate_filespec(struct diff_filespec *s, unsigned int flags)
2782 {
2783         int size_only = flags & CHECK_SIZE_ONLY;
2784         int err = 0;
2785         /*
2786          * demote FAIL to WARN to allow inspecting the situation
2787          * instead of refusing.
2788          */
2789         enum safe_crlf crlf_warn = (safe_crlf == SAFE_CRLF_FAIL
2790                                     ? SAFE_CRLF_WARN
2791                                     : safe_crlf);
2792
2793         if (!DIFF_FILE_VALID(s))
2794                 die("internal error: asking to populate invalid file.");
2795         if (S_ISDIR(s->mode))
2796                 return -1;
2797
2798         if (s->data)
2799                 return 0;
2800
2801         if (size_only && 0 < s->size)
2802                 return 0;
2803
2804         if (S_ISGITLINK(s->mode))
2805                 return diff_populate_gitlink(s, size_only);
2806
2807         if (!s->sha1_valid ||
2808             reuse_worktree_file(s->path, s->sha1, 0)) {
2809                 struct strbuf buf = STRBUF_INIT;
2810                 struct stat st;
2811                 int fd;
2812
2813                 if (lstat(s->path, &st) < 0) {
2814                         if (errno == ENOENT) {
2815                         err_empty:
2816                                 err = -1;
2817                         empty:
2818                                 s->data = (char *)"";
2819                                 s->size = 0;
2820                                 return err;
2821                         }
2822                 }
2823                 s->size = xsize_t(st.st_size);
2824                 if (!s->size)
2825                         goto empty;
2826                 if (S_ISLNK(st.st_mode)) {
2827                         struct strbuf sb = STRBUF_INIT;
2828
2829                         if (strbuf_readlink(&sb, s->path, s->size))
2830                                 goto err_empty;
2831                         s->size = sb.len;
2832                         s->data = strbuf_detach(&sb, NULL);
2833                         s->should_free = 1;
2834                         return 0;
2835                 }
2836                 if (size_only)
2837                         return 0;
2838                 if ((flags & CHECK_BINARY) &&
2839                     s->size > big_file_threshold && s->is_binary == -1) {
2840                         s->is_binary = 1;
2841                         return 0;
2842                 }
2843                 fd = open(s->path, O_RDONLY);
2844                 if (fd < 0)
2845                         goto err_empty;
2846                 s->data = xmmap(NULL, s->size, PROT_READ, MAP_PRIVATE, fd, 0);
2847                 close(fd);
2848                 s->should_munmap = 1;
2849
2850                 /*
2851                  * Convert from working tree format to canonical git format
2852                  */
2853                 if (convert_to_git(s->path, s->data, s->size, &buf, crlf_warn)) {
2854                         size_t size = 0;
2855                         munmap(s->data, s->size);
2856                         s->should_munmap = 0;
2857                         s->data = strbuf_detach(&buf, &size);
2858                         s->size = size;
2859                         s->should_free = 1;
2860                 }
2861         }
2862         else {
2863                 enum object_type type;
2864                 if (size_only || (flags & CHECK_BINARY)) {
2865                         type = sha1_object_info(s->sha1, &s->size);
2866                         if (type < 0)
2867                                 die("unable to read %s", sha1_to_hex(s->sha1));
2868                         if (size_only)
2869                                 return 0;
2870                         if (s->size > big_file_threshold && s->is_binary == -1) {
2871                                 s->is_binary = 1;
2872                                 return 0;
2873                         }
2874                 }
2875                 s->data = read_sha1_file(s->sha1, &type, &s->size);
2876                 if (!s->data)
2877                         die("unable to read %s", sha1_to_hex(s->sha1));
2878                 s->should_free = 1;
2879         }
2880         return 0;
2881 }
2882
2883 void diff_free_filespec_blob(struct diff_filespec *s)
2884 {
2885         if (s->should_free)
2886                 free(s->data);
2887         else if (s->should_munmap)
2888                 munmap(s->data, s->size);
2889
2890         if (s->should_free || s->should_munmap) {
2891                 s->should_free = s->should_munmap = 0;
2892                 s->data = NULL;
2893         }
2894 }
2895
2896 void diff_free_filespec_data(struct diff_filespec *s)
2897 {
2898         diff_free_filespec_blob(s);
2899         free(s->cnt_data);
2900         s->cnt_data = NULL;
2901 }
2902
2903 static void prep_temp_blob(const char *path, struct diff_tempfile *temp,
2904                            void *blob,
2905                            unsigned long size,
2906                            const unsigned char *sha1,
2907                            int mode)
2908 {
2909         int fd;
2910         struct strbuf buf = STRBUF_INIT;
2911         struct strbuf template = STRBUF_INIT;
2912         char *path_dup = xstrdup(path);
2913         const char *base = basename(path_dup);
2914
2915         /* Generate "XXXXXX_basename.ext" */
2916         strbuf_addstr(&template, "XXXXXX_");
2917         strbuf_addstr(&template, base);
2918
2919         fd = mks_tempfile_ts(&temp->tempfile, template.buf, strlen(base) + 1);
2920         if (fd < 0)
2921                 die_errno("unable to create temp-file");
2922         if (convert_to_working_tree(path,
2923                         (const char *)blob, (size_t)size, &buf)) {
2924                 blob = buf.buf;
2925                 size = buf.len;
2926         }
2927         if (write_in_full(fd, blob, size) != size)
2928                 die_errno("unable to write temp-file");
2929         close_tempfile(&temp->tempfile);
2930         temp->name = get_tempfile_path(&temp->tempfile);
2931         strcpy(temp->hex, sha1_to_hex(sha1));
2932         temp->hex[40] = 0;
2933         sprintf(temp->mode, "%06o", mode);
2934         strbuf_release(&buf);
2935         strbuf_release(&template);
2936         free(path_dup);
2937 }
2938
2939 static struct diff_tempfile *prepare_temp_file(const char *name,
2940                 struct diff_filespec *one)
2941 {
2942         struct diff_tempfile *temp = claim_diff_tempfile();
2943
2944         if (!DIFF_FILE_VALID(one)) {
2945         not_a_valid_file:
2946                 /* A '-' entry produces this for file-2, and
2947                  * a '+' entry produces this for file-1.
2948                  */
2949                 temp->name = "/dev/null";
2950                 strcpy(temp->hex, ".");
2951                 strcpy(temp->mode, ".");
2952                 return temp;
2953         }
2954
2955         if (!S_ISGITLINK(one->mode) &&
2956             (!one->sha1_valid ||
2957              reuse_worktree_file(name, one->sha1, 1))) {
2958                 struct stat st;
2959                 if (lstat(name, &st) < 0) {
2960                         if (errno == ENOENT)
2961                                 goto not_a_valid_file;
2962                         die_errno("stat(%s)", name);
2963                 }
2964                 if (S_ISLNK(st.st_mode)) {
2965                         struct strbuf sb = STRBUF_INIT;
2966                         if (strbuf_readlink(&sb, name, st.st_size) < 0)
2967                                 die_errno("readlink(%s)", name);
2968                         prep_temp_blob(name, temp, sb.buf, sb.len,
2969                                        (one->sha1_valid ?
2970                                         one->sha1 : null_sha1),
2971                                        (one->sha1_valid ?
2972                                         one->mode : S_IFLNK));
2973                         strbuf_release(&sb);
2974                 }
2975                 else {
2976                         /* we can borrow from the file in the work tree */
2977                         temp->name = name;
2978                         if (!one->sha1_valid)
2979                                 strcpy(temp->hex, sha1_to_hex(null_sha1));
2980                         else
2981                                 strcpy(temp->hex, sha1_to_hex(one->sha1));
2982                         /* Even though we may sometimes borrow the
2983                          * contents from the work tree, we always want
2984                          * one->mode.  mode is trustworthy even when
2985                          * !(one->sha1_valid), as long as
2986                          * DIFF_FILE_VALID(one).
2987                          */
2988                         sprintf(temp->mode, "%06o", one->mode);
2989                 }
2990                 return temp;
2991         }
2992         else {
2993                 if (diff_populate_filespec(one, 0))
2994                         die("cannot read data blob for %s", one->path);
2995                 prep_temp_blob(name, temp, one->data, one->size,
2996                                one->sha1, one->mode);
2997         }
2998         return temp;
2999 }
3000
3001 static void add_external_diff_name(struct argv_array *argv,
3002                                    const char *name,
3003                                    struct diff_filespec *df)
3004 {
3005         struct diff_tempfile *temp = prepare_temp_file(name, df);
3006         argv_array_push(argv, temp->name);
3007         argv_array_push(argv, temp->hex);
3008         argv_array_push(argv, temp->mode);
3009 }
3010
3011 /* An external diff command takes:
3012  *
3013  * diff-cmd name infile1 infile1-sha1 infile1-mode \
3014  *               infile2 infile2-sha1 infile2-mode [ rename-to ]
3015  *
3016  */
3017 static void run_external_diff(const char *pgm,
3018                               const char *name,
3019                               const char *other,
3020                               struct diff_filespec *one,
3021                               struct diff_filespec *two,
3022                               const char *xfrm_msg,
3023                               int complete_rewrite,
3024                               struct diff_options *o)
3025 {
3026         struct argv_array argv = ARGV_ARRAY_INIT;
3027         struct argv_array env = ARGV_ARRAY_INIT;
3028         struct diff_queue_struct *q = &diff_queued_diff;
3029
3030         argv_array_push(&argv, pgm);
3031         argv_array_push(&argv, name);
3032
3033         if (one && two) {
3034                 add_external_diff_name(&argv, name, one);
3035                 if (!other)
3036                         add_external_diff_name(&argv, name, two);
3037                 else {
3038                         add_external_diff_name(&argv, other, two);
3039                         argv_array_push(&argv, other);
3040                         argv_array_push(&argv, xfrm_msg);
3041                 }
3042         }
3043
3044         argv_array_pushf(&env, "GIT_DIFF_PATH_COUNTER=%d", ++o->diff_path_counter);
3045         argv_array_pushf(&env, "GIT_DIFF_PATH_TOTAL=%d", q->nr);
3046
3047         if (run_command_v_opt_cd_env(argv.argv, RUN_USING_SHELL, NULL, env.argv))
3048                 die(_("external diff died, stopping at %s"), name);
3049
3050         remove_tempfile();
3051         argv_array_clear(&argv);
3052         argv_array_clear(&env);
3053 }
3054
3055 static int similarity_index(struct diff_filepair *p)
3056 {
3057         return p->score * 100 / MAX_SCORE;
3058 }
3059
3060 static void fill_metainfo(struct strbuf *msg,
3061                           const char *name,
3062                           const char *other,
3063                           struct diff_filespec *one,
3064                           struct diff_filespec *two,
3065                           struct diff_options *o,
3066                           struct diff_filepair *p,
3067                           int *must_show_header,
3068                           int use_color)
3069 {
3070         const char *set = diff_get_color(use_color, DIFF_METAINFO);
3071         const char *reset = diff_get_color(use_color, DIFF_RESET);
3072         const char *line_prefix = diff_line_prefix(o);
3073
3074         *must_show_header = 1;
3075         strbuf_init(msg, PATH_MAX * 2 + 300);
3076         switch (p->status) {
3077         case DIFF_STATUS_COPIED:
3078                 strbuf_addf(msg, "%s%ssimilarity index %d%%",
3079                             line_prefix, set, similarity_index(p));
3080                 strbuf_addf(msg, "%s\n%s%scopy from ",
3081                             reset,  line_prefix, set);
3082                 quote_c_style(name, msg, NULL, 0);
