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

   1 /*
   2  * Copyright (C) 2005 Junio C Hamano
   3  */
   4 #include "cache.h"
   5 #include "quote.h"
   6 #include "diff.h"
   7 #include "diffcore.h"
   8 #include "delta.h"
   9 #include "xdiff-interface.h"
  10 #include "color.h"
  11 #include "attr.h"
  12 #include "run-command.h"
  13 #include "utf8.h"
  14 #include "userdiff.h"
  15
  16 #ifdef NO_FAST_WORKING_DIRECTORY
  17 #define FAST_WORKING_DIRECTORY 0
  18 #else
  19 #define FAST_WORKING_DIRECTORY 1
  20 #endif
  21
  22 static int diff_detect_rename_default;
  23 static int diff_rename_limit_default = 200;
  24 static int diff_suppress_blank_empty;
  25 int diff_use_color_default = -1;
  26 static const char *external_diff_cmd_cfg;
  27 int diff_auto_refresh_index = 1;
  28 static int diff_mnemonic_prefix;
  29
  30 static char diff_colors[][COLOR_MAXLEN] = {
  31         "\033[m",       /* reset */
  32         "",             /* PLAIN (normal) */
  33         "\033[1m",      /* METAINFO (bold) */
  34         "\033[36m",     /* FRAGINFO (cyan) */
  35         "\033[31m",     /* OLD (red) */
  36         "\033[32m",     /* NEW (green) */
  37         "\033[33m",     /* COMMIT (yellow) */
  38         "\033[41m",     /* WHITESPACE (red background) */
  39 };
  40
  41 static void diff_filespec_load_driver(struct diff_filespec *one);
  42 static char *run_textconv(const char *, struct diff_filespec *, size_t *);
  43
  44 static int parse_diff_color_slot(const char *var, int ofs)
  45 {
  46         if (!strcasecmp(var+ofs, "plain"))
  47                 return DIFF_PLAIN;
  48         if (!strcasecmp(var+ofs, "meta"))
  49                 return DIFF_METAINFO;
  50         if (!strcasecmp(var+ofs, "frag"))
  51                 return DIFF_FRAGINFO;
  52         if (!strcasecmp(var+ofs, "old"))
  53                 return DIFF_FILE_OLD;
  54         if (!strcasecmp(var+ofs, "new"))
  55                 return DIFF_FILE_NEW;
  56         if (!strcasecmp(var+ofs, "commit"))
  57                 return DIFF_COMMIT;
  58         if (!strcasecmp(var+ofs, "whitespace"))
  59                 return DIFF_WHITESPACE;
  60         die("bad config variable '%s'", var);
  61 }
  62
  63 /*
  64  * These are to give UI layer defaults.
  65  * The core-level commands such as git-diff-files should
  66  * never be affected by the setting of diff.renames
  67  * the user happens to have in the configuration file.
  68  */
  69 int git_diff_ui_config(const char *var, const char *value, void *cb)
  70 {
  71         if (!strcmp(var, "diff.color") || !strcmp(var, "color.diff")) {
  72                 diff_use_color_default = git_config_colorbool(var, value, -1);
  73                 return 0;
  74         }
  75         if (!strcmp(var, "diff.renames")) {
  76                 if (!value)
  77                         diff_detect_rename_default = DIFF_DETECT_RENAME;
  78                 else if (!strcasecmp(value, "copies") ||
  79                          !strcasecmp(value, "copy"))
  80                         diff_detect_rename_default = DIFF_DETECT_COPY;
  81                 else if (git_config_bool(var,value))
  82                         diff_detect_rename_default = DIFF_DETECT_RENAME;
  83                 return 0;
  84         }
  85         if (!strcmp(var, "diff.autorefreshindex")) {
  86                 diff_auto_refresh_index = git_config_bool(var, value);
  87                 return 0;
  88         }
  89         if (!strcmp(var, "diff.mnemonicprefix")) {
  90                 diff_mnemonic_prefix = git_config_bool(var, value);
  91                 return 0;
  92         }
  93         if (!strcmp(var, "diff.external"))
  94                 return git_config_string(&external_diff_cmd_cfg, var, value);
  95
  96         return git_diff_basic_config(var, value, cb);
  97 }
  98
  99 int git_diff_basic_config(const char *var, const char *value, void *cb)
 100 {
 101         if (!strcmp(var, "diff.renamelimit")) {
 102                 diff_rename_limit_default = git_config_int(var, value);
 103                 return 0;
 104         }
 105
 106         switch (userdiff_config(var, value)) {
 107                 case 0: break;
 108                 case -1: return -1;
 109                 default: return 0;
 110         }
 111
 112         if (!prefixcmp(var, "diff.color.") || !prefixcmp(var, "color.diff.")) {
 113                 int slot = parse_diff_color_slot(var, 11);
 114                 if (!value)
 115                         return config_error_nonbool(var);
 116                 color_parse(value, var, diff_colors[slot]);
 117                 return 0;
 118         }
 119
 120         /* like GNU diff's --suppress-blank-empty option  */
 121         if (!strcmp(var, "diff.suppressblankempty") ||
 122                         /* for backwards compatibility */
 123                         !strcmp(var, "diff.suppress-blank-empty")) {
 124                 diff_suppress_blank_empty = git_config_bool(var, value);
 125                 return 0;
 126         }
 127
 128         return git_color_default_config(var, value, cb);
 129 }
 130
 131 static char *quote_two(const char *one, const char *two)
 132 {
 133         int need_one = quote_c_style(one, NULL, NULL, 1);
 134         int need_two = quote_c_style(two, NULL, NULL, 1);
 135         struct strbuf res = STRBUF_INIT;
 136
 137         if (need_one + need_two) {
 138                 strbuf_addch(&res, '"');
 139                 quote_c_style(one, &res, NULL, 1);
 140                 quote_c_style(two, &res, NULL, 1);
 141                 strbuf_addch(&res, '"');
 142         } else {
 143                 strbuf_addstr(&res, one);
 144                 strbuf_addstr(&res, two);
 145         }
 146         return strbuf_detach(&res, NULL);
 147 }
 148
 149 static const char *external_diff(void)
 150 {
 151         static const char *external_diff_cmd = NULL;
 152         static int done_preparing = 0;
 153
 154         if (done_preparing)
 155                 return external_diff_cmd;
 156         external_diff_cmd = getenv("GIT_EXTERNAL_DIFF");
 157         if (!external_diff_cmd)
 158                 external_diff_cmd = external_diff_cmd_cfg;
 159         done_preparing = 1;
 160         return external_diff_cmd;
 161 }
 162
 163 static struct diff_tempfile {
 164         const char *name; /* filename external diff should read from */
 165         char hex[41];
 166         char mode[10];
 167         char tmp_path[PATH_MAX];
 168 } diff_temp[2];
 169
 170 static int count_lines(const char *data, int size)
 171 {
 172         int count, ch, completely_empty = 1, nl_just_seen = 0;
 173         count = 0;
 174         while (0 < size--) {
 175                 ch = *data++;
 176                 if (ch == '\n') {
 177                         count++;
 178                         nl_just_seen = 1;
 179                         completely_empty = 0;
 180                 }
 181                 else {
 182                         nl_just_seen = 0;
 183                         completely_empty = 0;
 184                 }
 185         }
 186         if (completely_empty)
 187                 return 0;
 188         if (!nl_just_seen)
 189                 count++; /* no trailing newline */
 190         return count;
 191 }
 192
 193 static void print_line_count(FILE *file, int count)
 194 {
 195         switch (count) {
 196         case 0:
 197                 fprintf(file, "0,0");
 198                 break;
 199         case 1:
 200                 fprintf(file, "1");
 201                 break;
 202         default:
 203                 fprintf(file, "1,%d", count);
 204                 break;
 205         }
 206 }
 207
 208 static void copy_file_with_prefix(FILE *file,
 209                                   int prefix, const char *data, int size,
 210                                   const char *set, const char *reset)
 211 {
 212         int ch, nl_just_seen = 1;
 213         while (0 < size--) {
 214                 ch = *data++;
 215                 if (nl_just_seen) {
 216                         fputs(set, file);
 217                         putc(prefix, file);
 218                 }
 219                 if (ch == '\n') {
 220                         nl_just_seen = 1;
 221                         fputs(reset, file);
 222                 } else
 223                         nl_just_seen = 0;
 224                 putc(ch, file);
 225         }
 226         if (!nl_just_seen)
 227                 fprintf(file, "%s\n\\ No newline at end of file\n", reset);
 228 }
 229
 230 static void emit_rewrite_diff(const char *name_a,
 231                               const char *name_b,
 232                               struct diff_filespec *one,
 233                               struct diff_filespec *two,
 234                               const char *textconv_one,
 235                               const char *textconv_two,
 236                               struct diff_options *o)
 237 {
 238         int lc_a, lc_b;
 239         int color_diff = DIFF_OPT_TST(o, COLOR_DIFF);
 240         const char *name_a_tab, *name_b_tab;
 241         const char *metainfo = diff_get_color(color_diff, DIFF_METAINFO);
 242         const char *fraginfo = diff_get_color(color_diff, DIFF_FRAGINFO);
 243         const char *old = diff_get_color(color_diff, DIFF_FILE_OLD);
 244         const char *new = diff_get_color(color_diff, DIFF_FILE_NEW);
 245         const char *reset = diff_get_color(color_diff, DIFF_RESET);
 246         static struct strbuf a_name = STRBUF_INIT, b_name = STRBUF_INIT;
 247         const char *a_prefix, *b_prefix;
 248         const char *data_one, *data_two;
 249         size_t size_one, size_two;
 250
 251         if (diff_mnemonic_prefix && DIFF_OPT_TST(o, REVERSE_DIFF)) {
 252                 a_prefix = o->b_prefix;
 253                 b_prefix = o->a_prefix;
 254         } else {
 255                 a_prefix = o->a_prefix;
 256                 b_prefix = o->b_prefix;
 257         }
 258
 259         name_a += (*name_a == '/');
 260         name_b += (*name_b == '/');
 261         name_a_tab = strchr(name_a, ' ') ? "\t" : "";
 262         name_b_tab = strchr(name_b, ' ') ? "\t" : "";
 263
 264         strbuf_reset(&a_name);
 265         strbuf_reset(&b_name);
 266         quote_two_c_style(&a_name, a_prefix, name_a, 0);
 267         quote_two_c_style(&b_name, b_prefix, name_b, 0);
 268
 269         diff_populate_filespec(one, 0);
 270         diff_populate_filespec(two, 0);
 271         if (textconv_one) {
 272                 data_one = run_textconv(textconv_one, one, &size_one);
 273                 if (!data_one)
 274                         die("unable to read files to diff");
 275         }
 276         else {
 277                 data_one = one->data;
 278                 size_one = one->size;
 279         }
 280         if (textconv_two) {
 281                 data_two = run_textconv(textconv_two, two, &size_two);
 282                 if (!data_two)
 283                         die("unable to read files to diff");
 284         }
 285         else {
 286                 data_two = two->data;
 287                 size_two = two->size;
 288         }
 289
 290         lc_a = count_lines(data_one, size_one);
 291         lc_b = count_lines(data_two, size_two);
 292         fprintf(o->file,
 293                 "%s--- %s%s%s\n%s+++ %s%s%s\n%s@@ -",
 294                 metainfo, a_name.buf, name_a_tab, reset,
 295                 metainfo, b_name.buf, name_b_tab, reset, fraginfo);
 296         print_line_count(o->file, lc_a);
 297         fprintf(o->file, " +");
 298         print_line_count(o->file, lc_b);
 299         fprintf(o->file, " @@%s\n", reset);
 300         if (lc_a)
 301                 copy_file_with_prefix(o->file, '-', data_one, size_one, old, reset);
 302         if (lc_b)
 303                 copy_file_with_prefix(o->file, '+', data_two, size_two, new, reset);
 304 }
 305
 306 static int fill_mmfile(mmfile_t *mf, struct diff_filespec *one)
 307 {
 308         if (!DIFF_FILE_VALID(one)) {
 309                 mf->ptr = (char *)""; /* does not matter */
 310                 mf->size = 0;
 311                 return 0;
 312         }
 313         else if (diff_populate_filespec(one, 0))
 314                 return -1;
 315
 316         mf->ptr = one->data;
 317         mf->size = one->size;
 318         return 0;
 319 }
 320
 321 struct diff_words_buffer {
 322         mmfile_t text;
 323         long alloc;
 324         long current; /* output pointer */
 325         int suppressed_newline;
 326 };
 327
 328 static void diff_words_append(char *line, unsigned long len,
 329                 struct diff_words_buffer *buffer)
 330 {
 331         if (buffer->text.size + len > buffer->alloc) {
 332                 buffer->alloc = (buffer->text.size + len) * 3 / 2;
 333                 buffer->text.ptr = xrealloc(buffer->text.ptr, buffer->alloc);
 334         }
 335         line++;
 336         len--;
 337         memcpy(buffer->text.ptr + buffer->text.size, line, len);
 338         buffer->text.size += len;
 339 }
 340
 341 struct diff_words_data {
 342         struct diff_words_buffer minus, plus;
 343         FILE *file;
 344 };
 345
 346 static void print_word(FILE *file, struct diff_words_buffer *buffer, int len, int color,
 347                 int suppress_newline)
 348 {
 349         const char *ptr;
 350         int eol = 0;
 351
 352         if (len == 0)
 353                 return;
 354
 355         ptr  = buffer->text.ptr + buffer->current;
 356         buffer->current += len;
 357
 358         if (ptr[len - 1] == '\n') {
 359                 eol = 1;
 360                 len--;
 361         }
 362
 363         fputs(diff_get_color(1, color), file);
 364         fwrite(ptr, len, 1, file);
 365         fputs(diff_get_color(1, DIFF_RESET), file);
 366
 367         if (eol) {
 368                 if (suppress_newline)
 369                         buffer->suppressed_newline = 1;
 370                 else
 371                         putc('\n', file);
 372         }
 373 }
 374
 375 static void fn_out_diff_words_aux(void *priv, char *line, unsigned long len)
 376 {
 377         struct diff_words_data *diff_words = priv;
 378
 379         if (diff_words->minus.suppressed_newline) {
 380                 if (line[0] != '+')
 381                         putc('\n', diff_words->file);
 382                 diff_words->minus.suppressed_newline = 0;
 383         }
 384
 385         len--;
 386         switch (line[0]) {
 387                 case '-':
 388                         print_word(diff_words->file,
 389                                    &diff_words->minus, len, DIFF_FILE_OLD, 1);
 390                         break;
 391                 case '+':
 392                         print_word(diff_words->file,
 393                                    &diff_words->plus, len, DIFF_FILE_NEW, 0);
 394                         break;
 395                 case ' ':
 396                         print_word(diff_words->file,
 397                                    &diff_words->plus, len, DIFF_PLAIN, 0);
 398                         diff_words->minus.current += len;
 399                         break;
 400         }
 401 }
 402
 403 /* this executes the word diff on the accumulated buffers */
 404 static void diff_words_show(struct diff_words_data *diff_words)
 405 {
 406         xpparam_t xpp;
 407         xdemitconf_t xecfg;
 408         xdemitcb_t ecb;
 409         mmfile_t minus, plus;
 410         int i;
 411
 412         memset(&xpp, 0, sizeof(xpp));
 413         memset(&xecfg, 0, sizeof(xecfg));
 414         minus.size = diff_words->minus.text.size;
 415         minus.ptr = xmalloc(minus.size);
 416         memcpy(minus.ptr, diff_words->minus.text.ptr, minus.size);
 417         for (i = 0; i < minus.size; i++)
 418                 if (isspace(minus.ptr[i]))
 419                         minus.ptr[i] = '\n';
 420         diff_words->minus.current = 0;
 421
 422         plus.size = diff_words->plus.text.size;
 423         plus.ptr = xmalloc(plus.size);
 424         memcpy(plus.ptr, diff_words->plus.text.ptr, plus.size);
 425         for (i = 0; i < plus.size; i++)
 426                 if (isspace(plus.ptr[i]))
 427                         plus.ptr[i] = '\n';
 428         diff_words->plus.current = 0;
 429
 430         xpp.flags = XDF_NEED_MINIMAL;
 431         xecfg.ctxlen = diff_words->minus.alloc + diff_words->plus.alloc;
 432         xdi_diff_outf(&minus, &plus, fn_out_diff_words_aux, diff_words,
 433                       &xpp, &xecfg, &ecb);
 434         free(minus.ptr);
 435         free(plus.ptr);
 436         diff_words->minus.text.size = diff_words->plus.text.size = 0;
 437
 438         if (diff_words->minus.suppressed_newline) {
 439                 putc('\n', diff_words->file);
 440                 diff_words->minus.suppressed_newline = 0;
 441         }
 442 }
 443
 444 typedef unsigned long (*sane_truncate_fn)(char *line, unsigned long len);
 445
 446 struct emit_callback {
 447         int nparents, color_diff;
 448         unsigned ws_rule;
 449         sane_truncate_fn truncate;
 450         const char **label_path;
 451         struct diff_words_data *diff_words;
 452         int *found_changesp;
 453         FILE *file;
 454 };
 455
 456 static void free_diff_words_data(struct emit_callback *ecbdata)
 457 {
 458         if (ecbdata->diff_words) {
 459                 /* flush buffers */
 460                 if (ecbdata->diff_words->minus.text.size ||
 461                                 ecbdata->diff_words->plus.text.size)
 462                         diff_words_show(ecbdata->diff_words);
 463
 464                 free (ecbdata->diff_words->minus.text.ptr);
 465                 free (ecbdata->diff_words->plus.text.ptr);
 466                 free(ecbdata->diff_words);
 467                 ecbdata->diff_words = NULL;
 468         }
 469 }
 470
 471 const char *diff_get_color(int diff_use_color, enum color_diff ix)
 472 {
 473         if (diff_use_color)
 474                 return diff_colors[ix];
 475         return "";
 476 }
 477
 478 static void emit_line(FILE *file, const char *set, const char *reset, const char *line, int len)
 479 {
 480         int has_trailing_newline, has_trailing_carriage_return;
 481
 482         has_trailing_newline = (len > 0 && line[len-1] == '\n');
 483         if (has_trailing_newline)
 484                 len--;
 485         has_trailing_carriage_return = (len > 0 && line[len-1] == '\r');
 486         if (has_trailing_carriage_return)
 487                 len--;
 488
 489         fputs(set, file);
 490         fwrite(line, len, 1, file);
 491         fputs(reset, file);
 492         if (has_trailing_carriage_return)
 493                 fputc('\r', file);
 494         if (has_trailing_newline)
 495                 fputc('\n', file);
 496 }
 497
 498 static void emit_add_line(const char *reset, struct emit_callback *ecbdata, const char *line, int len)
 499 {
 500         const char *ws = diff_get_color(ecbdata->color_diff, DIFF_WHITESPACE);
 501         const char *set = diff_get_color(ecbdata->color_diff, DIFF_FILE_NEW);
 502
 503         if (!*ws)
 504                 emit_line(ecbdata->file, set, reset, line, len);
 505         else {
 506                 /* Emit just the prefix, then the rest. */
 507                 emit_line(ecbdata->file, set, reset, line, ecbdata->nparents);
 508                 ws_check_emit(line + ecbdata->nparents,
 509                               len - ecbdata->nparents, ecbdata->ws_rule,
 510                               ecbdata->file, set, reset, ws);
 511         }
 512 }
 513
 514 static unsigned long sane_truncate_line(struct emit_callback *ecb, char *line, unsigned long len)
 515 {
 516         const char *cp;
 517         unsigned long allot;
 518         size_t l = len;
 519
 520         if (ecb->truncate)
 521                 return ecb->truncate(line, len);
 522         cp = line;
 523         allot = l;
 524         while (0 < l) {
 525                 (void) utf8_width(&cp, &l);
 526                 if (!cp)
 527                         break; /* truncated in the middle? */
 528         }
 529         return allot - l;
 530 }
 531
 532 static void fn_out_consume(void *priv, char *line, unsigned long len)
 533 {
 534         int i;
 535         int color;
 536         struct emit_callback *ecbdata = priv;
 537         const char *meta = diff_get_color(ecbdata->color_diff, DIFF_METAINFO);
 538         const char *plain = diff_get_color(ecbdata->color_diff, DIFF_PLAIN);
 539         const char *reset = diff_get_color(ecbdata->color_diff, DIFF_RESET);
 540
 541         *(ecbdata->found_changesp) = 1;
 542
 543         if (ecbdata->label_path[0]) {
 544                 const char *name_a_tab, *name_b_tab;
 545
 546                 name_a_tab = strchr(ecbdata->label_path[0], ' ') ? "\t" : "";
 547                 name_b_tab = strchr(ecbdata->label_path[1], ' ') ? "\t" : "";
 548
 549                 fprintf(ecbdata->file, "%s--- %s%s%s\n",
 550                         meta, ecbdata->label_path[0], reset, name_a_tab);
 551                 fprintf(ecbdata->file, "%s+++ %s%s%s\n",
 552                         meta, ecbdata->label_path[1], reset, name_b_tab);
 553                 ecbdata->label_path[0] = ecbdata->label_path[1] = NULL;
 554         }
 555
 556         if (diff_suppress_blank_empty
 557             && len == 2 && line[0] == ' ' && line[1] == '\n') {
 558                 line[0] = '\n';
 559                 len = 1;
 560         }
 561
 562         /* This is not really necessary for now because
 563          * this codepath only deals with two-way diffs.
