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