3083                 strbuf_addf(msg, "%s\n%s%scopy to ", reset, line_prefix, set);
3084                 quote_c_style(other, msg, NULL, 0);
3085                 strbuf_addf(msg, "%s\n", reset);
3086                 break;
3087         case DIFF_STATUS_RENAMED:
3088                 strbuf_addf(msg, "%s%ssimilarity index %d%%",
3089                             line_prefix, set, similarity_index(p));
3090                 strbuf_addf(msg, "%s\n%s%srename from ",
3091                             reset, line_prefix, set);
3092                 quote_c_style(name, msg, NULL, 0);
3093                 strbuf_addf(msg, "%s\n%s%srename to ",
3094                             reset, line_prefix, set);
3095                 quote_c_style(other, msg, NULL, 0);
3096                 strbuf_addf(msg, "%s\n", reset);
3097                 break;
3098         case DIFF_STATUS_MODIFIED:
3099                 if (p->score) {
3100                         strbuf_addf(msg, "%s%sdissimilarity index %d%%%s\n",
3101                                     line_prefix,
3102                                     set, similarity_index(p), reset);
3103                         break;
3104                 }
3105                 /* fallthru */
3106         default:
3107                 *must_show_header = 0;
3108         }
3109         if (one && two && hashcmp(one->sha1, two->sha1)) {
3110                 int abbrev = DIFF_OPT_TST(o, FULL_INDEX) ? 40 : DEFAULT_ABBREV;
3111
3112                 if (DIFF_OPT_TST(o, BINARY)) {
3113                         mmfile_t mf;
3114                         if ((!fill_mmfile(&mf, one) && diff_filespec_is_binary(one)) ||
3115                             (!fill_mmfile(&mf, two) && diff_filespec_is_binary(two)))
3116                                 abbrev = 40;
3117                 }
3118                 strbuf_addf(msg, "%s%sindex %s..", line_prefix, set,
3119                             find_unique_abbrev(one->sha1, abbrev));
3120                 strbuf_addstr(msg, find_unique_abbrev(two->sha1, abbrev));
3121                 if (one->mode == two->mode)
3122                         strbuf_addf(msg, " %06o", one->mode);
3123                 strbuf_addf(msg, "%s\n", reset);
3124         }
3125 }
3126
3127 static void run_diff_cmd(const char *pgm,
3128                          const char *name,
3129                          const char *other,
3130                          const char *attr_path,
3131                          struct diff_filespec *one,
3132                          struct diff_filespec *two,
3133                          struct strbuf *msg,
3134                          struct diff_options *o,
3135                          struct diff_filepair *p)
3136 {
3137         const char *xfrm_msg = NULL;
3138         int complete_rewrite = (p->status == DIFF_STATUS_MODIFIED) && p->score;
3139         int must_show_header = 0;
3140
3141
3142         if (DIFF_OPT_TST(o, ALLOW_EXTERNAL)) {
3143                 struct userdiff_driver *drv = userdiff_find_by_path(attr_path);
3144                 if (drv && drv->external)
3145                         pgm = drv->external;
3146         }
3147
3148         if (msg) {
3149                 /*
3150                  * don't use colors when the header is intended for an
3151                  * external diff driver
3152                  */
3153                 fill_metainfo(msg, name, other, one, two, o, p,
3154                               &must_show_header,
3155                               want_color(o->use_color) && !pgm);
3156                 xfrm_msg = msg->len ? msg->buf : NULL;
3157         }
3158
3159         if (pgm) {
3160                 run_external_diff(pgm, name, other, one, two, xfrm_msg,
3161                                   complete_rewrite, o);
3162                 return;
3163         }
3164         if (one && two)
3165                 builtin_diff(name, other ? other : name,
3166                              one, two, xfrm_msg, must_show_header,
3167                              o, complete_rewrite);
3168         else
3169                 fprintf(o->file, "* Unmerged path %s\n", name);
3170 }
3171
3172 static void diff_fill_sha1_info(struct diff_filespec *one)
3173 {
3174         if (DIFF_FILE_VALID(one)) {
3175                 if (!one->sha1_valid) {
3176                         struct stat st;
3177                         if (one->is_stdin) {
3178                                 hashcpy(one->sha1, null_sha1);
3179                                 return;
3180                         }
3181                         if (lstat(one->path, &st) < 0)
3182                                 die_errno("stat '%s'", one->path);
3183                         if (index_path(one->sha1, one->path, &st, 0))
3184                                 die("cannot hash %s", one->path);
3185                 }
3186         }
3187         else
3188                 hashclr(one->sha1);
3189 }
3190
3191 static void strip_prefix(int prefix_length, const char **namep, const char **otherp)
3192 {
3193         /* Strip the prefix but do not molest /dev/null and absolute paths */
3194         if (*namep && **namep != '/') {
3195                 *namep += prefix_length;
3196                 if (**namep == '/')
3197                         ++*namep;
3198         }
3199         if (*otherp && **otherp != '/') {
3200                 *otherp += prefix_length;
3201                 if (**otherp == '/')
3202                         ++*otherp;
3203         }
3204 }
3205
3206 static void run_diff(struct diff_filepair *p, struct diff_options *o)
3207 {
3208         const char *pgm = external_diff();
3209         struct strbuf msg;
3210         struct diff_filespec *one = p->one;
3211         struct diff_filespec *two = p->two;
3212         const char *name;
3213         const char *other;
3214         const char *attr_path;
3215
3216         name  = p->one->path;
3217         other = (strcmp(name, p->two->path) ? p->two->path : NULL);
3218         attr_path = name;
3219         if (o->prefix_length)
3220                 strip_prefix(o->prefix_length, &name, &other);
3221
3222         if (!DIFF_OPT_TST(o, ALLOW_EXTERNAL))
3223                 pgm = NULL;
3224
3225         if (DIFF_PAIR_UNMERGED(p)) {
3226                 run_diff_cmd(pgm, name, NULL, attr_path,
3227                              NULL, NULL, NULL, o, p);
3228                 return;
3229         }
3230
3231         diff_fill_sha1_info(one);
3232         diff_fill_sha1_info(two);
3233
3234         if (!pgm &&
3235             DIFF_FILE_VALID(one) && DIFF_FILE_VALID(two) &&
3236             (S_IFMT & one->mode) != (S_IFMT & two->mode)) {
3237                 /*
3238                  * a filepair that changes between file and symlink
3239                  * needs to be split into deletion and creation.
3240                  */
3241                 struct diff_filespec *null = alloc_filespec(two->path);
3242                 run_diff_cmd(NULL, name, other, attr_path,
3243                              one, null, &msg, o, p);
3244                 free(null);
3245                 strbuf_release(&msg);
3246
3247                 null = alloc_filespec(one->path);
3248                 run_diff_cmd(NULL, name, other, attr_path,
3249                              null, two, &msg, o, p);
3250                 free(null);
3251         }
3252         else
3253                 run_diff_cmd(pgm, name, other, attr_path,
3254                              one, two, &msg, o, p);
3255
3256         strbuf_release(&msg);
3257 }
3258
3259 static void run_diffstat(struct diff_filepair *p, struct diff_options *o,
3260                          struct diffstat_t *diffstat)
3261 {
3262         const char *name;
3263         const char *other;
3264
3265         if (DIFF_PAIR_UNMERGED(p)) {
3266                 /* unmerged */
3267                 builtin_diffstat(p->one->path, NULL, NULL, NULL, diffstat, o, p);
3268                 return;
3269         }
3270
3271         name = p->one->path;
3272         other = (strcmp(name, p->two->path) ? p->two->path : NULL);
3273
3274         if (o->prefix_length)
3275                 strip_prefix(o->prefix_length, &name, &other);
3276
3277         diff_fill_sha1_info(p->one);
3278         diff_fill_sha1_info(p->two);
3279
3280         builtin_diffstat(name, other, p->one, p->two, diffstat, o, p);
3281 }
3282
3283 static void run_checkdiff(struct diff_filepair *p, struct diff_options *o)
3284 {
3285         const char *name;
3286         const char *other;
3287         const char *attr_path;
3288
3289         if (DIFF_PAIR_UNMERGED(p)) {
3290                 /* unmerged */
3291                 return;
3292         }
3293
3294         name = p->one->path;
3295         other = (strcmp(name, p->two->path) ? p->two->path : NULL);
3296         attr_path = other ? other : name;
3297
3298         if (o->prefix_length)
3299                 strip_prefix(o->prefix_length, &name, &other);
3300
3301         diff_fill_sha1_info(p->one);
3302         diff_fill_sha1_info(p->two);
3303
3304         builtin_checkdiff(name, other, attr_path, p->one, p->two, o);
3305 }
3306
3307 void diff_setup(struct diff_options *options)
3308 {
3309         memcpy(options, &default_diff_options, sizeof(*options));
3310
3311         options->file = stdout;
3312
3313         options->line_termination = '\n';
3314         options->break_opt = -1;
3315         options->rename_limit = -1;
3316         options->dirstat_permille = diff_dirstat_permille_default;
3317         options->context = diff_context_default;
3318         options->ws_error_highlight = ws_error_highlight_default;
3319         DIFF_OPT_SET(options, RENAME_EMPTY);
3320
3321         /* pathchange left =NULL by default */
3322         options->change = diff_change;
3323         options->add_remove = diff_addremove;
3324         options->use_color = diff_use_color_default;
3325         options->detect_rename = diff_detect_rename_default;
3326         options->xdl_opts |= diff_algorithm;
3327
3328         options->orderfile = diff_order_file_cfg;
3329
3330         if (diff_no_prefix) {
3331                 options->a_prefix = options->b_prefix = "";
3332         } else if (!diff_mnemonic_prefix) {
3333                 options->a_prefix = "a/";
3334                 options->b_prefix = "b/";
3335         }
3336 }
3337
3338 void diff_setup_done(struct diff_options *options)
3339 {
3340         int count = 0;
3341
3342         if (options->set_default)
3343                 options->set_default(options);
3344
3345         if (options->output_format & DIFF_FORMAT_NAME)
3346                 count++;
3347         if (options->output_format & DIFF_FORMAT_NAME_STATUS)
3348                 count++;
3349         if (options->output_format & DIFF_FORMAT_CHECKDIFF)
3350                 count++;
3351         if (options->output_format & DIFF_FORMAT_NO_OUTPUT)
3352                 count++;
3353         if (count > 1)
3354                 die("--name-only, --name-status, --check and -s are mutually exclusive");
3355
3356         /*
3357          * Most of the time we can say "there are changes"
3358          * only by checking if there are changed paths, but
3359          * --ignore-whitespace* options force us to look
3360          * inside contents.
3361          */
3362
3363         if (DIFF_XDL_TST(options, IGNORE_WHITESPACE) ||
3364             DIFF_XDL_TST(options, IGNORE_WHITESPACE_CHANGE) ||
3365             DIFF_XDL_TST(options, IGNORE_WHITESPACE_AT_EOL))
3366                 DIFF_OPT_SET(options, DIFF_FROM_CONTENTS);
3367         else
3368                 DIFF_OPT_CLR(options, DIFF_FROM_CONTENTS);
3369
3370         if (DIFF_OPT_TST(options, FIND_COPIES_HARDER))
3371                 options->detect_rename = DIFF_DETECT_COPY;
3372
3373         if (!DIFF_OPT_TST(options, RELATIVE_NAME))
3374                 options->prefix = NULL;
3375         if (options->prefix)
3376                 options->prefix_length = strlen(options->prefix);
3377         else
3378                 options->prefix_length = 0;
3379
3380         if (options->output_format & (DIFF_FORMAT_NAME |
3381                                       DIFF_FORMAT_NAME_STATUS |
3382                                       DIFF_FORMAT_CHECKDIFF |
3383                                       DIFF_FORMAT_NO_OUTPUT))
3384                 options->output_format &= ~(DIFF_FORMAT_RAW |
3385                                             DIFF_FORMAT_NUMSTAT |
3386                                             DIFF_FORMAT_DIFFSTAT |
3387                                             DIFF_FORMAT_SHORTSTAT |
3388                                             DIFF_FORMAT_DIRSTAT |
3389                                             DIFF_FORMAT_SUMMARY |
3390                                             DIFF_FORMAT_PATCH);
3391
3392         /*
3393          * These cases always need recursive; we do not drop caller-supplied
3394          * recursive bits for other formats here.