 564          */
 565         for (i = 0; i < len && line[i] == '@'; i++)
 566                 ;
 567         if (2 <= i && i < len && line[i] == ' ') {
 568                 ecbdata->nparents = i - 1;
 569                 len = sane_truncate_line(ecbdata, line, len);
 570                 emit_line(ecbdata->file,
 571                           diff_get_color(ecbdata->color_diff, DIFF_FRAGINFO),
 572                           reset, line, len);
 573                 if (line[len-1] != '\n')
 574                         putc('\n', ecbdata->file);
 575                 return;
 576         }
 577
 578         if (len < ecbdata->nparents) {
 579                 emit_line(ecbdata->file, reset, reset, line, len);
 580                 return;
 581         }
 582
 583         color = DIFF_PLAIN;
 584         if (ecbdata->diff_words && ecbdata->nparents != 1)
 585                 /* fall back to normal diff */
 586                 free_diff_words_data(ecbdata);
 587         if (ecbdata->diff_words) {
 588                 if (line[0] == '-') {
 589                         diff_words_append(line, len,
 590                                           &ecbdata->diff_words->minus);
 591                         return;
 592                 } else if (line[0] == '+') {
 593                         diff_words_append(line, len,
 594                                           &ecbdata->diff_words->plus);
 595                         return;
 596                 }
 597                 if (ecbdata->diff_words->minus.text.size ||
 598                     ecbdata->diff_words->plus.text.size)
 599                         diff_words_show(ecbdata->diff_words);
 600                 line++;
 601                 len--;
 602                 emit_line(ecbdata->file, plain, reset, line, len);
 603                 return;
 604         }
 605         for (i = 0; i < ecbdata->nparents && len; i++) {
 606                 if (line[i] == '-')
 607                         color = DIFF_FILE_OLD;
 608                 else if (line[i] == '+')
 609                         color = DIFF_FILE_NEW;
 610         }
 611
 612         if (color != DIFF_FILE_NEW) {
 613                 emit_line(ecbdata->file,
 614                           diff_get_color(ecbdata->color_diff, color),
 615                           reset, line, len);
 616                 return;
 617         }
 618         emit_add_line(reset, ecbdata, line, len);
 619 }
 620
 621 static char *pprint_rename(const char *a, const char *b)
 622 {
 623         const char *old = a;
 624         const char *new = b;
 625         struct strbuf name = STRBUF_INIT;
 626         int pfx_length, sfx_length;
 627         int len_a = strlen(a);
 628         int len_b = strlen(b);
 629         int a_midlen, b_midlen;
 630         int qlen_a = quote_c_style(a, NULL, NULL, 0);
 631         int qlen_b = quote_c_style(b, NULL, NULL, 0);
 632
 633         if (qlen_a || qlen_b) {
 634                 quote_c_style(a, &name, NULL, 0);
 635                 strbuf_addstr(&name, " => ");
 636                 quote_c_style(b, &name, NULL, 0);
 637                 return strbuf_detach(&name, NULL);
 638         }
 639
 640         /* Find common prefix */
 641         pfx_length = 0;
 642         while (*old && *new && *old == *new) {
 643                 if (*old == '/')
 644                         pfx_length = old - a + 1;
 645                 old++;
 646                 new++;
 647         }
 648
 649         /* Find common suffix */
 650         old = a + len_a;
 651         new = b + len_b;
 652         sfx_length = 0;
 653         while (a <= old && b <= new && *old == *new) {
 654                 if (*old == '/')
 655                         sfx_length = len_a - (old - a);
 656                 old--;
 657                 new--;
 658         }
 659
 660         /*
 661          * pfx{mid-a => mid-b}sfx
 662          * {pfx-a => pfx-b}sfx
 663          * pfx{sfx-a => sfx-b}
 664          * name-a => name-b
 665          */
 666         a_midlen = len_a - pfx_length - sfx_length;
 667         b_midlen = len_b - pfx_length - sfx_length;
 668         if (a_midlen < 0)
 669                 a_midlen = 0;
 670         if (b_midlen < 0)
 671                 b_midlen = 0;
 672
 673         strbuf_grow(&name, pfx_length + a_midlen + b_midlen + sfx_length + 7);
 674         if (pfx_length + sfx_length) {
 675                 strbuf_add(&name, a, pfx_length);
 676                 strbuf_addch(&name, '{');
 677         }
 678         strbuf_add(&name, a + pfx_length, a_midlen);
 679         strbuf_addstr(&name, " => ");
 680         strbuf_add(&name, b + pfx_length, b_midlen);
 681         if (pfx_length + sfx_length) {
 682                 strbuf_addch(&name, '}');
 683                 strbuf_add(&name, a + len_a - sfx_length, sfx_length);
 684         }
 685         return strbuf_detach(&name, NULL);
 686 }
 687
 688 struct diffstat_t {
 689         int nr;
 690         int alloc;
 691         struct diffstat_file {
 692                 char *from_name;
 693                 char *name;
 694                 char *print_name;
 695                 unsigned is_unmerged:1;
 696                 unsigned is_binary:1;
 697                 unsigned is_renamed:1;
 698                 unsigned int added, deleted;
 699         } **files;
 700 };
 701
 702 static struct diffstat_file *diffstat_add(struct diffstat_t *diffstat,
 703                                           const char *name_a,
 704                                           const char *name_b)
 705 {
 706         struct diffstat_file *x;
 707         x = xcalloc(sizeof (*x), 1);
 708         if (diffstat->nr == diffstat->alloc) {
 709                 diffstat->alloc = alloc_nr(diffstat->alloc);
 710                 diffstat->files = xrealloc(diffstat->files,
 711                                 diffstat->alloc * sizeof(x));
 712         }
 713         diffstat->files[diffstat->nr++] = x;
 714         if (name_b) {
 715                 x->from_name = xstrdup(name_a);
 716                 x->name = xstrdup(name_b);
 717                 x->is_renamed = 1;
 718         }
 719         else {
 720                 x->from_name = NULL;
 721                 x->name = xstrdup(name_a);
 722         }
 723         return x;
 724 }
 725
 726 static void diffstat_consume(void *priv, char *line, unsigned long len)
 727 {
 728         struct diffstat_t *diffstat = priv;
 729         struct diffstat_file *x = diffstat->files[diffstat->nr - 1];
 730
 731         if (line[0] == '+')
 732                 x->added++;
 733         else if (line[0] == '-')
 734                 x->deleted++;
 735 }
 736
 737 const char mime_boundary_leader[] = "------------";
 738
 739 static int scale_linear(int it, int width, int max_change)
 740 {
 741         /*
 742          * make sure that at least one '-' is printed if there were deletions,
 743          * and likewise for '+'.
 744          */
 745         if (max_change < 2)
 746                 return it;
 747         return ((it - 1) * (width - 1) + max_change - 1) / (max_change - 1);
 748 }
 749
 750 static void show_name(FILE *file,
 751                       const char *prefix, const char *name, int len,
 752                       const char *reset, const char *set)
 753 {
 754         fprintf(file, " %s%s%-*s%s |", set, prefix, len, name, reset);
 755 }
 756
 757 static void show_graph(FILE *file, char ch, int cnt, const char *set, const char *reset)
 758 {
 759         if (cnt <= 0)
 760                 return;
 761         fprintf(file, "%s", set);
 762         while (cnt--)
 763                 putc(ch, file);
 764         fprintf(file, "%s", reset);
 765 }
 766
 767 static void fill_print_name(struct diffstat_file *file)
 768 {
 769         char *pname;
 770
 771         if (file->print_name)
 772                 return;
 773
 774         if (!file->is_renamed) {
 775                 struct strbuf buf = STRBUF_INIT;
 776                 if (quote_c_style(file->name, &buf, NULL, 0)) {
 777                         pname = strbuf_detach(&buf, NULL);
 778                 } else {
 779                         pname = file->name;
 780                         strbuf_release(&buf);
 781                 }
 782         } else {
 783                 pname = pprint_rename(file->from_name, file->name);
 784         }
 785         file->print_name = pname;
 786 }
 787
 788 static void show_stats(struct diffstat_t* data, struct diff_options *options)
 789 {
 790         int i, len, add, del, total, adds = 0, dels = 0;
 791         int max_change = 0, max_len = 0;
 792         int total_files = data->nr;
 793         int width, name_width;
 794         const char *reset, *set, *add_c, *del_c;
 795
 796         if (data->nr == 0)
 797                 return;
 798
 799         width = options->stat_width ? options->stat_width : 80;
 800         name_width = options->stat_name_width ? options->stat_name_width : 50;
 801
 802         /* Sanity: give at least 5 columns to the graph,
 803          * but leave at least 10 columns for the name.
 804          */
 805         if (width < 25)
 806                 width = 25;
 807         if (name_width < 10)
 808                 name_width = 10;
 809         else if (width < name_width + 15)
 810                 name_width = width - 15;
 811
 812         /* Find the longest filename and max number of changes */
 813         reset = diff_get_color_opt(options, DIFF_RESET);
 814         set   = diff_get_color_opt(options, DIFF_PLAIN);
 815         add_c = diff_get_color_opt(options, DIFF_FILE_NEW);
 816         del_c = diff_get_color_opt(options, DIFF_FILE_OLD);
 817
 818         for (i = 0; i < data->nr; i++) {
 819                 struct diffstat_file *file = data->files[i];
 820                 int change = file->added + file->deleted;
 821                 fill_print_name(file);
 822                 len = strlen(file->print_name);
 823                 if (max_len < len)
 824                         max_len = len;
 825
 826                 if (file->is_binary || file->is_unmerged)
 827                         continue;
 828                 if (max_change < change)
 829                         max_change = change;
 830         }
 831
 832         /* Compute the width of the graph part;
 833          * 10 is for one blank at the beginning of the line plus
 834          * " | count " between the name and the graph.
 835          *
 836          * From here on, name_width is the width of the name area,
 837          * and width is the width of the graph area.
