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

   1 /*
   2  * Copyright (C) 2005 Junio C Hamano
   3  */
   4 #include "cache.h"
   5 #include "config.h"
   6 #include "tempfile.h"
   7 #include "quote.h"
   8 #include "diff.h"
   9 #include "diffcore.h"
  10 #include "delta.h"
  11 #include "xdiff-interface.h"
  12 #include "color.h"
  13 #include "attr.h"
  14 #include "run-command.h"
  15 #include "utf8.h"
  16 #include "userdiff.h"
  17 #include "submodule-config.h"
  18 #include "submodule.h"
  19 #include "hashmap.h"
  20 #include "ll-merge.h"
  21 #include "string-list.h"
  22 #include "argv-array.h"
  23 #include "graph.h"
  24 #include "packfile.h"
  25
  26 #ifdef NO_FAST_WORKING_DIRECTORY
  27 #define FAST_WORKING_DIRECTORY 0
  28 #else
  29 #define FAST_WORKING_DIRECTORY 1
  30 #endif
  31
  32 static int diff_detect_rename_default;
  33 static int diff_indent_heuristic = 1;
  34 static int diff_rename_limit_default = 400;
  35 static int diff_suppress_blank_empty;
  36 static int diff_use_color_default = -1;
  37 static int diff_color_moved_default;
  38 static int diff_context_default = 3;
  39 static int diff_interhunk_context_default;
  40 static const char *diff_word_regex_cfg;
  41 static const char *external_diff_cmd_cfg;
  42 static const char *diff_order_file_cfg;
  43 int diff_auto_refresh_index = 1;
  44 static int diff_mnemonic_prefix;
  45 static int diff_no_prefix;
  46 static int diff_stat_graph_width;
  47 static int diff_dirstat_permille_default = 30;
  48 static struct diff_options default_diff_options;
  49 static long diff_algorithm;
  50 static unsigned ws_error_highlight_default = WSEH_NEW;
  51
  52 static char diff_colors[][COLOR_MAXLEN] = {
  53         GIT_COLOR_RESET,
  54         GIT_COLOR_NORMAL,       /* CONTEXT */
  55         GIT_COLOR_BOLD,         /* METAINFO */
  56         GIT_COLOR_CYAN,         /* FRAGINFO */
  57         GIT_COLOR_RED,          /* OLD */
  58         GIT_COLOR_GREEN,        /* NEW */
  59         GIT_COLOR_YELLOW,       /* COMMIT */
  60         GIT_COLOR_BG_RED,       /* WHITESPACE */
  61         GIT_COLOR_NORMAL,       /* FUNCINFO */
  62         GIT_COLOR_BOLD_MAGENTA, /* OLD_MOVED */
  63         GIT_COLOR_BOLD_BLUE,    /* OLD_MOVED ALTERNATIVE */
  64         GIT_COLOR_FAINT,        /* OLD_MOVED_DIM */
  65         GIT_COLOR_FAINT_ITALIC, /* OLD_MOVED_ALTERNATIVE_DIM */
  66         GIT_COLOR_BOLD_CYAN,    /* NEW_MOVED */
  67         GIT_COLOR_BOLD_YELLOW,  /* NEW_MOVED ALTERNATIVE */
  68         GIT_COLOR_FAINT,        /* NEW_MOVED_DIM */
  69         GIT_COLOR_FAINT_ITALIC, /* NEW_MOVED_ALTERNATIVE_DIM */
  70 };
  71
  72 static NORETURN void die_want_option(const char *option_name)
  73 {
  74         die(_("option '%s' requires a value"), option_name);
  75 }
  76
  77 static int parse_diff_color_slot(const char *var)
  78 {
  79         if (!strcasecmp(var, "context") || !strcasecmp(var, "plain"))
  80                 return DIFF_CONTEXT;
  81         if (!strcasecmp(var, "meta"))
  82                 return DIFF_METAINFO;
  83         if (!strcasecmp(var, "frag"))
  84                 return DIFF_FRAGINFO;
  85         if (!strcasecmp(var, "old"))
  86                 return DIFF_FILE_OLD;
  87         if (!strcasecmp(var, "new"))
  88                 return DIFF_FILE_NEW;
  89         if (!strcasecmp(var, "commit"))
  90                 return DIFF_COMMIT;
  91         if (!strcasecmp(var, "whitespace"))
  92                 return DIFF_WHITESPACE;
  93         if (!strcasecmp(var, "func"))
  94                 return DIFF_FUNCINFO;
  95         if (!strcasecmp(var, "oldmoved"))
  96                 return DIFF_FILE_OLD_MOVED;
  97         if (!strcasecmp(var, "oldmovedalternative"))
  98                 return DIFF_FILE_OLD_MOVED_ALT;
  99         if (!strcasecmp(var, "oldmoveddimmed"))
 100                 return DIFF_FILE_OLD_MOVED_DIM;
 101         if (!strcasecmp(var, "oldmovedalternativedimmed"))
 102                 return DIFF_FILE_OLD_MOVED_ALT_DIM;
 103         if (!strcasecmp(var, "newmoved"))
 104                 return DIFF_FILE_NEW_MOVED;
 105         if (!strcasecmp(var, "newmovedalternative"))
 106                 return DIFF_FILE_NEW_MOVED_ALT;
 107         if (!strcasecmp(var, "newmoveddimmed"))
 108                 return DIFF_FILE_NEW_MOVED_DIM;
 109         if (!strcasecmp(var, "newmovedalternativedimmed"))
 110                 return DIFF_FILE_NEW_MOVED_ALT_DIM;
 111         return -1;
 112 }
 113
 114 static int parse_dirstat_params(struct diff_options *options, const char *params_string,
 115                                 struct strbuf *errmsg)
 116 {
 117         char *params_copy = xstrdup(params_string);
 118         struct string_list params = STRING_LIST_INIT_NODUP;
 119         int ret = 0;
 120         int i;
 121
 122         if (*params_copy)
 123                 string_list_split_in_place(&params, params_copy, ',', -1);
 124         for (i = 0; i < params.nr; i++) {
 125                 const char *p = params.items[i].string;
 126                 if (!strcmp(p, "changes")) {
 127                         options->flags.dirstat_by_line = 0;
 128                         options->flags.dirstat_by_file = 0;
 129                 } else if (!strcmp(p, "lines")) {
 130                         options->flags.dirstat_by_line = 1;
 131                         options->flags.dirstat_by_file = 0;
 132                 } else if (!strcmp(p, "files")) {
 133                         options->flags.dirstat_by_line = 0;
 134                         options->flags.dirstat_by_file = 1;
 135                 } else if (!strcmp(p, "noncumulative")) {
 136                         options->flags.dirstat_cumulative = 0;
 137                 } else if (!strcmp(p, "cumulative")) {
 138                         options->flags.dirstat_cumulative = 1;
 139                 } else if (isdigit(*p)) {
 140                         char *end;
 141                         int permille = strtoul(p, &end, 10) * 10;
 142                         if (*end == '.' && isdigit(*++end)) {
 143                                 /* only use first digit */
 144                                 permille += *end - '0';
 145                                 /* .. and ignore any further digits */
 146                                 while (isdigit(*++end))
 147                                         ; /* nothing */
 148                         }
 149                         if (!*end)
 150                                 options->dirstat_permille = permille;
 151                         else {
 152                                 strbuf_addf(errmsg, _("  Failed to parse dirstat cut-off percentage '%s'\n"),
 153                                             p);
 154                                 ret++;
 155                         }
 156                 } else {
 157                         strbuf_addf(errmsg, _("  Unknown dirstat parameter '%s'\n"), p);
 158                         ret++;
 159                 }
 160
 161         }
 162         string_list_clear(&params, 0);
 163         free(params_copy);
 164         return ret;
 165 }
 166
 167 static int parse_submodule_params(struct diff_options *options, const char *value)
 168 {
 169         if (!strcmp(value, "log"))
 170                 options->submodule_format = DIFF_SUBMODULE_LOG;
 171         else if (!strcmp(value, "short"))
 172                 options->submodule_format = DIFF_SUBMODULE_SHORT;
 173         else if (!strcmp(value, "diff"))
 174                 options->submodule_format = DIFF_SUBMODULE_INLINE_DIFF;
 175         else
 176                 return -1;
 177         return 0;
 178 }
 179
 180 static int git_config_rename(const char *var, const char *value)
 181 {
 182         if (!value)
 183                 return DIFF_DETECT_RENAME;
 184         if (!strcasecmp(value, "copies") || !strcasecmp(value, "copy"))
 185                 return  DIFF_DETECT_COPY;
 186         return git_config_bool(var,value) ? DIFF_DETECT_RENAME : 0;
 187 }
 188
 189 long parse_algorithm_value(const char *value)
 190 {
 191         if (!value)
 192                 return -1;
 193         else if (!strcasecmp(value, "myers") || !strcasecmp(value, "default"))
 194                 return 0;
 195         else if (!strcasecmp(value, "minimal"))
 196                 return XDF_NEED_MINIMAL;
 197         else if (!strcasecmp(value, "patience"))
 198                 return XDF_PATIENCE_DIFF;
 199         else if (!strcasecmp(value, "histogram"))
 200                 return XDF_HISTOGRAM_DIFF;
 201         return -1;
 202 }
 203
 204 static int parse_one_token(const char **arg, const char *token)
 205 {
 206         const char *rest;
 207         if (skip_prefix(*arg, token, &rest) && (!*rest || *rest == ',')) {
 208                 *arg = rest;
 209                 return 1;
 210         }
 211         return 0;
 212 }
 213
 214 static int parse_ws_error_highlight(const char *arg)
 215 {
 216         const char *orig_arg = arg;
 217         unsigned val = 0;
 218
 219         while (*arg) {
 220                 if (parse_one_token(&arg, "none"))
 221                         val = 0;
 222                 else if (parse_one_token(&arg, "default"))
 223                         val = WSEH_NEW;
 224                 else if (parse_one_token(&arg, "all"))
 225                         val = WSEH_NEW | WSEH_OLD | WSEH_CONTEXT;
 226                 else if (parse_one_token(&arg, "new"))
 227                         val |= WSEH_NEW;
 228                 else if (parse_one_token(&arg, "old"))
 229                         val |= WSEH_OLD;
 230                 else if (parse_one_token(&arg, "context"))
 231                         val |= WSEH_CONTEXT;
 232                 else {
 233                         return -1 - (int)(arg - orig_arg);
 234                 }
 235                 if (*arg)
 236                         arg++;
 237         }
 238         return val;
 239 }
 240
 241 /*
 242  * These are to give UI layer defaults.
 243  * The core-level commands such as git-diff-files should
 244  * never be affected by the setting of diff.renames
 245  * the user happens to have in the configuration file.
 246  */
 247 void init_diff_ui_defaults(void)
 248 {
 249         diff_detect_rename_default = DIFF_DETECT_RENAME;
 250 }
 251
 252 int git_diff_heuristic_config(const char *var, const char *value, void *cb)
 253 {
 254         if (!strcmp(var, "diff.indentheuristic"))
 255                 diff_indent_heuristic = git_config_bool(var, value);
 256         return 0;
 257 }
 258
 259 static int parse_color_moved(const char *arg)
 260 {
 261         switch (git_parse_maybe_bool(arg)) {
 262         case 0:
 263                 return COLOR_MOVED_NO;
 264         case 1:
 265                 return COLOR_MOVED_DEFAULT;
 266         default:
 267                 break;
 268         }
 269
 270         if (!strcmp(arg, "no"))
 271                 return COLOR_MOVED_NO;
 272         else if (!strcmp(arg, "plain"))
 273                 return COLOR_MOVED_PLAIN;
 274         else if (!strcmp(arg, "zebra"))
 275                 return COLOR_MOVED_ZEBRA;
 276         else if (!strcmp(arg, "default"))
 277                 return COLOR_MOVED_DEFAULT;
 278         else if (!strcmp(arg, "dimmed-zebra"))
 279                 return COLOR_MOVED_ZEBRA_DIM;
 280         else if (!strcmp(arg, "dimmed_zebra"))
 281                 return COLOR_MOVED_ZEBRA_DIM;
 282         else
 283                 return error(_("color moved setting must be one of 'no', 'default', 'zebra', 'dimmed-zebra', 'plain'"));
 284 }
 285
 286 int git_diff_ui_config(const char *var, const char *value, void *cb)
 287 {
 288         if (!strcmp(var, "diff.color") || !strcmp(var, "color.diff")) {
 289                 diff_use_color_default = git_config_colorbool(var, value);
 290                 return 0;
 291         }
 292         if (!strcmp(var, "diff.colormoved")) {
 293                 int cm = parse_color_moved(value);
 294                 if (cm < 0)
 295                         return -1;
 296                 diff_color_moved_default = cm;
 297                 return 0;
 298         }
 299         if (!strcmp(var, "diff.context")) {
 300                 diff_context_default = git_config_int(var, value);
 301                 if (diff_context_default < 0)
 302                         return -1;
 303                 return 0;
 304         }
 305         if (!strcmp(var, "diff.interhunkcontext")) {
 306                 diff_interhunk_context_default = git_config_int(var, value);
 307                 if (diff_interhunk_context_default < 0)
 308                         return -1;
 309                 return 0;
 310         }
 311         if (!strcmp(var, "diff.renames")) {
 312                 diff_detect_rename_default = git_config_rename(var, value);
 313                 return 0;
 314         }
 315         if (!strcmp(var, "diff.autorefreshindex")) {
 316                 diff_auto_refresh_index = git_config_bool(var, value);
 317                 return 0;
 318         }
 319         if (!strcmp(var, "diff.mnemonicprefix")) {
 320                 diff_mnemonic_prefix = git_config_bool(var, value);
 321                 return 0;
 322         }
 323         if (!strcmp(var, "diff.noprefix")) {
 324                 diff_no_prefix = git_config_bool(var, value);
 325                 return 0;
 326         }
 327         if (!strcmp(var, "diff.statgraphwidth")) {
 328                 diff_stat_graph_width = git_config_int(var, value);
 329                 return 0;
 330         }
 331         if (!strcmp(var, "diff.external"))
 332                 return git_config_string(&external_diff_cmd_cfg, var, value);
 333         if (!strcmp(var, "diff.wordregex"))
 334                 return git_config_string(&diff_word_regex_cfg, var, value);
 335         if (!strcmp(var, "diff.orderfile"))
 336                 return git_config_pathname(&diff_order_file_cfg, var, value);
 337
 338         if (!strcmp(var, "diff.ignoresubmodules"))
 339                 handle_ignore_submodules_arg(&default_diff_options, value);
 340
 341         if (!strcmp(var, "diff.submodule")) {
 342                 if (parse_submodule_params(&default_diff_options, value))
 343                         warning(_("Unknown value for 'diff.submodule' config variable: '%s'"),
 344                                 value);
 345                 return 0;
 346         }
 347
 348         if (!strcmp(var, "diff.algorithm")) {
 349                 diff_algorithm = parse_algorithm_value(value);
 350                 if (diff_algorithm < 0)
 351                         return -1;
 352                 return 0;
 353         }
 354
 355         if (!strcmp(var, "diff.wserrorhighlight")) {
 356                 int val = parse_ws_error_highlight(value);
 357                 if (val < 0)
 358                         return -1;
 359                 ws_error_highlight_default = val;
 360                 return 0;
 361         }
 362
 363         if (git_color_config(var, value, cb) < 0)
 364                 return -1;
 365
 366         return git_diff_basic_config(var, value, cb);
 367 }
 368
 369 int git_diff_basic_config(const char *var, const char *value, void *cb)
 370 {
 371         const char *name;
 372
 373         if (!strcmp(var, "diff.renamelimit")) {
 374                 diff_rename_limit_default = git_config_int(var, value);
 375                 return 0;
 376         }
 377
 378         if (userdiff_config(var, value) < 0)
 379                 return -1;
 380
 381         if (skip_prefix(var, "diff.color.", &name) ||
 382             skip_prefix(var, "color.diff.", &name)) {
 383                 int slot = parse_diff_color_slot(name);
 384                 if (slot < 0)
 385                         return 0;
 386                 if (!value)
 387                         return config_error_nonbool(var);
 388                 return color_parse(value, diff_colors[slot]);
 389         }
 390
 391         /* like GNU diff's --suppress-blank-empty option  */
 392         if (!strcmp(var, "diff.suppressblankempty") ||
 393                         /* for backwards compatibility */
 394                         !strcmp(var, "diff.suppress-blank-empty")) {
 395                 diff_suppress_blank_empty = git_config_bool(var, value);
 396                 return 0;
 397         }
 398
 399         if (!strcmp(var, "diff.dirstat")) {
 400                 struct strbuf errmsg = STRBUF_INIT;
 401                 default_diff_options.dirstat_permille = diff_dirstat_permille_default;
 402                 if (parse_dirstat_params(&default_diff_options, value, &errmsg))
 403                         warning(_("Found errors in 'diff.dirstat' config variable:\n%s"),
 404                                 errmsg.buf);
 405                 strbuf_release(&errmsg);
 406                 diff_dirstat_permille_default = default_diff_options.dirstat_permille;
 407                 return 0;
 408         }
 409
 410         if (git_diff_heuristic_config(var, value, cb) < 0)
 411                 return -1;
 412
 413         return git_default_config(var, value, cb);
 414 }
 415
 416 static char *quote_two(const char *one, const char *two)
 417 {
 418         int need_one = quote_c_style(one, NULL, NULL, 1);
 419         int need_two = quote_c_style(two, NULL, NULL, 1);
 420         struct strbuf res = STRBUF_INIT;
 421
 422         if (need_one + need_two) {
 423                 strbuf_addch(&res, '"');
 424                 quote_c_style(one, &res, NULL, 1);
 425                 quote_c_style(two, &res, NULL, 1);
 426                 strbuf_addch(&res, '"');
 427         } else {
 428                 strbuf_addstr(&res, one);
 429                 strbuf_addstr(&res, two);
 430         }
 431         return strbuf_detach(&res, NULL);
 432 }
 433
 434 static const char *external_diff(void)
 435 {
 436         static const char *external_diff_cmd = NULL;
 437         static int done_preparing = 0;
 438
 439         if (done_preparing)
 440                 return external_diff_cmd;
 441         external_diff_cmd = getenv("GIT_EXTERNAL_DIFF");
 442         if (!external_diff_cmd)
 443                 external_diff_cmd = external_diff_cmd_cfg;
 444         done_preparing = 1;
 445         return external_diff_cmd;
 446 }
 447
 448 /*
 449  * Keep track of files used for diffing. Sometimes such an entry
 450  * refers to a temporary file, sometimes to an existing file, and
 451  * sometimes to "/dev/null".