3395          */
3396         if (options->output_format & (DIFF_FORMAT_PATCH |
3397                                       DIFF_FORMAT_NUMSTAT |
3398                                       DIFF_FORMAT_DIFFSTAT |
3399                                       DIFF_FORMAT_SHORTSTAT |
3400                                       DIFF_FORMAT_DIRSTAT |
3401                                       DIFF_FORMAT_SUMMARY |
3402                                       DIFF_FORMAT_CHECKDIFF))
3403                 DIFF_OPT_SET(options, RECURSIVE);
3404         /*
3405          * Also pickaxe would not work very well if you do not say recursive
3406          */
3407         if (options->pickaxe)
3408                 DIFF_OPT_SET(options, RECURSIVE);
3409         /*
3410          * When patches are generated, submodules diffed against the work tree
3411          * must be checked for dirtiness too so it can be shown in the output
3412          */
3413         if (options->output_format & DIFF_FORMAT_PATCH)
3414                 DIFF_OPT_SET(options, DIRTY_SUBMODULES);
3415
3416         if (options->detect_rename && options->rename_limit < 0)
3417                 options->rename_limit = diff_rename_limit_default;
3418         if (options->setup & DIFF_SETUP_USE_CACHE) {
3419                 if (!active_cache)
3420                         /* read-cache does not die even when it fails
3421                          * so it is safe for us to do this here.  Also
3422                          * it does not smudge active_cache or active_nr
3423                          * when it fails, so we do not have to worry about
3424                          * cleaning it up ourselves either.
3425                          */
3426                         read_cache();
3427         }
3428         if (options->abbrev <= 0 || 40 < options->abbrev)
3429                 options->abbrev = 40; /* full */
3430
3431         /*
3432          * It does not make sense to show the first hit we happened
3433          * to have found.  It does not make sense not to return with
3434          * exit code in such a case either.
3435          */
3436         if (DIFF_OPT_TST(options, QUICK)) {
3437                 options->output_format = DIFF_FORMAT_NO_OUTPUT;
3438                 DIFF_OPT_SET(options, EXIT_WITH_STATUS);
3439         }
3440
3441         options->diff_path_counter = 0;
3442
3443         if (DIFF_OPT_TST(options, FOLLOW_RENAMES) && options->pathspec.nr != 1)
3444                 die(_("--follow requires exactly one pathspec"));
3445 }
3446
3447 static int opt_arg(const char *arg, int arg_short, const char *arg_long, int *val)
3448 {
3449         char c, *eq;
3450         int len;
3451
3452         if (*arg != '-')
3453                 return 0;
3454         c = *++arg;
3455         if (!c)
3456                 return 0;
3457         if (c == arg_short) {
3458                 c = *++arg;
3459                 if (!c)
3460                         return 1;
3461                 if (val && isdigit(c)) {
3462                         char *end;
3463                         int n = strtoul(arg, &end, 10);
3464                         if (*end)
3465                                 return 0;
3466                         *val = n;
3467                         return 1;
3468                 }
3469                 return 0;
3470         }
3471         if (c != '-')
3472                 return 0;
3473         arg++;
3474         eq = strchrnul(arg, '=');
3475         len = eq - arg;
3476         if (!len || strncmp(arg, arg_long, len))
3477                 return 0;
3478         if (*eq) {
3479                 int n;
3480                 char *end;
3481                 if (!isdigit(*++eq))
3482                         return 0;
3483                 n = strtoul(eq, &end, 10);
3484                 if (*end)
3485                         return 0;
3486                 *val = n;
3487         }
3488         return 1;
3489 }
3490
3491 static int diff_scoreopt_parse(const char *opt);
3492
3493 static inline int short_opt(char opt, const char **argv,
3494                             const char **optarg)
3495 {
3496         const char *arg = argv[0];
3497         if (arg[0] != '-' || arg[1] != opt)
3498                 return 0;
3499         if (arg[2] != '\0') {
3500                 *optarg = arg + 2;
3501                 return 1;
3502         }
3503         if (!argv[1])
3504                 die("Option '%c' requires a value", opt);
3505         *optarg = argv[1];
3506         return 2;
3507 }
3508
3509 int parse_long_opt(const char *opt, const char **argv,
3510                    const char **optarg)
3511 {
3512         const char *arg = argv[0];
3513         if (!skip_prefix(arg, "--", &arg))
3514                 return 0;
3515         if (!skip_prefix(arg, opt, &arg))
3516                 return 0;
3517         if (*arg == '=') { /* stuck form: --option=value */
3518                 *optarg = arg + 1;
3519                 return 1;
3520         }
3521         if (*arg != '\0')
3522                 return 0;
3523         /* separate form: --option value */
3524         if (!argv[1])
3525                 die("Option '--%s' requires a value", opt);
3526         *optarg = argv[1];
3527         return 2;
3528 }
3529
3530 static int stat_opt(struct diff_options *options, const char **av)
3531 {
3532         const char *arg = av[0];
3533         char *end;
3534         int width = options->stat_width;
3535         int name_width = options->stat_name_width;
3536         int graph_width = options->stat_graph_width;
3537         int count = options->stat_count;
3538         int argcount = 1;
3539
3540         if (!skip_prefix(arg, "--stat", &arg))
3541                 die("BUG: stat option does not begin with --stat: %s", arg);
3542         end = (char *)arg;
3543
3544         switch (*arg) {
3545         case '-':
3546                 if (skip_prefix(arg, "-width", &arg)) {
3547                         if (*arg == '=')
3548                                 width = strtoul(arg + 1, &end, 10);
3549                         else if (!*arg && !av[1])
3550                                 die("Option '--stat-width' requires a value");
3551                         else if (!*arg) {
3552                                 width = strtoul(av[1], &end, 10);
3553                                 argcount = 2;
3554                         }
3555                 } else if (skip_prefix(arg, "-name-width", &arg)) {
3556                         if (*arg == '=')
3557                                 name_width = strtoul(arg + 1, &end, 10);
3558                         else if (!*arg && !av[1])
3559                                 die("Option '--stat-name-width' requires a value");
3560                         else if (!*arg) {
3561                                 name_width = strtoul(av[1], &end, 10);
3562                                 argcount = 2;
3563                         }
3564                 } else if (skip_prefix(arg, "-graph-width", &arg)) {
3565                         if (*arg == '=')
3566                                 graph_width = strtoul(arg + 1, &end, 10);
3567                         else if (!*arg && !av[1])
3568                                 die("Option '--stat-graph-width' requires a value");
3569                         else if (!*arg) {
3570                                 graph_width = strtoul(av[1], &end, 10);
3571                                 argcount = 2;
3572                         }
3573                 } else if (skip_prefix(arg, "-count", &arg)) {
3574                         if (*arg == '=')
3575                                 count = strtoul(arg + 1, &end, 10);
3576                         else if (!*arg && !av[1])
3577                                 die("Option '--stat-count' requires a value");
3578                         else if (!*arg) {
3579                                 count = strtoul(av[1], &end, 10);
3580                                 argcount = 2;
3581                         }
3582                 }
3583                 break;
3584         case '=':
3585                 width = strtoul(arg+1, &end, 10);
3586                 if (*end == ',')
3587                         name_width = strtoul(end+1, &end, 10);
3588                 if (*end == ',')
3589                         count = strtoul(end+1, &end, 10);
3590         }
3591
3592         /* Important! This checks all the error cases! */
3593         if (*end)
3594                 return 0;
3595         options->output_format |= DIFF_FORMAT_DIFFSTAT;
3596         options->stat_name_width = name_width;
3597         options->stat_graph_width = graph_width;
3598         options->stat_width = width;
3599         options->stat_count = count;
3600         return argcount;
3601 }
3602
3603 static int parse_dirstat_opt(struct diff_options *options, const char *params)
3604 {
3605         struct strbuf errmsg = STRBUF_INIT;
3606         if (parse_dirstat_params(options, params, &errmsg))
3607                 die(_("Failed to parse --dirstat/-X option parameter:\n%s"),
3608                     errmsg.buf);
3609         strbuf_release(&errmsg);
3610         /*
3611          * The caller knows a dirstat-related option is given from the command
3612          * line; allow it to say "return this_function();"
3613          */
3614         options->output_format |= DIFF_FORMAT_DIRSTAT;
3615         return 1;
3616 }
3617
3618 static int parse_submodule_opt(struct diff_options *options, const char *value)
3619 {
3620         if (parse_submodule_params(options, value))
3621                 die(_("Failed to parse --submodule option parameter: '%s'"),
3622                         value);
3623         return 1;
3624 }
3625
3626 static const char diff_status_letters[] = {
3627         DIFF_STATUS_ADDED,
3628         DIFF_STATUS_COPIED,
3629         DIFF_STATUS_DELETED,
3630         DIFF_STATUS_MODIFIED,
3631         DIFF_STATUS_RENAMED,
3632         DIFF_STATUS_TYPE_CHANGED,
3633         DIFF_STATUS_UNKNOWN,
3634         DIFF_STATUS_UNMERGED,
3635         DIFF_STATUS_FILTER_AON,
3636         DIFF_STATUS_FILTER_BROKEN,
3637         '\0',
3638 };
3639
3640 static unsigned int filter_bit['Z' + 1];
3641
3642 static void prepare_filter_bits(void)
3643 {
3644         int i;
3645
3646         if (!filter_bit[DIFF_STATUS_ADDED]) {
3647                 for (i = 0; diff_status_letters[i]; i++)
3648                         filter_bit[(int) diff_status_letters[i]] = (1 << i);
3649         }
3650 }
3651
3652 static unsigned filter_bit_tst(char status, const struct diff_options *opt)
3653 {
3654         return opt->filter & filter_bit[(int) status];
3655 }
3656
3657 static int parse_diff_filter_opt(const char *optarg, struct diff_options *opt)
3658 {
3659         int i, optch;
3660
3661         prepare_filter_bits();
3662
3663         /*
3664          * If there is a negation e.g. 'd' in the input, and we haven't
3665          * initialized the filter field with another --diff-filter, start
3666          * from full set of bits, except for AON.