 838          */
 839         name_width = (name_width < max_len) ? name_width : max_len;
 840         if (width < (name_width + 10) + max_change)
 841                 width = width - (name_width + 10);
 842         else
 843                 width = max_change;
 844
 845         for (i = 0; i < data->nr; i++) {
 846                 const char *prefix = "";
 847                 char *name = data->files[i]->print_name;
 848                 int added = data->files[i]->added;
 849                 int deleted = data->files[i]->deleted;
 850                 int name_len;
 851
 852                 /*
 853                  * "scale" the filename
 854                  */
 855                 len = name_width;
 856                 name_len = strlen(name);
 857                 if (name_width < name_len) {
 858                         char *slash;
 859                         prefix = "...";
 860                         len -= 3;
 861                         name += name_len - len;
 862                         slash = strchr(name, '/');
 863                         if (slash)
 864                                 name = slash;
 865                 }
 866
 867                 if (data->files[i]->is_binary) {
 868                         show_name(options->file, prefix, name, len, reset, set);
 869                         fprintf(options->file, "  Bin ");
 870                         fprintf(options->file, "%s%d%s", del_c, deleted, reset);
 871                         fprintf(options->file, " -> ");
 872                         fprintf(options->file, "%s%d%s", add_c, added, reset);
 873                         fprintf(options->file, " bytes");
 874                         fprintf(options->file, "\n");
 875                         continue;
 876                 }
 877                 else if (data->files[i]->is_unmerged) {
 878                         show_name(options->file, prefix, name, len, reset, set);
 879                         fprintf(options->file, "  Unmerged\n");
 880                         continue;
 881                 }
 882                 else if (!data->files[i]->is_renamed &&
 883                          (added + deleted == 0)) {
 884                         total_files--;
 885                         continue;
 886                 }
 887
 888                 /*
 889                  * scale the add/delete
 890                  */
 891                 add = added;
 892                 del = deleted;
 893                 total = add + del;
 894                 adds += add;
 895                 dels += del;
 896
 897                 if (width <= max_change) {
 898                         add = scale_linear(add, width, max_change);
 899                         del = scale_linear(del, width, max_change);
 900                         total = add + del;
 901                 }
 902                 show_name(options->file, prefix, name, len, reset, set);
 903                 fprintf(options->file, "%5d%s", added + deleted,
 904                                 added + deleted ? " " : "");
 905                 show_graph(options->file, '+', add, add_c, reset);
 906                 show_graph(options->file, '-', del, del_c, reset);
 907                 fprintf(options->file, "\n");
 908         }
 909         fprintf(options->file,
 910                "%s %d files changed, %d insertions(+), %d deletions(-)%s\n",
 911                set, total_files, adds, dels, reset);
 912 }
 913
 914 static void show_shortstats(struct diffstat_t* data, struct diff_options *options)
 915 {
 916         int i, adds = 0, dels = 0, total_files = data->nr;
 917
 918         if (data->nr == 0)
 919                 return;
 920
 921         for (i = 0; i < data->nr; i++) {
 922                 if (!data->files[i]->is_binary &&
 923                     !data->files[i]->is_unmerged) {
 924                         int added = data->files[i]->added;
 925                         int deleted= data->files[i]->deleted;
 926                         if (!data->files[i]->is_renamed &&
 927                             (added + deleted == 0)) {
 928                                 total_files--;
 929                         } else {
 930                                 adds += added;
 931                                 dels += deleted;
 932                         }
 933                 }
 934         }
 935         fprintf(options->file, " %d files changed, %d insertions(+), %d deletions(-)\n",
 936                total_files, adds, dels);
 937 }
 938
 939 static void show_numstat(struct diffstat_t* data, struct diff_options *options)
 940 {
 941         int i;
 942
 943         if (data->nr == 0)
 944                 return;
 945
 946         for (i = 0; i < data->nr; i++) {
 947                 struct diffstat_file *file = data->files[i];
 948
 949                 if (file->is_binary)
 950                         fprintf(options->file, "-\t-\t");
 951                 else
 952                         fprintf(options->file,
 953                                 "%d\t%d\t", file->added, file->deleted);
 954                 if (options->line_termination) {
 955                         fill_print_name(file);
 956                         if (!file->is_renamed)
 957                                 write_name_quoted(file->name, options->file,
 958                                                   options->line_termination);
 959                         else {
 960                                 fputs(file->print_name, options->file);
 961                                 putc(options->line_termination, options->file);
 962                         }
 963                 } else {
 964                         if (file->is_renamed) {
 965                                 putc('\0', options->file);
 966                                 write_name_quoted(file->from_name, options->file, '\0');
 967                         }
 968                         write_name_quoted(file->name, options->file, '\0');
 969                 }
 970         }
 971 }
 972
 973 struct dirstat_file {
 974         const char *name;
 975         unsigned long changed;
 976 };
 977
 978 struct dirstat_dir {
 979         struct dirstat_file *files;
 980         int alloc, nr, percent, cumulative;
 981 };
 982
 983 static long gather_dirstat(FILE *file, struct dirstat_dir *dir, unsigned long changed, const char *base, int baselen)
 984 {
 985         unsigned long this_dir = 0;
 986         unsigned int sources = 0;
 987
 988         while (dir->nr) {
 989                 struct dirstat_file *f = dir->files;
 990                 int namelen = strlen(f->name);
 991                 unsigned long this;
 992                 char *slash;
 993
 994                 if (namelen < baselen)
 995                         break;
 996                 if (memcmp(f->name, base, baselen))
 997                         break;
 998                 slash = strchr(f->name + baselen, '/');
 999                 if (slash) {
1000                         int newbaselen = slash + 1 - f->name;
1001                         this = gather_dirstat(file, dir, changed, f->name, newbaselen);
1002                         sources++;
1003                 } else {
1004                         this = f->changed;
1005                         dir->files++;
1006                         dir->nr--;
1007                         sources += 2;
1008                 }
1009                 this_dir += this;
1010         }
1011
1012         /*
1013          * We don't report dirstat's for
1014          *  - the top level
1015          *  - or cases where everything came from a single directory
1016          *    under this directory (sources == 1).
1017          */
1018         if (baselen && sources != 1) {
1019                 int permille = this_dir * 1000 / changed;
1020                 if (permille) {
1021                         int percent = permille / 10;
1022                         if (percent >= dir->percent) {
1023                                 fprintf(file, "%4d.%01d%% %.*s\n", percent, permille % 10, baselen, base);
1024                                 if (!dir->cumulative)
1025                                         return 0;
1026                         }
1027                 }
1028         }
1029         return this_dir;
1030 }
1031
1032 static int dirstat_compare(const void *_a, const void *_b)
1033 {
1034         const struct dirstat_file *a = _a;
1035         const struct dirstat_file *b = _b;
1036         return strcmp(a->name, b->name);
1037 }
1038
1039 static void show_dirstat(struct diff_options *options)
1040 {
1041         int i;
1042         unsigned long changed;
1043         struct dirstat_dir dir;
1044         struct diff_queue_struct *q = &diff_queued_diff;
1045
1046         dir.files = NULL;
1047         dir.alloc = 0;
1048         dir.nr = 0;
1049         dir.percent = options->dirstat_percent;
1050         dir.cumulative = DIFF_OPT_TST(options, DIRSTAT_CUMULATIVE);
1051
1052         changed = 0;
1053         for (i = 0; i < q->nr; i++) {
1054                 struct diff_filepair *p = q->queue[i];
1055                 const char *name;
1056                 unsigned long copied, added, damage;
1057
1058                 name = p->one->path ? p->one->path : p->two->path;
1059
1060                 if (DIFF_FILE_VALID(p->one) && DIFF_FILE_VALID(p->two)) {
1061                         diff_populate_filespec(p->one, 0);
1062                         diff_populate_filespec(p->two, 0);
1063                         diffcore_count_changes(p->one, p->two, NULL, NULL, 0,
1064                                                &copied, &added);
1065                         diff_free_filespec_data(p->one);
1066                         diff_free_filespec_data(p->two);
1067                 } else if (DIFF_FILE_VALID(p->one)) {
1068                         diff_populate_filespec(p->one, 1);
1069                         copied = added = 0;
1070                         diff_free_filespec_data(p->one);
1071                 } else if (DIFF_FILE_VALID(p->two)) {
1072                         diff_populate_filespec(p->two, 1);
1073                         copied = 0;
1074                         added = p->two->size;
1075                         diff_free_filespec_data(p->two);
1076                 } else
1077                         continue;
1078
1079                 /*
1080                  * Original minus copied is the removed material,
1081                  * added is the new material.  They are both damages
1082                  * made to the preimage. In --dirstat-by-file mode, count
1083                  * damaged files, not damaged lines. This is done by
1084                  * counting only a single damaged line per file.
1085                  */
1086                 damage = (p->one->size - copied) + added;
1087                 if (DIFF_OPT_TST(options, DIRSTAT_BY_FILE) && damage > 0)
1088                         damage = 1;
1089
1090                 ALLOC_GROW(dir.files, dir.nr + 1, dir.alloc);
1091                 dir.files[dir.nr].name = name;
1092                 dir.files[dir.nr].changed = damage;
1093                 changed += damage;
1094                 dir.nr++;
1095         }
1096
1097         /* This can happen even with many files, if everything was renames */
1098         if (!changed)
1099                 return;
1100
1101         /* Show all directories with more than x% of the changes */
1102         qsort(dir.files, dir.nr, sizeof(dir.files[0]), dirstat_compare);
1103         gather_dirstat(options->file, &dir, changed, "", 0);
1104 }
1105
1106 static void free_diffstat_info(struct diffstat_t *diffstat)
1107 {
1108         int i;
1109         for (i = 0; i < diffstat->nr; i++) {
1110                 struct diffstat_file *f = diffstat->files[i];
1111                 if (f->name != f->print_name)
1112                         free(f->print_name);
1113                 free(f->name);
1114                 free(f->from_name);
1115                 free(f);
1116         }
1117         free(diffstat->files);
1118 }
1119
1120 struct checkdiff_t {
1121         const char *filename;
1122         int lineno;
1123         struct diff_options *o;
1124         unsigned ws_rule;
1125         unsigned status;
1126         int trailing_blanks_start;
1127 };
1128
1129 static int is_conflict_marker(const char *line, unsigned long len)
1130 {
1131         char firstchar;
1132         int cnt;
1133
1134         if (len < 8)
1135                 return 0;
1136         firstchar = line[0];
1137         switch (firstchar) {
1138         case '=': case '>': case '<':
1139                 break;
1140         default:
1141                 return 0;
1142         }
1143         for (cnt = 1; cnt < 7; cnt++)
1144                 if (line[cnt] != firstchar)
1145                         return 0;
1146         /* line[0] thru line[6] are same as firstchar */
1147         if (firstchar == '=') {
1148                 /* divider between ours and theirs? */
1149                 if (len != 8 || line[7] != '\n')
1150                         return 0;
1151         } else if (len < 8 || !isspace(line[7])) {
1152                 /* not divider before ours nor after theirs */
1153                 return 0;
1154         }
1155         return 1;
1156 }
1157
1158 static void checkdiff_consume(void *priv, char *line, unsigned long len)
1159 {
1160         struct checkdiff_t *data = priv;
1161         int color_diff = DIFF_OPT_TST(data->o, COLOR_DIFF);
1162         const char *ws = diff_get_color(color_diff, DIFF_WHITESPACE);
1163         const char *reset = diff_get_color(color_diff, DIFF_RESET);
1164         const char *set = diff_get_color(color_diff, DIFF_FILE_NEW);
1165         char *err;
1166
1167         if (line[0] == '+') {
1168                 unsigned bad;
1169                 data->lineno++;
1170                 if (!ws_blank_line(line + 1, len - 1, data->ws_rule))
1171                         data->trailing_blanks_start = 0;
1172                 else if (!data->trailing_blanks_start)
1173                         data->trailing_blanks_start = data->lineno;
1174                 if (is_conflict_marker(line + 1, len - 1)) {
1175                         data->status |= 1;
1176                         fprintf(data->o->file,
1177                                 "%s:%d: leftover conflict marker\n",
1178                                 data->filename, data->lineno);
1179                 }
1180                 bad = ws_check(line + 1, len - 1, data->ws_rule);
1181                 if (!bad)
1182                         return;
1183                 data->status |= bad;
1184                 err = whitespace_error_string(bad);
1185                 fprintf(data->o->file, "%s:%d: %s.\n",
1186                         data->filename, data->lineno, err);
1187                 free(err);
1188                 emit_line(data->o->file, set, reset, line, 1);
1189                 ws_check_emit(line + 1, len - 1, data->ws_rule,
1190                               data->o->file, set, reset, ws);
1191         } else if (line[0] == ' ') {
1192                 data->lineno++;
1193                 data->trailing_blanks_start = 0;
1194         } else if (line[0] == '@') {
1195                 char *plus = strchr(line, '+');
1196                 if (plus)
1197                         data->lineno = strtol(plus, NULL, 10) - 1;
1198                 else
1199                         die("invalid diff");
1200                 data->trailing_blanks_start = 0;
1201         }
1202 }
1203
1204 static unsigned char *deflate_it(char *data,
1205                                  unsigned long size,
1206                                  unsigned long *result_size)
1207 {
1208         int bound;
1209         unsigned char *deflated;
1210         z_stream stream;
1211
1212         memset(&stream, 0, sizeof(stream));
1213         deflateInit(&stream, zlib_compression_level);
1214         bound = deflateBound(&stream, size);
1215         deflated = xmalloc(bound);
1216         stream.next_out = deflated;
1217         stream.avail_out = bound;
1218
1219         stream.next_in = (unsigned char *)data;
1220         stream.avail_in = size;
1221         while (deflate(&stream, Z_FINISH) == Z_OK)
1222                 ; /* nothing */
1223         deflateEnd(&stream);
1224         *result_size = stream.total_out;
1225         return deflated;
1226 }
1227
1228 static void emit_binary_diff_body(FILE *file, mmfile_t *one, mmfile_t *two)
1229 {
1230         void *cp;
1231         void *delta;
1232         void *deflated;
1233         void *data;
1234         unsigned long orig_size;
1235         unsigned long delta_size;
1236         unsigned long deflate_size;
1237         unsigned long data_size;
1238
1239         /* We could do deflated delta, or we could do just deflated two,
1240          * whichever is smaller.
1241          */
1242         delta = NULL;
1243         deflated = deflate_it(two->ptr, two->size, &deflate_size);
1244         if (one->size && two->size) {
1245                 delta = diff_delta(one->ptr, one->size,
1246                                    two->ptr, two->size,
1247                                    &delta_size, deflate_size);
1248                 if (delta) {
1249                         void *to_free = delta;
1250                         orig_size = delta_size;
1251                         delta = deflate_it(delta, delta_size, &delta_size);
1252                         free(to_free);
1253                 }
1254         }
1255
1256         if (delta && delta_size < deflate_size) {
1257                 fprintf(file, "delta %lu\n", orig_size);
1258                 free(deflated);
1259                 data = delta;
1260                 data_size = delta_size;
1261         }
1262         else {
1263                 fprintf(file, "literal %lu\n", two->size);
1264                 free(delta);
1265                 data = deflated;
1266                 data_size = deflate_size;
1267         }
1268
1269         /* emit data encoded in base85 */
1270         cp = data;
1271         while (data_size) {
1272                 int bytes = (52 < data_size) ? 52 : data_size;
1273                 char line[70];
1274                 data_size -= bytes;
1275                 if (bytes <= 26)
1276                         line[0] = bytes + 'A' - 1;
1277                 else
1278                         line[0] = bytes - 26 + 'a' - 1;
1279                 encode_85(line + 1, cp, bytes);
1280                 cp = (char *) cp + bytes;
1281                 fputs(line, file);
1282                 fputc('\n', file);
1283         }
1284         fprintf(file, "\n");
1285         free(data);
1286 }
1287
1288 static void emit_binary_diff(FILE *file, mmfile_t *one, mmfile_t *two)
1289 {
1290         fprintf(file, "GIT binary patch\n");
1291         emit_binary_diff_body(file, one, two);
1292         emit_binary_diff_body(file, two, one);
1293 }
1294
1295 static void diff_filespec_load_driver(struct diff_filespec *one)
1296 {
1297         if (!one->driver)
1298                 one->driver = userdiff_find_by_path(one->path);
1299         if (!one->driver)
1300                 one->driver = userdiff_find_by_name("default");
1301 }
1302
1303 int diff_filespec_is_binary(struct diff_filespec *one)
1304 {
1305         if (one->is_binary == -1) {
1306                 diff_filespec_load_driver(one);
1307                 if (one->driver->binary != -1)
1308                         one->is_binary = one->driver->binary;
1309                 else {
1310                         if (!one->data && DIFF_FILE_VALID(one))
1311                                 diff_populate_filespec(one, 0);
1312                         if (one->data)
1313                                 one->is_binary = buffer_is_binary(one->data,
1314                                                 one->size);
1315                         if (one->is_binary == -1)
1316                                 one->is_binary = 0;
1317                 }
1318         }
1319         return one->is_binary;
1320 }
1321
1322 static const struct userdiff_funcname *diff_funcname_pattern(struct diff_filespec *one)
1323 {
1324         diff_filespec_load_driver(one);
1325         return one->driver->funcname.pattern ? &one->driver->funcname : NULL;
1326 }
1327
1328 void diff_set_mnemonic_prefix(struct diff_options *options, const char *a, const char *b)
1329 {
1330         if (!options->a_prefix)
1331                 options->a_prefix = a;
1332         if (!options->b_prefix)
1333                 options->b_prefix = b;
1334 }
1335
1336 static const char *get_textconv(struct diff_filespec *one)
1337 {
1338         if (!DIFF_FILE_VALID(one))
1339                 return NULL;
1340         if (!S_ISREG(one->mode))
1341                 return NULL;
1342         diff_filespec_load_driver(one);
1343         return one->driver->textconv;
1344 }
1345
1346 static void builtin_diff(const char *name_a,
1347                          const char *name_b,
1348                          struct diff_filespec *one,
1349                          struct diff_filespec *two,
1350                          const char *xfrm_msg,
1351                          struct diff_options *o,
1352                          int complete_rewrite)
1353 {
1354         mmfile_t mf1, mf2;
1355         const char *lbl[2];
1356         char *a_one, *b_two;
1357         const char *set = diff_get_color_opt(o, DIFF_METAINFO);
1358         const char *reset = diff_get_color_opt(o, DIFF_RESET);
1359         const char *a_prefix, *b_prefix;
1360         const char *textconv_one = NULL, *textconv_two = NULL;
1361
1362         if (DIFF_OPT_TST(o, ALLOW_TEXTCONV)) {
1363                 textconv_one = get_textconv(one);
1364                 textconv_two = get_textconv(two);
1365         }
1366
1367         diff_set_mnemonic_prefix(o, "a/", "b/");
1368         if (DIFF_OPT_TST(o, REVERSE_DIFF)) {
1369                 a_prefix = o->b_prefix;
1370                 b_prefix = o->a_prefix;
1371         } else {
1372                 a_prefix = o->a_prefix;
1373                 b_prefix = o->b_prefix;
1374         }
1375
1376         /* Never use a non-valid filename anywhere if at all possible */
1377         name_a = DIFF_FILE_VALID(one) ? name_a : name_b;
1378         name_b = DIFF_FILE_VALID(two) ? name_b : name_a;
1379
1380         a_one = quote_two(a_prefix, name_a + (*name_a == '/'));
1381         b_two = quote_two(b_prefix, name_b + (*name_b == '/'));
1382         lbl[0] = DIFF_FILE_VALID(one) ? a_one : "/dev/null";
1383         lbl[1] = DIFF_FILE_VALID(two) ? b_two : "/dev/null";
1384         fprintf(o->file, "%sdiff --git %s %s%s\n", set, a_one, b_two, reset);
1385         if (lbl[0][0] == '/') {
1386                 /* /dev/null */
1387                 fprintf(o->file, "%snew file mode %06o%s\n", set, two->mode, reset);
1388                 if (xfrm_msg && xfrm_msg[0])
1389                         fprintf(o->file, "%s%s%s\n", set, xfrm_msg, reset);
1390         }
1391         else if (lbl[1][0] == '/') {
1392                 fprintf(o->file, "%sdeleted file mode %06o%s\n", set, one->mode, reset);
1393                 if (xfrm_msg && xfrm_msg[0])
1394                         fprintf(o->file, "%s%s%s\n", set, xfrm_msg, reset);
1395         }
1396         else {
1397                 if (one->mode != two->mode) {
1398                         fprintf(o->file, "%sold mode %06o%s\n", set, one->mode, reset);
1399                         fprintf(o->file, "%snew mode %06o%s\n", set, two->mode, reset);
1400                 }
1401                 if (xfrm_msg && xfrm_msg[0])
1402                         fprintf(o->file, "%s%s%s\n", set, xfrm_msg, reset);
1403                 /*
1404                  * we do not run diff between different kind
1405                  * of objects.