 452  */
 453 static struct diff_tempfile {
 454         /*
 455          * filename external diff should read from, or NULL if this
 456          * entry is currently not in use:
 457          */
 458         const char *name;
 459
 460         char hex[GIT_MAX_HEXSZ + 1];
 461         char mode[10];
 462
 463         /*
 464          * If this diff_tempfile instance refers to a temporary file,
 465          * this tempfile object is used to manage its lifetime.
 466          */
 467         struct tempfile *tempfile;
 468 } diff_temp[2];
 469
 470 struct emit_callback {
 471         int color_diff;
 472         unsigned ws_rule;
 473         int blank_at_eof_in_preimage;
 474         int blank_at_eof_in_postimage;
 475         int lno_in_preimage;
 476         int lno_in_postimage;
 477         const char **label_path;
 478         struct diff_words_data *diff_words;
 479         struct diff_options *opt;
 480         struct strbuf *header;
 481 };
 482
 483 static int count_lines(const char *data, int size)
 484 {
 485         int count, ch, completely_empty = 1, nl_just_seen = 0;
 486         count = 0;
 487         while (0 < size--) {
 488                 ch = *data++;
 489                 if (ch == '\n') {
 490                         count++;
 491                         nl_just_seen = 1;
 492                         completely_empty = 0;
 493                 }
 494                 else {
 495                         nl_just_seen = 0;
 496                         completely_empty = 0;
 497                 }
 498         }
 499         if (completely_empty)
 500                 return 0;
 501         if (!nl_just_seen)
 502                 count++; /* no trailing newline */
 503         return count;
 504 }
 505
 506 static int fill_mmfile(mmfile_t *mf, struct diff_filespec *one)
 507 {
 508         if (!DIFF_FILE_VALID(one)) {
 509                 mf->ptr = (char *)""; /* does not matter */
 510                 mf->size = 0;
 511                 return 0;
 512         }
 513         else if (diff_populate_filespec(one, 0))
 514                 return -1;
 515
 516         mf->ptr = one->data;
 517         mf->size = one->size;
 518         return 0;
 519 }
 520
 521 /* like fill_mmfile, but only for size, so we can avoid retrieving blob */
 522 static unsigned long diff_filespec_size(struct diff_filespec *one)
 523 {
 524         if (!DIFF_FILE_VALID(one))
 525                 return 0;
 526         diff_populate_filespec(one, CHECK_SIZE_ONLY);
 527         return one->size;
 528 }
 529
 530 static int count_trailing_blank(mmfile_t *mf, unsigned ws_rule)
 531 {
 532         char *ptr = mf->ptr;
 533         long size = mf->size;
 534         int cnt = 0;
 535
 536         if (!size)
 537                 return cnt;
 538         ptr += size - 1; /* pointing at the very end */
 539         if (*ptr != '\n')
 540                 ; /* incomplete line */
 541         else
 542                 ptr--; /* skip the last LF */
 543         while (mf->ptr < ptr) {
 544                 char *prev_eol;
 545                 for (prev_eol = ptr; mf->ptr <= prev_eol; prev_eol--)
 546                         if (*prev_eol == '\n')
 547                                 break;
 548                 if (!ws_blank_line(prev_eol + 1, ptr - prev_eol, ws_rule))
 549                         break;
 550                 cnt++;
 551                 ptr = prev_eol - 1;
 552         }
 553         return cnt;
 554 }
 555
 556 static void check_blank_at_eof(mmfile_t *mf1, mmfile_t *mf2,
 557                                struct emit_callback *ecbdata)
 558 {
 559         int l1, l2, at;
 560         unsigned ws_rule = ecbdata->ws_rule;
 561         l1 = count_trailing_blank(mf1, ws_rule);
 562         l2 = count_trailing_blank(mf2, ws_rule);
 563         if (l2 <= l1) {
 564                 ecbdata->blank_at_eof_in_preimage = 0;
 565                 ecbdata->blank_at_eof_in_postimage = 0;
 566                 return;
 567         }
 568         at = count_lines(mf1->ptr, mf1->size);
 569         ecbdata->blank_at_eof_in_preimage = (at - l1) + 1;
 570
 571         at = count_lines(mf2->ptr, mf2->size);
 572         ecbdata->blank_at_eof_in_postimage = (at - l2) + 1;
 573 }
 574
 575 static void emit_line_0(struct diff_options *o, const char *set, const char *reset,
 576                         int first, const char *line, int len)
 577 {
 578         int has_trailing_newline, has_trailing_carriage_return;
 579         int nofirst;
 580         FILE *file = o->file;
 581
 582         fputs(diff_line_prefix(o), file);
 583
 584         if (len == 0) {
 585                 has_trailing_newline = (first == '\n');
 586                 has_trailing_carriage_return = (!has_trailing_newline &&
 587                                                 (first == '\r'));
 588                 nofirst = has_trailing_newline || has_trailing_carriage_return;
 589         } else {
 590                 has_trailing_newline = (len > 0 && line[len-1] == '\n');
 591                 if (has_trailing_newline)
 592                         len--;
 593                 has_trailing_carriage_return = (len > 0 && line[len-1] == '\r');
 594                 if (has_trailing_carriage_return)
 595                         len--;
 596                 nofirst = 0;
 597         }
 598
 599         if (len || !nofirst) {
 600                 fputs(set, file);
 601                 if (!nofirst)
 602                         fputc(first, file);
 603                 fwrite(line, len, 1, file);
 604                 fputs(reset, file);
 605         }
 606         if (has_trailing_carriage_return)
 607                 fputc('\r', file);
 608         if (has_trailing_newline)
 609                 fputc('\n', file);
 610 }
 611
 612 static void emit_line(struct diff_options *o, const char *set, const char *reset,
 613                       const char *line, int len)
 614 {
 615         emit_line_0(o, set, reset, line[0], line+1, len-1);
 616 }
 617
 618 enum diff_symbol {
 619         DIFF_SYMBOL_BINARY_DIFF_HEADER,
 620         DIFF_SYMBOL_BINARY_DIFF_HEADER_DELTA,
 621         DIFF_SYMBOL_BINARY_DIFF_HEADER_LITERAL,
 622         DIFF_SYMBOL_BINARY_DIFF_BODY,
 623         DIFF_SYMBOL_BINARY_DIFF_FOOTER,
 624         DIFF_SYMBOL_STATS_SUMMARY_NO_FILES,
 625         DIFF_SYMBOL_STATS_SUMMARY_ABBREV,
 626         DIFF_SYMBOL_STATS_SUMMARY_INSERTS_DELETES,
 627         DIFF_SYMBOL_STATS_LINE,
 628         DIFF_SYMBOL_WORD_DIFF,
 629         DIFF_SYMBOL_STAT_SEP,
 630         DIFF_SYMBOL_SUMMARY,
 631         DIFF_SYMBOL_SUBMODULE_ADD,
 632         DIFF_SYMBOL_SUBMODULE_DEL,
 633         DIFF_SYMBOL_SUBMODULE_UNTRACKED,
 634         DIFF_SYMBOL_SUBMODULE_MODIFIED,
 635         DIFF_SYMBOL_SUBMODULE_HEADER,
 636         DIFF_SYMBOL_SUBMODULE_ERROR,
 637         DIFF_SYMBOL_SUBMODULE_PIPETHROUGH,
 638         DIFF_SYMBOL_REWRITE_DIFF,
 639         DIFF_SYMBOL_BINARY_FILES,
 640         DIFF_SYMBOL_HEADER,
 641         DIFF_SYMBOL_FILEPAIR_PLUS,
 642         DIFF_SYMBOL_FILEPAIR_MINUS,
 643         DIFF_SYMBOL_WORDS_PORCELAIN,
 644         DIFF_SYMBOL_WORDS,
 645         DIFF_SYMBOL_CONTEXT,
 646         DIFF_SYMBOL_CONTEXT_INCOMPLETE,
 647         DIFF_SYMBOL_PLUS,
 648         DIFF_SYMBOL_MINUS,
 649         DIFF_SYMBOL_NO_LF_EOF,
 650         DIFF_SYMBOL_CONTEXT_FRAGINFO,
 651         DIFF_SYMBOL_CONTEXT_MARKER,
 652         DIFF_SYMBOL_SEPARATOR
 653 };
 654 /*
 655  * Flags for content lines:
 656  * 0..12 are whitespace rules
 657  * 13-15 are WSEH_NEW | WSEH_OLD | WSEH_CONTEXT
 658  * 16 is marking if the line is blank at EOF
 659  */
 660 #define DIFF_SYMBOL_CONTENT_BLANK_LINE_EOF      (1<<16)
 661 #define DIFF_SYMBOL_MOVED_LINE                  (1<<17)
 662 #define DIFF_SYMBOL_MOVED_LINE_ALT              (1<<18)
 663 #define DIFF_SYMBOL_MOVED_LINE_UNINTERESTING    (1<<19)
 664 #define DIFF_SYMBOL_CONTENT_WS_MASK (WSEH_NEW | WSEH_OLD | WSEH_CONTEXT | WS_RULE_MASK)
 665
 666 /*
 667  * This struct is used when we need to buffer the output of the diff output.
 668  *
 669  * NEEDSWORK: Instead of storing a copy of the line, add an offset pointer
 670  * into the pre/post image file. This pointer could be a union with the
 671  * line pointer. By storing an offset into the file instead of the literal line,
 672  * we can decrease the memory footprint for the buffered output. At first we
 673  * may want to only have indirection for the content lines, but we could also
 674  * enhance the state for emitting prefabricated lines, e.g. the similarity
 675  * score line or hunk/file headers would only need to store a number or path
 676  * and then the output can be constructed later on depending on state.
 677  */
 678 struct emitted_diff_symbol {
 679         const char *line;
 680         int len;
 681         int flags;
 682         enum diff_symbol s;
 683 };
 684 #define EMITTED_DIFF_SYMBOL_INIT {NULL}
 685
 686 struct emitted_diff_symbols {
 687         struct emitted_diff_symbol *buf;
 688         int nr, alloc;
 689 };
 690 #define EMITTED_DIFF_SYMBOLS_INIT {NULL, 0, 0}
 691
 692 static void append_emitted_diff_symbol(struct diff_options *o,
 693                                        struct emitted_diff_symbol *e)
 694 {
 695         struct emitted_diff_symbol *f;
 696
 697         ALLOC_GROW(o->emitted_symbols->buf,
 698                    o->emitted_symbols->nr + 1,
 699                    o->emitted_symbols->alloc);
 700         f = &o->emitted_symbols->buf[o->emitted_symbols->nr++];
 701
 702         memcpy(f, e, sizeof(struct emitted_diff_symbol));
 703         f->line = e->line ? xmemdupz(e->line, e->len) : NULL;
 704 }
 705
 706 struct moved_entry {
 707         struct hashmap_entry ent;
 708         const struct emitted_diff_symbol *es;
 709         struct moved_entry *next_line;
 710 };
 711
 712 static int moved_entry_cmp(const struct diff_options *diffopt,
 713                            const struct moved_entry *a,
 714                            const struct moved_entry *b,
 715                            const void *keydata)
 716 {
 717         return !xdiff_compare_lines(a->es->line, a->es->len,
 718                                     b->es->line, b->es->len,
 719                                     diffopt->xdl_opts);
 720 }
 721
 722 static struct moved_entry *prepare_entry(struct diff_options *o,
 723                                          int line_no)
 724 {
 725         struct moved_entry *ret = xmalloc(sizeof(*ret));
 726         struct emitted_diff_symbol *l = &o->emitted_symbols->buf[line_no];
 727
 728         ret->ent.hash = xdiff_hash_string(l->line, l->len, o->xdl_opts);
 729         ret->es = l;
 730         ret->next_line = NULL;
 731
 732         return ret;
 733 }
 734
 735 static void add_lines_to_move_detection(struct diff_options *o,
 736                                         struct hashmap *add_lines,
 737                                         struct hashmap *del_lines)
 738 {
 739         struct moved_entry *prev_line = NULL;
 740
 741         int n;
 742         for (n = 0; n < o->emitted_symbols->nr; n++) {
 743                 struct hashmap *hm;
 744                 struct moved_entry *key;
 745
 746                 switch (o->emitted_symbols->buf[n].s) {
 747                 case DIFF_SYMBOL_PLUS:
 748                         hm = add_lines;
 749                         break;
 750                 case DIFF_SYMBOL_MINUS:
 751                         hm = del_lines;
 752                         break;
 753                 default:
 754                         prev_line = NULL;
 755                         continue;
 756                 }
 757
 758                 key = prepare_entry(o, n);
 759                 if (prev_line && prev_line->es->s == o->emitted_symbols->buf[n].s)
 760                         prev_line->next_line = key;
 761
 762                 hashmap_add(hm, key);
 763                 prev_line = key;
 764         }
 765 }
 766
 767 static int shrink_potential_moved_blocks(struct moved_entry **pmb,
 768                                          int pmb_nr)
 769 {
 770         int lp, rp;
 771
 772         /* Shrink the set of potential block to the remaining running */
 773         for (lp = 0, rp = pmb_nr - 1; lp <= rp;) {
 774                 while (lp < pmb_nr && pmb[lp])
 775                         lp++;
 776                 /* lp points at the first NULL now */
 777
 778                 while (rp > -1 && !pmb[rp])
 779                         rp--;
 780                 /* rp points at the last non-NULL */
 781
 782                 if (lp < pmb_nr && rp > -1 && lp < rp) {
 783                         pmb[lp] = pmb[rp];
 784                         pmb[rp] = NULL;
 785                         rp--;
 786                         lp++;
 787                 }
 788         }
 789
 790         /* Remember the number of running sets */
 791         return rp + 1;
 792 }
 793
 794 /*
 795  * If o->color_moved is COLOR_MOVED_PLAIN, this function does nothing.
 796  *
 797  * Otherwise, if the last block has fewer alphanumeric characters than
 798  * COLOR_MOVED_MIN_ALNUM_COUNT, unset DIFF_SYMBOL_MOVED_LINE on all lines in
 799  * that block.
 800  *
 801  * The last block consists of the (n - block_length)'th line up to but not
 802  * including the nth line.
 803  *
 804  * NEEDSWORK: This uses the same heuristic as blame_entry_score() in blame.c.
 805  * Think of a way to unify them.
 806  */
 807 static void adjust_last_block(struct diff_options *o, int n, int block_length)
 808 {
 809         int i, alnum_count = 0;
 810         if (o->color_moved == COLOR_MOVED_PLAIN)
 811                 return;
 812         for (i = 1; i < block_length + 1; i++) {
 813                 const char *c = o->emitted_symbols->buf[n - i].line;
 814                 for (; *c; c++) {
 815                         if (!isalnum(*c))
 816                                 continue;
 817                         alnum_count++;
 818                         if (alnum_count >= COLOR_MOVED_MIN_ALNUM_COUNT)
 819                                 return;
 820                 }
 821         }
 822         for (i = 1; i < block_length + 1; i++)
 823                 o->emitted_symbols->buf[n - i].flags &= ~DIFF_SYMBOL_MOVED_LINE;
 824 }
 825
 826 /* Find blocks of moved code, delegate actual coloring decision to helper */
 827 static void mark_color_as_moved(struct diff_options *o,
 828                                 struct hashmap *add_lines,
 829                                 struct hashmap *del_lines)
 830 {
 831         struct moved_entry **pmb = NULL; /* potentially moved blocks */
 832         int pmb_nr = 0, pmb_alloc = 0;
 833         int n, flipped_block = 1, block_length = 0;
 834
 835
 836         for (n = 0; n < o->emitted_symbols->nr; n++) {
 837                 struct hashmap *hm = NULL;
 838                 struct moved_entry *key;
 839                 struct moved_entry *match = NULL;
 840                 struct emitted_diff_symbol *l = &o->emitted_symbols->buf[n];
 841                 int i;
 842
 843                 switch (l->s) {
 844                 case DIFF_SYMBOL_PLUS:
 845                         hm = del_lines;
 846                         key = prepare_entry(o, n);
 847                         match = hashmap_get(hm, key, o);
 848                         free(key);
 849                         break;
 850                 case DIFF_SYMBOL_MINUS:
 851                         hm = add_lines;
 852                         key = prepare_entry(o, n);
 853                         match = hashmap_get(hm, key, o);
 854                         free(key);
 855                         break;
 856                 default:
 857                         flipped_block = 1;
 858                 }
 859
 860                 if (!match) {
 861                         adjust_last_block(o, n, block_length);
 862                         pmb_nr = 0;
 863                         block_length = 0;
 864                         continue;
 865                 }
 866
 867                 l->flags |= DIFF_SYMBOL_MOVED_LINE;
 868
 869                 if (o->color_moved == COLOR_MOVED_PLAIN)
 870                         continue;
 871
 872                 /* Check any potential block runs, advance each or nullify */
 873                 for (i = 0; i < pmb_nr; i++) {
 874                         struct moved_entry *p = pmb[i];
 875                         struct moved_entry *pnext = (p && p->next_line) ?
 876                                         p->next_line : NULL;
 877                         if (pnext && !hm->cmpfn(o, pnext, match, NULL)) {
 878                                 pmb[i] = p->next_line;
 879                         } else {
 880                                 pmb[i] = NULL;
 881                         }
 882                 }
 883
 884                 pmb_nr = shrink_potential_moved_blocks(pmb, pmb_nr);
 885
 886                 if (pmb_nr == 0) {
 887                         /*
 888                          * The current line is the start of a new block.
 889                          * Setup the set of potential blocks.