3667          */
3668         if (!opt->filter) {
3669                 for (i = 0; (optch = optarg[i]) != '\0'; i++) {
3670                         if (optch < 'a' || 'z' < optch)
3671                                 continue;
3672                         opt->filter = (1 << (ARRAY_SIZE(diff_status_letters) - 1)) - 1;
3673                         opt->filter &= ~filter_bit[DIFF_STATUS_FILTER_AON];
3674                         break;
3675                 }
3676         }
3677
3678         for (i = 0; (optch = optarg[i]) != '\0'; i++) {
3679                 unsigned int bit;
3680                 int negate;
3681
3682                 if ('a' <= optch && optch <= 'z') {
3683                         negate = 1;
3684                         optch = toupper(optch);
3685                 } else {
3686                         negate = 0;
3687                 }
3688
3689                 bit = (0 <= optch && optch <= 'Z') ? filter_bit[optch] : 0;
3690                 if (!bit)
3691                         return optarg[i];
3692                 if (negate)
3693                         opt->filter &= ~bit;
3694                 else
3695                         opt->filter |= bit;
3696         }
3697         return 0;
3698 }
3699
3700 static void enable_patch_output(int *fmt) {
3701         *fmt &= ~DIFF_FORMAT_NO_OUTPUT;
3702         *fmt |= DIFF_FORMAT_PATCH;
3703 }
3704
3705 static int parse_ws_error_highlight_opt(struct diff_options *opt, const char *arg)
3706 {
3707         int val = parse_ws_error_highlight(arg);
3708
3709         if (val < 0) {
3710                 error("unknown value after ws-error-highlight=%.*s",
3711                       -1 - val, arg);
3712                 return 0;
3713         }
3714         opt->ws_error_highlight = val;
3715         return 1;
3716 }
3717
3718 int diff_opt_parse(struct diff_options *options, const char **av, int ac)
3719 {
3720         const char *arg = av[0];
3721         const char *optarg;
3722         int argcount;
3723
3724         /* Output format options */
3725         if (!strcmp(arg, "-p") || !strcmp(arg, "-u") || !strcmp(arg, "--patch")
3726             || opt_arg(arg, 'U', "unified", &options->context))
3727                 enable_patch_output(&options->output_format);
3728         else if (!strcmp(arg, "--raw"))
3729                 options->output_format |= DIFF_FORMAT_RAW;
3730         else if (!strcmp(arg, "--patch-with-raw")) {
3731                 enable_patch_output(&options->output_format);
3732                 options->output_format |= DIFF_FORMAT_RAW;
3733         } else if (!strcmp(arg, "--numstat"))
3734                 options->output_format |= DIFF_FORMAT_NUMSTAT;
3735         else if (!strcmp(arg, "--shortstat"))
3736                 options->output_format |= DIFF_FORMAT_SHORTSTAT;
3737         else if (!strcmp(arg, "-X") || !strcmp(arg, "--dirstat"))
3738                 return parse_dirstat_opt(options, "");
3739         else if (skip_prefix(arg, "-X", &arg))
3740                 return parse_dirstat_opt(options, arg);
3741         else if (skip_prefix(arg, "--dirstat=", &arg))
3742                 return parse_dirstat_opt(options, arg);
3743         else if (!strcmp(arg, "--cumulative"))
3744                 return parse_dirstat_opt(options, "cumulative");
3745         else if (!strcmp(arg, "--dirstat-by-file"))
3746                 return parse_dirstat_opt(options, "files");
3747         else if (skip_prefix(arg, "--dirstat-by-file=", &arg)) {
3748                 parse_dirstat_opt(options, "files");
3749                 return parse_dirstat_opt(options, arg);
3750         }
3751         else if (!strcmp(arg, "--check"))
3752                 options->output_format |= DIFF_FORMAT_CHECKDIFF;
3753         else if (!strcmp(arg, "--summary"))
3754                 options->output_format |= DIFF_FORMAT_SUMMARY;
3755         else if (!strcmp(arg, "--patch-with-stat")) {
3756                 enable_patch_output(&options->output_format);
3757                 options->output_format |= DIFF_FORMAT_DIFFSTAT;
3758         } else if (!strcmp(arg, "--name-only"))
3759                 options->output_format |= DIFF_FORMAT_NAME;
3760         else if (!strcmp(arg, "--name-status"))
3761                 options->output_format |= DIFF_FORMAT_NAME_STATUS;
3762         else if (!strcmp(arg, "-s") || !strcmp(arg, "--no-patch"))
3763                 options->output_format |= DIFF_FORMAT_NO_OUTPUT;
3764         else if (starts_with(arg, "--stat"))
3765                 /* --stat, --stat-width, --stat-name-width, or --stat-count */
3766                 return stat_opt(options, av);
3767
3768         /* renames options */
3769         else if (starts_with(arg, "-B") || starts_with(arg, "--break-rewrites=") ||
3770                  !strcmp(arg, "--break-rewrites")) {
3771                 if ((options->break_opt = diff_scoreopt_parse(arg)) == -1)
3772                         return error("invalid argument to -B: %s", arg+2);
3773         }
3774         else if (starts_with(arg, "-M") || starts_with(arg, "--find-renames=") ||
3775                  !strcmp(arg, "--find-renames")) {
3776                 if ((options->rename_score = diff_scoreopt_parse(arg)) == -1)
3777                         return error("invalid argument to -M: %s", arg+2);
3778                 options->detect_rename = DIFF_DETECT_RENAME;
3779         }
3780         else if (!strcmp(arg, "-D") || !strcmp(arg, "--irreversible-delete")) {
3781                 options->irreversible_delete = 1;
3782         }
3783         else if (starts_with(arg, "-C") || starts_with(arg, "--find-copies=") ||
3784                  !strcmp(arg, "--find-copies")) {
3785                 if (options->detect_rename == DIFF_DETECT_COPY)
3786                         DIFF_OPT_SET(options, FIND_COPIES_HARDER);
3787                 if ((options->rename_score = diff_scoreopt_parse(arg)) == -1)
3788                         return error("invalid argument to -C: %s", arg+2);
3789                 options->detect_rename = DIFF_DETECT_COPY;
3790         }
3791         else if (!strcmp(arg, "--no-renames"))
3792                 options->detect_rename = 0;
3793         else if (!strcmp(arg, "--rename-empty"))
3794                 DIFF_OPT_SET(options, RENAME_EMPTY);
3795         else if (!strcmp(arg, "--no-rename-empty"))
3796                 DIFF_OPT_CLR(options, RENAME_EMPTY);
3797         else if (!strcmp(arg, "--relative"))
3798                 DIFF_OPT_SET(options, RELATIVE_NAME);
3799         else if (skip_prefix(arg, "--relative=", &arg)) {
3800                 DIFF_OPT_SET(options, RELATIVE_NAME);
3801                 options->prefix = arg;
3802         }
3803
3804         /* xdiff options */
3805         else if (!strcmp(arg, "--minimal"))
3806                 DIFF_XDL_SET(options, NEED_MINIMAL);
3807         else if (!strcmp(arg, "--no-minimal"))
3808                 DIFF_XDL_CLR(options, NEED_MINIMAL);
3809         else if (!strcmp(arg, "-w") || !strcmp(arg, "--ignore-all-space"))
3810                 DIFF_XDL_SET(options, IGNORE_WHITESPACE);
3811         else if (!strcmp(arg, "-b") || !strcmp(arg, "--ignore-space-change"))
3812                 DIFF_XDL_SET(options, IGNORE_WHITESPACE_CHANGE);
3813         else if (!strcmp(arg, "--ignore-space-at-eol"))
3814                 DIFF_XDL_SET(options, IGNORE_WHITESPACE_AT_EOL);
3815         else if (!strcmp(arg, "--ignore-blank-lines"))
3816                 DIFF_XDL_SET(options, IGNORE_BLANK_LINES);
3817         else if (!strcmp(arg, "--patience"))
3818                 options->xdl_opts = DIFF_WITH_ALG(options, PATIENCE_DIFF);
3819         else if (!strcmp(arg, "--histogram"))
3820                 options->xdl_opts = DIFF_WITH_ALG(options, HISTOGRAM_DIFF);
3821         else if ((argcount = parse_long_opt("diff-algorithm", av, &optarg))) {
3822                 long value = parse_algorithm_value(optarg);
3823                 if (value < 0)
3824                         return error("option diff-algorithm accepts \"myers\", "
3825                                      "\"minimal\", \"patience\" and \"histogram\"");
3826                 /* clear out previous settings */
3827                 DIFF_XDL_CLR(options, NEED_MINIMAL);
3828                 options->xdl_opts &= ~XDF_DIFF_ALGORITHM_MASK;
3829                 options->xdl_opts |= value;
3830                 return argcount;
3831         }
3832
3833         /* flags options */
3834         else if (!strcmp(arg, "--binary")) {
3835                 enable_patch_output(&options->output_format);
3836                 DIFF_OPT_SET(options, BINARY);
3837         }
3838         else if (!strcmp(arg, "--full-index"))
3839                 DIFF_OPT_SET(options, FULL_INDEX);
3840         else if (!strcmp(arg, "-a") || !strcmp(arg, "--text"))
3841                 DIFF_OPT_SET(options, TEXT);
3842         else if (!strcmp(arg, "-R"))
3843                 DIFF_OPT_SET(options, REVERSE_DIFF);
3844         else if (!strcmp(arg, "--find-copies-harder"))
3845                 DIFF_OPT_SET(options, FIND_COPIES_HARDER);
3846         else if (!strcmp(arg, "--follow"))
3847                 DIFF_OPT_SET(options, FOLLOW_RENAMES);
3848         else if (!strcmp(arg, "--no-follow")) {
3849                 DIFF_OPT_CLR(options, FOLLOW_RENAMES);
3850                 DIFF_OPT_CLR(options, DEFAULT_FOLLOW_RENAMES);
3851         } else if (!strcmp(arg, "--color"))
3852                 options->use_color = 1;
3853         else if (skip_prefix(arg, "--color=", &arg)) {
3854                 int value = git_config_colorbool(NULL, arg);
3855                 if (value < 0)
3856                         return error("option `color' expects \"always\", \"auto\", or \"never\"");
3857                 options->use_color = value;
3858         }
3859         else if (!strcmp(arg, "--no-color"))
3860                 options->use_color = 0;
3861         else if (!strcmp(arg, "--color-words")) {
3862                 options->use_color = 1;
3863                 options->word_diff = DIFF_WORDS_COLOR;
3864         }
3865         else if (skip_prefix(arg, "--color-words=", &arg)) {
3866                 options->use_color = 1;
3867                 options->word_diff = DIFF_WORDS_COLOR;
3868                 options->word_regex = arg;
3869         }
3870         else if (!strcmp(arg, "--word-diff")) {
3871                 if (options->word_diff == DIFF_WORDS_NONE)
3872                         options->word_diff = DIFF_WORDS_PLAIN;
3873         }
3874         else if (skip_prefix(arg, "--word-diff=", &arg)) {
3875                 if (!strcmp(arg, "plain"))
3876                         options->word_diff = DIFF_WORDS_PLAIN;
3877                 else if (!strcmp(arg, "color")) {
3878                         options->use_color = 1;
3879                         options->word_diff = DIFF_WORDS_COLOR;
3880                 }
3881                 else if (!strcmp(arg, "porcelain"))
3882                         options->word_diff = DIFF_WORDS_PORCELAIN;
3883                 else if (!strcmp(arg, "none"))
3884                         options->word_diff = DIFF_WORDS_NONE;
3885                 else
3886                         die("bad --word-diff argument: %s", arg);
3887         }
3888         else if ((argcount = parse_long_opt("word-diff-regex", av, &optarg))) {
3889                 if (options->word_diff == DIFF_WORDS_NONE)
3890                         options->word_diff = DIFF_WORDS_PLAIN;
3891                 options->word_regex = optarg;
3892                 return argcount;
3893         }
3894         else if (!strcmp(arg, "--exit-code"))
3895                 DIFF_OPT_SET(options, EXIT_WITH_STATUS);
3896         else if (!strcmp(arg, "--quiet"))
3897                 DIFF_OPT_SET(options, QUICK);
3898         else if (!strcmp(arg, "--ext-diff"))
3899                 DIFF_OPT_SET(options, ALLOW_EXTERNAL);
3900         else if (!strcmp(arg, "--no-ext-diff"))
3901                 DIFF_OPT_CLR(options, ALLOW_EXTERNAL);
3902         else if (!strcmp(arg, "--textconv"))
3903                 DIFF_OPT_SET(options, ALLOW_TEXTCONV);
3904         else if (!strcmp(arg, "--no-textconv"))
3905                 DIFF_OPT_CLR(options, ALLOW_TEXTCONV);
3906         else if (!strcmp(arg, "--ignore-submodules")) {
3907                 DIFF_OPT_SET(options, OVERRIDE_SUBMODULE_CONFIG);
3908                 handle_ignore_submodules_arg(options, "all");
3909         } else if (skip_prefix(arg, "--ignore-submodules=", &arg)) {
3910                 DIFF_OPT_SET(options, OVERRIDE_SUBMODULE_CONFIG);
3911                 handle_ignore_submodules_arg(options, arg);
3912         } else if (!strcmp(arg, "--submodule"))
3913                 DIFF_OPT_SET(options, SUBMODULE_LOG);
3914         else if (skip_prefix(arg, "--submodule=", &arg))
3915                 return parse_submodule_opt(options, arg);
3916         else if (skip_prefix(arg, "--ws-error-highlight=", &arg))
3917                 return parse_ws_error_highlight_opt(options, arg);
3918
3919         /* misc options */
3920         else if (!strcmp(arg, "-z"))
3921                 options->line_termination = 0;
3922         else if ((argcount = short_opt('l', av, &optarg))) {
3923                 options->rename_limit = strtoul(optarg, NULL, 10);
3924                 return argcount;
3925         }
3926         else if ((argcount = short_opt('S', av, &optarg))) {
3927                 options->pickaxe = optarg;
3928                 options->pickaxe_opts |= DIFF_PICKAXE_KIND_S;
3929                 return argcount;
3930         } else if ((argcount = short_opt('G', av, &optarg))) {
3931                 options->pickaxe = optarg;
3932                 options->pickaxe_opts |= DIFF_PICKAXE_KIND_G;
3933                 return argcount;
3934         }
3935         else if (!strcmp(arg, "--pickaxe-all"))
3936                 options->pickaxe_opts |= DIFF_PICKAXE_ALL;
3937         else if (!strcmp(arg, "--pickaxe-regex"))
3938                 options->pickaxe_opts |= DIFF_PICKAXE_REGEX;
3939         else if ((argcount = short_opt('O', av, &optarg))) {
