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