1406                  */
1407                 if ((one->mode ^ two->mode) & S_IFMT)
1408                         goto free_ab_and_return;
1409                 if (complete_rewrite &&
1410                     (textconv_one || !diff_filespec_is_binary(one)) &&
1411                     (textconv_two || !diff_filespec_is_binary(two))) {
1412                         emit_rewrite_diff(name_a, name_b, one, two,
1413                                                 textconv_one, textconv_two, o);
1414                         o->found_changes = 1;
1415                         goto free_ab_and_return;
1416                 }
1417         }
1418
1419         if (fill_mmfile(&mf1, one) < 0 || fill_mmfile(&mf2, two) < 0)
1420                 die("unable to read files to diff");
1421
1422         if (!DIFF_OPT_TST(o, TEXT) &&
1423             ( (diff_filespec_is_binary(one) && !textconv_one) ||
1424               (diff_filespec_is_binary(two) && !textconv_two) )) {
1425                 /* Quite common confusing case */
1426                 if (mf1.size == mf2.size &&
1427                     !memcmp(mf1.ptr, mf2.ptr, mf1.size))
1428                         goto free_ab_and_return;
1429                 if (DIFF_OPT_TST(o, BINARY))
1430                         emit_binary_diff(o->file, &mf1, &mf2);
1431                 else
1432                         fprintf(o->file, "Binary files %s and %s differ\n",
1433                                 lbl[0], lbl[1]);
1434                 o->found_changes = 1;
1435         }
1436         else {
1437                 /* Crazy xdl interfaces.. */
1438                 const char *diffopts = getenv("GIT_DIFF_OPTS");
1439                 xpparam_t xpp;
1440                 xdemitconf_t xecfg;
1441                 xdemitcb_t ecb;
1442                 struct emit_callback ecbdata;
1443                 const struct userdiff_funcname *pe;
1444
1445                 if (textconv_one) {
1446                         size_t size;
1447                         mf1.ptr = run_textconv(textconv_one, one, &size);
1448                         if (!mf1.ptr)
1449                                 die("unable to read files to diff");
1450                         mf1.size = size;
1451                 }
1452                 if (textconv_two) {
1453                         size_t size;
1454                         mf2.ptr = run_textconv(textconv_two, two, &size);
1455                         if (!mf2.ptr)
1456                                 die("unable to read files to diff");
1457                         mf2.size = size;
1458                 }
1459
1460                 pe = diff_funcname_pattern(one);
1461                 if (!pe)
1462                         pe = diff_funcname_pattern(two);
1463
1464                 memset(&xpp, 0, sizeof(xpp));
1465                 memset(&xecfg, 0, sizeof(xecfg));
1466                 memset(&ecbdata, 0, sizeof(ecbdata));
1467                 ecbdata.label_path = lbl;
1468                 ecbdata.color_diff = DIFF_OPT_TST(o, COLOR_DIFF);
1469                 ecbdata.found_changesp = &o->found_changes;
1470                 ecbdata.ws_rule = whitespace_rule(name_b ? name_b : name_a);
1471                 ecbdata.file = o->file;
1472                 xpp.flags = XDF_NEED_MINIMAL | o->xdl_opts;
1473                 xecfg.ctxlen = o->context;
1474                 xecfg.flags = XDL_EMIT_FUNCNAMES;
1475                 if (pe)
1476                         xdiff_set_find_func(&xecfg, pe->pattern, pe->cflags);
1477                 if (!diffopts)
1478                         ;
1479                 else if (!prefixcmp(diffopts, "--unified="))
1480                         xecfg.ctxlen = strtoul(diffopts + 10, NULL, 10);
1481                 else if (!prefixcmp(diffopts, "-u"))
1482                         xecfg.ctxlen = strtoul(diffopts + 2, NULL, 10);
1483                 if (DIFF_OPT_TST(o, COLOR_DIFF_WORDS)) {
1484                         ecbdata.diff_words =
1485                                 xcalloc(1, sizeof(struct diff_words_data));
1486                         ecbdata.diff_words->file = o->file;
1487                 }
1488                 xdi_diff_outf(&mf1, &mf2, fn_out_consume, &ecbdata,
1489                               &xpp, &xecfg, &ecb);
1490                 if (DIFF_OPT_TST(o, COLOR_DIFF_WORDS))
1491                         free_diff_words_data(&ecbdata);
1492                 if (textconv_one)
1493                         free(mf1.ptr);
1494                 if (textconv_two)
1495                         free(mf2.ptr);
1496         }
1497
1498  free_ab_and_return:
1499         diff_free_filespec_data(one);
1500         diff_free_filespec_data(two);
1501         free(a_one);
1502         free(b_two);
1503         return;
1504 }
1505
1506 static void builtin_diffstat(const char *name_a, const char *name_b,
1507                              struct diff_filespec *one,
1508                              struct diff_filespec *two,
1509                              struct diffstat_t *diffstat,
1510                              struct diff_options *o,
1511                              int complete_rewrite)
1512 {
1513         mmfile_t mf1, mf2;
1514         struct diffstat_file *data;
1515
1516         data = diffstat_add(diffstat, name_a, name_b);
1517
1518         if (!one || !two) {
1519                 data->is_unmerged = 1;
1520                 return;
1521         }
1522         if (complete_rewrite) {
1523                 diff_populate_filespec(one, 0);
1524                 diff_populate_filespec(two, 0);
1525                 data->deleted = count_lines(one->data, one->size);
1526                 data->added = count_lines(two->data, two->size);
1527                 goto free_and_return;
1528         }
1529         if (fill_mmfile(&mf1, one) < 0 || fill_mmfile(&mf2, two) < 0)
1530                 die("unable to read files to diff");
1531
1532         if (diff_filespec_is_binary(one) || diff_filespec_is_binary(two)) {
1533                 data->is_binary = 1;
1534                 data->added = mf2.size;
1535                 data->deleted = mf1.size;
1536         } else {
1537                 /* Crazy xdl interfaces.. */
1538                 xpparam_t xpp;
1539                 xdemitconf_t xecfg;
1540                 xdemitcb_t ecb;
1541
1542                 memset(&xpp, 0, sizeof(xpp));
1543                 memset(&xecfg, 0, sizeof(xecfg));
1544                 xpp.flags = XDF_NEED_MINIMAL | o->xdl_opts;
1545                 xdi_diff_outf(&mf1, &mf2, diffstat_consume, diffstat,
1546                               &xpp, &xecfg, &ecb);
1547         }
1548
1549  free_and_return:
1550         diff_free_filespec_data(one);
1551         diff_free_filespec_data(two);
1552 }
1553
1554 static void builtin_checkdiff(const char *name_a, const char *name_b,
1555                               const char *attr_path,
1556                               struct diff_filespec *one,
1557                               struct diff_filespec *two,
1558                               struct diff_options *o)
1559 {
1560         mmfile_t mf1, mf2;
1561         struct checkdiff_t data;
1562
1563         if (!two)
1564                 return;
1565
1566         memset(&data, 0, sizeof(data));
1567         data.filename = name_b ? name_b : name_a;
1568         data.lineno = 0;
1569         data.o = o;
1570         data.ws_rule = whitespace_rule(attr_path);
1571
1572         if (fill_mmfile(&mf1, one) < 0 || fill_mmfile(&mf2, two) < 0)
1573                 die("unable to read files to diff");
1574
1575         /*
1576          * All the other codepaths check both sides, but not checking
1577          * the "old" side here is deliberate.  We are checking the newly
1578          * introduced changes, and as long as the "new" side is text, we
1579          * can and should check what it introduces.
1580          */
1581         if (diff_filespec_is_binary(two))
1582                 goto free_and_return;
1583         else {
1584                 /* Crazy xdl interfaces.. */
1585                 xpparam_t xpp;
1586                 xdemitconf_t xecfg;
1587                 xdemitcb_t ecb;
1588
1589                 memset(&xpp, 0, sizeof(xpp));
1590                 memset(&xecfg, 0, sizeof(xecfg));
1591                 xecfg.ctxlen = 1; /* at least one context line */
1592                 xpp.flags = XDF_NEED_MINIMAL;
1593                 xdi_diff_outf(&mf1, &mf2, checkdiff_consume, &data,
1594                               &xpp, &xecfg, &ecb);
1595
1596                 if ((data.ws_rule & WS_TRAILING_SPACE) &&
1597                     data.trailing_blanks_start) {
1598                         fprintf(o->file, "%s:%d: ends with blank lines.\n",
1599                                 data.filename, data.trailing_blanks_start);
1600                         data.status = 1; /* report errors */
1601                 }
1602         }
1603  free_and_return:
1604         diff_free_filespec_data(one);
1605         diff_free_filespec_data(two);
1606         if (data.status)
1607                 DIFF_OPT_SET(o, CHECK_FAILED);
1608 }
1609
1610 struct diff_filespec *alloc_filespec(const char *path)
1611 {
1612         int namelen = strlen(path);
1613         struct diff_filespec *spec = xmalloc(sizeof(*spec) + namelen + 1);
1614
1615         memset(spec, 0, sizeof(*spec));
1616         spec->path = (char *)(spec + 1);
1617         memcpy(spec->path, path, namelen+1);
1618         spec->count = 1;
1619         spec->is_binary = -1;
1620         return spec;
1621 }
1622
1623 void free_filespec(struct diff_filespec *spec)
1624 {
1625         if (!--spec->count) {
1626                 diff_free_filespec_data(spec);
1627                 free(spec);
1628         }
1629 }
1630
1631 void fill_filespec(struct diff_filespec *spec, const unsigned char *sha1,
1632                    unsigned short mode)
1633 {
1634         if (mode) {
1635                 spec->mode = canon_mode(mode);
1636                 hashcpy(spec->sha1, sha1);
1637                 spec->sha1_valid = !is_null_sha1(sha1);
1638         }
1639 }
1640
1641 /*
1642  * Given a name and sha1 pair, if the index tells us the file in
1643  * the work tree has that object contents, return true, so that
1644  * prepare_temp_file() does not have to inflate and extract.
1645  */
1646 static int reuse_worktree_file(const char *name, const unsigned char *sha1, int want_file)
1647 {
1648         struct cache_entry *ce;
1649         struct stat st;
1650         int pos, len;
1651
1652         /*
1653          * We do not read the cache ourselves here, because the
1654          * benchmark with my previous version that always reads cache
1655          * shows that it makes things worse for diff-tree comparing
1656          * two linux-2.6 kernel trees in an already checked out work
1657          * tree.  This is because most diff-tree comparisons deal with
1658          * only a small number of files, while reading the cache is
1659          * expensive for a large project, and its cost outweighs the
1660          * savings we get by not inflating the object to a temporary
1661          * file.  Practically, this code only helps when we are used
1662          * by diff-cache --cached, which does read the cache before
1663          * calling us.
1664          */
1665         if (!active_cache)
1666                 return 0;
1667
1668         /* We want to avoid the working directory if our caller
1669          * doesn't need the data in a normal file, this system
1670          * is rather slow with its stat/open/mmap/close syscalls,
1671          * and the object is contained in a pack file.  The pack
1672          * is probably already open and will be faster to obtain
1673          * the data through than the working directory.  Loose
1674          * objects however would tend to be slower as they need
1675          * to be individually opened and inflated.
1676          */
1677         if (!FAST_WORKING_DIRECTORY && !want_file && has_sha1_pack(sha1))
1678                 return 0;
1679
1680         len = strlen(name);
1681         pos = cache_name_pos(name, len);
1682         if (pos < 0)
1683                 return 0;
1684         ce = active_cache[pos];
1685
1686         /*
1687          * This is not the sha1 we are looking for, or
1688          * unreusable because it is not a regular file.
1689          */
1690         if (hashcmp(sha1, ce->sha1) || !S_ISREG(ce->ce_mode))
1691                 return 0;
1692
1693         /*
1694          * If ce is marked as "assume unchanged", there is no
1695          * guarantee that work tree matches what we are looking for.
1696          */
1697         if (ce->ce_flags & CE_VALID)
1698                 return 0;
1699
1700         /*
1701          * If ce matches the file in the work tree, we can reuse it.
1702          */
1703         if (ce_uptodate(ce) ||
1704             (!lstat(name, &st) && !ce_match_stat(ce, &st, 0)))
1705                 return 1;
1706
1707         return 0;
1708 }
1709
1710 static int populate_from_stdin(struct diff_filespec *s)
1711 {
1712         struct strbuf buf = STRBUF_INIT;
1713         size_t size = 0;
1714
1715         if (strbuf_read(&buf, 0, 0) < 0)
1716                 return error("error while reading from stdin %s",
1717                                      strerror(errno));
1718
1719         s->should_munmap = 0;
1720         s->data = strbuf_detach(&buf, &size);
1721         s->size = size;
1722         s->should_free = 1;
1723         return 0;
1724 }
1725
1726 static int diff_populate_gitlink(struct diff_filespec *s, int size_only)
1727 {
1728         int len;
1729         char *data = xmalloc(100);
1730         len = snprintf(data, 100,
1731                 "Subproject commit %s\n", sha1_to_hex(s->sha1));
1732         s->data = data;
1733         s->size = len;
1734         s->should_free = 1;
1735         if (size_only) {
1736                 s->data = NULL;
1737                 free(data);
1738         }
1739         return 0;
1740 }
1741
1742 /*
1743  * While doing rename detection and pickaxe operation, we may need to
1744  * grab the data for the blob (or file) for our own in-core comparison.
1745  * diff_filespec has data and size fields for this purpose.