3940                 options->orderfile = optarg;
3941                 return argcount;
3942         }
3943         else if ((argcount = parse_long_opt("diff-filter", av, &optarg))) {
3944                 int offending = parse_diff_filter_opt(optarg, options);
3945                 if (offending)
3946                         die("unknown change class '%c' in --diff-filter=%s",
3947                             offending, optarg);
3948                 return argcount;
3949         }
3950         else if (!strcmp(arg, "--abbrev"))
3951                 options->abbrev = DEFAULT_ABBREV;
3952         else if (skip_prefix(arg, "--abbrev=", &arg)) {
3953                 options->abbrev = strtoul(arg, NULL, 10);
3954                 if (options->abbrev < MINIMUM_ABBREV)
3955                         options->abbrev = MINIMUM_ABBREV;
3956                 else if (40 < options->abbrev)
3957                         options->abbrev = 40;
3958         }
3959         else if ((argcount = parse_long_opt("src-prefix", av, &optarg))) {
3960                 options->a_prefix = optarg;
3961                 return argcount;
3962         }
3963         else if ((argcount = parse_long_opt("dst-prefix", av, &optarg))) {
3964                 options->b_prefix = optarg;
3965                 return argcount;
3966         }
3967         else if (!strcmp(arg, "--no-prefix"))
3968                 options->a_prefix = options->b_prefix = "";
3969         else if (opt_arg(arg, '\0', "inter-hunk-context",
3970                          &options->interhunkcontext))
3971                 ;
3972         else if (!strcmp(arg, "-W"))
3973                 DIFF_OPT_SET(options, FUNCCONTEXT);
3974         else if (!strcmp(arg, "--function-context"))
3975                 DIFF_OPT_SET(options, FUNCCONTEXT);
3976         else if (!strcmp(arg, "--no-function-context"))
3977                 DIFF_OPT_CLR(options, FUNCCONTEXT);
3978         else if ((argcount = parse_long_opt("output", av, &optarg))) {
3979                 options->file = fopen(optarg, "w");
3980                 if (!options->file)
3981                         die_errno("Could not open '%s'", optarg);
3982                 options->close_file = 1;
3983                 return argcount;
3984         } else
3985                 return 0;
3986         return 1;
3987 }
3988
3989 int parse_rename_score(const char **cp_p)
3990 {
3991         unsigned long num, scale;
3992         int ch, dot;
3993         const char *cp = *cp_p;
3994
3995         num = 0;
3996         scale = 1;
3997         dot = 0;
3998         for (;;) {
3999                 ch = *cp;
4000                 if ( !dot && ch == '.' ) {
4001                         scale = 1;
4002                         dot = 1;
4003                 } else if ( ch == '%' ) {
4004                         scale = dot ? scale*100 : 100;
4005                         cp++;   /* % is always at the end */
4006                         break;
4007                 } else if ( ch >= '0' && ch <= '9' ) {
4008                         if ( scale < 100000 ) {
4009                                 scale *= 10;
4010                                 num = (num*10) + (ch-'0');
4011                         }
4012                 } else {
4013                         break;
4014                 }
4015                 cp++;
4016         }
4017         *cp_p = cp;
4018
4019         /* user says num divided by scale and we say internally that
4020          * is MAX_SCORE * num / scale.
4021          */
4022         return (int)((num >= scale) ? MAX_SCORE : (MAX_SCORE * num / scale));
4023 }
4024
4025 static int diff_scoreopt_parse(const char *opt)
4026 {
4027         int opt1, opt2, cmd;
4028
4029         if (*opt++ != '-')
4030                 return -1;
4031         cmd = *opt++;
4032         if (cmd == '-') {
4033                 /* convert the long-form arguments into short-form versions */
4034                 if (skip_prefix(opt, "break-rewrites", &opt)) {
4035                         if (*opt == 0 || *opt++ == '=')
4036                                 cmd = 'B';
4037                 } else if (skip_prefix(opt, "find-copies", &opt)) {
4038                         if (*opt == 0 || *opt++ == '=')
4039                                 cmd = 'C';
4040                 } else if (skip_prefix(opt, "find-renames", &opt)) {
4041                         if (*opt == 0 || *opt++ == '=')
4042                                 cmd = 'M';
4043                 }
4044         }
4045         if (cmd != 'M' && cmd != 'C' && cmd != 'B')
4046                 return -1; /* that is not a -M, -C, or -B option */
4047
4048         opt1 = parse_rename_score(&opt);
4049         if (cmd != 'B')
4050                 opt2 = 0;
4051         else {
4052                 if (*opt == 0)
4053                         opt2 = 0;
4054                 else if (*opt != '/')
4055                         return -1; /* we expect -B80/99 or -B80 */
4056                 else {
4057                         opt++;
4058                         opt2 = parse_rename_score(&opt);
4059                 }
4060         }
4061         if (*opt != 0)
4062                 return -1;
4063         return opt1 | (opt2 << 16);
4064 }
4065
4066 struct diff_queue_struct diff_queued_diff;
4067
4068 void diff_q(struct diff_queue_struct *queue, struct diff_filepair *dp)
4069 {
4070         ALLOC_GROW(queue->queue, queue->nr + 1, queue->alloc);
4071         queue->queue[queue->nr++] = dp;
4072 }
4073
4074 struct diff_filepair *diff_queue(struct diff_queue_struct *queue,
4075                                  struct diff_filespec *one,
4076                                  struct diff_filespec *two)
4077 {
4078         struct diff_filepair *dp = xcalloc(1, sizeof(*dp));
4079         dp->one = one;
4080         dp->two = two;
4081         if (queue)
4082                 diff_q(queue, dp);
4083         return dp;
4084 }
4085
4086 void diff_free_filepair(struct diff_filepair *p)
4087 {
4088         free_filespec(p->one);
4089         free_filespec(p->two);
4090         free(p);
4091 }
4092
4093 /* This is different from find_unique_abbrev() in that
4094  * it stuffs the result with dots for alignment.
4095  */
4096 const char *diff_unique_abbrev(const unsigned char *sha1, int len)
4097 {
4098         int abblen;
4099         const char *abbrev;
4100         if (len == 40)
4101                 return sha1_to_hex(sha1);
4102
4103         abbrev = find_unique_abbrev(sha1, len);
4104         abblen = strlen(abbrev);
4105         if (abblen < 37) {
4106                 static char hex[41];
4107                 if (len < abblen && abblen <= len + 2)
4108                         sprintf(hex, "%s%.*s", abbrev, len+3-abblen, "..");
4109                 else
4110                         sprintf(hex, "%s...", abbrev);
4111                 return hex;
4112         }
4113         return sha1_to_hex(sha1);
4114 }
4115
4116 static void diff_flush_raw(struct diff_filepair *p, struct diff_options *opt)
4117 {
4118         int line_termination = opt->line_termination;
4119         int inter_name_termination = line_termination ? '\t' : '\0';
4120
4121         fprintf(opt->file, "%s", diff_line_prefix(opt));
4122         if (!(opt->output_format & DIFF_FORMAT_NAME_STATUS)) {
4123                 fprintf(opt->file, ":%06o %06o %s ", p->one->mode, p->two->mode,
4124                         diff_unique_abbrev(p->one->sha1, opt->abbrev));
4125                 fprintf(opt->file, "%s ", diff_unique_abbrev(p->two->sha1, opt->abbrev));
4126         }
4127         if (p->score) {
4128                 fprintf(opt->file, "%c%03d%c", p->status, similarity_index(p),
4129                         inter_name_termination);
4130         } else {
4131                 fprintf(opt->file, "%c%c", p->status, inter_name_termination);
4132         }
4133
4134         if (p->status == DIFF_STATUS_COPIED ||
4135             p->status == DIFF_STATUS_RENAMED) {
4136                 const char *name_a, *name_b;
4137                 name_a = p->one->path;
4138                 name_b = p->two->path;
4139                 strip_prefix(opt->prefix_length, &name_a, &name_b);
4140                 write_name_quoted(name_a, opt->file, inter_name_termination);
4141                 write_name_quoted(name_b, opt->file, line_termination);
4142         } else {
4143                 const char *name_a, *name_b;
4144                 name_a = p->one->mode ? p->one->path : p->two->path;
4145                 name_b = NULL;
4146                 strip_prefix(opt->prefix_length, &name_a, &name_b);
4147                 write_name_quoted(name_a, opt->file, line_termination);
4148         }
4149 }
4150
4151 int diff_unmodified_pair(struct diff_filepair *p)
4152 {
4153         /* This function is written stricter than necessary to support
4154          * the currently implemented transformers, but the idea is to
4155          * let transformers to produce diff_filepairs any way they want,
4156          * and filter and clean them up here before producing the output.
4157          */
4158         struct diff_filespec *one = p->one, *two = p->two;
4159
4160         if (DIFF_PAIR_UNMERGED(p))
4161                 return 0; /* unmerged is interesting */
4162
4163         /* deletion, addition, mode or type change
4164          * and rename are all interesting.
4165          */
4166         if (DIFF_FILE_VALID(one) != DIFF_FILE_VALID(two) ||
4167             DIFF_PAIR_MODE_CHANGED(p) ||
4168             strcmp(one->path, two->path))
4169                 return 0;
4170
4171         /* both are valid and point at the same path.  that is, we are
4172          * dealing with a change.
4173          */
4174         if (one->sha1_valid && two->sha1_valid &&
4175             !hashcmp(one->sha1, two->sha1) &&
4176             !one->dirty_submodule && !two->dirty_submodule)
4177                 return 1; /* no change */
4178         if (!one->sha1_valid && !two->sha1_valid)
4179                 return 1; /* both look at the same file on the filesystem. */
4180         return 0;
4181 }
4182
4183 static void diff_flush_patch(struct diff_filepair *p, struct diff_options *o)
4184 {
4185         if (diff_unmodified_pair(p))
4186                 return;
4187
4188         if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
4189             (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
4190                 return; /* no tree diffs in patch format */
4191
4192         run_diff(p, o);
4193 }
4194
4195 static void diff_flush_stat(struct diff_filepair *p, struct diff_options *o,
4196                             struct diffstat_t *diffstat)
4197 {
4198         if (diff_unmodified_pair(p))
4199                 return;
4200
4201         if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
4202             (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
4203                 return; /* no useful stat for tree diffs */
4204
4205         run_diffstat(p, o, diffstat);
4206 }
4207
4208 static void diff_flush_checkdiff(struct diff_filepair *p,
4209                 struct diff_options *o)
4210 {
4211         if (diff_unmodified_pair(p))
4212                 return;
4213
4214         if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
4215             (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
4216                 return; /* nothing to check in tree diffs */
4217
4218         run_checkdiff(p, o);
4219 }
4220
4221 int diff_queue_is_empty(void)
4222 {
4223         struct diff_queue_struct *q = &diff_queued_diff;
4224         int i;
4225         for (i = 0; i < q->nr; i++)
4226                 if (!diff_unmodified_pair(q->queue[i]))
4227                         return 0;
4228         return 1;
4229 }
4230
4231 #if DIFF_DEBUG
4232 void diff_debug_filespec(struct diff_filespec *s, int x, const char *one)
4233 {
4234         fprintf(stderr, "queue[%d] %s (%s) %s %06o %s\n",
4235                 x, one ? one : "",
4236                 s->path,
4237                 DIFF_FILE_VALID(s) ? "valid" : "invalid",
4238                 s->mode,
4239                 s->sha1_valid ? sha1_to_hex(s->sha1) : "");
4240         fprintf(stderr, "queue[%d] %s size %lu\n",
4241                 x, one ? one : "",
4242                 s->size);
4243 }
4244
4245 void diff_debug_filepair(const struct diff_filepair *p, int i)
4246 {
4247         diff_debug_filespec(p->one, i, "one");
4248         diff_debug_filespec(p->two, i, "two");
4249         fprintf(stderr, "score %d, status %c rename_used %d broken %d\n",
4250                 p->score, p->status ? p->status : '?',
4251                 p->one->rename_used, p->broken_pair);
4252 }
4253
4254 void diff_debug_queue(const char *msg, struct diff_queue_struct *q)
4255 {
4256         int i;
4257         if (msg)
4258                 fprintf(stderr, "%s\n", msg);
4259         fprintf(stderr, "q->nr = %d\n", q->nr);
4260         for (i = 0; i < q->nr; i++) {
4261                 struct diff_filepair *p = q->queue[i];
4262                 diff_debug_filepair(p, i);
4263         }
4264 }
4265 #endif
4266
4267 static void diff_resolve_rename_copy(void)
4268 {
4269         int i;
4270         struct diff_filepair *p;
4271         struct diff_queue_struct *q = &diff_queued_diff;
4272
4273         diff_debug_queue("resolve-rename-copy", q);
4274
4275         for (i = 0; i < q->nr; i++) {
4276                 p = q->queue[i];
4277                 p->status = 0; /* undecided */
4278                 if (DIFF_PAIR_UNMERGED(p))
4279                         p->status = DIFF_STATUS_UNMERGED;
4280                 else if (!DIFF_FILE_VALID(p->one))
4281                         p->status = DIFF_STATUS_ADDED;
4282                 else if (!DIFF_FILE_VALID(p->two))
4283                         p->status = DIFF_STATUS_DELETED;
4284                 else if (DIFF_PAIR_TYPE_CHANGED(p))
4285                         p->status = DIFF_STATUS_TYPE_CHANGED;
4286
4287                 /* from this point on, we are dealing with a pair
4288                  * whose both sides are valid and of the same type, i.e.