 890                          */
 891                         for (; match; match = hashmap_get_next(hm, match)) {
 892                                 ALLOC_GROW(pmb, pmb_nr + 1, pmb_alloc);
 893                                 pmb[pmb_nr++] = match;
 894                         }
 895
 896                         flipped_block = (flipped_block + 1) % 2;
 897
 898                         adjust_last_block(o, n, block_length);
 899                         block_length = 0;
 900                 }
 901
 902                 block_length++;
 903
 904                 if (flipped_block)
 905                         l->flags |= DIFF_SYMBOL_MOVED_LINE_ALT;
 906         }
 907         adjust_last_block(o, n, block_length);
 908
 909         free(pmb);
 910 }
 911
 912 #define DIFF_SYMBOL_MOVED_LINE_ZEBRA_MASK \
 913   (DIFF_SYMBOL_MOVED_LINE | DIFF_SYMBOL_MOVED_LINE_ALT)
 914 static void dim_moved_lines(struct diff_options *o)
 915 {
 916         int n;
 917         for (n = 0; n < o->emitted_symbols->nr; n++) {
 918                 struct emitted_diff_symbol *prev = (n != 0) ?
 919                                 &o->emitted_symbols->buf[n - 1] : NULL;
 920                 struct emitted_diff_symbol *l = &o->emitted_symbols->buf[n];
 921                 struct emitted_diff_symbol *next =
 922                                 (n < o->emitted_symbols->nr - 1) ?
 923                                 &o->emitted_symbols->buf[n + 1] : NULL;
 924
 925                 /* Not a plus or minus line? */
 926                 if (l->s != DIFF_SYMBOL_PLUS && l->s != DIFF_SYMBOL_MINUS)
 927                         continue;
 928
 929                 /* Not a moved line? */
 930                 if (!(l->flags & DIFF_SYMBOL_MOVED_LINE))
 931                         continue;
 932
 933                 /*
 934                  * If prev or next are not a plus or minus line,
 935                  * pretend they don't exist
 936                  */
 937                 if (prev && prev->s != DIFF_SYMBOL_PLUS &&
 938                             prev->s != DIFF_SYMBOL_MINUS)
 939                         prev = NULL;
 940                 if (next && next->s != DIFF_SYMBOL_PLUS &&
 941                             next->s != DIFF_SYMBOL_MINUS)
 942                         next = NULL;
 943
 944                 /* Inside a block? */
 945                 if ((prev &&
 946                     (prev->flags & DIFF_SYMBOL_MOVED_LINE_ZEBRA_MASK) ==
 947                     (l->flags & DIFF_SYMBOL_MOVED_LINE_ZEBRA_MASK)) &&
 948                     (next &&
 949                     (next->flags & DIFF_SYMBOL_MOVED_LINE_ZEBRA_MASK) ==
 950                     (l->flags & DIFF_SYMBOL_MOVED_LINE_ZEBRA_MASK))) {
 951                         l->flags |= DIFF_SYMBOL_MOVED_LINE_UNINTERESTING;
 952                         continue;
 953                 }
 954
 955                 /* Check if we are at an interesting bound: */
 956                 if (prev && (prev->flags & DIFF_SYMBOL_MOVED_LINE) &&
 957                     (prev->flags & DIFF_SYMBOL_MOVED_LINE_ALT) !=
 958                        (l->flags & DIFF_SYMBOL_MOVED_LINE_ALT))
 959                         continue;
 960                 if (next && (next->flags & DIFF_SYMBOL_MOVED_LINE) &&
 961                     (next->flags & DIFF_SYMBOL_MOVED_LINE_ALT) !=
 962                        (l->flags & DIFF_SYMBOL_MOVED_LINE_ALT))
 963                         continue;
 964
 965                 /*
 966                  * The boundary to prev and next are not interesting,
 967                  * so this line is not interesting as a whole
 968                  */
 969                 l->flags |= DIFF_SYMBOL_MOVED_LINE_UNINTERESTING;
 970         }
 971 }
 972
 973 static void emit_line_ws_markup(struct diff_options *o,
 974                                 const char *set, const char *reset,
 975                                 const char *line, int len, char sign,
 976                                 unsigned ws_rule, int blank_at_eof)
 977 {
 978         const char *ws = NULL;
 979
 980         if (o->ws_error_highlight & ws_rule) {
 981                 ws = diff_get_color_opt(o, DIFF_WHITESPACE);
 982                 if (!*ws)
 983                         ws = NULL;
 984         }
 985
 986         if (!ws)
 987                 emit_line_0(o, set, reset, sign, line, len);
 988         else if (blank_at_eof)
 989                 /* Blank line at EOF - paint '+' as well */
 990                 emit_line_0(o, ws, reset, sign, line, len);
 991         else {
 992                 /* Emit just the prefix, then the rest. */
 993                 emit_line_0(o, set, reset, sign, "", 0);
 994                 ws_check_emit(line, len, ws_rule,
 995                               o->file, set, reset, ws);
 996         }
 997 }
 998
 999 static void emit_diff_symbol_from_struct(struct diff_options *o,
1000                                          struct emitted_diff_symbol *eds)
1001 {
1002         static const char *nneof = " No newline at end of file\n";
1003         const char *context, *reset, *set, *meta, *fraginfo;
1004         struct strbuf sb = STRBUF_INIT;
1005
1006         enum diff_symbol s = eds->s;
1007         const char *line = eds->line;
1008         int len = eds->len;
1009         unsigned flags = eds->flags;
1010
1011         switch (s) {
1012         case DIFF_SYMBOL_NO_LF_EOF:
1013                 context = diff_get_color_opt(o, DIFF_CONTEXT);
1014                 reset = diff_get_color_opt(o, DIFF_RESET);
1015                 putc('\n', o->file);
1016                 emit_line_0(o, context, reset, '\\',
1017                             nneof, strlen(nneof));
1018                 break;
1019         case DIFF_SYMBOL_SUBMODULE_HEADER:
1020         case DIFF_SYMBOL_SUBMODULE_ERROR:
1021         case DIFF_SYMBOL_SUBMODULE_PIPETHROUGH:
1022         case DIFF_SYMBOL_STATS_SUMMARY_INSERTS_DELETES:
1023         case DIFF_SYMBOL_SUMMARY:
1024         case DIFF_SYMBOL_STATS_LINE:
1025         case DIFF_SYMBOL_BINARY_DIFF_BODY:
1026         case DIFF_SYMBOL_CONTEXT_FRAGINFO:
1027                 emit_line(o, "", "", line, len);
1028                 break;
1029         case DIFF_SYMBOL_CONTEXT_INCOMPLETE:
1030         case DIFF_SYMBOL_CONTEXT_MARKER:
1031                 context = diff_get_color_opt(o, DIFF_CONTEXT);
1032                 reset = diff_get_color_opt(o, DIFF_RESET);
1033                 emit_line(o, context, reset, line, len);
1034                 break;
1035         case DIFF_SYMBOL_SEPARATOR:
1036                 fprintf(o->file, "%s%c",
1037                         diff_line_prefix(o),
1038                         o->line_termination);
1039                 break;
1040         case DIFF_SYMBOL_CONTEXT:
1041                 set = diff_get_color_opt(o, DIFF_CONTEXT);
1042                 reset = diff_get_color_opt(o, DIFF_RESET);
1043                 emit_line_ws_markup(o, set, reset, line, len, ' ',
1044                                     flags & (DIFF_SYMBOL_CONTENT_WS_MASK), 0);
1045                 break;
1046         case DIFF_SYMBOL_PLUS:
1047                 switch (flags & (DIFF_SYMBOL_MOVED_LINE |
1048                                  DIFF_SYMBOL_MOVED_LINE_ALT |
1049                                  DIFF_SYMBOL_MOVED_LINE_UNINTERESTING)) {
1050                 case DIFF_SYMBOL_MOVED_LINE |
1051                      DIFF_SYMBOL_MOVED_LINE_ALT |
1052                      DIFF_SYMBOL_MOVED_LINE_UNINTERESTING:
1053                         set = diff_get_color_opt(o, DIFF_FILE_NEW_MOVED_ALT_DIM);
1054                         break;
1055                 case DIFF_SYMBOL_MOVED_LINE |
1056                      DIFF_SYMBOL_MOVED_LINE_ALT:
1057                         set = diff_get_color_opt(o, DIFF_FILE_NEW_MOVED_ALT);
1058                         break;
1059                 case DIFF_SYMBOL_MOVED_LINE |
1060                      DIFF_SYMBOL_MOVED_LINE_UNINTERESTING:
1061                         set = diff_get_color_opt(o, DIFF_FILE_NEW_MOVED_DIM);
1062                         break;
1063                 case DIFF_SYMBOL_MOVED_LINE:
1064                         set = diff_get_color_opt(o, DIFF_FILE_NEW_MOVED);
1065                         break;
1066                 default:
1067                         set = diff_get_color_opt(o, DIFF_FILE_NEW);
1068                 }
1069                 reset = diff_get_color_opt(o, DIFF_RESET);
1070                 emit_line_ws_markup(o, set, reset, line, len, '+',
1071                                     flags & DIFF_SYMBOL_CONTENT_WS_MASK,
1072                                     flags & DIFF_SYMBOL_CONTENT_BLANK_LINE_EOF);
1073                 break;
1074         case DIFF_SYMBOL_MINUS:
1075                 switch (flags & (DIFF_SYMBOL_MOVED_LINE |
1076                                  DIFF_SYMBOL_MOVED_LINE_ALT |
1077                                  DIFF_SYMBOL_MOVED_LINE_UNINTERESTING)) {
1078                 case DIFF_SYMBOL_MOVED_LINE |
1079                      DIFF_SYMBOL_MOVED_LINE_ALT |
1080                      DIFF_SYMBOL_MOVED_LINE_UNINTERESTING:
1081                         set = diff_get_color_opt(o, DIFF_FILE_OLD_MOVED_ALT_DIM);
1082                         break;
1083                 case DIFF_SYMBOL_MOVED_LINE |
1084                      DIFF_SYMBOL_MOVED_LINE_ALT:
1085                         set = diff_get_color_opt(o, DIFF_FILE_OLD_MOVED_ALT);
1086                         break;
1087                 case DIFF_SYMBOL_MOVED_LINE |
1088                      DIFF_SYMBOL_MOVED_LINE_UNINTERESTING:
1089                         set = diff_get_color_opt(o, DIFF_FILE_OLD_MOVED_DIM);
1090                         break;
1091                 case DIFF_SYMBOL_MOVED_LINE:
1092                         set = diff_get_color_opt(o, DIFF_FILE_OLD_MOVED);
1093                         break;
1094                 default:
1095                         set = diff_get_color_opt(o, DIFF_FILE_OLD);
1096                 }
1097                 reset = diff_get_color_opt(o, DIFF_RESET);
1098                 emit_line_ws_markup(o, set, reset, line, len, '-',
1099                                     flags & DIFF_SYMBOL_CONTENT_WS_MASK, 0);
1100                 break;
1101         case DIFF_SYMBOL_WORDS_PORCELAIN:
1102                 context = diff_get_color_opt(o, DIFF_CONTEXT);
1103                 reset = diff_get_color_opt(o, DIFF_RESET);
1104                 emit_line(o, context, reset, line, len);
1105                 fputs("~\n", o->file);
1106                 break;
1107         case DIFF_SYMBOL_WORDS:
1108                 context = diff_get_color_opt(o, DIFF_CONTEXT);
1109                 reset = diff_get_color_opt(o, DIFF_RESET);
1110                 /*
1111                  * Skip the prefix character, if any.  With
1112                  * diff_suppress_blank_empty, there may be
1113                  * none.
1114                  */
1115                 if (line[0] != '\n') {
1116                         line++;
1117                         len--;
1118                 }
1119                 emit_line(o, context, reset, line, len);
1120                 break;
1121         case DIFF_SYMBOL_FILEPAIR_PLUS:
1122                 meta = diff_get_color_opt(o, DIFF_METAINFO);
1123                 reset = diff_get_color_opt(o, DIFF_RESET);
1124                 fprintf(o->file, "%s%s+++ %s%s%s\n", diff_line_prefix(o), meta,
1125                         line, reset,
1126                         strchr(line, ' ') ? "\t" : "");
1127                 break;
1128         case DIFF_SYMBOL_FILEPAIR_MINUS:
1129                 meta = diff_get_color_opt(o, DIFF_METAINFO);
1130                 reset = diff_get_color_opt(o, DIFF_RESET);
1131                 fprintf(o->file, "%s%s--- %s%s%s\n", diff_line_prefix(o), meta,
1132                         line, reset,
1133                         strchr(line, ' ') ? "\t" : "");
1134                 break;
1135         case DIFF_SYMBOL_BINARY_FILES:
1136         case DIFF_SYMBOL_HEADER:
1137                 fprintf(o->file, "%s", line);
1138                 break;
1139         case DIFF_SYMBOL_BINARY_DIFF_HEADER:
1140                 fprintf(o->file, "%sGIT binary patch\n", diff_line_prefix(o));
1141                 break;
1142         case DIFF_SYMBOL_BINARY_DIFF_HEADER_DELTA:
1143                 fprintf(o->file, "%sdelta %s\n", diff_line_prefix(o), line);
1144                 break;
1145         case DIFF_SYMBOL_BINARY_DIFF_HEADER_LITERAL:
1146                 fprintf(o->file, "%sliteral %s\n", diff_line_prefix(o), line);
1147                 break;
1148         case DIFF_SYMBOL_BINARY_DIFF_FOOTER:
1149                 fputs(diff_line_prefix(o), o->file);
1150                 fputc('\n', o->file);
1151                 break;
1152         case DIFF_SYMBOL_REWRITE_DIFF:
1153                 fraginfo = diff_get_color(o->use_color, DIFF_FRAGINFO);
1154                 reset = diff_get_color_opt(o, DIFF_RESET);
1155                 emit_line(o, fraginfo, reset, line, len);
1156                 break;
1157         case DIFF_SYMBOL_SUBMODULE_ADD:
1158                 set = diff_get_color_opt(o, DIFF_FILE_NEW);
1159                 reset = diff_get_color_opt(o, DIFF_RESET);
1160                 emit_line(o, set, reset, line, len);
1161                 break;
1162         case DIFF_SYMBOL_SUBMODULE_DEL:
1163                 set = diff_get_color_opt(o, DIFF_FILE_OLD);
1164                 reset = diff_get_color_opt(o, DIFF_RESET);
1165                 emit_line(o, set, reset, line, len);
1166                 break;
1167         case DIFF_SYMBOL_SUBMODULE_UNTRACKED:
1168                 fprintf(o->file, "%sSubmodule %s contains untracked content\n",
1169                         diff_line_prefix(o), line);
1170                 break;
1171         case DIFF_SYMBOL_SUBMODULE_MODIFIED:
1172                 fprintf(o->file, "%sSubmodule %s contains modified content\n",
1173                         diff_line_prefix(o), line);
1174                 break;
1175         case DIFF_SYMBOL_STATS_SUMMARY_NO_FILES:
1176                 emit_line(o, "", "", " 0 files changed\n",
1177                           strlen(" 0 files changed\n"));
1178                 break;
1179         case DIFF_SYMBOL_STATS_SUMMARY_ABBREV:
1180                 emit_line(o, "", "", " ...\n", strlen(" ...\n"));
1181                 break;
1182         case DIFF_SYMBOL_WORD_DIFF:
1183                 fprintf(o->file, "%.*s", len, line);
1184                 break;
1185         case DIFF_SYMBOL_STAT_SEP:
1186                 fputs(o->stat_sep, o->file);
1187                 break;
1188         default:
1189                 die("BUG: unknown diff symbol");
1190         }
1191         strbuf_release(&sb);
1192 }
1193
1194 static void emit_diff_symbol(struct diff_options *o, enum diff_symbol s,
1195                              const char *line, int len, unsigned flags)
1196 {
1197         struct emitted_diff_symbol e = {line, len, flags, s};
1198
1199         if (o->emitted_symbols)
1200                 append_emitted_diff_symbol(o, &e);
1201         else
1202                 emit_diff_symbol_from_struct(o, &e);
1203 }
1204
1205 void diff_emit_submodule_del(struct diff_options *o, const char *line)
1206 {
1207         emit_diff_symbol(o, DIFF_SYMBOL_SUBMODULE_DEL, line, strlen(line), 0);
1208 }
1209
1210 void diff_emit_submodule_add(struct diff_options *o, const char *line)
1211 {
1212         emit_diff_symbol(o, DIFF_SYMBOL_SUBMODULE_ADD, line, strlen(line), 0);
1213 }
1214
1215 void diff_emit_submodule_untracked(struct diff_options *o, const char *path)
1216 {
1217         emit_diff_symbol(o, DIFF_SYMBOL_SUBMODULE_UNTRACKED,
1218                          path, strlen(path), 0);
1219 }
1220
1221 void diff_emit_submodule_modified(struct diff_options *o, const char *path)
1222 {
1223         emit_diff_symbol(o, DIFF_SYMBOL_SUBMODULE_MODIFIED,
1224                          path, strlen(path), 0);
1225 }
1226
1227 void diff_emit_submodule_header(struct diff_options *o, const char *header)
1228 {
1229         emit_diff_symbol(o, DIFF_SYMBOL_SUBMODULE_HEADER,
1230                          header, strlen(header), 0);
1231 }
1232
1233 void diff_emit_submodule_error(struct diff_options *o, const char *err)
1234 {
1235         emit_diff_symbol(o, DIFF_SYMBOL_SUBMODULE_ERROR, err, strlen(err), 0);
1236 }
1237
1238 void diff_emit_submodule_pipethrough(struct diff_options *o,
1239                                      const char *line, int len)
1240 {
1241         emit_diff_symbol(o, DIFF_SYMBOL_SUBMODULE_PIPETHROUGH, line, len, 0);
1242 }
1243
1244 static int new_blank_line_at_eof(struct emit_callback *ecbdata, const char *line, int len)
1245 {
1246         if (!((ecbdata->ws_rule & WS_BLANK_AT_EOF) &&
1247               ecbdata->blank_at_eof_in_preimage &&
1248               ecbdata->blank_at_eof_in_postimage &&
1249               ecbdata->blank_at_eof_in_preimage <= ecbdata->lno_in_preimage &&
1250               ecbdata->blank_at_eof_in_postimage <= ecbdata->lno_in_postimage))
1251                 return 0;
1252         return ws_blank_line(line, len, ecbdata->ws_rule);
1253 }
1254
1255 static void emit_add_line(const char *reset,
1256                           struct emit_callback *ecbdata,
1257                           const char *line, int len)
1258 {
1259         unsigned flags = WSEH_NEW | ecbdata->ws_rule;
1260         if (new_blank_line_at_eof(ecbdata, line, len))
1261                 flags |= DIFF_SYMBOL_CONTENT_BLANK_LINE_EOF;
1262
1263         emit_diff_symbol(ecbdata->opt, DIFF_SYMBOL_PLUS, line, len, flags);
1264 }
1265
1266 static void emit_del_line(const char *reset,
1267                           struct emit_callback *ecbdata,
1268                           const char *line, int len)
1269 {
1270         unsigned flags = WSEH_OLD | ecbdata->ws_rule;
1271         emit_diff_symbol(ecbdata->opt, DIFF_SYMBOL_MINUS, line, len, flags);
1272 }
1273
1274 static void emit_context_line(const char *reset,
1275                               struct emit_callback *ecbdata,
1276                               const char *line, int len)
1277 {
1278         unsigned flags = WSEH_CONTEXT | ecbdata->ws_rule;
1279         emit_diff_symbol(ecbdata->opt, DIFF_SYMBOL_CONTEXT, line, len, flags);
1280 }
1281
1282 static void emit_hunk_header(struct emit_callback *ecbdata,
1283                              const char *line, int len)
1284 {
1285         const char *context = diff_get_color(ecbdata->color_diff, DIFF_CONTEXT);
1286         const char *frag = diff_get_color(ecbdata->color_diff, DIFF_FRAGINFO);
1287         const char *func = diff_get_color(ecbdata->color_diff, DIFF_FUNCINFO);
1288         const char *reset = diff_get_color(ecbdata->color_diff, DIFF_RESET);
1289         static const char atat[2] = { '@', '@' };
1290         const char *cp, *ep;
1291         struct strbuf msgbuf = STRBUF_INIT;
1292         int org_len = len;
1293         int i = 1;
1294
1295         /*
1296          * As a hunk header must begin with "@@ -<old>, +<new> @@",
1297          * it always is at least 10 bytes long.