1746  */
1747 int diff_populate_filespec(struct diff_filespec *s, int size_only)
1748 {
1749         int err = 0;
1750         if (!DIFF_FILE_VALID(s))
1751                 die("internal error: asking to populate invalid file.");
1752         if (S_ISDIR(s->mode))
1753                 return -1;
1754
1755         if (s->data)
1756                 return 0;
1757
1758         if (size_only && 0 < s->size)
1759                 return 0;
1760
1761         if (S_ISGITLINK(s->mode))
1762                 return diff_populate_gitlink(s, size_only);
1763
1764         if (!s->sha1_valid ||
1765             reuse_worktree_file(s->path, s->sha1, 0)) {
1766                 struct strbuf buf = STRBUF_INIT;
1767                 struct stat st;
1768                 int fd;
1769
1770                 if (!strcmp(s->path, "-"))
1771                         return populate_from_stdin(s);
1772
1773                 if (lstat(s->path, &st) < 0) {
1774                         if (errno == ENOENT) {
1775                         err_empty:
1776                                 err = -1;
1777                         empty:
1778                                 s->data = (char *)"";
1779                                 s->size = 0;
1780                                 return err;
1781                         }
1782                 }
1783                 s->size = xsize_t(st.st_size);
1784                 if (!s->size)
1785                         goto empty;
1786                 if (S_ISLNK(st.st_mode)) {
1787                         struct strbuf sb = STRBUF_INIT;
1788
1789                         if (strbuf_readlink(&sb, s->path, s->size))
1790                                 goto err_empty;
1791                         s->size = sb.len;
1792                         s->data = strbuf_detach(&sb, NULL);
1793                         s->should_free = 1;
1794                         return 0;
1795                 }
1796                 if (size_only)
1797                         return 0;
1798                 fd = open(s->path, O_RDONLY);
1799                 if (fd < 0)
1800                         goto err_empty;
1801                 s->data = xmmap(NULL, s->size, PROT_READ, MAP_PRIVATE, fd, 0);
1802                 close(fd);
1803                 s->should_munmap = 1;
1804
1805                 /*
1806                  * Convert from working tree format to canonical git format
1807                  */
1808                 if (convert_to_git(s->path, s->data, s->size, &buf, safe_crlf)) {
1809                         size_t size = 0;
1810                         munmap(s->data, s->size);
1811                         s->should_munmap = 0;
1812                         s->data = strbuf_detach(&buf, &size);
1813                         s->size = size;
1814                         s->should_free = 1;
1815                 }
1816         }
1817         else {
1818                 enum object_type type;
1819                 if (size_only)
1820                         type = sha1_object_info(s->sha1, &s->size);
1821                 else {
1822                         s->data = read_sha1_file(s->sha1, &type, &s->size);
1823                         s->should_free = 1;
1824                 }
1825         }
1826         return 0;
1827 }
1828
1829 void diff_free_filespec_blob(struct diff_filespec *s)
1830 {
1831         if (s->should_free)
1832                 free(s->data);
1833         else if (s->should_munmap)
1834                 munmap(s->data, s->size);
1835
1836         if (s->should_free || s->should_munmap) {
1837                 s->should_free = s->should_munmap = 0;
1838                 s->data = NULL;
1839         }
1840 }
1841
1842 void diff_free_filespec_data(struct diff_filespec *s)
1843 {
1844         diff_free_filespec_blob(s);
1845         free(s->cnt_data);
1846         s->cnt_data = NULL;
1847 }
1848
1849 static void prep_temp_blob(struct diff_tempfile *temp,
1850                            void *blob,
1851                            unsigned long size,
1852                            const unsigned char *sha1,
1853                            int mode)
1854 {
1855         int fd;
1856
1857         fd = git_mkstemp(temp->tmp_path, PATH_MAX, ".diff_XXXXXX");
1858         if (fd < 0)
1859                 die("unable to create temp-file: %s", strerror(errno));
1860         if (write_in_full(fd, blob, size) != size)
1861                 die("unable to write temp-file");
1862         close(fd);
1863         temp->name = temp->tmp_path;
1864         strcpy(temp->hex, sha1_to_hex(sha1));
1865         temp->hex[40] = 0;
1866         sprintf(temp->mode, "%06o", mode);
1867 }
1868
1869 static void prepare_temp_file(const char *name,
1870                               struct diff_tempfile *temp,
1871                               struct diff_filespec *one)
1872 {
1873         if (!DIFF_FILE_VALID(one)) {
1874         not_a_valid_file:
1875                 /* A '-' entry produces this for file-2, and
1876                  * a '+' entry produces this for file-1.
1877                  */
1878                 temp->name = "/dev/null";
1879                 strcpy(temp->hex, ".");
1880                 strcpy(temp->mode, ".");
1881                 return;
1882         }
1883
1884         if (!one->sha1_valid ||
1885             reuse_worktree_file(name, one->sha1, 1)) {
1886                 struct stat st;
1887                 if (lstat(name, &st) < 0) {
1888                         if (errno == ENOENT)
1889                                 goto not_a_valid_file;
1890                         die("stat(%s): %s", name, strerror(errno));
1891                 }
1892                 if (S_ISLNK(st.st_mode)) {
1893                         int ret;
1894                         char buf[PATH_MAX + 1]; /* ought to be SYMLINK_MAX */
1895                         ret = readlink(name, buf, sizeof(buf));
1896                         if (ret < 0)
1897                                 die("readlink(%s)", name);
1898                         if (ret == sizeof(buf))
1899                                 die("symlink too long: %s", name);
1900                         prep_temp_blob(temp, buf, ret,
1901                                        (one->sha1_valid ?
1902                                         one->sha1 : null_sha1),
1903                                        (one->sha1_valid ?
1904                                         one->mode : S_IFLNK));
1905                 }
1906                 else {
1907                         /* we can borrow from the file in the work tree */
1908                         temp->name = name;
1909                         if (!one->sha1_valid)
1910                                 strcpy(temp->hex, sha1_to_hex(null_sha1));
1911                         else
1912                                 strcpy(temp->hex, sha1_to_hex(one->sha1));
1913                         /* Even though we may sometimes borrow the
1914                          * contents from the work tree, we always want
1915                          * one->mode.  mode is trustworthy even when
1916                          * !(one->sha1_valid), as long as
1917                          * DIFF_FILE_VALID(one).
1918                          */
1919                         sprintf(temp->mode, "%06o", one->mode);
1920                 }
1921                 return;
1922         }
1923         else {
1924                 if (diff_populate_filespec(one, 0))
1925                         die("cannot read data blob for %s", one->path);
1926                 prep_temp_blob(temp, one->data, one->size,
1927                                one->sha1, one->mode);
1928         }
1929 }
1930
1931 static void remove_tempfile(void)
1932 {
1933         int i;
1934
1935         for (i = 0; i < 2; i++)
1936                 if (diff_temp[i].name == diff_temp[i].tmp_path) {
1937                         unlink(diff_temp[i].name);
1938                         diff_temp[i].name = NULL;
1939                 }
1940 }
1941
1942 static void remove_tempfile_on_signal(int signo)
1943 {
1944         remove_tempfile();
1945         signal(SIGINT, SIG_DFL);
1946         raise(signo);
1947 }
1948
1949 /* An external diff command takes:
1950  *
1951  * diff-cmd name infile1 infile1-sha1 infile1-mode \
1952  *               infile2 infile2-sha1 infile2-mode [ rename-to ]
1953  *
1954  */
1955 static void run_external_diff(const char *pgm,
1956                               const char *name,
1957                               const char *other,
1958                               struct diff_filespec *one,
1959                               struct diff_filespec *two,
1960                               const char *xfrm_msg,
1961                               int complete_rewrite)
1962 {
1963         const char *spawn_arg[10];
1964         struct diff_tempfile *temp = diff_temp;
1965         int retval;
1966         static int atexit_asked = 0;
1967         const char *othername;
1968         const char **arg = &spawn_arg[0];
1969
1970         othername = (other? other : name);
1971         if (one && two) {
1972                 prepare_temp_file(name, &temp[0], one);
1973                 prepare_temp_file(othername, &temp[1], two);
1974                 if (! atexit_asked &&
1975                     (temp[0].name == temp[0].tmp_path ||
1976                      temp[1].name == temp[1].tmp_path)) {
1977                         atexit_asked = 1;
1978                         atexit(remove_tempfile);
1979                 }
1980                 signal(SIGINT, remove_tempfile_on_signal);
1981         }
1982
1983         if (one && two) {
1984                 *arg++ = pgm;
1985                 *arg++ = name;
1986                 *arg++ = temp[0].name;
1987                 *arg++ = temp[0].hex;
1988                 *arg++ = temp[0].mode;
1989                 *arg++ = temp[1].name;
1990                 *arg++ = temp[1].hex;
1991                 *arg++ = temp[1].mode;
1992                 if (other) {
1993                         *arg++ = other;
1994                         *arg++ = xfrm_msg;
1995                 }
1996         } else {
1997                 *arg++ = pgm;
1998                 *arg++ = name;
1999         }
2000         *arg = NULL;
2001         fflush(NULL);
2002         retval = run_command_v_opt(spawn_arg, 0);
2003         remove_tempfile();
2004         if (retval) {
2005                 fprintf(stderr, "external diff died, stopping at %s.\n", name);
2006                 exit(1);
2007         }
2008 }
2009
2010 static int similarity_index(struct diff_filepair *p)
2011 {
2012         return p->score * 100 / MAX_SCORE;
2013 }
2014
2015 static void fill_metainfo(struct strbuf *msg,
2016                           const char *name,
2017                           const char *other,
2018                           struct diff_filespec *one,
2019                           struct diff_filespec *two,
2020                           struct diff_options *o,
2021                           struct diff_filepair *p)
2022 {
2023         strbuf_init(msg, PATH_MAX * 2 + 300);
2024         switch (p->status) {
2025         case DIFF_STATUS_COPIED:
2026                 strbuf_addf(msg, "similarity index %d%%", similarity_index(p));
2027                 strbuf_addstr(msg, "\ncopy from ");
2028                 quote_c_style(name, msg, NULL, 0);
2029                 strbuf_addstr(msg, "\ncopy to ");
2030                 quote_c_style(other, msg, NULL, 0);
2031                 strbuf_addch(msg, '\n');
2032                 break;
2033         case DIFF_STATUS_RENAMED:
2034                 strbuf_addf(msg, "similarity index %d%%", similarity_index(p));
2035                 strbuf_addstr(msg, "\nrename from ");
2036                 quote_c_style(name, msg, NULL, 0);
2037                 strbuf_addstr(msg, "\nrename to ");
2038                 quote_c_style(other, msg, NULL, 0);
2039                 strbuf_addch(msg, '\n');
2040                 break;
2041         case DIFF_STATUS_MODIFIED:
2042                 if (p->score) {
2043                         strbuf_addf(msg, "dissimilarity index %d%%\n",
2044                                     similarity_index(p));
2045                         break;
2046                 }
2047                 /* fallthru */
2048         default:
2049                 /* nothing */
2050                 ;
2051         }
2052         if (one && two && hashcmp(one->sha1, two->sha1)) {
2053                 int abbrev = DIFF_OPT_TST(o, FULL_INDEX) ? 40 : DEFAULT_ABBREV;
2054
2055                 if (DIFF_OPT_TST(o, BINARY)) {
2056                         mmfile_t mf;
2057                         if ((!fill_mmfile(&mf, one) && diff_filespec_is_binary(one)) ||
2058                             (!fill_mmfile(&mf, two) && diff_filespec_is_binary(two)))
2059                                 abbrev = 40;
2060                 }
2061                 strbuf_addf(msg, "index %.*s..%.*s",
2062                             abbrev, sha1_to_hex(one->sha1),
2063                             abbrev, sha1_to_hex(two->sha1));
2064                 if (one->mode == two->mode)
2065                         strbuf_addf(msg, " %06o", one->mode);
2066                 strbuf_addch(msg, '\n');
2067         }
2068         if (msg->len)
2069                 strbuf_setlen(msg, msg->len - 1);
2070 }
2071
2072 static void run_diff_cmd(const char *pgm,
2073                          const char *name,
2074                          const char *other,
2075                          const char *attr_path,
2076                          struct diff_filespec *one,
2077                          struct diff_filespec *two,
2078                          struct strbuf *msg,
2079                          struct diff_options *o,
2080                          struct diff_filepair *p)
2081 {
2082         const char *xfrm_msg = NULL;
2083         int complete_rewrite = (p->status == DIFF_STATUS_MODIFIED) && p->score;
2084
2085         if (msg) {
2086                 fill_metainfo(msg, name, other, one, two, o, p);
2087                 xfrm_msg = msg->len ? msg->buf : NULL;
2088         }
2089
2090         if (!DIFF_OPT_TST(o, ALLOW_EXTERNAL))
2091                 pgm = NULL;
2092         else {
2093                 struct userdiff_driver *drv = userdiff_find_by_path(attr_path);
2094                 if (drv && drv->external)
2095                         pgm = drv->external;
2096         }
2097
2098         if (pgm) {
2099                 run_external_diff(pgm, name, other, one, two, xfrm_msg,
2100                                   complete_rewrite);
2101                 return;
2102         }
2103         if (one && two)
2104                 builtin_diff(name, other ? other : name,
2105                              one, two, xfrm_msg, o, complete_rewrite);
2106         else
2107                 fprintf(o->file, "* Unmerged path %s\n", name);
2108 }
2109
2110 static void diff_fill_sha1_info(struct diff_filespec *one)
2111 {
2112         if (DIFF_FILE_VALID(one)) {
2113                 if (!one->sha1_valid) {
2114                         struct stat st;
2115                         if (!strcmp(one->path, "-")) {
2116                                 hashcpy(one->sha1, null_sha1);
2117                                 return;
2118                         }
2119                         if (lstat(one->path, &st) < 0)
2120                                 die("stat %s", one->path);
2121                         if (index_path(one->sha1, one->path, &st, 0))
2122                                 die("cannot hash %s\n", one->path);
2123                 }
2124         }
2125         else
2126                 hashclr(one->sha1);
2127 }
2128
2129 static void strip_prefix(int prefix_length, const char **namep, const char **otherp)
2130 {
2131         /* Strip the prefix but do not molest /dev/null and absolute paths */
2132         if (*namep && **namep != '/')
2133                 *namep += prefix_length;
2134         if (*otherp && **otherp != '/')
2135                 *otherp += prefix_length;
2136 }
2137
2138 static void run_diff(struct diff_filepair *p, struct diff_options *o)
2139 {
2140         const char *pgm = external_diff();
2141         struct strbuf msg;
2142         struct diff_filespec *one = p->one;
2143         struct diff_filespec *two = p->two;
2144         const char *name;
2145         const char *other;
2146         const char *attr_path;
2147
2148         name  = p->one->path;
2149         other = (strcmp(name, p->two->path) ? p->two->path : NULL);
2150         attr_path = name;
2151         if (o->prefix_length)
2152                 strip_prefix(o->prefix_length, &name, &other);
2153
2154         if (DIFF_PAIR_UNMERGED(p)) {
2155                 run_diff_cmd(pgm, name, NULL, attr_path,
2156                              NULL, NULL, NULL, o, p);
2157                 return;
2158         }
2159
2160         diff_fill_sha1_info(one);
2161         diff_fill_sha1_info(two);
2162
2163         if (!pgm &&
2164             DIFF_FILE_VALID(one) && DIFF_FILE_VALID(two) &&
2165             (S_IFMT & one->mode) != (S_IFMT & two->mode)) {
2166                 /*
2167                  * a filepair that changes between file and symlink
2168                  * needs to be split into deletion and creation.
2169                  */
2170                 struct diff_filespec *null = alloc_filespec(two->path);
2171                 run_diff_cmd(NULL, name, other, attr_path,
2172                              one, null, &msg, o, p);
2173                 free(null);
2174                 strbuf_release(&msg);
2175
2176                 null = alloc_filespec(one->path);
2177                 run_diff_cmd(NULL, name, other, attr_path,
2178                              null, two, &msg, o, p);
2179                 free(null);
2180         }
2181         else
2182                 run_diff_cmd(pgm, name, other, attr_path,
2183                              one, two, &msg, o, p);
2184
2185         strbuf_release(&msg);
2186 }
2187
2188 static void run_diffstat(struct diff_filepair *p, struct diff_options *o,
2189                          struct diffstat_t *diffstat)
2190 {
2191         const char *name;
2192         const char *other;
2193         int complete_rewrite = 0;
2194
2195         if (DIFF_PAIR_UNMERGED(p)) {
2196                 /* unmerged */
2197                 builtin_diffstat(p->one->path, NULL, NULL, NULL, diffstat, o, 0);
2198                 return;
2199         }
2200
2201         name = p->one->path;
2202         other = (strcmp(name, p->two->path) ? p->two->path : NULL);
2203
2204         if (o->prefix_length)
2205                 strip_prefix(o->prefix_length, &name, &other);
2206
2207         diff_fill_sha1_info(p->one);
2208         diff_fill_sha1_info(p->two);
2209
2210         if (p->status == DIFF_STATUS_MODIFIED && p->score)
2211                 complete_rewrite = 1;
2212         builtin_diffstat(name, other, p->one, p->two, diffstat, o, complete_rewrite);
2213 }
2214
2215 static void run_checkdiff(struct diff_filepair *p, struct diff_options *o)
2216 {
2217         const char *name;
2218         const char *other;
2219         const char *attr_path;
2220
2221         if (DIFF_PAIR_UNMERGED(p)) {
2222                 /* unmerged */
2223                 return;
2224         }
2225
2226         name = p->one->path;
2227         other = (strcmp(name, p->two->path) ? p->two->path : NULL);
2228         attr_path = other ? other : name;
2229
2230         if (o->prefix_length)
2231                 strip_prefix(o->prefix_length, &name, &other);
2232
2233         diff_fill_sha1_info(p->one);
2234         diff_fill_sha1_info(p->two);
2235
2236         builtin_checkdiff(name, other, attr_path, p->one, p->two, o);
2237 }
2238
2239 void diff_setup(struct diff_options *options)
2240 {
2241         memset(options, 0, sizeof(*options));
2242
2243         options->file = stdout;
2244
2245         options->line_termination = '\n';
2246         options->break_opt = -1;
2247         options->rename_limit = -1;
2248         options->dirstat_percent = 3;
2249         DIFF_OPT_CLR(options, DIRSTAT_CUMULATIVE);
2250         options->context = 3;
2251
2252         options->change = diff_change;
2253         options->add_remove = diff_addremove;
2254         if (diff_use_color_default > 0)
2255                 DIFF_OPT_SET(options, COLOR_DIFF);
2256         else
2257                 DIFF_OPT_CLR(options, COLOR_DIFF);
2258         options->detect_rename = diff_detect_rename_default;
2259
2260         if (!diff_mnemonic_prefix) {
2261                 options->a_prefix = "a/";
2262                 options->b_prefix = "b/";
2263         }
2264 }
2265
2266 int diff_setup_done(struct diff_options *options)
2267 {
2268         int count = 0;
2269
2270         if (options->output_format & DIFF_FORMAT_NAME)
2271                 count++;
2272         if (options->output_format & DIFF_FORMAT_NAME_STATUS)
2273                 count++;
2274         if (options->output_format & DIFF_FORMAT_CHECKDIFF)
2275                 count++;
2276         if (options->output_format & DIFF_FORMAT_NO_OUTPUT)
2277                 count++;
2278         if (count > 1)
2279                 die("--name-only, --name-status, --check and -s are mutually exclusive");
2280
2281         if (DIFF_OPT_TST(options, FIND_COPIES_HARDER))
2282                 options->detect_rename = DIFF_DETECT_COPY;
2283
2284         if (!DIFF_OPT_TST(options, RELATIVE_NAME))
2285                 options->prefix = NULL;
2286         if (options->prefix)
2287                 options->prefix_length = strlen(options->prefix);
2288         else
2289                 options->prefix_length = 0;
2290
2291         if (options->output_format & (DIFF_FORMAT_NAME |
2292                                       DIFF_FORMAT_NAME_STATUS |
2293                                       DIFF_FORMAT_CHECKDIFF |
2294                                       DIFF_FORMAT_NO_OUTPUT))
2295                 options->output_format &= ~(DIFF_FORMAT_RAW |
2296                                             DIFF_FORMAT_NUMSTAT |
2297                                             DIFF_FORMAT_DIFFSTAT |
2298                                             DIFF_FORMAT_SHORTSTAT |
2299                                             DIFF_FORMAT_DIRSTAT |
2300                                             DIFF_FORMAT_SUMMARY |
2301                                             DIFF_FORMAT_PATCH);
2302
2303         /*
2304          * These cases always need recursive; we do not drop caller-supplied
2305          * recursive bits for other formats here.