4289                  * either in-place edit or rename/copy edit.
4290                  */
4291                 else if (DIFF_PAIR_RENAME(p)) {
4292                         /*
4293                          * A rename might have re-connected a broken
4294                          * pair up, causing the pathnames to be the
4295                          * same again. If so, that's not a rename at
4296                          * all, just a modification..
4297                          *
4298                          * Otherwise, see if this source was used for
4299                          * multiple renames, in which case we decrement
4300                          * the count, and call it a copy.
4301                          */
4302                         if (!strcmp(p->one->path, p->two->path))
4303                                 p->status = DIFF_STATUS_MODIFIED;
4304                         else if (--p->one->rename_used > 0)
4305                                 p->status = DIFF_STATUS_COPIED;
4306                         else
4307                                 p->status = DIFF_STATUS_RENAMED;
4308                 }
4309                 else if (hashcmp(p->one->sha1, p->two->sha1) ||
4310                          p->one->mode != p->two->mode ||
4311                          p->one->dirty_submodule ||
4312                          p->two->dirty_submodule ||
4313                          is_null_sha1(p->one->sha1))
4314                         p->status = DIFF_STATUS_MODIFIED;
4315                 else {
4316                         /* This is a "no-change" entry and should not
4317                          * happen anymore, but prepare for broken callers.
4318                          */
4319                         error("feeding unmodified %s to diffcore",
4320                               p->one->path);
4321                         p->status = DIFF_STATUS_UNKNOWN;
4322                 }
4323         }
4324         diff_debug_queue("resolve-rename-copy done", q);
4325 }
4326
4327 static int check_pair_status(struct diff_filepair *p)
4328 {
4329         switch (p->status) {
4330         case DIFF_STATUS_UNKNOWN:
4331                 return 0;
4332         case 0:
4333                 die("internal error in diff-resolve-rename-copy");
4334         default:
4335                 return 1;
4336         }
4337 }
4338
4339 static void flush_one_pair(struct diff_filepair *p, struct diff_options *opt)
4340 {
4341         int fmt = opt->output_format;
4342
4343         if (fmt & DIFF_FORMAT_CHECKDIFF)
4344                 diff_flush_checkdiff(p, opt);
4345         else if (fmt & (DIFF_FORMAT_RAW | DIFF_FORMAT_NAME_STATUS))
4346                 diff_flush_raw(p, opt);
4347         else if (fmt & DIFF_FORMAT_NAME) {
4348                 const char *name_a, *name_b;
4349                 name_a = p->two->path;
4350                 name_b = NULL;
4351                 strip_prefix(opt->prefix_length, &name_a, &name_b);
4352                 write_name_quoted(name_a, opt->file, opt->line_termination);
4353         }
4354 }
4355
4356 static void show_file_mode_name(FILE *file, const char *newdelete, struct diff_filespec *fs)
4357 {
4358         if (fs->mode)
4359                 fprintf(file, " %s mode %06o ", newdelete, fs->mode);
4360         else
4361                 fprintf(file, " %s ", newdelete);
4362         write_name_quoted(fs->path, file, '\n');
4363 }
4364
4365
4366 static void show_mode_change(FILE *file, struct diff_filepair *p, int show_name,
4367                 const char *line_prefix)
4368 {
4369         if (p->one->mode && p->two->mode && p->one->mode != p->two->mode) {
4370                 fprintf(file, "%s mode change %06o => %06o%c", line_prefix, p->one->mode,
4371                         p->two->mode, show_name ? ' ' : '\n');
4372                 if (show_name) {
4373                         write_name_quoted(p->two->path, file, '\n');
4374                 }
4375         }
4376 }
4377
4378 static void show_rename_copy(FILE *file, const char *renamecopy, struct diff_filepair *p,
4379                         const char *line_prefix)
4380 {
4381         char *names = pprint_rename(p->one->path, p->two->path);
4382
4383         fprintf(file, " %s %s (%d%%)\n", renamecopy, names, similarity_index(p));
4384         free(names);
4385         show_mode_change(file, p, 0, line_prefix);
4386 }
4387
4388 static void diff_summary(struct diff_options *opt, struct diff_filepair *p)
4389 {
4390         FILE *file = opt->file;
4391         const char *line_prefix = diff_line_prefix(opt);
4392
4393         switch(p->status) {
4394         case DIFF_STATUS_DELETED:
4395                 fputs(line_prefix, file);
4396                 show_file_mode_name(file, "delete", p->one);
4397                 break;
4398         case DIFF_STATUS_ADDED:
4399                 fputs(line_prefix, file);
4400                 show_file_mode_name(file, "create", p->two);
4401                 break;
4402         case DIFF_STATUS_COPIED:
4403                 fputs(line_prefix, file);
4404                 show_rename_copy(file, "copy", p, line_prefix);
4405                 break;
4406         case DIFF_STATUS_RENAMED:
4407                 fputs(line_prefix, file);
4408                 show_rename_copy(file, "rename", p, line_prefix);
4409                 break;
4410         default:
4411                 if (p->score) {
4412                         fprintf(file, "%s rewrite ", line_prefix);
4413                         write_name_quoted(p->two->path, file, ' ');
4414                         fprintf(file, "(%d%%)\n", similarity_index(p));
4415                 }
4416                 show_mode_change(file, p, !p->score, line_prefix);
4417                 break;
4418         }
4419 }
4420
4421 struct patch_id_t {
4422         git_SHA_CTX *ctx;
4423         int patchlen;
4424 };
4425
4426 static int remove_space(char *line, int len)
4427 {
4428         int i;
4429         char *dst = line;
4430         unsigned char c;
4431
4432         for (i = 0; i < len; i++)
4433                 if (!isspace((c = line[i])))
4434                         *dst++ = c;
4435
4436         return dst - line;
4437 }
4438
4439 static void patch_id_consume(void *priv, char *line, unsigned long len)
4440 {
4441         struct patch_id_t *data = priv;
4442         int new_len;
4443
4444         /* Ignore line numbers when computing the SHA1 of the patch */
4445         if (starts_with(line, "@@ -"))
4446                 return;
4447
4448         new_len = remove_space(line, len);
4449
4450         git_SHA1_Update(data->ctx, line, new_len);
4451         data->patchlen += new_len;
4452 }
4453
4454 /* returns 0 upon success, and writes result into sha1 */
4455 static int diff_get_patch_id(struct diff_options *options, unsigned char *sha1)
4456 {
4457         struct diff_queue_struct *q = &diff_queued_diff;
4458         int i;
4459         git_SHA_CTX ctx;
4460         struct patch_id_t data;
4461         char buffer[PATH_MAX * 4 + 20];
4462
4463         git_SHA1_Init(&ctx);
4464         memset(&data, 0, sizeof(struct patch_id_t));
4465         data.ctx = &ctx;
4466
4467         for (i = 0; i < q->nr; i++) {
4468                 xpparam_t xpp;
4469                 xdemitconf_t xecfg;
4470                 mmfile_t mf1, mf2;
4471                 struct diff_filepair *p = q->queue[i];
4472                 int len1, len2;
4473
4474                 memset(&xpp, 0, sizeof(xpp));
4475                 memset(&xecfg, 0, sizeof(xecfg));
4476                 if (p->status == 0)
4477                         return error("internal diff status error");
4478                 if (p->status == DIFF_STATUS_UNKNOWN)
4479                         continue;
4480                 if (diff_unmodified_pair(p))
4481                         continue;
4482                 if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
4483                     (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
4484                         continue;
4485                 if (DIFF_PAIR_UNMERGED(p))
4486                         continue;
4487
4488                 diff_fill_sha1_info(p->one);
4489                 diff_fill_sha1_info(p->two);
4490                 if (fill_mmfile(&mf1, p->one) < 0 ||
4491                                 fill_mmfile(&mf2, p->two) < 0)
4492                         return error("unable to read files to diff");
4493
4494                 len1 = remove_space(p->one->path, strlen(p->one->path));
4495                 len2 = remove_space(p->two->path, strlen(p->two->path));
4496                 if (p->one->mode == 0)
4497                         len1 = snprintf(buffer, sizeof(buffer),
4498                                         "diff--gita/%.*sb/%.*s"
4499                                         "newfilemode%06o"
4500                                         "---/dev/null"
4501                                         "+++b/%.*s",
4502                                         len1, p->one->path,
4503                                         len2, p->two->path,
4504                                         p->two->mode,
4505                                         len2, p->two->path);
4506                 else if (p->two->mode == 0)
4507                         len1 = snprintf(buffer, sizeof(buffer),
4508                                         "diff--gita/%.*sb/%.*s"
4509                                         "deletedfilemode%06o"
4510                                         "---a/%.*s"
4511                                         "+++/dev/null",
4512                                         len1, p->one->path,
4513                                         len2, p->two->path,
4514                                         p->one->mode,
4515                                         len1, p->one->path);
4516                 else
4517                         len1 = snprintf(buffer, sizeof(buffer),
4518                                         "diff--gita/%.*sb/%.*s"
4519                                         "---a/%.*s"
4520                                         "+++b/%.*s",
4521                                         len1, p->one->path,
4522                                         len2, p->two->path,
4523                                         len1, p->one->path,
4524                                         len2, p->two->path);
4525                 git_SHA1_Update(&ctx, buffer, len1);
4526
4527                 if (diff_filespec_is_binary(p->one) ||
4528                     diff_filespec_is_binary(p->two)) {
4529                         git_SHA1_Update(&ctx, sha1_to_hex(p->one->sha1), 40);
4530                         git_SHA1_Update(&ctx, sha1_to_hex(p->two->sha1), 40);
4531                         continue;
4532                 }
4533
4534                 xpp.flags = 0;
4535                 xecfg.ctxlen = 3;
4536                 xecfg.flags = 0;
4537                 if (xdi_diff_outf(&mf1, &mf2, patch_id_consume, &data,
4538                                   &xpp, &xecfg))
4539                         return error("unable to generate patch-id diff for %s",
4540                                      p->one->path);
4541         }
4542
4543         git_SHA1_Final(sha1, &ctx);
4544         return 0;
4545 }
4546
4547 int diff_flush_patch_id(struct diff_options *options, unsigned char *sha1)
4548 {
4549         struct diff_queue_struct *q = &diff_queued_diff;
4550         int i;
4551         int result = diff_get_patch_id(options, sha1);
4552
4553         for (i = 0; i < q->nr; i++)
4554                 diff_free_filepair(q->queue[i]);
4555
4556         free(q->queue);
4557         DIFF_QUEUE_CLEAR(q);
4558
4559         return result;
4560 }
4561
4562 static int is_summary_empty(const struct diff_queue_struct *q)
4563 {
4564         int i;
4565
4566         for (i = 0; i < q->nr; i++) {
4567                 const struct diff_filepair *p = q->queue[i];
4568
4569                 switch (p->status) {
4570                 case DIFF_STATUS_DELETED:
4571                 case DIFF_STATUS_ADDED:
4572                 case DIFF_STATUS_COPIED:
4573                 case DIFF_STATUS_RENAMED:
4574                         return 0;
4575                 default:
4576                         if (p->score)
4577                                 return 0;
4578                         if (p->one->mode && p->two->mode &&
4579                             p->one->mode != p->two->mode)
4580                                 return 0;
4581                         break;
4582                 }
4583         }
4584         return 1;
4585 }
4586
4587 static const char rename_limit_warning[] =
4588 "inexact rename detection was skipped due to too many files.";
4589
4590 static const char degrade_cc_to_c_warning[] =
4591 "only found copies from modified paths due to too many files.";
4592
4593 static const char rename_limit_advice[] =
4594 "you may want to set your %s variable to at least "
4595 "%d and retry the command.";
4596
4597 void diff_warn_rename_limit(const char *varname, int needed, int degraded_cc)
4598 {
4599         if (degraded_cc)
4600                 warning(degrade_cc_to_c_warning);
4601         else if (needed)
4602                 warning(rename_limit_warning);
4603         else
4604                 return;
4605         if (0 < needed && needed < 32767)
4606                 warning(rename_limit_advice, varname, needed);
4607 }
4608
4609 void diff_flush(struct diff_options *options)
4610 {
4611         struct diff_queue_struct *q = &diff_queued_diff;
4612         int i, output_format = options->output_format;
4613         int separator = 0;
4614         int dirstat_by_line = 0;
4615
4616         /*
4617          * Order: raw, stat, summary, patch
4618          * or:    name/name-status/checkdiff (other bits clear)
4619          */
4620         if (!q->nr)
4621                 goto free_queue;
4622
4623         if (output_format & (DIFF_FORMAT_RAW |
4624                              DIFF_FORMAT_NAME |
4625                              DIFF_FORMAT_NAME_STATUS |
4626                              DIFF_FORMAT_CHECKDIFF)) {
4627                 for (i = 0; i < q->nr; i++) {
4628                         struct diff_filepair *p = q->queue[i];
4629                         if (check_pair_status(p))
4630                                 flush_one_pair(p, options);
4631                 }
4632                 separator++;
4633         }
4634
4635         if (output_format & DIFF_FORMAT_DIRSTAT && DIFF_OPT_TST(options, DIRSTAT_BY_LINE))
4636                 dirstat_by_line = 1;
4637
4638         if (output_format & (DIFF_FORMAT_DIFFSTAT|DIFF_FORMAT_SHORTSTAT|DIFF_FORMAT_NUMSTAT) ||
4639             dirstat_by_line) {
4640                 struct diffstat_t diffstat;
4641
4642                 memset(&diffstat, 0, sizeof(struct diffstat_t));
4643                 for (i = 0; i < q->nr; i++) {
4644                         struct diff_filepair *p = q->queue[i];
4645                         if (check_pair_status(p))
4646                                 diff_flush_stat(p, options, &diffstat);
4647                 }
4648                 if (output_format & DIFF_FORMAT_NUMSTAT)
4649                         show_numstat(&diffstat, options);
4650                 if (output_format & DIFF_FORMAT_DIFFSTAT)
4651                         show_stats(&diffstat, options);
4652                 if (output_format & DIFF_FORMAT_SHORTSTAT)
4653                         show_shortstats(&diffstat, options);
4654                 if (output_format & DIFF_FORMAT_DIRSTAT && dirstat_by_line)
4655                         show_dirstat_by_line(&diffstat, options);
4656                 free_diffstat_info(&diffstat);
4657                 separator++;
4658         }
4659         if ((output_format & DIFF_FORMAT_DIRSTAT) && !dirstat_by_line)
4660                 show_dirstat(options);
4661
4662         if (output_format & DIFF_FORMAT_SUMMARY && !is_summary_empty(q)) {
4663                 for (i = 0; i < q->nr; i++) {
4664                         diff_summary(options, q->queue[i]);
4665                 }
4666                 separator++;
4667         }
4668
4669         if (output_format & DIFF_FORMAT_NO_OUTPUT &&
4670             DIFF_OPT_TST(options, EXIT_WITH_STATUS) &&
4671             DIFF_OPT_TST(options, DIFF_FROM_CONTENTS)) {
4672                 /*
4673                  * run diff_flush_patch for the exit status. setting
4674                  * options->file to /dev/null should be safe, because we
4675                  * aren't supposed to produce any output anyway.
4676                  */
4677                 if (options->close_file)
4678                         fclose(options->file);
4679                 options->file = fopen("/dev/null", "w");
4680                 if (!options->file)
4681                         die_errno("Could not open /dev/null");
4682                 options->close_file = 1;
4683                 for (i = 0; i < q->nr; i++) {
4684                         struct diff_filepair *p = q->queue[i];
4685                         if (check_pair_status(p))
4686                                 diff_flush_patch(p, options);
4687                         if (options->found_changes)
4688                                 break;
4689                 }
4690         }
4691
4692         if (output_format & DIFF_FORMAT_PATCH) {
4693                 if (separator) {
4694                         fprintf(options->file, "%s%c",
4695                                 diff_line_prefix(options),
4696                                 options->line_termination);
4697                         if (options->stat_sep) {
4698                                 /* attach patch instead of inline */
4699                                 fputs(options->stat_sep, options->file);
4700                         }
4701                 }
4702
4703                 for (i = 0; i < q->nr; i++) {
4704                         struct diff_filepair *p = q->queue[i];
4705                         if (check_pair_status(p))
4706                                 diff_flush_patch(p, options);
4707                 }
4708         }
4709
4710         if (output_format & DIFF_FORMAT_CALLBACK)
4711                 options->format_callback(q, options, options->format_callback_data);
4712
4713         for (i = 0; i < q->nr; i++)
4714                 diff_free_filepair(q->queue[i]);
4715 free_queue:
4716         free(q->queue);
4717         DIFF_QUEUE_CLEAR(q);
4718         if (options->close_file)
4719                 fclose(options->file);
4720
4721         /*
4722          * Report the content-level differences with HAS_CHANGES;
4723          * diff_addremove/diff_change does not set the bit when
4724          * DIFF_FROM_CONTENTS is in effect (e.g. with -w).
4725          */
4726         if (DIFF_OPT_TST(options, DIFF_FROM_CONTENTS)) {
4727                 if (options->found_changes)
4728                         DIFF_OPT_SET(options, HAS_CHANGES);
4729                 else
4730                         DIFF_OPT_CLR(options, HAS_CHANGES);
4731         }
4732 }
4733
4734 static int match_filter(const struct diff_options *options, const struct diff_filepair *p)
4735 {
4736         return (((p->status == DIFF_STATUS_MODIFIED) &&
4737                  ((p->score &&
4738                    filter_bit_tst(DIFF_STATUS_FILTER_BROKEN, options)) ||
4739                   (!p->score &&
4740                    filter_bit_tst(DIFF_STATUS_MODIFIED, options)))) ||
4741                 ((p->status != DIFF_STATUS_MODIFIED) &&
4742                  filter_bit_tst(p->status, options)));
4743 }
4744
4745 static void diffcore_apply_filter(struct diff_options *options)
4746 {
4747         int i;
4748         struct diff_queue_struct *q = &diff_queued_diff;
4749         struct diff_queue_struct outq;
4750
4751         DIFF_QUEUE_CLEAR(&outq);
4752
4753         if (!options->filter)
4754                 return;
4755
4756         if (filter_bit_tst(DIFF_STATUS_FILTER_AON, options)) {
4757                 int found;
4758                 for (i = found = 0; !found && i < q->nr; i++) {
4759                         if (match_filter(options, q->queue[i]))
4760                                 found++;
4761                 }
4762                 if (found)
4763                         return;
4764
4765                 /* otherwise we will clear the whole queue
4766                  * by copying the empty outq at the end of this
4767                  * function, but first clear the current entries
4768                  * in the queue.
4769                  */
4770                 for (i = 0; i < q->nr; i++)
4771                         diff_free_filepair(q->queue[i]);
4772         }
4773         else {
4774                 /* Only the matching ones */
4775                 for (i = 0; i < q->nr; i++) {
4776                         struct diff_filepair *p = q->queue[i];
4777                         if (match_filter(options, p))
4778                                 diff_q(&outq, p);
4779                         else
4780                                 diff_free_filepair(p);
4781                 }
4782         }
4783         free(q->queue);
4784         *q = outq;
4785 }
4786
4787 /* Check whether two filespecs with the same mode and size are identical */
4788 static int diff_filespec_is_identical(struct diff_filespec *one,
4789                                       struct diff_filespec *two)
4790 {
4791         if (S_ISGITLINK(one->mode))
4792                 return 0;
4793         if (diff_populate_filespec(one, 0))
4794                 return 0;
4795         if (diff_populate_filespec(two, 0))
4796                 return 0;
4797         return !memcmp(one->data, two->data, one->size);
4798 }
4799
4800 static int diff_filespec_check_stat_unmatch(struct diff_filepair *p)
4801 {
4802         if (p->done_skip_stat_unmatch)
4803                 return p->skip_stat_unmatch_result;
4804
4805         p->done_skip_stat_unmatch = 1;
4806         p->skip_stat_unmatch_result = 0;
4807         /*
4808          * 1. Entries that come from stat info dirtiness
4809          *    always have both sides (iow, not create/delete),
4810          *    one side of the object name is unknown, with
4811          *    the same mode and size.  Keep the ones that
4812          *    do not match these criteria.  They have real
4813          *    differences.
4814          *
4815          * 2. At this point, the file is known to be modified,
4816          *    with the same mode and size, and the object
4817          *    name of one side is unknown.  Need to inspect
4818          *    the identical contents.
4819          */
4820         if (!DIFF_FILE_VALID(p->one) || /* (1) */
4821             !DIFF_FILE_VALID(p->two) ||
4822             (p->one->sha1_valid && p->two->sha1_valid) ||
4823             (p->one->mode != p->two->mode) ||
4824             diff_populate_filespec(p->one, CHECK_SIZE_ONLY) ||
4825             diff_populate_filespec(p->two, CHECK_SIZE_ONLY) ||
4826             (p->one->size != p->two->size) ||
4827             !diff_filespec_is_identical(p->one, p->two)) /* (2) */
4828                 p->skip_stat_unmatch_result = 1;
4829         return p->skip_stat_unmatch_result;
4830 }
4831
4832 static void diffcore_skip_stat_unmatch(struct diff_options *diffopt)
4833 {
4834         int i;
4835         struct diff_queue_struct *q = &diff_queued_diff;
4836         struct diff_queue_struct outq;
4837         DIFF_QUEUE_CLEAR(&outq);
4838
4839         for (i = 0; i < q->nr; i++) {
4840                 struct diff_filepair *p = q->queue[i];
4841
4842                 if (diff_filespec_check_stat_unmatch(p))
4843                         diff_q(&outq, p);
4844                 else {
4845                         /*
4846                          * The caller can subtract 1 from skip_stat_unmatch
4847                          * to determine how many paths were dirty only
4848                          * due to stat info mismatch.