1298          */
1299         if (len < 10 ||
1300             memcmp(line, atat, 2) ||
1301             !(ep = memmem(line + 2, len - 2, atat, 2))) {
1302                 emit_diff_symbol(ecbdata->opt,
1303                                  DIFF_SYMBOL_CONTEXT_MARKER, line, len, 0);
1304                 return;
1305         }
1306         ep += 2; /* skip over @@ */
1307
1308         /* The hunk header in fraginfo color */
1309         strbuf_addstr(&msgbuf, frag);
1310         strbuf_add(&msgbuf, line, ep - line);
1311         strbuf_addstr(&msgbuf, reset);
1312
1313         /*
1314          * trailing "\r\n"
1315          */
1316         for ( ; i < 3; i++)
1317                 if (line[len - i] == '\r' || line[len - i] == '\n')
1318                         len--;
1319
1320         /* blank before the func header */
1321         for (cp = ep; ep - line < len; ep++)
1322                 if (*ep != ' ' && *ep != '\t')
1323                         break;
1324         if (ep != cp) {
1325                 strbuf_addstr(&msgbuf, context);
1326                 strbuf_add(&msgbuf, cp, ep - cp);
1327                 strbuf_addstr(&msgbuf, reset);
1328         }
1329
1330         if (ep < line + len) {
1331                 strbuf_addstr(&msgbuf, func);
1332                 strbuf_add(&msgbuf, ep, line + len - ep);
1333                 strbuf_addstr(&msgbuf, reset);
1334         }
1335
1336         strbuf_add(&msgbuf, line + len, org_len - len);
1337         strbuf_complete_line(&msgbuf);
1338         emit_diff_symbol(ecbdata->opt,
1339                          DIFF_SYMBOL_CONTEXT_FRAGINFO, msgbuf.buf, msgbuf.len, 0);
1340         strbuf_release(&msgbuf);
1341 }
1342
1343 static struct diff_tempfile *claim_diff_tempfile(void) {
1344         int i;
1345         for (i = 0; i < ARRAY_SIZE(diff_temp); i++)
1346                 if (!diff_temp[i].name)
1347                         return diff_temp + i;
1348         die("BUG: diff is failing to clean up its tempfiles");
1349 }
1350
1351 static void remove_tempfile(void)
1352 {
1353         int i;
1354         for (i = 0; i < ARRAY_SIZE(diff_temp); i++) {
1355                 if (is_tempfile_active(diff_temp[i].tempfile))
1356                         delete_tempfile(&diff_temp[i].tempfile);
1357                 diff_temp[i].name = NULL;
1358         }
1359 }
1360
1361 static void add_line_count(struct strbuf *out, int count)
1362 {
1363         switch (count) {
1364         case 0:
1365                 strbuf_addstr(out, "0,0");
1366                 break;
1367         case 1:
1368                 strbuf_addstr(out, "1");
1369                 break;
1370         default:
1371                 strbuf_addf(out, "1,%d", count);
1372                 break;
1373         }
1374 }
1375
1376 static void emit_rewrite_lines(struct emit_callback *ecb,
1377                                int prefix, const char *data, int size)
1378 {
1379         const char *endp = NULL;
1380         const char *reset = diff_get_color(ecb->color_diff, DIFF_RESET);
1381
1382         while (0 < size) {
1383                 int len;
1384
1385                 endp = memchr(data, '\n', size);
1386                 len = endp ? (endp - data + 1) : size;
1387                 if (prefix != '+') {
1388                         ecb->lno_in_preimage++;
1389                         emit_del_line(reset, ecb, data, len);
1390                 } else {
1391                         ecb->lno_in_postimage++;
1392                         emit_add_line(reset, ecb, data, len);
1393                 }
1394                 size -= len;
1395                 data += len;
1396         }
1397         if (!endp)
1398                 emit_diff_symbol(ecb->opt, DIFF_SYMBOL_NO_LF_EOF, NULL, 0, 0);
1399 }
1400
1401 static void emit_rewrite_diff(const char *name_a,
1402                               const char *name_b,
1403                               struct diff_filespec *one,
1404                               struct diff_filespec *two,
1405                               struct userdiff_driver *textconv_one,
1406                               struct userdiff_driver *textconv_two,
1407                               struct diff_options *o)
1408 {
1409         int lc_a, lc_b;
1410         static struct strbuf a_name = STRBUF_INIT, b_name = STRBUF_INIT;
1411         const char *a_prefix, *b_prefix;
1412         char *data_one, *data_two;
1413         size_t size_one, size_two;
1414         struct emit_callback ecbdata;
1415         struct strbuf out = STRBUF_INIT;
1416
1417         if (diff_mnemonic_prefix && o->flags.reverse_diff) {
1418                 a_prefix = o->b_prefix;
1419                 b_prefix = o->a_prefix;
1420         } else {
1421                 a_prefix = o->a_prefix;
1422                 b_prefix = o->b_prefix;
1423         }
1424
1425         name_a += (*name_a == '/');
1426         name_b += (*name_b == '/');
1427
1428         strbuf_reset(&a_name);
1429         strbuf_reset(&b_name);
1430         quote_two_c_style(&a_name, a_prefix, name_a, 0);
1431         quote_two_c_style(&b_name, b_prefix, name_b, 0);
1432
1433         size_one = fill_textconv(textconv_one, one, &data_one);
1434         size_two = fill_textconv(textconv_two, two, &data_two);
1435
1436         memset(&ecbdata, 0, sizeof(ecbdata));
1437         ecbdata.color_diff = want_color(o->use_color);
1438         ecbdata.ws_rule = whitespace_rule(name_b);
1439         ecbdata.opt = o;
1440         if (ecbdata.ws_rule & WS_BLANK_AT_EOF) {
1441                 mmfile_t mf1, mf2;
1442                 mf1.ptr = (char *)data_one;
1443                 mf2.ptr = (char *)data_two;
1444                 mf1.size = size_one;
1445                 mf2.size = size_two;
1446                 check_blank_at_eof(&mf1, &mf2, &ecbdata);
1447         }
1448         ecbdata.lno_in_preimage = 1;
1449         ecbdata.lno_in_postimage = 1;
1450
1451         lc_a = count_lines(data_one, size_one);
1452         lc_b = count_lines(data_two, size_two);
1453
1454         emit_diff_symbol(o, DIFF_SYMBOL_FILEPAIR_MINUS,
1455                          a_name.buf, a_name.len, 0);
1456         emit_diff_symbol(o, DIFF_SYMBOL_FILEPAIR_PLUS,
1457                          b_name.buf, b_name.len, 0);
1458
1459         strbuf_addstr(&out, "@@ -");
1460         if (!o->irreversible_delete)
1461                 add_line_count(&out, lc_a);
1462         else
1463                 strbuf_addstr(&out, "?,?");
1464         strbuf_addstr(&out, " +");
1465         add_line_count(&out, lc_b);
1466         strbuf_addstr(&out, " @@\n");
1467         emit_diff_symbol(o, DIFF_SYMBOL_REWRITE_DIFF, out.buf, out.len, 0);
1468         strbuf_release(&out);
1469
1470         if (lc_a && !o->irreversible_delete)
1471                 emit_rewrite_lines(&ecbdata, '-', data_one, size_one);
1472         if (lc_b)
1473                 emit_rewrite_lines(&ecbdata, '+', data_two, size_two);
1474         if (textconv_one)
1475                 free((char *)data_one);
1476         if (textconv_two)
1477                 free((char *)data_two);
1478 }
1479
1480 struct diff_words_buffer {
1481         mmfile_t text;
1482         unsigned long alloc;
1483         struct diff_words_orig {
1484                 const char *begin, *end;
1485         } *orig;
1486         int orig_nr, orig_alloc;
1487 };
1488
1489 static void diff_words_append(char *line, unsigned long len,
1490                 struct diff_words_buffer *buffer)
1491 {
1492         ALLOC_GROW(buffer->text.ptr, buffer->text.size + len, buffer->alloc);
1493         line++;
1494         len--;
1495         memcpy(buffer->text.ptr + buffer->text.size, line, len);
1496         buffer->text.size += len;
1497         buffer->text.ptr[buffer->text.size] = '\0';
1498 }
1499
1500 struct diff_words_style_elem {
1501         const char *prefix;
1502         const char *suffix;
1503         const char *color; /* NULL; filled in by the setup code if
1504                             * color is enabled */
1505 };
1506
1507 struct diff_words_style {
1508         enum diff_words_type type;
1509         struct diff_words_style_elem new, old, ctx;
1510         const char *newline;
1511 };
1512
1513 static struct diff_words_style diff_words_styles[] = {
1514         { DIFF_WORDS_PORCELAIN, {"+", "\n"}, {"-", "\n"}, {" ", "\n"}, "~\n" },
1515         { DIFF_WORDS_PLAIN, {"{+", "+}"}, {"[-", "-]"}, {"", ""}, "\n" },
1516         { DIFF_WORDS_COLOR, {"", ""}, {"", ""}, {"", ""}, "\n" }
1517 };
1518
1519 struct diff_words_data {
1520         struct diff_words_buffer minus, plus;
1521         const char *current_plus;
1522         int last_minus;
1523         struct diff_options *opt;
1524         regex_t *word_regex;
1525         enum diff_words_type type;
1526         struct diff_words_style *style;
1527 };
1528
1529 static int fn_out_diff_words_write_helper(struct diff_options *o,
1530                                           struct diff_words_style_elem *st_el,
1531                                           const char *newline,
1532                                           size_t count, const char *buf)
1533 {
1534         int print = 0;
1535         struct strbuf sb = STRBUF_INIT;
1536
1537         while (count) {
1538                 char *p = memchr(buf, '\n', count);
1539                 if (print)
1540                         strbuf_addstr(&sb, diff_line_prefix(o));
1541
1542                 if (p != buf) {
1543                         const char *reset = st_el->color && *st_el->color ?
1544                                             GIT_COLOR_RESET : NULL;
1545                         if (st_el->color && *st_el->color)
1546                                 strbuf_addstr(&sb, st_el->color);
1547                         strbuf_addstr(&sb, st_el->prefix);
1548                         strbuf_add(&sb, buf, p ? p - buf : count);
1549                         strbuf_addstr(&sb, st_el->suffix);
1550                         if (reset)
1551                                 strbuf_addstr(&sb, reset);
1552                 }
1553                 if (!p)
1554                         goto out;
1555
1556                 strbuf_addstr(&sb, newline);
1557                 count -= p + 1 - buf;
1558                 buf = p + 1;
1559                 print = 1;
1560                 if (count) {
1561                         emit_diff_symbol(o, DIFF_SYMBOL_WORD_DIFF,
1562                                          sb.buf, sb.len, 0);
1563                         strbuf_reset(&sb);
1564                 }
1565         }
1566
1567 out:
1568         if (sb.len)
1569                 emit_diff_symbol(o, DIFF_SYMBOL_WORD_DIFF,
1570                                  sb.buf, sb.len, 0);
1571         strbuf_release(&sb);
1572         return 0;
1573 }
1574
1575 /*
1576  * '--color-words' algorithm can be described as:
1577  *
1578  *   1. collect the minus/plus lines of a diff hunk, divided into
1579  *      minus-lines and plus-lines;
1580  *
1581  *   2. break both minus-lines and plus-lines into words and
1582  *      place them into two mmfile_t with one word for each line;
1583  *
1584  *   3. use xdiff to run diff on the two mmfile_t to get the words level diff;
1585  *
1586  * And for the common parts of the both file, we output the plus side text.
1587  * diff_words->current_plus is used to trace the current position of the plus file
1588  * which printed. diff_words->last_minus is used to trace the last minus word
1589  * printed.
1590  *
1591  * For '--graph' to work with '--color-words', we need to output the graph prefix
1592  * on each line of color words output. Generally, there are two conditions on
1593  * which we should output the prefix.
1594  *
1595  *   1. diff_words->last_minus == 0 &&
1596  *      diff_words->current_plus == diff_words->plus.text.ptr
1597  *
1598  *      that is: the plus text must start as a new line, and if there is no minus
1599  *      word printed, a graph prefix must be printed.
1600  *
1601  *   2. diff_words->current_plus > diff_words->plus.text.ptr &&
1602  *      *(diff_words->current_plus - 1) == '\n'
1603  *
1604  *      that is: a graph prefix must be printed following a '\n'
1605  */
1606 static int color_words_output_graph_prefix(struct diff_words_data *diff_words)
1607 {
1608         if ((diff_words->last_minus == 0 &&
1609                 diff_words->current_plus == diff_words->plus.text.ptr) ||
1610                 (diff_words->current_plus > diff_words->plus.text.ptr &&
1611                 *(diff_words->current_plus - 1) == '\n')) {
1612                 return 1;
1613         } else {
1614                 return 0;
1615         }
1616 }
1617
1618 static void fn_out_diff_words_aux(void *priv, char *line, unsigned long len)
1619 {
1620         struct diff_words_data *diff_words = priv;
1621         struct diff_words_style *style = diff_words->style;
1622         int minus_first, minus_len, plus_first, plus_len;
1623         const char *minus_begin, *minus_end, *plus_begin, *plus_end;
1624         struct diff_options *opt = diff_words->opt;
1625         const char *line_prefix;
1626
1627         if (line[0] != '@' || parse_hunk_header(line, len,
1628                         &minus_first, &minus_len, &plus_first, &plus_len))
1629                 return;
1630
1631         assert(opt);
1632         line_prefix = diff_line_prefix(opt);
1633
1634         /* POSIX requires that first be decremented by one if len == 0... */
1635         if (minus_len) {
1636                 minus_begin = diff_words->minus.orig[minus_first].begin;
1637                 minus_end =
1638                         diff_words->minus.orig[minus_first + minus_len - 1].end;
1639         } else
1640                 minus_begin = minus_end =
1641                         diff_words->minus.orig[minus_first].end;
1642
1643         if (plus_len) {
1644                 plus_begin = diff_words->plus.orig[plus_first].begin;
1645                 plus_end = diff_words->plus.orig[plus_first + plus_len - 1].end;
1646         } else
1647                 plus_begin = plus_end = diff_words->plus.orig[plus_first].end;
1648
1649         if (color_words_output_graph_prefix(diff_words)) {
1650                 fputs(line_prefix, diff_words->opt->file);
1651         }
1652         if (diff_words->current_plus != plus_begin) {
1653                 fn_out_diff_words_write_helper(diff_words->opt,
1654                                 &style->ctx, style->newline,
1655                                 plus_begin - diff_words->current_plus,
1656                                 diff_words->current_plus);
1657         }
1658         if (minus_begin != minus_end) {
1659                 fn_out_diff_words_write_helper(diff_words->opt,
1660                                 &style->old, style->newline,
1661                                 minus_end - minus_begin, minus_begin);
1662         }
1663         if (plus_begin != plus_end) {
1664                 fn_out_diff_words_write_helper(diff_words->opt,
1665                                 &style->new, style->newline,
1666                                 plus_end - plus_begin, plus_begin);
1667         }
1668
1669         diff_words->current_plus = plus_end;
1670         diff_words->last_minus = minus_first;
1671 }
1672
1673 /* This function starts looking at *begin, and returns 0 iff a word was found. */
1674 static int find_word_boundaries(mmfile_t *buffer, regex_t *word_regex,
1675                 int *begin, int *end)
1676 {
1677         if (word_regex && *begin < buffer->size) {
1678                 regmatch_t match[1];
1679                 if (!regexec_buf(word_regex, buffer->ptr + *begin,
1680                                  buffer->size - *begin, 1, match, 0)) {
1681                         char *p = memchr(buffer->ptr + *begin + match[0].rm_so,
1682                                         '\n', match[0].rm_eo - match[0].rm_so);
1683                         *end = p ? p - buffer->ptr : match[0].rm_eo + *begin;
1684                         *begin += match[0].rm_so;
1685                         return *begin >= *end;
1686                 }
1687                 return -1;
1688         }
1689
1690         /* find the next word */
1691         while (*begin < buffer->size && isspace(buffer->ptr[*begin]))
1692                 (*begin)++;
1693         if (*begin >= buffer->size)
1694                 return -1;
1695
1696         /* find the end of the word */
1697         *end = *begin + 1;
1698         while (*end < buffer->size && !isspace(buffer->ptr[*end]))
1699                 (*end)++;
1700
1701         return 0;
1702 }
1703
1704 /*
1705  * This function splits the words in buffer->text, stores the list with
1706  * newline separator into out, and saves the offsets of the original words
1707  * in buffer->orig.