2306          */
2307         if (options->output_format & (DIFF_FORMAT_PATCH |
2308                                       DIFF_FORMAT_NUMSTAT |
2309                                       DIFF_FORMAT_DIFFSTAT |
2310                                       DIFF_FORMAT_SHORTSTAT |
2311                                       DIFF_FORMAT_DIRSTAT |
2312                                       DIFF_FORMAT_SUMMARY |
2313                                       DIFF_FORMAT_CHECKDIFF))
2314                 DIFF_OPT_SET(options, RECURSIVE);
2315         /*
2316          * Also pickaxe would not work very well if you do not say recursive
2317          */
2318         if (options->pickaxe)
2319                 DIFF_OPT_SET(options, RECURSIVE);
2320
2321         if (options->detect_rename && options->rename_limit < 0)
2322                 options->rename_limit = diff_rename_limit_default;
2323         if (options->setup & DIFF_SETUP_USE_CACHE) {
2324                 if (!active_cache)
2325                         /* read-cache does not die even when it fails
2326                          * so it is safe for us to do this here.  Also
2327                          * it does not smudge active_cache or active_nr
2328                          * when it fails, so we do not have to worry about
2329                          * cleaning it up ourselves either.
2330                          */
2331                         read_cache();
2332         }
2333         if (options->abbrev <= 0 || 40 < options->abbrev)
2334                 options->abbrev = 40; /* full */
2335
2336         /*
2337          * It does not make sense to show the first hit we happened
2338          * to have found.  It does not make sense not to return with
2339          * exit code in such a case either.
2340          */
2341         if (DIFF_OPT_TST(options, QUIET)) {
2342                 options->output_format = DIFF_FORMAT_NO_OUTPUT;
2343                 DIFF_OPT_SET(options, EXIT_WITH_STATUS);
2344         }
2345
2346         return 0;
2347 }
2348
2349 static int opt_arg(const char *arg, int arg_short, const char *arg_long, int *val)
2350 {
2351         char c, *eq;
2352         int len;
2353
2354         if (*arg != '-')
2355                 return 0;
2356         c = *++arg;
2357         if (!c)
2358                 return 0;
2359         if (c == arg_short) {
2360                 c = *++arg;
2361                 if (!c)
2362                         return 1;
2363                 if (val && isdigit(c)) {
2364                         char *end;
2365                         int n = strtoul(arg, &end, 10);
2366                         if (*end)
2367                                 return 0;
2368                         *val = n;
2369                         return 1;
2370                 }
2371                 return 0;
2372         }
2373         if (c != '-')
2374                 return 0;
2375         arg++;
2376         eq = strchr(arg, '=');
2377         if (eq)
2378                 len = eq - arg;
2379         else
2380                 len = strlen(arg);
2381         if (!len || strncmp(arg, arg_long, len))
2382                 return 0;
2383         if (eq) {
2384                 int n;
2385                 char *end;
2386                 if (!isdigit(*++eq))
2387                         return 0;
2388                 n = strtoul(eq, &end, 10);
2389                 if (*end)
2390                         return 0;
2391                 *val = n;
2392         }
2393         return 1;
2394 }
2395
2396 static int diff_scoreopt_parse(const char *opt);
2397
2398 int diff_opt_parse(struct diff_options *options, const char **av, int ac)
2399 {
2400         const char *arg = av[0];
2401
2402         /* Output format options */
2403         if (!strcmp(arg, "-p") || !strcmp(arg, "-u"))
2404                 options->output_format |= DIFF_FORMAT_PATCH;
2405         else if (opt_arg(arg, 'U', "unified", &options->context))
2406                 options->output_format |= DIFF_FORMAT_PATCH;
2407         else if (!strcmp(arg, "--raw"))
2408                 options->output_format |= DIFF_FORMAT_RAW;
2409         else if (!strcmp(arg, "--patch-with-raw"))
2410                 options->output_format |= DIFF_FORMAT_PATCH | DIFF_FORMAT_RAW;
2411         else if (!strcmp(arg, "--numstat"))
2412                 options->output_format |= DIFF_FORMAT_NUMSTAT;
2413         else if (!strcmp(arg, "--shortstat"))
2414                 options->output_format |= DIFF_FORMAT_SHORTSTAT;
2415         else if (opt_arg(arg, 'X', "dirstat", &options->dirstat_percent))
2416                 options->output_format |= DIFF_FORMAT_DIRSTAT;
2417         else if (!strcmp(arg, "--cumulative")) {
2418                 options->output_format |= DIFF_FORMAT_DIRSTAT;
2419                 DIFF_OPT_SET(options, DIRSTAT_CUMULATIVE);
2420         } else if (opt_arg(arg, 0, "dirstat-by-file",
2421                            &options->dirstat_percent)) {
2422                 options->output_format |= DIFF_FORMAT_DIRSTAT;
2423                 DIFF_OPT_SET(options, DIRSTAT_BY_FILE);
2424         }
2425         else if (!strcmp(arg, "--check"))
2426                 options->output_format |= DIFF_FORMAT_CHECKDIFF;
2427         else if (!strcmp(arg, "--summary"))
2428                 options->output_format |= DIFF_FORMAT_SUMMARY;
2429         else if (!strcmp(arg, "--patch-with-stat"))
2430                 options->output_format |= DIFF_FORMAT_PATCH | DIFF_FORMAT_DIFFSTAT;
2431         else if (!strcmp(arg, "--name-only"))
2432                 options->output_format |= DIFF_FORMAT_NAME;
2433         else if (!strcmp(arg, "--name-status"))
2434                 options->output_format |= DIFF_FORMAT_NAME_STATUS;
2435         else if (!strcmp(arg, "-s"))
2436                 options->output_format |= DIFF_FORMAT_NO_OUTPUT;
2437         else if (!prefixcmp(arg, "--stat")) {
2438                 char *end;
2439                 int width = options->stat_width;
2440                 int name_width = options->stat_name_width;
2441                 arg += 6;
2442                 end = (char *)arg;
2443
2444                 switch (*arg) {
2445                 case '-':
2446                         if (!prefixcmp(arg, "-width="))
2447                                 width = strtoul(arg + 7, &end, 10);
2448                         else if (!prefixcmp(arg, "-name-width="))
2449                                 name_width = strtoul(arg + 12, &end, 10);
2450                         break;
2451                 case '=':
2452                         width = strtoul(arg+1, &end, 10);
2453                         if (*end == ',')
2454                                 name_width = strtoul(end+1, &end, 10);
2455                 }
2456
2457                 /* Important! This checks all the error cases! */
2458                 if (*end)
2459                         return 0;
2460                 options->output_format |= DIFF_FORMAT_DIFFSTAT;
2461                 options->stat_name_width = name_width;
2462                 options->stat_width = width;
2463         }
2464
2465         /* renames options */
2466         else if (!prefixcmp(arg, "-B")) {
2467                 if ((options->break_opt = diff_scoreopt_parse(arg)) == -1)
2468                         return -1;
2469         }
2470         else if (!prefixcmp(arg, "-M")) {
2471                 if ((options->rename_score = diff_scoreopt_parse(arg)) == -1)
2472                         return -1;
2473                 options->detect_rename = DIFF_DETECT_RENAME;
2474         }
2475         else if (!prefixcmp(arg, "-C")) {
2476                 if (options->detect_rename == DIFF_DETECT_COPY)
2477                         DIFF_OPT_SET(options, FIND_COPIES_HARDER);
2478                 if ((options->rename_score = diff_scoreopt_parse(arg)) == -1)
2479                         return -1;
2480                 options->detect_rename = DIFF_DETECT_COPY;
2481         }
2482         else if (!strcmp(arg, "--no-renames"))
2483                 options->detect_rename = 0;
2484         else if (!strcmp(arg, "--relative"))
2485                 DIFF_OPT_SET(options, RELATIVE_NAME);
2486         else if (!prefixcmp(arg, "--relative=")) {
2487                 DIFF_OPT_SET(options, RELATIVE_NAME);
2488                 options->prefix = arg + 11;
2489         }
2490
2491         /* xdiff options */
2492         else if (!strcmp(arg, "-w") || !strcmp(arg, "--ignore-all-space"))
2493                 options->xdl_opts |= XDF_IGNORE_WHITESPACE;
2494         else if (!strcmp(arg, "-b") || !strcmp(arg, "--ignore-space-change"))
2495                 options->xdl_opts |= XDF_IGNORE_WHITESPACE_CHANGE;
2496         else if (!strcmp(arg, "--ignore-space-at-eol"))
2497                 options->xdl_opts |= XDF_IGNORE_WHITESPACE_AT_EOL;
2498
2499         /* flags options */
2500         else if (!strcmp(arg, "--binary")) {
2501                 options->output_format |= DIFF_FORMAT_PATCH;
2502                 DIFF_OPT_SET(options, BINARY);
2503         }
2504         else if (!strcmp(arg, "--full-index"))
2505                 DIFF_OPT_SET(options, FULL_INDEX);
2506         else if (!strcmp(arg, "-a") || !strcmp(arg, "--text"))
2507                 DIFF_OPT_SET(options, TEXT);
2508         else if (!strcmp(arg, "-R"))
2509                 DIFF_OPT_SET(options, REVERSE_DIFF);
2510         else if (!strcmp(arg, "--find-copies-harder"))
2511                 DIFF_OPT_SET(options, FIND_COPIES_HARDER);
2512         else if (!strcmp(arg, "--follow"))
2513                 DIFF_OPT_SET(options, FOLLOW_RENAMES);
2514         else if (!strcmp(arg, "--color"))
2515                 DIFF_OPT_SET(options, COLOR_DIFF);
2516         else if (!strcmp(arg, "--no-color"))
2517                 DIFF_OPT_CLR(options, COLOR_DIFF);
2518         else if (!strcmp(arg, "--color-words"))
2519                 options->flags |= DIFF_OPT_COLOR_DIFF | DIFF_OPT_COLOR_DIFF_WORDS;
2520         else if (!strcmp(arg, "--exit-code"))
2521                 DIFF_OPT_SET(options, EXIT_WITH_STATUS);
2522         else if (!strcmp(arg, "--quiet"))
2523                 DIFF_OPT_SET(options, QUIET);
2524         else if (!strcmp(arg, "--ext-diff"))
2525                 DIFF_OPT_SET(options, ALLOW_EXTERNAL);
2526         else if (!strcmp(arg, "--no-ext-diff"))
2527                 DIFF_OPT_CLR(options, ALLOW_EXTERNAL);
2528         else if (!strcmp(arg, "--textconv"))
2529                 DIFF_OPT_SET(options, ALLOW_TEXTCONV);
2530         else if (!strcmp(arg, "--no-textconv"))
2531                 DIFF_OPT_CLR(options, ALLOW_TEXTCONV);
2532         else if (!strcmp(arg, "--ignore-submodules"))
2533                 DIFF_OPT_SET(options, IGNORE_SUBMODULES);
2534
2535         /* misc options */
2536         else if (!strcmp(arg, "-z"))
2537                 options->line_termination = 0;
2538         else if (!prefixcmp(arg, "-l"))
2539                 options->rename_limit = strtoul(arg+2, NULL, 10);
2540         else if (!prefixcmp(arg, "-S"))
2541                 options->pickaxe = arg + 2;
2542         else if (!strcmp(arg, "--pickaxe-all"))
2543                 options->pickaxe_opts = DIFF_PICKAXE_ALL;
2544         else if (!strcmp(arg, "--pickaxe-regex"))
2545                 options->pickaxe_opts = DIFF_PICKAXE_REGEX;
2546         else if (!prefixcmp(arg, "-O"))
2547                 options->orderfile = arg + 2;
2548         else if (!prefixcmp(arg, "--diff-filter="))
2549                 options->filter = arg + 14;
2550         else if (!strcmp(arg, "--abbrev"))
2551                 options->abbrev = DEFAULT_ABBREV;
2552         else if (!prefixcmp(arg, "--abbrev=")) {
2553                 options->abbrev = strtoul(arg + 9, NULL, 10);
2554                 if (options->abbrev < MINIMUM_ABBREV)
2555                         options->abbrev = MINIMUM_ABBREV;
2556                 else if (40 < options->abbrev)
2557                         options->abbrev = 40;
2558         }
2559         else if (!prefixcmp(arg, "--src-prefix="))
2560                 options->a_prefix = arg + 13;
2561         else if (!prefixcmp(arg, "--dst-prefix="))
2562                 options->b_prefix = arg + 13;
2563         else if (!strcmp(arg, "--no-prefix"))
2564                 options->a_prefix = options->b_prefix = "";
2565         else if (!prefixcmp(arg, "--output=")) {
2566                 options->file = fopen(arg + strlen("--output="), "w");
2567                 options->close_file = 1;
2568         } else
2569                 return 0;
2570         return 1;
2571 }
2572
2573 static int parse_num(const char **cp_p)
2574 {
2575         unsigned long num, scale;
2576         int ch, dot;
2577         const char *cp = *cp_p;
2578
2579         num = 0;
2580         scale = 1;
2581         dot = 0;
2582         for(;;) {
2583                 ch = *cp;
2584                 if ( !dot && ch == '.' ) {
2585                         scale = 1;
2586                         dot = 1;
2587                 } else if ( ch == '%' ) {
2588                         scale = dot ? scale*100 : 100;
2589                         cp++;   /* % is always at the end */
2590                         break;
2591                 } else if ( ch >= '0' && ch <= '9' ) {
2592                         if ( scale < 100000 ) {
2593                                 scale *= 10;
2594                                 num = (num*10) + (ch-'0');
2595                         }
2596                 } else {
2597                         break;
2598                 }
2599                 cp++;
2600         }
2601         *cp_p = cp;
2602
2603         /* user says num divided by scale and we say internally that
2604          * is MAX_SCORE * num / scale.