4849                          */
4850                         if (!DIFF_OPT_TST(diffopt, NO_INDEX))
4851                                 diffopt->skip_stat_unmatch++;
4852                         diff_free_filepair(p);
4853                 }
4854         }
4855         free(q->queue);
4856         *q = outq;
4857 }
4858
4859 static int diffnamecmp(const void *a_, const void *b_)
4860 {
4861         const struct diff_filepair *a = *((const struct diff_filepair **)a_);
4862         const struct diff_filepair *b = *((const struct diff_filepair **)b_);
4863         const char *name_a, *name_b;
4864
4865         name_a = a->one ? a->one->path : a->two->path;
4866         name_b = b->one ? b->one->path : b->two->path;
4867         return strcmp(name_a, name_b);
4868 }
4869
4870 void diffcore_fix_diff_index(struct diff_options *options)
4871 {
4872         struct diff_queue_struct *q = &diff_queued_diff;
4873         qsort(q->queue, q->nr, sizeof(q->queue[0]), diffnamecmp);
4874 }
4875
4876 void diffcore_std(struct diff_options *options)
4877 {
4878         /* NOTE please keep the following in sync with diff_tree_combined() */
4879         if (options->skip_stat_unmatch)
4880                 diffcore_skip_stat_unmatch(options);
4881         if (!options->found_follow) {
4882                 /* See try_to_follow_renames() in tree-diff.c */
4883                 if (options->break_opt != -1)
4884                         diffcore_break(options->break_opt);
4885                 if (options->detect_rename)
4886                         diffcore_rename(options);
4887                 if (options->break_opt != -1)
4888                         diffcore_merge_broken();
4889         }
4890         if (options->pickaxe)
4891                 diffcore_pickaxe(options);
4892         if (options->orderfile)
4893                 diffcore_order(options->orderfile);
4894         if (!options->found_follow)
4895                 /* See try_to_follow_renames() in tree-diff.c */
4896                 diff_resolve_rename_copy();
4897         diffcore_apply_filter(options);
4898
4899         if (diff_queued_diff.nr && !DIFF_OPT_TST(options, DIFF_FROM_CONTENTS))
4900                 DIFF_OPT_SET(options, HAS_CHANGES);
4901         else
4902                 DIFF_OPT_CLR(options, HAS_CHANGES);
4903
4904         options->found_follow = 0;
4905 }
4906
4907 int diff_result_code(struct diff_options *opt, int status)
4908 {
4909         int result = 0;
4910
4911         diff_warn_rename_limit("diff.renameLimit",
4912                                opt->needed_rename_limit,
4913                                opt->degraded_cc_to_c);
4914         if (!DIFF_OPT_TST(opt, EXIT_WITH_STATUS) &&
4915             !(opt->output_format & DIFF_FORMAT_CHECKDIFF))
4916                 return status;
4917         if (DIFF_OPT_TST(opt, EXIT_WITH_STATUS) &&
4918             DIFF_OPT_TST(opt, HAS_CHANGES))
4919                 result |= 01;
4920         if ((opt->output_format & DIFF_FORMAT_CHECKDIFF) &&
4921             DIFF_OPT_TST(opt, CHECK_FAILED))
4922                 result |= 02;
4923         return result;
4924 }
4925
4926 int diff_can_quit_early(struct diff_options *opt)
4927 {
4928         return (DIFF_OPT_TST(opt, QUICK) &&
4929                 !opt->filter &&
4930                 DIFF_OPT_TST(opt, HAS_CHANGES));
4931 }
4932
4933 /*
4934  * Shall changes to this submodule be ignored?
4935  *
4936  * Submodule changes can be configured to be ignored separately for each path,
4937  * but that configuration can be overridden from the command line.
4938  */
4939 static int is_submodule_ignored(const char *path, struct diff_options *options)
4940 {
4941         int ignored = 0;
4942         unsigned orig_flags = options->flags;
4943         if (!DIFF_OPT_TST(options, OVERRIDE_SUBMODULE_CONFIG))
4944                 set_diffopt_flags_from_submodule_config(options, path);
4945         if (DIFF_OPT_TST(options, IGNORE_SUBMODULES))
4946                 ignored = 1;
4947         options->flags = orig_flags;
4948         return ignored;
4949 }
4950
4951 void diff_addremove(struct diff_options *options,
4952                     int addremove, unsigned mode,
4953                     const unsigned char *sha1,
4954                     int sha1_valid,
4955                     const char *concatpath, unsigned dirty_submodule)
4956 {
4957         struct diff_filespec *one, *two;
4958
4959         if (S_ISGITLINK(mode) && is_submodule_ignored(concatpath, options))
4960                 return;
4961
4962         /* This may look odd, but it is a preparation for
4963          * feeding "there are unchanged files which should
4964          * not produce diffs, but when you are doing copy
4965          * detection you would need them, so here they are"
4966          * entries to the diff-core.  They will be prefixed
4967          * with something like '=' or '*' (I haven't decided
4968          * which but should not make any difference).
4969          * Feeding the same new and old to diff_change()
4970          * also has the same effect.
4971          * Before the final output happens, they are pruned after
4972          * merged into rename/copy pairs as appropriate.
4973          */
4974         if (DIFF_OPT_TST(options, REVERSE_DIFF))
4975                 addremove = (addremove == '+' ? '-' :
4976                              addremove == '-' ? '+' : addremove);
4977
4978         if (options->prefix &&
4979             strncmp(concatpath, options->prefix, options->prefix_length))
4980                 return;
4981
4982         one = alloc_filespec(concatpath);
4983         two = alloc_filespec(concatpath);
4984
4985         if (addremove != '+')
4986                 fill_filespec(one, sha1, sha1_valid, mode);
4987         if (addremove != '-') {
4988                 fill_filespec(two, sha1, sha1_valid, mode);
4989                 two->dirty_submodule = dirty_submodule;
4990         }
4991
4992         diff_queue(&diff_queued_diff, one, two);
4993         if (!DIFF_OPT_TST(options, DIFF_FROM_CONTENTS))
4994                 DIFF_OPT_SET(options, HAS_CHANGES);
4995 }
4996
4997 void diff_change(struct diff_options *options,
4998                  unsigned old_mode, unsigned new_mode,
4999                  const unsigned char *old_sha1,
5000                  const unsigned char *new_sha1,
5001                  int old_sha1_valid, int new_sha1_valid,
5002                  const char *concatpath,
5003                  unsigned old_dirty_submodule, unsigned new_dirty_submodule)
5004 {
5005         struct diff_filespec *one, *two;
5006         struct diff_filepair *p;
5007
5008         if (S_ISGITLINK(old_mode) && S_ISGITLINK(new_mode) &&
5009             is_submodule_ignored(concatpath, options))
5010                 return;
5011
5012         if (DIFF_OPT_TST(options, REVERSE_DIFF)) {
5013                 unsigned tmp;
5014                 const unsigned char *tmp_c;
5015                 tmp = old_mode; old_mode = new_mode; new_mode = tmp;
5016                 tmp_c = old_sha1; old_sha1 = new_sha1; new_sha1 = tmp_c;
5017                 tmp = old_sha1_valid; old_sha1_valid = new_sha1_valid;
5018                         new_sha1_valid = tmp;
5019                 tmp = old_dirty_submodule; old_dirty_submodule = new_dirty_submodule;
5020                         new_dirty_submodule = tmp;
5021         }
5022
5023         if (options->prefix &&
5024             strncmp(concatpath, options->prefix, options->prefix_length))
5025                 return;
5026
5027         one = alloc_filespec(concatpath);
5028         two = alloc_filespec(concatpath);
5029         fill_filespec(one, old_sha1, old_sha1_valid, old_mode);
5030         fill_filespec(two, new_sha1, new_sha1_valid, new_mode);
5031         one->dirty_submodule = old_dirty_submodule;
5032         two->dirty_submodule = new_dirty_submodule;
5033         p = diff_queue(&diff_queued_diff, one, two);
5034
5035         if (DIFF_OPT_TST(options, DIFF_FROM_CONTENTS))
5036                 return;
5037
5038         if (DIFF_OPT_TST(options, QUICK) && options->skip_stat_unmatch &&
5039             !diff_filespec_check_stat_unmatch(p))
5040                 return;
5041
5042         DIFF_OPT_SET(options, HAS_CHANGES);
5043 }
5044
5045 struct diff_filepair *diff_unmerge(struct diff_options *options, const char *path)
5046 {
5047         struct diff_filepair *pair;
5048         struct diff_filespec *one, *two;
5049
5050         if (options->prefix &&
5051             strncmp(path, options->prefix, options->prefix_length))
5052                 return NULL;
5053
5054         one = alloc_filespec(path);
5055         two = alloc_filespec(path);
5056         pair = diff_queue(&diff_queued_diff, one, two);
5057         pair->is_unmerged = 1;
5058         return pair;
5059 }
5060
5061 static char *run_textconv(const char *pgm, struct diff_filespec *spec,
5062                 size_t *outsize)
5063 {
5064         struct diff_tempfile *temp;
5065         const char *argv[3];
5066         const char **arg = argv;
5067         struct child_process child = CHILD_PROCESS_INIT;
5068         struct strbuf buf = STRBUF_INIT;
5069         int err = 0;
5070
5071         temp = prepare_temp_file(spec->path, spec);
5072         *arg++ = pgm;
5073         *arg++ = temp->name;
5074         *arg = NULL;
5075
5076         child.use_shell = 1;
5077         child.argv = argv;
5078         child.out = -1;
5079         if (start_command(&child)) {
5080                 remove_tempfile();
5081                 return NULL;
5082         }
5083
5084         if (strbuf_read(&buf, child.out, 0) < 0)
5085                 err = error("error reading from textconv command '%s'", pgm);
5086         close(child.out);
5087
5088         if (finish_command(&child) || err) {
5089                 strbuf_release(&buf);
5090                 remove_tempfile();
5091                 return NULL;
5092         }
5093         remove_tempfile();
5094
5095         return strbuf_detach(&buf, outsize);
5096 }
5097
5098 size_t fill_textconv(struct userdiff_driver *driver,
5099                      struct diff_filespec *df,
5100                      char **outbuf)
5101 {
5102         size_t size;
5103
5104         if (!driver || !driver->textconv) {
5105                 if (!DIFF_FILE_VALID(df)) {
5106                         *outbuf = "";
5107                         return 0;
5108                 }
5109                 if (diff_populate_filespec(df, 0))
5110                         die("unable to read files to diff");
5111                 *outbuf = df->data;
5112                 return df->size;
5113         }
5114
5115         if (driver->textconv_cache && df->sha1_valid) {
5116                 *outbuf = notes_cache_get(driver->textconv_cache, df->sha1,
5117                                           &size);
5118                 if (*outbuf)
5119                         return size;
5120         }
5121
5122         *outbuf = run_textconv(driver->textconv, df, &size);
5123         if (!*outbuf)
5124                 die("unable to read files to diff");
5125
5126         if (driver->textconv_cache && df->sha1_valid) {
5127                 /* ignore errors, as we might be in a readonly repository */
5128                 notes_cache_put(driver->textconv_cache, df->sha1, *outbuf,
5129                                 size);
5130                 /*
5131                  * we could save up changes and flush them all at the end,
5132                  * but we would need an extra call after all diffing is done.
5133                  * Since generating a cache entry is the slow path anyway,
5134                  * this extra overhead probably isn't a big deal.
5135                  */
5136                 notes_cache_write(driver->textconv_cache);
5137         }
5138
5139         return size;
5140 }
5141
5142 void setup_diff_pager(struct diff_options *opt)
5143 {
5144         /*
5145          * If the user asked for our exit code, then either they want --quiet
5146          * or --exit-code. We should definitely not bother with a pager in the
5147          * former case, as we will generate no output. Since we still properly
5148          * report our exit code even when a pager is run, we _could_ run a
5149          * pager with --exit-code. But since we have not done so historically,
5150          * and because it is easy to find people oneline advising "git diff
5151          * --exit-code" in hooks and other scripts, we do not do so.
5152          */
5153         if (!DIFF_OPT_TST(opt, EXIT_WITH_STATUS) &&
5154             check_pager_config("diff") != 0)
5155                 setup_pager();
5156 }