1708  */
1709 static void diff_words_fill(struct diff_words_buffer *buffer, mmfile_t *out,
1710                 regex_t *word_regex)
1711 {
1712         int i, j;
1713         long alloc = 0;
1714
1715         out->size = 0;
1716         out->ptr = NULL;
1717
1718         /* fake an empty "0th" word */
1719         ALLOC_GROW(buffer->orig, 1, buffer->orig_alloc);
1720         buffer->orig[0].begin = buffer->orig[0].end = buffer->text.ptr;
1721         buffer->orig_nr = 1;
1722
1723         for (i = 0; i < buffer->text.size; i++) {
1724                 if (find_word_boundaries(&buffer->text, word_regex, &i, &j))
1725                         return;
1726
1727                 /* store original boundaries */
1728                 ALLOC_GROW(buffer->orig, buffer->orig_nr + 1,
1729                                 buffer->orig_alloc);
1730                 buffer->orig[buffer->orig_nr].begin = buffer->text.ptr + i;
1731                 buffer->orig[buffer->orig_nr].end = buffer->text.ptr + j;
1732                 buffer->orig_nr++;
1733
1734                 /* store one word */
1735                 ALLOC_GROW(out->ptr, out->size + j - i + 1, alloc);
1736                 memcpy(out->ptr + out->size, buffer->text.ptr + i, j - i);
1737                 out->ptr[out->size + j - i] = '\n';
1738                 out->size += j - i + 1;
1739
1740                 i = j - 1;
1741         }
1742 }
1743
1744 /* this executes the word diff on the accumulated buffers */
1745 static void diff_words_show(struct diff_words_data *diff_words)
1746 {
1747         xpparam_t xpp;
1748         xdemitconf_t xecfg;
1749         mmfile_t minus, plus;
1750         struct diff_words_style *style = diff_words->style;
1751
1752         struct diff_options *opt = diff_words->opt;
1753         const char *line_prefix;
1754
1755         assert(opt);
1756         line_prefix = diff_line_prefix(opt);
1757
1758         /* special case: only removal */
1759         if (!diff_words->plus.text.size) {
1760                 emit_diff_symbol(diff_words->opt, DIFF_SYMBOL_WORD_DIFF,
1761                                  line_prefix, strlen(line_prefix), 0);
1762                 fn_out_diff_words_write_helper(diff_words->opt,
1763                         &style->old, style->newline,
1764                         diff_words->minus.text.size,
1765                         diff_words->minus.text.ptr);
1766                 diff_words->minus.text.size = 0;
1767                 return;
1768         }
1769
1770         diff_words->current_plus = diff_words->plus.text.ptr;
1771         diff_words->last_minus = 0;
1772
1773         memset(&xpp, 0, sizeof(xpp));
1774         memset(&xecfg, 0, sizeof(xecfg));
1775         diff_words_fill(&diff_words->minus, &minus, diff_words->word_regex);
1776         diff_words_fill(&diff_words->plus, &plus, diff_words->word_regex);
1777         xpp.flags = 0;
1778         /* as only the hunk header will be parsed, we need a 0-context */
1779         xecfg.ctxlen = 0;
1780         if (xdi_diff_outf(&minus, &plus, fn_out_diff_words_aux, diff_words,
1781                           &xpp, &xecfg))
1782                 die("unable to generate word diff");
1783         free(minus.ptr);
1784         free(plus.ptr);
1785         if (diff_words->current_plus != diff_words->plus.text.ptr +
1786                         diff_words->plus.text.size) {
1787                 if (color_words_output_graph_prefix(diff_words))
1788                         emit_diff_symbol(diff_words->opt, DIFF_SYMBOL_WORD_DIFF,
1789                                          line_prefix, strlen(line_prefix), 0);
1790                 fn_out_diff_words_write_helper(diff_words->opt,
1791                         &style->ctx, style->newline,
1792                         diff_words->plus.text.ptr + diff_words->plus.text.size
1793                         - diff_words->current_plus, diff_words->current_plus);
1794         }
1795         diff_words->minus.text.size = diff_words->plus.text.size = 0;
1796 }
1797
1798 /* In "color-words" mode, show word-diff of words accumulated in the buffer */
1799 static void diff_words_flush(struct emit_callback *ecbdata)
1800 {
1801         struct diff_options *wo = ecbdata->diff_words->opt;
1802
1803         if (ecbdata->diff_words->minus.text.size ||
1804             ecbdata->diff_words->plus.text.size)
1805                 diff_words_show(ecbdata->diff_words);
1806
1807         if (wo->emitted_symbols) {
1808                 struct diff_options *o = ecbdata->opt;
1809                 struct emitted_diff_symbols *wol = wo->emitted_symbols;
1810                 int i;
1811
1812                 /*
1813                  * NEEDSWORK:
1814                  * Instead of appending each, concat all words to a line?
1815                  */
1816                 for (i = 0; i < wol->nr; i++)
1817                         append_emitted_diff_symbol(o, &wol->buf[i]);
1818
1819                 for (i = 0; i < wol->nr; i++)
1820                         free((void *)wol->buf[i].line);
1821
1822                 wol->nr = 0;
1823         }
1824 }
1825
1826 static void diff_filespec_load_driver(struct diff_filespec *one)
1827 {
1828         /* Use already-loaded driver */
1829         if (one->driver)
1830                 return;
1831
1832         if (S_ISREG(one->mode))
1833                 one->driver = userdiff_find_by_path(one->path);
1834
1835         /* Fallback to default settings */
1836         if (!one->driver)
1837                 one->driver = userdiff_find_by_name("default");
1838 }
1839
1840 static const char *userdiff_word_regex(struct diff_filespec *one)
1841 {
1842         diff_filespec_load_driver(one);
1843         return one->driver->word_regex;
1844 }
1845
1846 static void init_diff_words_data(struct emit_callback *ecbdata,
1847                                  struct diff_options *orig_opts,
1848                                  struct diff_filespec *one,
1849                                  struct diff_filespec *two)
1850 {
1851         int i;
1852         struct diff_options *o = xmalloc(sizeof(struct diff_options));
1853         memcpy(o, orig_opts, sizeof(struct diff_options));
1854
1855         ecbdata->diff_words =
1856                 xcalloc(1, sizeof(struct diff_words_data));
1857         ecbdata->diff_words->type = o->word_diff;
1858         ecbdata->diff_words->opt = o;
1859
1860         if (orig_opts->emitted_symbols)
1861                 o->emitted_symbols =
1862                         xcalloc(1, sizeof(struct emitted_diff_symbols));
1863
1864         if (!o->word_regex)
1865                 o->word_regex = userdiff_word_regex(one);
1866         if (!o->word_regex)
1867                 o->word_regex = userdiff_word_regex(two);
1868         if (!o->word_regex)
1869                 o->word_regex = diff_word_regex_cfg;
1870         if (o->word_regex) {
1871                 ecbdata->diff_words->word_regex = (regex_t *)
1872                         xmalloc(sizeof(regex_t));
1873                 if (regcomp(ecbdata->diff_words->word_regex,
1874                             o->word_regex,
1875                             REG_EXTENDED | REG_NEWLINE))
1876                         die ("Invalid regular expression: %s",
1877                              o->word_regex);
1878         }
1879         for (i = 0; i < ARRAY_SIZE(diff_words_styles); i++) {
1880                 if (o->word_diff == diff_words_styles[i].type) {
1881                         ecbdata->diff_words->style =
1882                                 &diff_words_styles[i];
1883                         break;
1884                 }
1885         }
1886         if (want_color(o->use_color)) {
1887                 struct diff_words_style *st = ecbdata->diff_words->style;
1888                 st->old.color = diff_get_color_opt(o, DIFF_FILE_OLD);
1889                 st->new.color = diff_get_color_opt(o, DIFF_FILE_NEW);
1890                 st->ctx.color = diff_get_color_opt(o, DIFF_CONTEXT);
1891         }
1892 }
1893
1894 static void free_diff_words_data(struct emit_callback *ecbdata)
1895 {
1896         if (ecbdata->diff_words) {
1897                 diff_words_flush(ecbdata);
1898                 free (ecbdata->diff_words->opt->emitted_symbols);
1899                 free (ecbdata->diff_words->opt);
1900                 free (ecbdata->diff_words->minus.text.ptr);
1901                 free (ecbdata->diff_words->minus.orig);
1902                 free (ecbdata->diff_words->plus.text.ptr);
1903                 free (ecbdata->diff_words->plus.orig);
1904                 if (ecbdata->diff_words->word_regex) {
1905                         regfree(ecbdata->diff_words->word_regex);
1906                         free(ecbdata->diff_words->word_regex);
1907                 }
1908                 FREE_AND_NULL(ecbdata->diff_words);
1909         }
1910 }
1911
1912 const char *diff_get_color(int diff_use_color, enum color_diff ix)
1913 {
1914         if (want_color(diff_use_color))
1915                 return diff_colors[ix];
1916         return "";
1917 }
1918
1919 const char *diff_line_prefix(struct diff_options *opt)
1920 {
1921         struct strbuf *msgbuf;
1922         if (!opt->output_prefix)
1923                 return "";
1924
1925         msgbuf = opt->output_prefix(opt, opt->output_prefix_data);
1926         return msgbuf->buf;
1927 }
1928
1929 static unsigned long sane_truncate_line(struct emit_callback *ecb, char *line, unsigned long len)
1930 {
1931         const char *cp;
1932         unsigned long allot;
1933         size_t l = len;
1934
1935         cp = line;
1936         allot = l;
1937         while (0 < l) {
1938                 (void) utf8_width(&cp, &l);
1939                 if (!cp)
1940                         break; /* truncated in the middle? */
1941         }
1942         return allot - l;
1943 }
1944
1945 static void find_lno(const char *line, struct emit_callback *ecbdata)
1946 {
1947         const char *p;
1948         ecbdata->lno_in_preimage = 0;
1949         ecbdata->lno_in_postimage = 0;
1950         p = strchr(line, '-');
1951         if (!p)
1952                 return; /* cannot happen */
1953         ecbdata->lno_in_preimage = strtol(p + 1, NULL, 10);
1954         p = strchr(p, '+');
1955         if (!p)
1956                 return; /* cannot happen */
1957         ecbdata->lno_in_postimage = strtol(p + 1, NULL, 10);
1958 }
1959
1960 static void fn_out_consume(void *priv, char *line, unsigned long len)
1961 {
1962         struct emit_callback *ecbdata = priv;
1963         const char *reset = diff_get_color(ecbdata->color_diff, DIFF_RESET);
1964         struct diff_options *o = ecbdata->opt;
1965
1966         o->found_changes = 1;
1967
1968         if (ecbdata->header) {
1969                 emit_diff_symbol(o, DIFF_SYMBOL_HEADER,
1970                                  ecbdata->header->buf, ecbdata->header->len, 0);
1971                 strbuf_reset(ecbdata->header);
1972                 ecbdata->header = NULL;
1973         }
1974
1975         if (ecbdata->label_path[0]) {
1976                 emit_diff_symbol(o, DIFF_SYMBOL_FILEPAIR_MINUS,
1977                                  ecbdata->label_path[0],
1978                                  strlen(ecbdata->label_path[0]), 0);
1979                 emit_diff_symbol(o, DIFF_SYMBOL_FILEPAIR_PLUS,
1980                                  ecbdata->label_path[1],
1981                                  strlen(ecbdata->label_path[1]), 0);
1982                 ecbdata->label_path[0] = ecbdata->label_path[1] = NULL;
1983         }
1984
1985         if (diff_suppress_blank_empty
1986             && len == 2 && line[0] == ' ' && line[1] == '\n') {
1987                 line[0] = '\n';
1988                 len = 1;
1989         }
1990
1991         if (line[0] == '@') {
1992                 if (ecbdata->diff_words)
1993                         diff_words_flush(ecbdata);
1994                 len = sane_truncate_line(ecbdata, line, len);
1995                 find_lno(line, ecbdata);
1996                 emit_hunk_header(ecbdata, line, len);
1997                 return;
1998         }
1999
2000         if (ecbdata->diff_words) {
2001                 enum diff_symbol s =
2002                         ecbdata->diff_words->type == DIFF_WORDS_PORCELAIN ?
2003                         DIFF_SYMBOL_WORDS_PORCELAIN : DIFF_SYMBOL_WORDS;
2004                 if (line[0] == '-') {
2005                         diff_words_append(line, len,
2006                                           &ecbdata->diff_words->minus);
2007                         return;
2008                 } else if (line[0] == '+') {
2009                         diff_words_append(line, len,
2010                                           &ecbdata->diff_words->plus);
2011                         return;
2012                 } else if (starts_with(line, "\\ ")) {
2013                         /*
2014                          * Eat the "no newline at eof" marker as if we
2015                          * saw a "+" or "-" line with nothing on it,
2016                          * and return without diff_words_flush() to
2017                          * defer processing. If this is the end of
2018                          * preimage, more "+" lines may come after it.
2019                          */
2020                         return;
2021                 }
2022                 diff_words_flush(ecbdata);
2023                 emit_diff_symbol(o, s, line, len, 0);
2024                 return;
2025         }
2026
2027         switch (line[0]) {
2028         case '+':
2029                 ecbdata->lno_in_postimage++;
2030                 emit_add_line(reset, ecbdata, line + 1, len - 1);
2031                 break;
2032         case '-':
2033                 ecbdata->lno_in_preimage++;
2034                 emit_del_line(reset, ecbdata, line + 1, len - 1);
2035                 break;
2036         case ' ':
2037                 ecbdata->lno_in_postimage++;
2038                 ecbdata->lno_in_preimage++;
2039                 emit_context_line(reset, ecbdata, line + 1, len - 1);
2040                 break;
2041         default:
2042                 /* incomplete line at the end */
2043                 ecbdata->lno_in_preimage++;
2044                 emit_diff_symbol(o, DIFF_SYMBOL_CONTEXT_INCOMPLETE,
2045                                  line, len, 0);
2046                 break;
2047         }
2048 }
2049
2050 static char *pprint_rename(const char *a, const char *b)
2051 {
2052         const char *old = a;
2053         const char *new = b;
2054         struct strbuf name = STRBUF_INIT;
2055         int pfx_length, sfx_length;
2056         int pfx_adjust_for_slash;
2057         int len_a = strlen(a);
2058         int len_b = strlen(b);
2059         int a_midlen, b_midlen;
2060         int qlen_a = quote_c_style(a, NULL, NULL, 0);
2061         int qlen_b = quote_c_style(b, NULL, NULL, 0);
2062
2063         if (qlen_a || qlen_b) {
2064                 quote_c_style(a, &name, NULL, 0);
2065                 strbuf_addstr(&name, " => ");
2066                 quote_c_style(b, &name, NULL, 0);
2067                 return strbuf_detach(&name, NULL);
2068         }
2069
2070         /* Find common prefix */
2071         pfx_length = 0;
2072         while (*old && *new && *old == *new) {
2073                 if (*old == '/')
2074                         pfx_length = old - a + 1;
2075                 old++;
2076                 new++;
2077         }
2078
2079         /* Find common suffix */
2080         old = a + len_a;
2081         new = b + len_b;
2082         sfx_length = 0;
2083         /*
2084          * If there is a common prefix, it must end in a slash.  In
2085          * that case we let this loop run 1 into the prefix to see the
2086          * same slash.
2087          *
2088          * If there is no common prefix, we cannot do this as it would
2089          * underrun the input strings.
2090          */
2091         pfx_adjust_for_slash = (pfx_length ? 1 : 0);
2092         while (a + pfx_length - pfx_adjust_for_slash <= old &&
2093                b + pfx_length - pfx_adjust_for_slash <= new &&
2094                *old == *new) {
2095                 if (*old == '/')
2096                         sfx_length = len_a - (old - a);
2097                 old--;
2098                 new--;
2099         }
2100
2101         /*
2102          * pfx{mid-a => mid-b}sfx
2103          * {pfx-a => pfx-b}sfx
2104          * pfx{sfx-a => sfx-b}
2105          * name-a => name-b
2106          */
2107         a_midlen = len_a - pfx_length - sfx_length;
2108         b_midlen = len_b - pfx_length - sfx_length;
2109         if (a_midlen < 0)
2110                 a_midlen = 0;
2111         if (b_midlen < 0)
2112                 b_midlen = 0;
2113
2114         strbuf_grow(&name, pfx_length + a_midlen + b_midlen + sfx_length + 7);
2115         if (pfx_length + sfx_length) {
2116                 strbuf_add(&name, a, pfx_length);
2117                 strbuf_addch(&name, '{');
2118         }
2119         strbuf_add(&name, a + pfx_length, a_midlen);
2120         strbuf_addstr(&name, " => ");
2121         strbuf_add(&name, b + pfx_length, b_midlen);
2122         if (pfx_length + sfx_length) {
2123                 strbuf_addch(&name, '}');
2124                 strbuf_add(&name, a + len_a - sfx_length, sfx_length);
2125         }
2126         return strbuf_detach(&name, NULL);
2127 }
2128
2129 struct diffstat_t {
2130         int nr;
2131         int alloc;
2132         struct diffstat_file {
2133                 char *from_name;
2134                 char *name;
2135                 char *print_name;
2136                 unsigned is_unmerged:1;
2137                 unsigned is_binary:1;
2138                 unsigned is_renamed:1;
2139                 unsigned is_interesting:1;
2140                 uintmax_t added, deleted;
2141         } **files;
2142 };
2143
2144 static struct diffstat_file *diffstat_add(struct diffstat_t *diffstat,
2145                                           const char *name_a,
2146                                           const char *name_b)
2147 {
2148         struct diffstat_file *x;
2149         x = xcalloc(1, sizeof(*x));
2150         ALLOC_GROW(diffstat->files, diffstat->nr + 1, diffstat->alloc);
2151         diffstat->files[diffstat->nr++] = x;
2152         if (name_b) {
2153                 x->from_name = xstrdup(name_a);
2154                 x->name = xstrdup(name_b);
2155                 x->is_renamed = 1;
2156         }
2157         else {
2158                 x->from_name = NULL;
2159                 x->name = xstrdup(name_a);
2160         }
2161         return x;
2162 }
2163
2164 static void diffstat_consume(void *priv, char *line, unsigned long len)
2165 {
2166         struct diffstat_t *diffstat = priv;
2167         struct diffstat_file *x = diffstat->files[diffstat->nr - 1];
2168
2169         if (line[0] == '+')
2170                 x->added++;
2171         else if (line[0] == '-')
2172                 x->deleted++;
2173 }
2174
2175 const char mime_boundary_leader[] = "------------";
2176
2177 static int scale_linear(int it, int width, int max_change)
2178 {
2179         if (!it)
2180                 return 0;
2181         /*
2182          * make sure that at least one '-' or '+' is printed if
2183          * there is any change to this path. The easiest way is to
2184          * scale linearly as if the alloted width is one column shorter
2185          * than it is, and then add 1 to the result.