2605          */
2606         return (int)((num >= scale) ? MAX_SCORE : (MAX_SCORE * num / scale));
2607 }
2608
2609 static int diff_scoreopt_parse(const char *opt)
2610 {
2611         int opt1, opt2, cmd;
2612
2613         if (*opt++ != '-')
2614                 return -1;
2615         cmd = *opt++;
2616         if (cmd != 'M' && cmd != 'C' && cmd != 'B')
2617                 return -1; /* that is not a -M, -C nor -B option */
2618
2619         opt1 = parse_num(&opt);
2620         if (cmd != 'B')
2621                 opt2 = 0;
2622         else {
2623                 if (*opt == 0)
2624                         opt2 = 0;
2625                 else if (*opt != '/')
2626                         return -1; /* we expect -B80/99 or -B80 */
2627                 else {
2628                         opt++;
2629                         opt2 = parse_num(&opt);
2630                 }
2631         }
2632         if (*opt != 0)
2633                 return -1;
2634         return opt1 | (opt2 << 16);
2635 }
2636
2637 struct diff_queue_struct diff_queued_diff;
2638
2639 void diff_q(struct diff_queue_struct *queue, struct diff_filepair *dp)
2640 {
2641         if (queue->alloc <= queue->nr) {
2642                 queue->alloc = alloc_nr(queue->alloc);
2643                 queue->queue = xrealloc(queue->queue,
2644                                         sizeof(dp) * queue->alloc);
2645         }
2646         queue->queue[queue->nr++] = dp;
2647 }
2648
2649 struct diff_filepair *diff_queue(struct diff_queue_struct *queue,
2650                                  struct diff_filespec *one,
2651                                  struct diff_filespec *two)
2652 {
2653         struct diff_filepair *dp = xcalloc(1, sizeof(*dp));
2654         dp->one = one;
2655         dp->two = two;
2656         if (queue)
2657                 diff_q(queue, dp);
2658         return dp;
2659 }
2660
2661 void diff_free_filepair(struct diff_filepair *p)
2662 {
2663         free_filespec(p->one);
2664         free_filespec(p->two);
2665         free(p);
2666 }
2667
2668 /* This is different from find_unique_abbrev() in that
2669  * it stuffs the result with dots for alignment.
2670  */
2671 const char *diff_unique_abbrev(const unsigned char *sha1, int len)
2672 {
2673         int abblen;
2674         const char *abbrev;
2675         if (len == 40)
2676                 return sha1_to_hex(sha1);
2677
2678         abbrev = find_unique_abbrev(sha1, len);
2679         abblen = strlen(abbrev);
2680         if (abblen < 37) {
2681                 static char hex[41];
2682                 if (len < abblen && abblen <= len + 2)
2683                         sprintf(hex, "%s%.*s", abbrev, len+3-abblen, "..");
2684                 else
2685                         sprintf(hex, "%s...", abbrev);
2686                 return hex;
2687         }
2688         return sha1_to_hex(sha1);
2689 }
2690
2691 static void diff_flush_raw(struct diff_filepair *p, struct diff_options *opt)
2692 {
2693         int line_termination = opt->line_termination;
2694         int inter_name_termination = line_termination ? '\t' : '\0';
2695
2696         if (!(opt->output_format & DIFF_FORMAT_NAME_STATUS)) {
2697                 fprintf(opt->file, ":%06o %06o %s ", p->one->mode, p->two->mode,
2698                         diff_unique_abbrev(p->one->sha1, opt->abbrev));
2699                 fprintf(opt->file, "%s ", diff_unique_abbrev(p->two->sha1, opt->abbrev));
2700         }
2701         if (p->score) {
2702                 fprintf(opt->file, "%c%03d%c", p->status, similarity_index(p),
2703                         inter_name_termination);
2704         } else {
2705                 fprintf(opt->file, "%c%c", p->status, inter_name_termination);
2706         }
2707
2708         if (p->status == DIFF_STATUS_COPIED ||
2709             p->status == DIFF_STATUS_RENAMED) {
2710                 const char *name_a, *name_b;
2711                 name_a = p->one->path;
2712                 name_b = p->two->path;
2713                 strip_prefix(opt->prefix_length, &name_a, &name_b);
2714                 write_name_quoted(name_a, opt->file, inter_name_termination);
2715                 write_name_quoted(name_b, opt->file, line_termination);
2716         } else {
2717                 const char *name_a, *name_b;
2718                 name_a = p->one->mode ? p->one->path : p->two->path;
2719                 name_b = NULL;
2720                 strip_prefix(opt->prefix_length, &name_a, &name_b);
2721                 write_name_quoted(name_a, opt->file, line_termination);
2722         }
2723 }
2724
2725 int diff_unmodified_pair(struct diff_filepair *p)
2726 {
2727         /* This function is written stricter than necessary to support
2728          * the currently implemented transformers, but the idea is to
2729          * let transformers to produce diff_filepairs any way they want,
2730          * and filter and clean them up here before producing the output.
2731          */
2732         struct diff_filespec *one = p->one, *two = p->two;
2733
2734         if (DIFF_PAIR_UNMERGED(p))
2735                 return 0; /* unmerged is interesting */
2736
2737         /* deletion, addition, mode or type change
2738          * and rename are all interesting.
2739          */
2740         if (DIFF_FILE_VALID(one) != DIFF_FILE_VALID(two) ||
2741             DIFF_PAIR_MODE_CHANGED(p) ||
2742             strcmp(one->path, two->path))
2743                 return 0;
2744
2745         /* both are valid and point at the same path.  that is, we are
2746          * dealing with a change.
2747          */
2748         if (one->sha1_valid && two->sha1_valid &&
2749             !hashcmp(one->sha1, two->sha1))
2750                 return 1; /* no change */
2751         if (!one->sha1_valid && !two->sha1_valid)
2752                 return 1; /* both look at the same file on the filesystem. */
2753         return 0;
2754 }
2755
2756 static void diff_flush_patch(struct diff_filepair *p, struct diff_options *o)
2757 {
2758         if (diff_unmodified_pair(p))
2759                 return;
2760
2761         if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
2762             (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
2763                 return; /* no tree diffs in patch format */
2764
2765         run_diff(p, o);
2766 }
2767
2768 static void diff_flush_stat(struct diff_filepair *p, struct diff_options *o,
2769                             struct diffstat_t *diffstat)
2770 {
2771         if (diff_unmodified_pair(p))
2772                 return;
2773
2774         if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
2775             (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
2776                 return; /* no tree diffs in patch format */
2777
2778         run_diffstat(p, o, diffstat);
2779 }
2780
2781 static void diff_flush_checkdiff(struct diff_filepair *p,
2782                 struct diff_options *o)
2783 {
2784         if (diff_unmodified_pair(p))
2785                 return;
2786
2787         if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
2788             (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
2789                 return; /* no tree diffs in patch format */
2790
2791         run_checkdiff(p, o);
2792 }
2793
2794 int diff_queue_is_empty(void)
2795 {
2796         struct diff_queue_struct *q = &diff_queued_diff;
2797         int i;
2798         for (i = 0; i < q->nr; i++)
2799                 if (!diff_unmodified_pair(q->queue[i]))
2800                         return 0;
2801         return 1;
2802 }
2803
2804 #if DIFF_DEBUG
2805 void diff_debug_filespec(struct diff_filespec *s, int x, const char *one)
2806 {
2807         fprintf(stderr, "queue[%d] %s (%s) %s %06o %s\n",
2808                 x, one ? one : "",
2809                 s->path,
2810                 DIFF_FILE_VALID(s) ? "valid" : "invalid",
2811                 s->mode,
2812                 s->sha1_valid ? sha1_to_hex(s->sha1) : "");
2813         fprintf(stderr, "queue[%d] %s size %lu flags %d\n",
2814                 x, one ? one : "",
2815                 s->size, s->xfrm_flags);
2816 }
2817
2818 void diff_debug_filepair(const struct diff_filepair *p, int i)
2819 {
2820         diff_debug_filespec(p->one, i, "one");
2821         diff_debug_filespec(p->two, i, "two");
2822         fprintf(stderr, "score %d, status %c rename_used %d broken %d\n",
2823                 p->score, p->status ? p->status : '?',
2824                 p->one->rename_used, p->broken_pair);
2825 }
2826
2827 void diff_debug_queue(const char *msg, struct diff_queue_struct *q)
2828 {
2829         int i;
2830         if (msg)
2831                 fprintf(stderr, "%s\n", msg);
2832         fprintf(stderr, "q->nr = %d\n", q->nr);
2833         for (i = 0; i < q->nr; i++) {
2834                 struct diff_filepair *p = q->queue[i];
2835                 diff_debug_filepair(p, i);
2836         }
2837 }
2838 #endif
2839
2840 static void diff_resolve_rename_copy(void)
2841 {
2842         int i;
2843         struct diff_filepair *p;
2844         struct diff_queue_struct *q = &diff_queued_diff;
2845
2846         diff_debug_queue("resolve-rename-copy", q);
2847
2848         for (i = 0; i < q->nr; i++) {
2849                 p = q->queue[i];
2850                 p->status = 0; /* undecided */
2851                 if (DIFF_PAIR_UNMERGED(p))
2852                         p->status = DIFF_STATUS_UNMERGED;
2853                 else if (!DIFF_FILE_VALID(p->one))
2854                         p->status = DIFF_STATUS_ADDED;
2855                 else if (!DIFF_FILE_VALID(p->two))
2856                         p->status = DIFF_STATUS_DELETED;
2857                 else if (DIFF_PAIR_TYPE_CHANGED(p))
2858                         p->status = DIFF_STATUS_TYPE_CHANGED;
2859
2860                 /* from this point on, we are dealing with a pair
2861                  * whose both sides are valid and of the same type, i.e.
2862                  * either in-place edit or rename/copy edit.
2863                  */
2864                 else if (DIFF_PAIR_RENAME(p)) {
2865                         /*
2866                          * A rename might have re-connected a broken
2867                          * pair up, causing the pathnames to be the
2868                          * same again. If so, that's not a rename at
2869                          * all, just a modification..
2870                          *
2871                          * Otherwise, see if this source was used for
2872                          * multiple renames, in which case we decrement
2873                          * the count, and call it a copy.
2874                          */
2875                         if (!strcmp(p->one->path, p->two->path))
2876                                 p->status = DIFF_STATUS_MODIFIED;
2877                         else if (--p->one->rename_used > 0)
2878                                 p->status = DIFF_STATUS_COPIED;
2879                         else
2880                                 p->status = DIFF_STATUS_RENAMED;
2881                 }
2882                 else if (hashcmp(p->one->sha1, p->two->sha1) ||
2883                          p->one->mode != p->two->mode ||
2884                          is_null_sha1(p->one->sha1))
2885                         p->status = DIFF_STATUS_MODIFIED;
2886                 else {
2887                         /* This is a "no-change" entry and should not
2888                          * happen anymore, but prepare for broken callers.
2889                          */
2890                         error("feeding unmodified %s to diffcore",
2891                               p->one->path);
2892                         p->status = DIFF_STATUS_UNKNOWN;
2893                 }
2894         }
2895         diff_debug_queue("resolve-rename-copy done", q);
2896 }
2897
2898 static int check_pair_status(struct diff_filepair *p)
2899 {
2900         switch (p->status) {
2901         case DIFF_STATUS_UNKNOWN:
2902                 return 0;
2903         case 0:
2904                 die("internal error in diff-resolve-rename-copy");
2905         default:
2906                 return 1;
2907         }
2908 }
2909
2910 static void flush_one_pair(struct diff_filepair *p, struct diff_options *opt)
2911 {
2912         int fmt = opt->output_format;
2913
2914         if (fmt & DIFF_FORMAT_CHECKDIFF)
2915                 diff_flush_checkdiff(p, opt);
2916         else if (fmt & (DIFF_FORMAT_RAW | DIFF_FORMAT_NAME_STATUS))
2917                 diff_flush_raw(p, opt);
2918         else if (fmt & DIFF_FORMAT_NAME) {
2919                 const char *name_a, *name_b;
2920                 name_a = p->two->path;
2921                 name_b = NULL;
2922                 strip_prefix(opt->prefix_length, &name_a, &name_b);
2923                 write_name_quoted(name_a, opt->file, opt->line_termination);
2924         }
2925 }
2926
2927 static void show_file_mode_name(FILE *file, const char *newdelete, struct diff_filespec *fs)
2928 {
2929         if (fs->mode)
2930                 fprintf(file, " %s mode %06o ", newdelete, fs->mode);
2931         else
2932                 fprintf(file, " %s ", newdelete);
2933         write_name_quoted(fs->path, file, '\n');
2934 }
2935
2936
2937 static void show_mode_change(FILE *file, struct diff_filepair *p, int show_name)
2938 {
2939         if (p->one->mode && p->two->mode && p->one->mode != p->two->mode) {
2940                 fprintf(file, " mode change %06o => %06o%c", p->one->mode, p->two->mode,
2941                         show_name ? ' ' : '\n');
2942                 if (show_name) {
2943                         write_name_quoted(p->two->path, file, '\n');
2944                 }
2945         }
2946 }
2947
2948 static void show_rename_copy(FILE *file, const char *renamecopy, struct diff_filepair *p)
2949 {
2950         char *names = pprint_rename(p->one->path, p->two->path);
2951
2952         fprintf(file, " %s %s (%d%%)\n", renamecopy, names, similarity_index(p));
2953         free(names);
2954         show_mode_change(file, p, 0);
2955 }
2956
2957 static void diff_summary(FILE *file, struct diff_filepair *p)
2958 {
2959         switch(p->status) {
2960         case DIFF_STATUS_DELETED:
2961                 show_file_mode_name(file, "delete", p->one);
2962                 break;
2963         case DIFF_STATUS_ADDED:
2964                 show_file_mode_name(file, "create", p->two);
2965                 break;
2966         case DIFF_STATUS_COPIED:
2967                 show_rename_copy(file, "copy", p);
2968                 break;
2969         case DIFF_STATUS_RENAMED:
2970                 show_rename_copy(file, "rename", p);
2971                 break;
2972         default:
2973                 if (p->score) {
2974                         fputs(" rewrite ", file);
2975                         write_name_quoted(p->two->path, file, ' ');
2976                         fprintf(file, "(%d%%)\n", similarity_index(p));
2977                 }
2978                 show_mode_change(file, p, !p->score);
2979                 break;
2980         }
2981 }
2982
2983 struct patch_id_t {
2984         git_SHA_CTX *ctx;
2985         int patchlen;
2986 };
2987
2988 static int remove_space(char *line, int len)
2989 {
2990         int i;
2991         char *dst = line;
2992         unsigned char c;
2993
2994         for (i = 0; i < len; i++)
2995                 if (!isspace((c = line[i])))
2996                         *dst++ = c;
2997
2998         return dst - line;
2999 }
3000
3001 static void patch_id_consume(void *priv, char *line, unsigned long len)
3002 {
3003         struct patch_id_t *data = priv;
3004         int new_len;
3005
3006         /* Ignore line numbers when computing the SHA1 of the patch */
3007         if (!prefixcmp(line, "@@ -"))
3008                 return;
3009
3010         new_len = remove_space(line, len);
3011
3012         git_SHA1_Update(data->ctx, line, new_len);
3013         data->patchlen += new_len;
3014 }
3015
3016 /* returns 0 upon success, and writes result into sha1 */
3017 static int diff_get_patch_id(struct diff_options *options, unsigned char *sha1)
3018 {
3019         struct diff_queue_struct *q = &diff_queued_diff;
3020         int i;
3021         git_SHA_CTX ctx;
3022         struct patch_id_t data;
3023         char buffer[PATH_MAX * 4 + 20];
3024
3025         git_SHA1_Init(&ctx);
3026         memset(&data, 0, sizeof(struct patch_id_t));
3027         data.ctx = &ctx;
3028
3029         for (i = 0; i < q->nr; i++) {
3030                 xpparam_t xpp;
3031                 xdemitconf_t xecfg;
3032                 xdemitcb_t ecb;
3033                 mmfile_t mf1, mf2;
3034                 struct diff_filepair *p = q->queue[i];
3035                 int len1, len2;
3036
3037                 memset(&xpp, 0, sizeof(xpp));
3038                 memset(&xecfg, 0, sizeof(xecfg));
3039                 if (p->status == 0)
3040                         return error("internal diff status error");
3041                 if (p->status == DIFF_STATUS_UNKNOWN)
3042                         continue;
3043                 if (diff_unmodified_pair(p))
3044                         continue;
3045                 if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
3046                     (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
3047                         continue;
3048                 if (DIFF_PAIR_UNMERGED(p))
3049                         continue;
3050
3051                 diff_fill_sha1_info(p->one);
3052                 diff_fill_sha1_info(p->two);
3053                 if (fill_mmfile(&mf1, p->one) < 0 ||
3054                                 fill_mmfile(&mf2, p->two) < 0)
3055                         return error("unable to read files to diff");
3056
3057                 len1 = remove_space(p->one->path, strlen(p->one->path));
3058                 len2 = remove_space(p->two->path, strlen(p->two->path));
3059                 if (p->one->mode == 0)
3060                         len1 = snprintf(buffer, sizeof(buffer),
3061                                         "diff--gita/%.*sb/%.*s"
3062                                         "newfilemode%06o"
3063                                         "---/dev/null"
3064                                         "+++b/%.*s",