2186          */
2187         return 1 + (it * (width - 1) / max_change);
2188 }
2189
2190 static void show_graph(struct strbuf *out, char ch, int cnt,
2191                        const char *set, const char *reset)
2192 {
2193         if (cnt <= 0)
2194                 return;
2195         strbuf_addstr(out, set);
2196         strbuf_addchars(out, ch, cnt);
2197         strbuf_addstr(out, reset);
2198 }
2199
2200 static void fill_print_name(struct diffstat_file *file)
2201 {
2202         char *pname;
2203
2204         if (file->print_name)
2205                 return;
2206
2207         if (!file->is_renamed) {
2208                 struct strbuf buf = STRBUF_INIT;
2209                 if (quote_c_style(file->name, &buf, NULL, 0)) {
2210                         pname = strbuf_detach(&buf, NULL);
2211                 } else {
2212                         pname = file->name;
2213                         strbuf_release(&buf);
2214                 }
2215         } else {
2216                 pname = pprint_rename(file->from_name, file->name);
2217         }
2218         file->print_name = pname;
2219 }
2220
2221 static void print_stat_summary_inserts_deletes(struct diff_options *options,
2222                 int files, int insertions, int deletions)
2223 {
2224         struct strbuf sb = STRBUF_INIT;
2225
2226         if (!files) {
2227                 assert(insertions == 0 && deletions == 0);
2228                 emit_diff_symbol(options, DIFF_SYMBOL_STATS_SUMMARY_NO_FILES,
2229                                  NULL, 0, 0);
2230                 return;
2231         }
2232
2233         strbuf_addf(&sb,
2234                     (files == 1) ? " %d file changed" : " %d files changed",
2235                     files);
2236
2237         /*
2238          * For binary diff, the caller may want to print "x files
2239          * changed" with insertions == 0 && deletions == 0.
2240          *
2241          * Not omitting "0 insertions(+), 0 deletions(-)" in this case
2242          * is probably less confusing (i.e skip over "2 files changed
2243          * but nothing about added/removed lines? Is this a bug in Git?").
2244          */
2245         if (insertions || deletions == 0) {
2246                 strbuf_addf(&sb,
2247                             (insertions == 1) ? ", %d insertion(+)" : ", %d insertions(+)",
2248                             insertions);
2249         }
2250
2251         if (deletions || insertions == 0) {
2252                 strbuf_addf(&sb,
2253                             (deletions == 1) ? ", %d deletion(-)" : ", %d deletions(-)",
2254                             deletions);
2255         }
2256         strbuf_addch(&sb, '\n');
2257         emit_diff_symbol(options, DIFF_SYMBOL_STATS_SUMMARY_INSERTS_DELETES,
2258                          sb.buf, sb.len, 0);
2259         strbuf_release(&sb);
2260 }
2261
2262 void print_stat_summary(FILE *fp, int files,
2263                         int insertions, int deletions)
2264 {
2265         struct diff_options o;
2266         memset(&o, 0, sizeof(o));
2267         o.file = fp;
2268
2269         print_stat_summary_inserts_deletes(&o, files, insertions, deletions);
2270 }
2271
2272 static void show_stats(struct diffstat_t *data, struct diff_options *options)
2273 {
2274         int i, len, add, del, adds = 0, dels = 0;
2275         uintmax_t max_change = 0, max_len = 0;
2276         int total_files = data->nr, count;
2277         int width, name_width, graph_width, number_width = 0, bin_width = 0;
2278         const char *reset, *add_c, *del_c;
2279         int extra_shown = 0;
2280         const char *line_prefix = diff_line_prefix(options);
2281         struct strbuf out = STRBUF_INIT;
2282
2283         if (data->nr == 0)
2284                 return;
2285
2286         count = options->stat_count ? options->stat_count : data->nr;
2287
2288         reset = diff_get_color_opt(options, DIFF_RESET);
2289         add_c = diff_get_color_opt(options, DIFF_FILE_NEW);
2290         del_c = diff_get_color_opt(options, DIFF_FILE_OLD);
2291
2292         /*
2293          * Find the longest filename and max number of changes
2294          */
2295         for (i = 0; (i < count) && (i < data->nr); i++) {
2296                 struct diffstat_file *file = data->files[i];
2297                 uintmax_t change = file->added + file->deleted;
2298
2299                 if (!file->is_interesting && (change == 0)) {
2300                         count++; /* not shown == room for one more */
2301                         continue;
2302                 }
2303                 fill_print_name(file);
2304                 len = strlen(file->print_name);
2305                 if (max_len < len)
2306                         max_len = len;
2307
2308                 if (file->is_unmerged) {
2309                         /* "Unmerged" is 8 characters */
2310                         bin_width = bin_width < 8 ? 8 : bin_width;
2311                         continue;
2312                 }
2313                 if (file->is_binary) {
2314                         /* "Bin XXX -> YYY bytes" */
2315                         int w = 14 + decimal_width(file->added)
2316                                 + decimal_width(file->deleted);
2317                         bin_width = bin_width < w ? w : bin_width;
2318                         /* Display change counts aligned with "Bin" */
2319                         number_width = 3;
2320                         continue;
2321                 }
2322
2323                 if (max_change < change)
2324                         max_change = change;
2325         }
2326         count = i; /* where we can stop scanning in data->files[] */
2327
2328         /*
2329          * We have width = stat_width or term_columns() columns total.
2330          * We want a maximum of min(max_len, stat_name_width) for the name part.
2331          * We want a maximum of min(max_change, stat_graph_width) for the +- part.
2332          * We also need 1 for " " and 4 + decimal_width(max_change)
2333          * for " | NNNN " and one the empty column at the end, altogether
2334          * 6 + decimal_width(max_change).
2335          *
2336          * If there's not enough space, we will use the smaller of
2337          * stat_name_width (if set) and 5/8*width for the filename,
2338          * and the rest for constant elements + graph part, but no more
2339          * than stat_graph_width for the graph part.
2340          * (5/8 gives 50 for filename and 30 for the constant parts + graph
2341          * for the standard terminal size).
2342          *
2343          * In other words: stat_width limits the maximum width, and
2344          * stat_name_width fixes the maximum width of the filename,
2345          * and is also used to divide available columns if there
2346          * aren't enough.
2347          *
2348          * Binary files are displayed with "Bin XXX -> YYY bytes"
2349          * instead of the change count and graph. This part is treated
2350          * similarly to the graph part, except that it is not
2351          * "scaled". If total width is too small to accommodate the
2352          * guaranteed minimum width of the filename part and the
2353          * separators and this message, this message will "overflow"
2354          * making the line longer than the maximum width.
2355          */
2356
2357         if (options->stat_width == -1)
2358                 width = term_columns() - strlen(line_prefix);
2359         else
2360                 width = options->stat_width ? options->stat_width : 80;
2361         number_width = decimal_width(max_change) > number_width ?
2362                 decimal_width(max_change) : number_width;
2363
2364         if (options->stat_graph_width == -1)
2365                 options->stat_graph_width = diff_stat_graph_width;
2366
2367         /*
2368          * Guarantee 3/8*16==6 for the graph part
2369          * and 5/8*16==10 for the filename part
2370          */
2371         if (width < 16 + 6 + number_width)
2372                 width = 16 + 6 + number_width;
2373
2374         /*
2375          * First assign sizes that are wanted, ignoring available width.
2376          * strlen("Bin XXX -> YYY bytes") == bin_width, and the part
2377          * starting from "XXX" should fit in graph_width.
2378          */
2379         graph_width = max_change + 4 > bin_width ? max_change : bin_width - 4;
2380         if (options->stat_graph_width &&
2381             options->stat_graph_width < graph_width)
2382                 graph_width = options->stat_graph_width;
2383
2384         name_width = (options->stat_name_width > 0 &&
2385                       options->stat_name_width < max_len) ?
2386                 options->stat_name_width : max_len;
2387
2388         /*
2389          * Adjust adjustable widths not to exceed maximum width
2390          */
2391         if (name_width + number_width + 6 + graph_width > width) {
2392                 if (graph_width > width * 3/8 - number_width - 6) {
2393                         graph_width = width * 3/8 - number_width - 6;
2394                         if (graph_width < 6)
2395                                 graph_width = 6;
2396                 }
2397
2398                 if (options->stat_graph_width &&
2399                     graph_width > options->stat_graph_width)
2400                         graph_width = options->stat_graph_width;
2401                 if (name_width > width - number_width - 6 - graph_width)
2402                         name_width = width - number_width - 6 - graph_width;
2403                 else
2404                         graph_width = width - number_width - 6 - name_width;
2405         }
2406
2407         /*
2408          * From here name_width is the width of the name area,
2409          * and graph_width is the width of the graph area.
2410          * max_change is used to scale graph properly.
2411          */
2412         for (i = 0; i < count; i++) {
2413                 const char *prefix = "";
2414                 struct diffstat_file *file = data->files[i];
2415                 char *name = file->print_name;
2416                 uintmax_t added = file->added;
2417                 uintmax_t deleted = file->deleted;
2418                 int name_len;
2419
2420                 if (!file->is_interesting && (added + deleted == 0))
2421                         continue;
2422
2423                 /*
2424                  * "scale" the filename
2425                  */
2426                 len = name_width;
2427                 name_len = strlen(name);
2428                 if (name_width < name_len) {
2429                         char *slash;
2430                         prefix = "...";
2431                         len -= 3;
2432                         name += name_len - len;
2433                         slash = strchr(name, '/');
2434                         if (slash)
2435                                 name = slash;
2436                 }
2437
2438                 if (file->is_binary) {
2439                         strbuf_addf(&out, " %s%-*s |", prefix, len, name);
2440                         strbuf_addf(&out, " %*s", number_width, "Bin");
2441                         if (!added && !deleted) {
2442                                 strbuf_addch(&out, '\n');
2443                                 emit_diff_symbol(options, DIFF_SYMBOL_STATS_LINE,
2444                                                  out.buf, out.len, 0);
2445                                 strbuf_reset(&out);
2446                                 continue;
2447                         }
2448                         strbuf_addf(&out, " %s%"PRIuMAX"%s",
2449                                 del_c, deleted, reset);
2450                         strbuf_addstr(&out, " -> ");
2451                         strbuf_addf(&out, "%s%"PRIuMAX"%s",
2452                                 add_c, added, reset);
2453                         strbuf_addstr(&out, " bytes\n");
2454                         emit_diff_symbol(options, DIFF_SYMBOL_STATS_LINE,
2455                                          out.buf, out.len, 0);
2456                         strbuf_reset(&out);
2457                         continue;
2458                 }
2459                 else if (file->is_unmerged) {
2460                         strbuf_addf(&out, " %s%-*s |", prefix, len, name);
2461                         strbuf_addstr(&out, " Unmerged\n");
2462                         emit_diff_symbol(options, DIFF_SYMBOL_STATS_LINE,
2463                                          out.buf, out.len, 0);
2464                         strbuf_reset(&out);
2465                         continue;
2466                 }
2467
2468                 /*
2469                  * scale the add/delete
2470                  */
2471                 add = added;
2472                 del = deleted;
2473
2474                 if (graph_width <= max_change) {
2475                         int total = scale_linear(add + del, graph_width, max_change);
2476                         if (total < 2 && add && del)
2477                                 /* width >= 2 due to the sanity check */
2478                                 total = 2;
2479                         if (add < del) {
2480                                 add = scale_linear(add, graph_width, max_change);
2481                                 del = total - add;
2482                         } else {
2483                                 del = scale_linear(del, graph_width, max_change);
2484                                 add = total - del;
2485                         }
2486                 }
2487                 strbuf_addf(&out, " %s%-*s |", prefix, len, name);
2488                 strbuf_addf(&out, " %*"PRIuMAX"%s",
2489                         number_width, added + deleted,
2490                         added + deleted ? " " : "");
2491                 show_graph(&out, '+', add, add_c, reset);
2492                 show_graph(&out, '-', del, del_c, reset);
2493                 strbuf_addch(&out, '\n');
2494                 emit_diff_symbol(options, DIFF_SYMBOL_STATS_LINE,
2495                                  out.buf, out.len, 0);
2496                 strbuf_reset(&out);
2497         }
2498
2499         for (i = 0; i < data->nr; i++) {
2500                 struct diffstat_file *file = data->files[i];
2501                 uintmax_t added = file->added;
2502                 uintmax_t deleted = file->deleted;
2503
2504                 if (file->is_unmerged ||
2505                     (!file->is_interesting && (added + deleted == 0))) {
2506                         total_files--;
2507                         continue;
2508                 }
2509
2510                 if (!file->is_binary) {
2511                         adds += added;
2512                         dels += deleted;
2513                 }
2514                 if (i < count)
2515                         continue;
2516                 if (!extra_shown)
2517                         emit_diff_symbol(options,
2518                                          DIFF_SYMBOL_STATS_SUMMARY_ABBREV,
2519                                          NULL, 0, 0);
2520                 extra_shown = 1;
2521         }
2522
2523         print_stat_summary_inserts_deletes(options, total_files, adds, dels);
2524         strbuf_release(&out);
2525 }
2526
2527 static void show_shortstats(struct diffstat_t *data, struct diff_options *options)
2528 {
2529         int i, adds = 0, dels = 0, total_files = data->nr;
2530
2531         if (data->nr == 0)
2532                 return;
2533
2534         for (i = 0; i < data->nr; i++) {
2535                 int added = data->files[i]->added;
2536                 int deleted = data->files[i]->deleted;
2537
2538                 if (data->files[i]->is_unmerged ||
2539                     (!data->files[i]->is_interesting && (added + deleted == 0))) {
2540                         total_files--;
2541                 } else if (!data->files[i]->is_binary) { /* don't count bytes */
2542                         adds += added;
2543                         dels += deleted;
2544                 }
2545         }
2546         print_stat_summary_inserts_deletes(options, total_files, adds, dels);
2547 }
2548
2549 static void show_numstat(struct diffstat_t *data, struct diff_options *options)
2550 {
2551         int i;
2552
2553         if (data->nr == 0)
2554                 return;
2555
2556         for (i = 0; i < data->nr; i++) {
2557                 struct diffstat_file *file = data->files[i];
2558
2559                 fprintf(options->file, "%s", diff_line_prefix(options));
2560
2561                 if (file->is_binary)
2562                         fprintf(options->file, "-\t-\t");
2563                 else
2564                         fprintf(options->file,
2565                                 "%"PRIuMAX"\t%"PRIuMAX"\t",
2566                                 file->added, file->deleted);
2567                 if (options->line_termination) {
2568                         fill_print_name(file);
2569                         if (!file->is_renamed)
2570                                 write_name_quoted(file->name, options->file,
2571                                                   options->line_termination);
2572                         else {
2573                                 fputs(file->print_name, options->file);
2574                                 putc(options->line_termination, options->file);
2575                         }
2576                 } else {
2577                         if (file->is_renamed) {
2578                                 putc('\0', options->file);
2579                                 write_name_quoted(file->from_name, options->file, '\0');
2580                         }
2581                         write_name_quoted(file->name, options->file, '\0');
2582                 }
2583         }
2584 }
2585
2586 struct dirstat_file {
2587         const char *name;
2588         unsigned long changed;
2589 };
2590
2591 struct dirstat_dir {
2592         struct dirstat_file *files;
2593         int alloc, nr, permille, cumulative;
2594 };
2595
2596 static long gather_dirstat(struct diff_options *opt, struct dirstat_dir *dir,
2597                 unsigned long changed, const char *base, int baselen)
2598 {
2599         unsigned long this_dir = 0;
2600         unsigned int sources = 0;
2601         const char *line_prefix = diff_line_prefix(opt);
2602
2603         while (dir->nr) {
2604                 struct dirstat_file *f = dir->files;
2605                 int namelen = strlen(f->name);
2606                 unsigned long this;
2607                 char *slash;
2608
2609                 if (namelen < baselen)
2610                         break;
2611                 if (memcmp(f->name, base, baselen))
2612                         break;
2613                 slash = strchr(f->name + baselen, '/');
2614                 if (slash) {
2615                         int newbaselen = slash + 1 - f->name;
2616                         this = gather_dirstat(opt, dir, changed, f->name, newbaselen);
2617                         sources++;
2618                 } else {
2619                         this = f->changed;
2620                         dir->files++;
2621                         dir->nr--;
2622                         sources += 2;
2623                 }
2624                 this_dir += this;
2625         }
2626
2627         /*
2628          * We don't report dirstat's for
2629          *  - the top level
2630          *  - or cases where everything came from a single directory
2631          *    under this directory (sources == 1).
2632          */
2633         if (baselen && sources != 1) {
2634                 if (this_dir) {
2635                         int permille = this_dir * 1000 / changed;
2636                         if (permille >= dir->permille) {
2637                                 fprintf(opt->file, "%s%4d.%01d%% %.*s\n", line_prefix,
2638                                         permille / 10, permille % 10, baselen, base);
2639                                 if (!dir->cumulative)
2640                                         return 0;
2641                         }
2642                 }
2643         }
2644         return this_dir;
2645 }
2646
2647 static int dirstat_compare(const void *_a, const void *_b)
2648 {
2649         const struct dirstat_file *a = _a;
2650         const struct dirstat_file *b = _b;
2651         return strcmp(a->name, b->name);
2652 }
2653
2654 static void show_dirstat(struct diff_options *options)
2655 {
2656         int i;
2657         unsigned long changed;
2658         struct dirstat_dir dir;
2659         struct diff_queue_struct *q = &diff_queued_diff;
2660
2661         dir.files = NULL;
2662         dir.alloc = 0;
2663         dir.nr = 0;
2664         dir.permille = options->dirstat_permille;
2665         dir.cumulative = options->flags.dirstat_cumulative;
2666
2667         changed = 0;
2668         for (i = 0; i < q->nr; i++) {
2669                 struct diff_filepair *p = q->queue[i];
2670                 const char *name;
2671                 unsigned long copied, added, damage;
2672                 int content_changed;
2673
2674                 name = p->two->path ? p->two->path : p->one->path;
2675
2676                 if (p->one->oid_valid && p->two->oid_valid)
2677                         content_changed = oidcmp(&p->one->oid, &p->two->oid);
2678                 else
2679                         content_changed = 1;
2680
2681                 if (!content_changed) {
2682                         /*
2683                          * The SHA1 has not changed, so pre-/post-content is
2684                          * identical. We can therefore skip looking at the
2685                          * file contents altogether.
2686                          */
2687                         damage = 0;
2688                         goto found_damage;
2689                 }
2690
2691                 if (options->flags.dirstat_by_file) {
2692                         /*
2693                          * In --dirstat-by-file mode, we don't really need to
2694                          * look at the actual file contents at all.
2695                          * The fact that the SHA1 changed is enough for us to
2696                          * add this file to the list of results
2697                          * (with each file contributing equal damage).
2698                          */
2699                         damage = 1;
2700                         goto found_damage;
2701                 }
2702
2703                 if (DIFF_FILE_VALID(p->one) && DIFF_FILE_VALID(p->two)) {
2704                         diff_populate_filespec(p->one, 0);
2705                         diff_populate_filespec(p->two, 0);
2706                         diffcore_count_changes(p->one, p->two, NULL, NULL,
2707                                                &copied, &added);
2708                         diff_free_filespec_data(p->one);
2709                         diff_free_filespec_data(p->two);
2710                 } else if (DIFF_FILE_VALID(p->one)) {
2711                         diff_populate_filespec(p->one, CHECK_SIZE_ONLY);
2712                         copied = added = 0;
2713                         diff_free_filespec_data(p->one);
2714                 } else if (DIFF_FILE_VALID(p->two)) {
2715                         diff_populate_filespec(p->two, CHECK_SIZE_ONLY);
2716                         copied = 0;
2717                         added = p->two->size;
2718                         diff_free_filespec_data(p->two);
2719                 } else
2720                         continue;
2721
2722                 /*
2723                  * Original minus copied is the removed material,
2724                  * added is the new material.  They are both damages
2725                  * made to the preimage.
2726                  * If the resulting damage is zero, we know that
2727                  * diffcore_count_changes() considers the two entries to
2728                  * be identical, but since content_changed is true, we
2729                  * know that there must have been _some_ kind of change,
2730                  * so we force all entries to have damage > 0.
2731                  */
2732                 damage = (p->one->size - copied) + added;
2733                 if (!damage)
2734                         damage = 1;
2735
2736 found_damage:
2737                 ALLOC_GROW(dir.files, dir.nr + 1, dir.alloc);
2738                 dir.files[dir.nr].name = name;
2739                 dir.files[dir.nr].changed = damage;
2740                 changed += damage;
2741                 dir.nr++;
2742         }
2743
2744         /* This can happen even with many files, if everything was renames */
2745         if (!changed)
2746                 return;
2747
2748         /* Show all directories with more than x% of the changes */
2749         QSORT(dir.files, dir.nr, dirstat_compare);
2750         gather_dirstat(options, &dir, changed, "", 0);
2751 }
2752
2753 static void show_dirstat_by_line(struct diffstat_t *data, struct diff_options *options)
2754 {
2755         int i;
2756         unsigned long changed;
2757         struct dirstat_dir dir;
2758
2759         if (data->nr == 0)
2760                 return;
2761
2762         dir.files = NULL;
2763         dir.alloc = 0;
2764         dir.nr = 0;
2765         dir.permille = options->dirstat_permille;
2766         dir.cumulative = options->flags.dirstat_cumulative;
2767
2768         changed = 0;
2769         for (i = 0; i < data->nr; i++) {
2770                 struct diffstat_file *file = data->files[i];
2771                 unsigned long damage = file->added + file->deleted;
2772                 if (file->is_binary)
2773                         /*
2774                          * binary files counts bytes, not lines. Must find some
2775                          * way to normalize binary bytes vs. textual lines.
2776                          * The following heuristic assumes that there are 64
2777                          * bytes per "line".
2778                          * This is stupid and ugly, but very cheap...
2779                          */
2780                         damage = DIV_ROUND_UP(damage, 64);
2781                 ALLOC_GROW(dir.files, dir.nr + 1, dir.alloc);
2782                 dir.files[dir.nr].name = file->name;
2783                 dir.files[dir.nr].changed = damage;
2784                 changed += damage;
2785                 dir.nr++;
2786         }
2787
2788         /* This can happen even with many files, if everything was renames */
2789         if (!changed)
2790                 return;
2791
2792         /* Show all directories with more than x% of the changes */
2793         QSORT(dir.files, dir.nr, dirstat_compare);
2794         gather_dirstat(options, &dir, changed, "", 0);
2795 }
2796
2797 static void free_diffstat_info(struct diffstat_t *diffstat)
2798 {
2799         int i;
2800         for (i = 0; i < diffstat->nr; i++) {
2801                 struct diffstat_file *f = diffstat->files[i];
2802                 if (f->name != f->print_name)
2803                         free(f->print_name);
2804                 free(f->name);
2805                 free(f->from_name);
2806                 free(f);
2807         }
2808         free(diffstat->files);
2809 }
2810
2811 struct checkdiff_t {
2812         const char *filename;
2813         int lineno;
2814         int conflict_marker_size;
2815         struct diff_options *o;
2816         unsigned ws_rule;
2817         unsigned status;
2818 };
2819
2820 static int is_conflict_marker(const char *line, int marker_size, unsigned long len)
2821 {
2822         char firstchar;
2823         int cnt;
2824
2825         if (len < marker_size + 1)
2826                 return 0;
2827         firstchar = line[0];
2828         switch (firstchar) {
2829         case '=': case '>': case '<': case '|':
2830                 break;
2831         default:
2832                 return 0;
2833         }
2834         for (cnt = 1; cnt < marker_size; cnt++)
2835                 if (line[cnt] != firstchar)
2836                         return 0;
2837         /* line[1] thru line[marker_size-1] are same as firstchar */
2838         if (len < marker_size + 1 || !isspace(line[marker_size]))
2839                 return 0;
2840         return 1;
2841 }
2842
2843 static void checkdiff_consume(void *priv, char *line, unsigned long len)
2844 {
2845         struct checkdiff_t *data = priv;
2846         int marker_size = data->conflict_marker_size;
2847         const char *ws = diff_get_color(data->o->use_color, DIFF_WHITESPACE);
2848         const char *reset = diff_get_color(data->o->use_color, DIFF_RESET);
2849         const char *set = diff_get_color(data->o->use_color, DIFF_FILE_NEW);
2850         char *err;
2851         const char *line_prefix;
2852
2853         assert(data->o);
2854         line_prefix = diff_line_prefix(data->o);
2855
2856         if (line[0] == '+') {
2857                 unsigned bad;
2858                 data->lineno++;
2859                 if (is_conflict_marker(line + 1, marker_size, len - 1)) {
2860                         data->status |= 1;
2861                         fprintf(data->o->file,
2862                                 "%s%s:%d: leftover conflict marker\n",
2863                                 line_prefix, data->filename, data->lineno);
2864                 }
2865                 bad = ws_check(line + 1, len - 1, data->ws_rule);
2866                 if (!bad)
2867                         return;
2868                 data->status |= bad;
2869                 err = whitespace_error_string(bad);
2870                 fprintf(data->o->file, "%s%s:%d: %s.\n",
2871                         line_prefix, data->filename, data->lineno, err);
2872                 free(err);
2873                 emit_line(data->o, set, reset, line, 1);
2874                 ws_check_emit(line + 1, len - 1, data->ws_rule,
2875                               data->o->file, set, reset, ws);
2876         } else if (line[0] == ' ') {
2877                 data->lineno++;
2878         } else if (line[0] == '@') {
2879                 char *plus = strchr(line, '+');
2880                 if (plus)
2881                         data->lineno = strtol(plus, NULL, 10) - 1;
2882                 else
2883                         die("invalid diff");
2884         }
2885 }
2886
2887 static unsigned char *deflate_it(char *data,
2888                                  unsigned long size,
2889                                  unsigned long *result_size)
2890 {
2891         int bound;
2892         unsigned char *deflated;
2893         git_zstream stream;
2894
2895         git_deflate_init(&stream, zlib_compression_level);
2896         bound = git_deflate_bound(&stream, size);
2897         deflated = xmalloc(bound);
2898         stream.next_out = deflated;
2899         stream.avail_out = bound;
2900
2901         stream.next_in = (unsigned char *)data;
2902         stream.avail_in = size;
2903         while (git_deflate(&stream, Z_FINISH) == Z_OK)
2904                 ; /* nothing */
2905         git_deflate_end(&stream);
2906         *result_size = stream.total_out;
2907         return deflated;
2908 }
2909
2910 static void emit_binary_diff_body(struct diff_options *o,
2911                                   mmfile_t *one, mmfile_t *two)
2912 {
2913         void *cp;
2914         void *delta;
2915         void *deflated;
2916         void *data;
2917         unsigned long orig_size;
2918         unsigned long delta_size;
2919         unsigned long deflate_size;
2920         unsigned long data_size;
2921
2922         /* We could do deflated delta, or we could do just deflated two,
2923          * whichever is smaller.
2924          */
2925         delta = NULL;
2926         deflated = deflate_it(two->ptr, two->size, &deflate_size);
2927         if (one->size && two->size) {
2928                 delta = diff_delta(one->ptr, one->size,
2929                                    two->ptr, two->size,
2930                                    &delta_size, deflate_size);
2931                 if (delta) {
2932                         void *to_free = delta;
2933                         orig_size = delta_size;
2934                         delta = deflate_it(delta, delta_size, &delta_size);
2935                         free(to_free);
2936                 }
2937         }
2938
2939         if (delta && delta_size < deflate_size) {
2940                 char *s = xstrfmt("%lu", orig_size);
2941                 emit_diff_symbol(o, DIFF_SYMBOL_BINARY_DIFF_HEADER_DELTA,
2942                                  s, strlen(s), 0);
2943                 free(s);
2944                 free(deflated);
2945                 data = delta;
2946                 data_size = delta_size;
2947         } else {
2948                 char *s = xstrfmt("%lu", two->size);
2949                 emit_diff_symbol(o, DIFF_SYMBOL_BINARY_DIFF_HEADER_LITERAL,
2950                                  s, strlen(s), 0);
2951                 free(s);
2952                 free(delta);
2953                 data = deflated;
2954                 data_size = deflate_size;
2955         }
2956
2957         /* emit data encoded in base85 */
2958         cp = data;
2959         while (data_size) {
2960                 int len;
2961                 int bytes = (52 < data_size) ? 52 : data_size;
2962                 char line[71];
2963                 data_size -= bytes;
2964                 if (bytes <= 26)
2965                         line[0] = bytes + 'A' - 1;
2966                 else
2967                         line[0] = bytes - 26 + 'a' - 1;
2968                 encode_85(line + 1, cp, bytes);
2969                 cp = (char *) cp + bytes;
2970
2971                 len = strlen(line);
2972                 line[len++] = '\n';
2973                 line[len] = '\0';
2974
2975                 emit_diff_symbol(o, DIFF_SYMBOL_BINARY_DIFF_BODY,
2976                                  line, len, 0);
2977         }
2978         emit_diff_symbol(o, DIFF_SYMBOL_BINARY_DIFF_FOOTER, NULL, 0, 0);
2979         free(data);
2980 }
2981
2982 static void emit_binary_diff(struct diff_options *o,
2983                              mmfile_t *one, mmfile_t *two)
2984 {
2985         emit_diff_symbol(o, DIFF_SYMBOL_BINARY_DIFF_HEADER, NULL, 0, 0);
2986         emit_binary_diff_body(o, one, two);
2987         emit_binary_diff_body(o, two, one);
2988 }
2989
2990 int diff_filespec_is_binary(struct diff_filespec *one)
2991 {
2992         if (one->is_binary == -1) {
2993                 diff_filespec_load_driver(one);
2994                 if (one->driver->binary != -1)
2995                         one->is_binary = one->driver->binary;
2996                 else {
2997                         if (!one->data && DIFF_FILE_VALID(one))
2998                                 diff_populate_filespec(one, CHECK_BINARY);
2999                         if (one->is_binary == -1 && one->data)
3000                                 one->is_binary = buffer_is_binary(one->data,
3001                                                 one->size);
3002                         if (one->is_binary == -1)
3003                                 one->is_binary = 0;
3004                 }
3005         }
3006         return one->is_binary;
3007 }
3008
3009 static const struct userdiff_funcname *diff_funcname_pattern(struct diff_filespec *one)
3010 {
3011         diff_filespec_load_driver(one);
3012         return one->driver->funcname.pattern ? &one->driver->funcname : NULL;
3013 }
3014
3015 void diff_set_mnemonic_prefix(struct diff_options *options, const char *a, const char *b)
3016 {
3017         if (!options->a_prefix)
3018                 options->a_prefix = a;
3019         if (!options->b_prefix)
3020                 options->b_prefix = b;
3021 }
3022
3023 struct userdiff_driver *get_textconv(struct diff_filespec *one)
3024 {
3025         if (!DIFF_FILE_VALID(one))
3026                 return NULL;
3027
3028         diff_filespec_load_driver(one);
3029         return userdiff_get_textconv(one->driver);
3030 }
3031
3032 static void builtin_diff(const char *name_a,
3033                          const char *name_b,
3034                          struct diff_filespec *one,
3035                          struct diff_filespec *two,
3036                          const char *xfrm_msg,
3037                          int must_show_header,
3038                          struct diff_options *o,
3039                          int complete_rewrite)
3040 {
3041         mmfile_t mf1, mf2;
3042         const char *lbl[2];
3043         char *a_one, *b_two;
3044         const char *meta = diff_get_color_opt(o, DIFF_METAINFO);
3045         const char *reset = diff_get_color_opt(o, DIFF_RESET);
3046         const char *a_prefix, *b_prefix;
3047         struct userdiff_driver *textconv_one = NULL;
3048         struct userdiff_driver *textconv_two = NULL;
3049         struct strbuf header = STRBUF_INIT;
3050         const char *line_prefix = diff_line_prefix(o);
3051
3052         diff_set_mnemonic_prefix(o, "a/", "b/");
3053         if (o->flags.reverse_diff) {
3054                 a_prefix = o->b_prefix;
3055                 b_prefix = o->a_prefix;
3056         } else {
3057                 a_prefix = o->a_prefix;
3058                 b_prefix = o->b_prefix;
3059         }
3060
3061         if (o->submodule_format == DIFF_SUBMODULE_LOG &&
3062             (!one->mode || S_ISGITLINK(one->mode)) &&
3063             (!two->mode || S_ISGITLINK(two->mode))) {
3064                 show_submodule_summary(o, one->path ? one->path : two->path,
3065                                 &one->oid, &two->oid,
3066                                 two->dirty_submodule);
3067                 return;
3068         } else if (o->submodule_format == DIFF_SUBMODULE_INLINE_DIFF &&
3069                    (!one->mode || S_ISGITLINK(one->mode)) &&
3070                    (!two->mode || S_ISGITLINK(two->mode))) {
3071                 show_submodule_inline_diff(o, one->path ? one->path : two->path,
3072                                 &one->oid, &two->oid,
3073                                 two->dirty_submodule);
3074                 return;
3075         }
3076
3077         if (o->flags.allow_textconv) {
3078                 textconv_one = get_textconv(one);
3079                 textconv_two = get_textconv(two);
3080         }
3081
3082         /* Never use a non-valid filename anywhere if at all possible */
3083         name_a = DIFF_FILE_VALID(one) ? name_a : name_b;
3084         name_b = DIFF_FILE_VALID(two) ? name_b : name_a;
3085
3086         a_one = quote_two(a_prefix, name_a + (*name_a == '/'));
3087         b_two = quote_two(b_prefix, name_b + (*name_b == '/'));
3088         lbl[0] = DIFF_FILE_VALID(one) ? a_one : "/dev/null";
3089         lbl[1] = DIFF_FILE_VALID(two) ? b_two : "/dev/null";
3090         strbuf_addf(&header, "%s%sdiff --git %s %s%s\n", line_prefix, meta, a_one, b_two, reset);
3091         if (lbl[0][0] == '/') {
3092                 /* /dev/null */
3093                 strbuf_addf(&header, "%s%snew file mode %06o%s\n", line_prefix, meta, two->mode, reset);
3094                 if (xfrm_msg)
3095                         strbuf_addstr(&header, xfrm_msg);
3096                 must_show_header = 1;
3097         }
3098         else if (lbl[1][0] == '/') {
3099                 strbuf_addf(&header, "%s%sdeleted file mode %06o%s\n", line_prefix, meta, one->mode, reset);
3100                 if (xfrm_msg)
3101                         strbuf_addstr(&header, xfrm_msg);
3102                 must_show_header = 1;
3103         }
3104         else {
3105                 if (one->mode != two->mode) {
3106                         strbuf_addf(&header, "%s%sold mode %06o%s\n", line_prefix, meta, one->mode, reset);
3107                         strbuf_addf(&header, "%s%snew mode %06o%s\n", line_prefix, meta, two->mode, reset);
3108                         must_show_header = 1;
3109                 }
3110                 if (xfrm_msg)
3111                         strbuf_addstr(&header, xfrm_msg);
3112
3113                 /*
3114                  * we do not run diff between different kind
3115                  * of objects.