3065                                         len1, p->one->path,
3066                                         len2, p->two->path,
3067                                         p->two->mode,
3068                                         len2, p->two->path);
3069                 else if (p->two->mode == 0)
3070                         len1 = snprintf(buffer, sizeof(buffer),
3071                                         "diff--gita/%.*sb/%.*s"
3072                                         "deletedfilemode%06o"
3073                                         "---a/%.*s"
3074                                         "+++/dev/null",
3075                                         len1, p->one->path,
3076                                         len2, p->two->path,
3077                                         p->one->mode,
3078                                         len1, p->one->path);
3079                 else
3080                         len1 = snprintf(buffer, sizeof(buffer),
3081                                         "diff--gita/%.*sb/%.*s"
3082                                         "---a/%.*s"
3083                                         "+++b/%.*s",
3084                                         len1, p->one->path,
3085                                         len2, p->two->path,
3086                                         len1, p->one->path,
3087                                         len2, p->two->path);
3088                 git_SHA1_Update(&ctx, buffer, len1);
3089
3090                 xpp.flags = XDF_NEED_MINIMAL;
3091                 xecfg.ctxlen = 3;
3092                 xecfg.flags = XDL_EMIT_FUNCNAMES;
3093                 xdi_diff_outf(&mf1, &mf2, patch_id_consume, &data,
3094                               &xpp, &xecfg, &ecb);
3095         }
3096
3097         git_SHA1_Final(sha1, &ctx);
3098         return 0;
3099 }
3100
3101 int diff_flush_patch_id(struct diff_options *options, unsigned char *sha1)
3102 {
3103         struct diff_queue_struct *q = &diff_queued_diff;
3104         int i;
3105         int result = diff_get_patch_id(options, sha1);
3106
3107         for (i = 0; i < q->nr; i++)
3108                 diff_free_filepair(q->queue[i]);
3109
3110         free(q->queue);
3111         q->queue = NULL;
3112         q->nr = q->alloc = 0;
3113
3114         return result;
3115 }
3116
3117 static int is_summary_empty(const struct diff_queue_struct *q)
3118 {
3119         int i;
3120
3121         for (i = 0; i < q->nr; i++) {
3122                 const struct diff_filepair *p = q->queue[i];
3123
3124                 switch (p->status) {
3125                 case DIFF_STATUS_DELETED:
3126                 case DIFF_STATUS_ADDED:
3127                 case DIFF_STATUS_COPIED:
3128                 case DIFF_STATUS_RENAMED:
3129                         return 0;
3130                 default:
3131                         if (p->score)
3132                                 return 0;
3133                         if (p->one->mode && p->two->mode &&
3134                             p->one->mode != p->two->mode)
3135                                 return 0;
3136                         break;
3137                 }
3138         }
3139         return 1;
3140 }
3141
3142 void diff_flush(struct diff_options *options)
3143 {
3144         struct diff_queue_struct *q = &diff_queued_diff;
3145         int i, output_format = options->output_format;
3146         int separator = 0;
3147
3148         /*
3149          * Order: raw, stat, summary, patch
3150          * or:    name/name-status/checkdiff (other bits clear)
3151          */
3152         if (!q->nr)
3153                 goto free_queue;
3154
3155         if (output_format & (DIFF_FORMAT_RAW |
3156                              DIFF_FORMAT_NAME |
3157                              DIFF_FORMAT_NAME_STATUS |
3158                              DIFF_FORMAT_CHECKDIFF)) {
3159                 for (i = 0; i < q->nr; i++) {
3160                         struct diff_filepair *p = q->queue[i];
3161                         if (check_pair_status(p))
3162                                 flush_one_pair(p, options);
3163                 }
3164                 separator++;
3165         }
3166
3167         if (output_format & (DIFF_FORMAT_DIFFSTAT|DIFF_FORMAT_SHORTSTAT|DIFF_FORMAT_NUMSTAT)) {
3168                 struct diffstat_t diffstat;
3169
3170                 memset(&diffstat, 0, sizeof(struct diffstat_t));
3171                 for (i = 0; i < q->nr; i++) {
3172                         struct diff_filepair *p = q->queue[i];
3173                         if (check_pair_status(p))
3174                                 diff_flush_stat(p, options, &diffstat);
3175                 }
3176                 if (output_format & DIFF_FORMAT_NUMSTAT)
3177                         show_numstat(&diffstat, options);
3178                 if (output_format & DIFF_FORMAT_DIFFSTAT)
3179                         show_stats(&diffstat, options);
3180                 if (output_format & DIFF_FORMAT_SHORTSTAT)
3181                         show_shortstats(&diffstat, options);
3182                 free_diffstat_info(&diffstat);
3183                 separator++;
3184         }
3185         if (output_format & DIFF_FORMAT_DIRSTAT)
3186                 show_dirstat(options);
3187
3188         if (output_format & DIFF_FORMAT_SUMMARY && !is_summary_empty(q)) {
3189                 for (i = 0; i < q->nr; i++)
3190                         diff_summary(options->file, q->queue[i]);
3191                 separator++;
3192         }
3193
3194         if (output_format & DIFF_FORMAT_PATCH) {
3195                 if (separator) {
3196                         putc(options->line_termination, options->file);
3197                         if (options->stat_sep) {
3198                                 /* attach patch instead of inline */
3199                                 fputs(options->stat_sep, options->file);
3200                         }
3201                 }
3202
3203                 for (i = 0; i < q->nr; i++) {
3204                         struct diff_filepair *p = q->queue[i];
3205                         if (check_pair_status(p))
3206                                 diff_flush_patch(p, options);
3207                 }
3208         }
3209
3210         if (output_format & DIFF_FORMAT_CALLBACK)
3211                 options->format_callback(q, options, options->format_callback_data);
3212
3213         for (i = 0; i < q->nr; i++)
3214                 diff_free_filepair(q->queue[i]);
3215 free_queue:
3216         free(q->queue);
3217         q->queue = NULL;
3218         q->nr = q->alloc = 0;
3219         if (options->close_file)
3220                 fclose(options->file);
3221 }
3222
3223 static void diffcore_apply_filter(const char *filter)
3224 {
3225         int i;
3226         struct diff_queue_struct *q = &diff_queued_diff;
3227         struct diff_queue_struct outq;
3228         outq.queue = NULL;
3229         outq.nr = outq.alloc = 0;
3230
3231         if (!filter)
3232                 return;
3233
3234         if (strchr(filter, DIFF_STATUS_FILTER_AON)) {
3235                 int found;
3236                 for (i = found = 0; !found && i < q->nr; i++) {
3237                         struct diff_filepair *p = q->queue[i];
3238                         if (((p->status == DIFF_STATUS_MODIFIED) &&
3239                              ((p->score &&
3240                                strchr(filter, DIFF_STATUS_FILTER_BROKEN)) ||
3241                               (!p->score &&
3242                                strchr(filter, DIFF_STATUS_MODIFIED)))) ||
3243                             ((p->status != DIFF_STATUS_MODIFIED) &&
3244                              strchr(filter, p->status)))
3245                                 found++;
3246                 }
3247                 if (found)
3248                         return;
3249
3250                 /* otherwise we will clear the whole queue
3251                  * by copying the empty outq at the end of this
3252                  * function, but first clear the current entries
3253                  * in the queue.
3254                  */
3255                 for (i = 0; i < q->nr; i++)
3256                         diff_free_filepair(q->queue[i]);
3257         }
3258         else {
3259                 /* Only the matching ones */
3260                 for (i = 0; i < q->nr; i++) {
3261                         struct diff_filepair *p = q->queue[i];
3262
3263                         if (((p->status == DIFF_STATUS_MODIFIED) &&
3264                              ((p->score &&
3265                                strchr(filter, DIFF_STATUS_FILTER_BROKEN)) ||
3266                               (!p->score &&
3267                                strchr(filter, DIFF_STATUS_MODIFIED)))) ||
3268                             ((p->status != DIFF_STATUS_MODIFIED) &&
3269                              strchr(filter, p->status)))
3270                                 diff_q(&outq, p);
3271                         else
3272                                 diff_free_filepair(p);
3273                 }
3274         }
3275         free(q->queue);
3276         *q = outq;
3277 }
3278
3279 /* Check whether two filespecs with the same mode and size are identical */
3280 static int diff_filespec_is_identical(struct diff_filespec *one,
3281                                       struct diff_filespec *two)
3282 {
3283         if (S_ISGITLINK(one->mode))
3284                 return 0;
3285         if (diff_populate_filespec(one, 0))
3286                 return 0;
3287         if (diff_populate_filespec(two, 0))
3288                 return 0;
3289         return !memcmp(one->data, two->data, one->size);
3290 }
3291
3292 static void diffcore_skip_stat_unmatch(struct diff_options *diffopt)
3293 {
3294         int i;
3295         struct diff_queue_struct *q = &diff_queued_diff;
3296         struct diff_queue_struct outq;
3297         outq.queue = NULL;
3298         outq.nr = outq.alloc = 0;
3299
3300         for (i = 0; i < q->nr; i++) {
3301                 struct diff_filepair *p = q->queue[i];
3302
3303                 /*
3304                  * 1. Entries that come from stat info dirtyness
3305                  *    always have both sides (iow, not create/delete),
3306                  *    one side of the object name is unknown, with
3307                  *    the same mode and size.  Keep the ones that
3308                  *    do not match these criteria.  They have real
3309                  *    differences.
3310                  *
3311                  * 2. At this point, the file is known to be modified,
3312                  *    with the same mode and size, and the object
3313                  *    name of one side is unknown.  Need to inspect
3314                  *    the identical contents.
3315                  */
3316                 if (!DIFF_FILE_VALID(p->one) || /* (1) */
3317                     !DIFF_FILE_VALID(p->two) ||
3318                     (p->one->sha1_valid && p->two->sha1_valid) ||
3319                     (p->one->mode != p->two->mode) ||
3320                     diff_populate_filespec(p->one, 1) ||
3321                     diff_populate_filespec(p->two, 1) ||
3322                     (p->one->size != p->two->size) ||
3323                     !diff_filespec_is_identical(p->one, p->two)) /* (2) */
3324                         diff_q(&outq, p);
3325                 else {
3326                         /*
3327                          * The caller can subtract 1 from skip_stat_unmatch
3328                          * to determine how many paths were dirty only
3329                          * due to stat info mismatch.
3330                          */
3331                         if (!DIFF_OPT_TST(diffopt, NO_INDEX))
3332                                 diffopt->skip_stat_unmatch++;
3333                         diff_free_filepair(p);
3334                 }
3335         }
3336         free(q->queue);
3337         *q = outq;
3338 }
3339
3340 void diffcore_std(struct diff_options *options)
3341 {
3342         if (options->skip_stat_unmatch)
3343                 diffcore_skip_stat_unmatch(options);
3344         if (options->break_opt != -1)
3345                 diffcore_break(options->break_opt);
3346         if (options->detect_rename)
3347                 diffcore_rename(options);
3348         if (options->break_opt != -1)
3349                 diffcore_merge_broken();
3350         if (options->pickaxe)
3351                 diffcore_pickaxe(options->pickaxe, options->pickaxe_opts);
3352         if (options->orderfile)
3353                 diffcore_order(options->orderfile);
3354         diff_resolve_rename_copy();
3355         diffcore_apply_filter(options->filter);
3356
3357         if (diff_queued_diff.nr)
3358                 DIFF_OPT_SET(options, HAS_CHANGES);
3359         else
3360                 DIFF_OPT_CLR(options, HAS_CHANGES);
3361 }
3362
3363 int diff_result_code(struct diff_options *opt, int status)
3364 {
3365         int result = 0;
3366         if (!DIFF_OPT_TST(opt, EXIT_WITH_STATUS) &&
3367             !(opt->output_format & DIFF_FORMAT_CHECKDIFF))
3368                 return status;
3369         if (DIFF_OPT_TST(opt, EXIT_WITH_STATUS) &&
3370             DIFF_OPT_TST(opt, HAS_CHANGES))
3371                 result |= 01;
3372         if ((opt->output_format & DIFF_FORMAT_CHECKDIFF) &&
3373             DIFF_OPT_TST(opt, CHECK_FAILED))
3374                 result |= 02;
3375         return result;
3376 }
3377
3378 void diff_addremove(struct diff_options *options,
3379                     int addremove, unsigned mode,
3380                     const unsigned char *sha1,
3381                     const char *concatpath)
3382 {
3383         struct diff_filespec *one, *two;
3384
3385         if (DIFF_OPT_TST(options, IGNORE_SUBMODULES) && S_ISGITLINK(mode))
3386                 return;
3387
3388         /* This may look odd, but it is a preparation for
3389          * feeding "there are unchanged files which should
3390          * not produce diffs, but when you are doing copy
3391          * detection you would need them, so here they are"
3392          * entries to the diff-core.  They will be prefixed
3393          * with something like '=' or '*' (I haven't decided
3394          * which but should not make any difference).
3395          * Feeding the same new and old to diff_change()
3396          * also has the same effect.
3397          * Before the final output happens, they are pruned after
3398          * merged into rename/copy pairs as appropriate.
3399          */
3400         if (DIFF_OPT_TST(options, REVERSE_DIFF))
3401                 addremove = (addremove == '+' ? '-' :
3402                              addremove == '-' ? '+' : addremove);
3403
3404         if (options->prefix &&
3405             strncmp(concatpath, options->prefix, options->prefix_length))
3406                 return;
3407
3408         one = alloc_filespec(concatpath);
3409         two = alloc_filespec(concatpath);
3410
3411         if (addremove != '+')
3412                 fill_filespec(one, sha1, mode);
3413         if (addremove != '-')
3414                 fill_filespec(two, sha1, mode);
3415
3416         diff_queue(&diff_queued_diff, one, two);
3417         DIFF_OPT_SET(options, HAS_CHANGES);
3418 }
3419
3420 void diff_change(struct diff_options *options,
3421                  unsigned old_mode, unsigned new_mode,
3422                  const unsigned char *old_sha1,
3423                  const unsigned char *new_sha1,
3424                  const char *concatpath)
3425 {
3426         struct diff_filespec *one, *two;
3427
3428         if (DIFF_OPT_TST(options, IGNORE_SUBMODULES) && S_ISGITLINK(old_mode)
3429                         && S_ISGITLINK(new_mode))
3430                 return;
3431
3432         if (DIFF_OPT_TST(options, REVERSE_DIFF)) {
3433                 unsigned tmp;
3434                 const unsigned char *tmp_c;
3435                 tmp = old_mode; old_mode = new_mode; new_mode = tmp;
3436                 tmp_c = old_sha1; old_sha1 = new_sha1; new_sha1 = tmp_c;
3437         }
3438
3439         if (options->prefix &&
3440             strncmp(concatpath, options->prefix, options->prefix_length))
3441                 return;
3442
3443         one = alloc_filespec(concatpath);
3444         two = alloc_filespec(concatpath);
3445         fill_filespec(one, old_sha1, old_mode);
3446         fill_filespec(two, new_sha1, new_mode);
3447
3448         diff_queue(&diff_queued_diff, one, two);
3449         DIFF_OPT_SET(options, HAS_CHANGES);
3450 }
3451
3452 void diff_unmerge(struct diff_options *options,
3453                   const char *path,
3454                   unsigned mode, const unsigned char *sha1)
3455 {
3456         struct diff_filespec *one, *two;
3457
3458         if (options->prefix &&
3459             strncmp(path, options->prefix, options->prefix_length))
3460                 return;
3461
3462         one = alloc_filespec(path);
3463         two = alloc_filespec(path);
3464         fill_filespec(one, sha1, mode);
3465         diff_queue(&diff_queued_diff, one, two)->is_unmerged = 1;
3466 }
3467
3468 static char *run_textconv(const char *pgm, struct diff_filespec *spec,
3469                 size_t *outsize)
3470 {
3471         struct diff_tempfile temp;
3472         const char *argv[3];
3473         const char **arg = argv;
3474         struct child_process child;
3475         struct strbuf buf = STRBUF_INIT;
3476
3477         prepare_temp_file(spec->path, &temp, spec);
3478         *arg++ = pgm;
3479         *arg++ = temp.name;
3480         *arg = NULL;
3481
3482         memset(&child, 0, sizeof(child));
3483         child.argv = argv;
3484         child.out = -1;
3485         if (start_command(&child) != 0 ||
3486             strbuf_read(&buf, child.out, 0) < 0 ||
3487             finish_command(&child) != 0) {
3488                 if (temp.name == temp.tmp_path)
3489                         unlink(temp.name);
3490                 error("error running textconv command '%s'", pgm);
3491                 return NULL;
3492         }
3493         if (temp.name == temp.tmp_path)
3494                 unlink(temp.name);
3495
3496         return strbuf_detach(&buf, outsize);
3497 }