3116                  */
3117                 if ((one->mode ^ two->mode) & S_IFMT)
3118                         goto free_ab_and_return;
3119                 if (complete_rewrite &&
3120                     (textconv_one || !diff_filespec_is_binary(one)) &&
3121                     (textconv_two || !diff_filespec_is_binary(two))) {
3122                         emit_diff_symbol(o, DIFF_SYMBOL_HEADER,
3123                                          header.buf, header.len, 0);
3124                         strbuf_reset(&header);
3125                         emit_rewrite_diff(name_a, name_b, one, two,
3126                                                 textconv_one, textconv_two, o);
3127                         o->found_changes = 1;
3128                         goto free_ab_and_return;
3129                 }
3130         }
3131
3132         if (o->irreversible_delete && lbl[1][0] == '/') {
3133                 emit_diff_symbol(o, DIFF_SYMBOL_HEADER, header.buf,
3134                                  header.len, 0);
3135                 strbuf_reset(&header);
3136                 goto free_ab_and_return;
3137         } else if (!o->flags.text &&
3138             ( (!textconv_one && diff_filespec_is_binary(one)) ||
3139               (!textconv_two && diff_filespec_is_binary(two)) )) {
3140                 struct strbuf sb = STRBUF_INIT;
3141                 if (!one->data && !two->data &&
3142                     S_ISREG(one->mode) && S_ISREG(two->mode) &&
3143                     !o->flags.binary) {
3144                         if (!oidcmp(&one->oid, &two->oid)) {
3145                                 if (must_show_header)
3146                                         emit_diff_symbol(o, DIFF_SYMBOL_HEADER,
3147                                                          header.buf, header.len,
3148                                                          0);
3149                                 goto free_ab_and_return;
3150                         }
3151                         emit_diff_symbol(o, DIFF_SYMBOL_HEADER,
3152                                          header.buf, header.len, 0);
3153                         strbuf_addf(&sb, "%sBinary files %s and %s differ\n",
3154                                     diff_line_prefix(o), lbl[0], lbl[1]);
3155                         emit_diff_symbol(o, DIFF_SYMBOL_BINARY_FILES,
3156                                          sb.buf, sb.len, 0);
3157                         strbuf_release(&sb);
3158                         goto free_ab_and_return;
3159                 }
3160                 if (fill_mmfile(&mf1, one) < 0 || fill_mmfile(&mf2, two) < 0)
3161                         die("unable to read files to diff");
3162                 /* Quite common confusing case */
3163                 if (mf1.size == mf2.size &&
3164                     !memcmp(mf1.ptr, mf2.ptr, mf1.size)) {
3165                         if (must_show_header)
3166                                 emit_diff_symbol(o, DIFF_SYMBOL_HEADER,
3167                                                  header.buf, header.len, 0);
3168                         goto free_ab_and_return;
3169                 }
3170                 emit_diff_symbol(o, DIFF_SYMBOL_HEADER, header.buf, header.len, 0);
3171                 strbuf_reset(&header);
3172                 if (o->flags.binary)
3173                         emit_binary_diff(o, &mf1, &mf2);
3174                 else {
3175                         strbuf_addf(&sb, "%sBinary files %s and %s differ\n",
3176                                     diff_line_prefix(o), lbl[0], lbl[1]);
3177                         emit_diff_symbol(o, DIFF_SYMBOL_BINARY_FILES,
3178                                          sb.buf, sb.len, 0);
3179                         strbuf_release(&sb);
3180                 }
3181                 o->found_changes = 1;
3182         } else {
3183                 /* Crazy xdl interfaces.. */
3184                 const char *diffopts = getenv("GIT_DIFF_OPTS");
3185                 const char *v;
3186                 xpparam_t xpp;
3187                 xdemitconf_t xecfg;
3188                 struct emit_callback ecbdata;
3189                 const struct userdiff_funcname *pe;
3190
3191                 if (must_show_header) {
3192                         emit_diff_symbol(o, DIFF_SYMBOL_HEADER,
3193                                          header.buf, header.len, 0);
3194                         strbuf_reset(&header);
3195                 }
3196
3197                 mf1.size = fill_textconv(textconv_one, one, &mf1.ptr);
3198                 mf2.size = fill_textconv(textconv_two, two, &mf2.ptr);
3199
3200                 pe = diff_funcname_pattern(one);
3201                 if (!pe)
3202                         pe = diff_funcname_pattern(two);
3203
3204                 memset(&xpp, 0, sizeof(xpp));
3205                 memset(&xecfg, 0, sizeof(xecfg));
3206                 memset(&ecbdata, 0, sizeof(ecbdata));
3207                 ecbdata.label_path = lbl;
3208                 ecbdata.color_diff = want_color(o->use_color);
3209                 ecbdata.ws_rule = whitespace_rule(name_b);
3210                 if (ecbdata.ws_rule & WS_BLANK_AT_EOF)
3211                         check_blank_at_eof(&mf1, &mf2, &ecbdata);
3212                 ecbdata.opt = o;
3213                 ecbdata.header = header.len ? &header : NULL;
3214                 xpp.flags = o->xdl_opts;
3215                 xpp.anchors = o->anchors;
3216                 xpp.anchors_nr = o->anchors_nr;
3217                 xecfg.ctxlen = o->context;
3218                 xecfg.interhunkctxlen = o->interhunkcontext;
3219                 xecfg.flags = XDL_EMIT_FUNCNAMES;
3220                 if (o->flags.funccontext)
3221                         xecfg.flags |= XDL_EMIT_FUNCCONTEXT;
3222                 if (pe)
3223                         xdiff_set_find_func(&xecfg, pe->pattern, pe->cflags);
3224                 if (!diffopts)
3225                         ;
3226                 else if (skip_prefix(diffopts, "--unified=", &v))
3227                         xecfg.ctxlen = strtoul(v, NULL, 10);
3228                 else if (skip_prefix(diffopts, "-u", &v))
3229                         xecfg.ctxlen = strtoul(v, NULL, 10);
3230                 if (o->word_diff)
3231                         init_diff_words_data(&ecbdata, o, one, two);
3232                 if (xdi_diff_outf(&mf1, &mf2, fn_out_consume, &ecbdata,
3233                                   &xpp, &xecfg))
3234                         die("unable to generate diff for %s", one->path);
3235                 if (o->word_diff)
3236                         free_diff_words_data(&ecbdata);
3237                 if (textconv_one)
3238                         free(mf1.ptr);
3239                 if (textconv_two)
3240                         free(mf2.ptr);
3241                 xdiff_clear_find_func(&xecfg);
3242         }
3243
3244  free_ab_and_return:
3245         strbuf_release(&header);
3246         diff_free_filespec_data(one);
3247         diff_free_filespec_data(two);
3248         free(a_one);
3249         free(b_two);
3250         return;
3251 }
3252
3253 static void builtin_diffstat(const char *name_a, const char *name_b,
3254                              struct diff_filespec *one,
3255                              struct diff_filespec *two,
3256                              struct diffstat_t *diffstat,
3257                              struct diff_options *o,
3258                              struct diff_filepair *p)
3259 {
3260         mmfile_t mf1, mf2;
3261         struct diffstat_file *data;
3262         int same_contents;
3263         int complete_rewrite = 0;
3264
3265         if (!DIFF_PAIR_UNMERGED(p)) {
3266                 if (p->status == DIFF_STATUS_MODIFIED && p->score)
3267                         complete_rewrite = 1;
3268         }
3269
3270         data = diffstat_add(diffstat, name_a, name_b);
3271         data->is_interesting = p->status != DIFF_STATUS_UNKNOWN;
3272
3273         if (!one || !two) {
3274                 data->is_unmerged = 1;
3275                 return;
3276         }
3277
3278         same_contents = !oidcmp(&one->oid, &two->oid);
3279
3280         if (diff_filespec_is_binary(one) || diff_filespec_is_binary(two)) {
3281                 data->is_binary = 1;
3282                 if (same_contents) {
3283                         data->added = 0;
3284                         data->deleted = 0;
3285                 } else {
3286                         data->added = diff_filespec_size(two);
3287                         data->deleted = diff_filespec_size(one);
3288                 }
3289         }
3290
3291         else if (complete_rewrite) {
3292                 diff_populate_filespec(one, 0);
3293                 diff_populate_filespec(two, 0);
3294                 data->deleted = count_lines(one->data, one->size);
3295                 data->added = count_lines(two->data, two->size);
3296         }
3297
3298         else if (!same_contents) {
3299                 /* Crazy xdl interfaces.. */
3300                 xpparam_t xpp;
3301                 xdemitconf_t xecfg;
3302
3303                 if (fill_mmfile(&mf1, one) < 0 || fill_mmfile(&mf2, two) < 0)
3304                         die("unable to read files to diff");
3305
3306                 memset(&xpp, 0, sizeof(xpp));
3307                 memset(&xecfg, 0, sizeof(xecfg));
3308                 xpp.flags = o->xdl_opts;
3309                 xpp.anchors = o->anchors;
3310                 xpp.anchors_nr = o->anchors_nr;
3311                 xecfg.ctxlen = o->context;
3312                 xecfg.interhunkctxlen = o->interhunkcontext;
3313                 if (xdi_diff_outf(&mf1, &mf2, diffstat_consume, diffstat,
3314                                   &xpp, &xecfg))
3315                         die("unable to generate diffstat for %s", one->path);
3316         }
3317
3318         diff_free_filespec_data(one);
3319         diff_free_filespec_data(two);
3320 }
3321
3322 static void builtin_checkdiff(const char *name_a, const char *name_b,
3323                               const char *attr_path,
3324                               struct diff_filespec *one,
3325                               struct diff_filespec *two,
3326                               struct diff_options *o)
3327 {
3328         mmfile_t mf1, mf2;
3329         struct checkdiff_t data;
3330
3331         if (!two)
3332                 return;
3333
3334         memset(&data, 0, sizeof(data));
3335         data.filename = name_b ? name_b : name_a;
3336         data.lineno = 0;
3337         data.o = o;
3338         data.ws_rule = whitespace_rule(attr_path);
3339         data.conflict_marker_size = ll_merge_marker_size(attr_path);
3340
3341         if (fill_mmfile(&mf1, one) < 0 || fill_mmfile(&mf2, two) < 0)
3342                 die("unable to read files to diff");
3343
3344         /*
3345          * All the other codepaths check both sides, but not checking
3346          * the "old" side here is deliberate.  We are checking the newly
3347          * introduced changes, and as long as the "new" side is text, we
3348          * can and should check what it introduces.
3349          */
3350         if (diff_filespec_is_binary(two))
3351                 goto free_and_return;
3352         else {
3353                 /* Crazy xdl interfaces.. */
3354                 xpparam_t xpp;
3355                 xdemitconf_t xecfg;
3356
3357                 memset(&xpp, 0, sizeof(xpp));
3358                 memset(&xecfg, 0, sizeof(xecfg));
3359                 xecfg.ctxlen = 1; /* at least one context line */
3360                 xpp.flags = 0;
3361                 if (xdi_diff_outf(&mf1, &mf2, checkdiff_consume, &data,
3362                                   &xpp, &xecfg))
3363                         die("unable to generate checkdiff for %s", one->path);
3364
3365                 if (data.ws_rule & WS_BLANK_AT_EOF) {
3366                         struct emit_callback ecbdata;
3367                         int blank_at_eof;
3368
3369                         ecbdata.ws_rule = data.ws_rule;
3370                         check_blank_at_eof(&mf1, &mf2, &ecbdata);
3371                         blank_at_eof = ecbdata.blank_at_eof_in_postimage;
3372
3373                         if (blank_at_eof) {
3374                                 static char *err;
3375                                 if (!err)
3376                                         err = whitespace_error_string(WS_BLANK_AT_EOF);
3377                                 fprintf(o->file, "%s:%d: %s.\n",
3378                                         data.filename, blank_at_eof, err);
3379                                 data.status = 1; /* report errors */
3380                         }
3381                 }
3382         }
3383  free_and_return:
3384         diff_free_filespec_data(one);
3385         diff_free_filespec_data(two);
3386         if (data.status)
3387                 o->flags.check_failed = 1;
3388 }
3389
3390 struct diff_filespec *alloc_filespec(const char *path)
3391 {
3392         struct diff_filespec *spec;
3393
3394         FLEXPTR_ALLOC_STR(spec, path, path);
3395         spec->count = 1;
3396         spec->is_binary = -1;
3397         return spec;
3398 }
3399
3400 void free_filespec(struct diff_filespec *spec)
3401 {
3402         if (!--spec->count) {
3403                 diff_free_filespec_data(spec);
3404                 free(spec);
3405         }
3406 }
3407
3408 void fill_filespec(struct diff_filespec *spec, const struct object_id *oid,
3409                    int oid_valid, unsigned short mode)
3410 {
3411         if (mode) {
3412                 spec->mode = canon_mode(mode);
3413                 oidcpy(&spec->oid, oid);
3414                 spec->oid_valid = oid_valid;
3415         }
3416 }
3417
3418 /*
3419  * Given a name and sha1 pair, if the index tells us the file in
3420  * the work tree has that object contents, return true, so that
3421  * prepare_temp_file() does not have to inflate and extract.
3422  */
3423 static int reuse_worktree_file(const char *name, const struct object_id *oid, int want_file)
3424 {
3425         const struct cache_entry *ce;
3426         struct stat st;
3427         int pos, len;
3428
3429         /*
3430          * We do not read the cache ourselves here, because the
3431          * benchmark with my previous version that always reads cache
3432          * shows that it makes things worse for diff-tree comparing
3433          * two linux-2.6 kernel trees in an already checked out work
3434          * tree.  This is because most diff-tree comparisons deal with
3435          * only a small number of files, while reading the cache is
3436          * expensive for a large project, and its cost outweighs the
3437          * savings we get by not inflating the object to a temporary
3438          * file.  Practically, this code only helps when we are used
3439          * by diff-cache --cached, which does read the cache before
3440          * calling us.
3441          */
3442         if (!active_cache)
3443                 return 0;
3444
3445         /* We want to avoid the working directory if our caller
3446          * doesn't need the data in a normal file, this system
3447          * is rather slow with its stat/open/mmap/close syscalls,
3448          * and the object is contained in a pack file.  The pack
3449          * is probably already open and will be faster to obtain
3450          * the data through than the working directory.  Loose
3451          * objects however would tend to be slower as they need
3452          * to be individually opened and inflated.
3453          */
3454         if (!FAST_WORKING_DIRECTORY && !want_file && has_sha1_pack(oid->hash))
3455                 return 0;
3456
3457         /*
3458          * Similarly, if we'd have to convert the file contents anyway, that
3459          * makes the optimization not worthwhile.
3460          */
3461         if (!want_file && would_convert_to_git(&the_index, name))
3462                 return 0;
3463
3464         len = strlen(name);
3465         pos = cache_name_pos(name, len);
3466         if (pos < 0)
3467                 return 0;
3468         ce = active_cache[pos];
3469
3470         /*
3471          * This is not the sha1 we are looking for, or
3472          * unreusable because it is not a regular file.
3473          */
3474         if (oidcmp(oid, &ce->oid) || !S_ISREG(ce->ce_mode))
3475                 return 0;
3476
3477         /*
3478          * If ce is marked as "assume unchanged", there is no
3479          * guarantee that work tree matches what we are looking for.
3480          */
3481         if ((ce->ce_flags & CE_VALID) || ce_skip_worktree(ce))
3482                 return 0;
3483
3484         /*
3485          * If ce matches the file in the work tree, we can reuse it.
3486          */
3487         if (ce_uptodate(ce) ||
3488             (!lstat(name, &st) && !ce_match_stat(ce, &st, 0)))
3489                 return 1;
3490
3491         return 0;
3492 }
3493
3494 static int diff_populate_gitlink(struct diff_filespec *s, int size_only)
3495 {
3496         struct strbuf buf = STRBUF_INIT;
3497         char *dirty = "";
3498
3499         /* Are we looking at the work tree? */
3500         if (s->dirty_submodule)
3501                 dirty = "-dirty";
3502
3503         strbuf_addf(&buf, "Subproject commit %s%s\n",
3504                     oid_to_hex(&s->oid), dirty);
3505         s->size = buf.len;
3506         if (size_only) {
3507                 s->data = NULL;
3508                 strbuf_release(&buf);
3509         } else {
3510                 s->data = strbuf_detach(&buf, NULL);
3511                 s->should_free = 1;
3512         }
3513         return 0;
3514 }
3515
3516 /*
3517  * While doing rename detection and pickaxe operation, we may need to
3518  * grab the data for the blob (or file) for our own in-core comparison.
3519  * diff_filespec has data and size fields for this purpose.
3520  */
3521 int diff_populate_filespec(struct diff_filespec *s, unsigned int flags)
3522 {
3523         int size_only = flags & CHECK_SIZE_ONLY;
3524         int err = 0;
3525         /*
3526          * demote FAIL to WARN to allow inspecting the situation
3527          * instead of refusing.
3528          */
3529         enum safe_crlf crlf_warn = (safe_crlf == SAFE_CRLF_FAIL
3530                                     ? SAFE_CRLF_WARN
3531                                     : safe_crlf);
3532
3533         if (!DIFF_FILE_VALID(s))
3534                 die("internal error: asking to populate invalid file.");
3535         if (S_ISDIR(s->mode))
3536                 return -1;
3537
3538         if (s->data)
3539                 return 0;
3540
3541         if (size_only && 0 < s->size)
3542                 return 0;
3543
3544         if (S_ISGITLINK(s->mode))
3545                 return diff_populate_gitlink(s, size_only);
3546
3547         if (!s->oid_valid ||
3548             reuse_worktree_file(s->path, &s->oid, 0)) {
3549                 struct strbuf buf = STRBUF_INIT;
3550                 struct stat st;
3551                 int fd;
3552
3553                 if (lstat(s->path, &st) < 0) {
3554                 err_empty:
3555                         err = -1;
3556                 empty:
3557                         s->data = (char *)"";
3558                         s->size = 0;
3559                         return err;
3560                 }
3561                 s->size = xsize_t(st.st_size);
3562                 if (!s->size)
3563                         goto empty;
3564                 if (S_ISLNK(st.st_mode)) {
3565                         struct strbuf sb = STRBUF_INIT;
3566
3567                         if (strbuf_readlink(&sb, s->path, s->size))
3568                                 goto err_empty;
3569                         s->size = sb.len;
3570                         s->data = strbuf_detach(&sb, NULL);
3571                         s->should_free = 1;
3572                         return 0;
3573                 }
3574
3575                 /*
3576                  * Even if the caller would be happy with getting
3577                  * only the size, we cannot return early at this
3578                  * point if the path requires us to run the content
3579                  * conversion.
3580                  */
3581                 if (size_only && !would_convert_to_git(&the_index, s->path))
3582                         return 0;
3583
3584                 /*
3585                  * Note: this check uses xsize_t(st.st_size) that may
3586                  * not be the true size of the blob after it goes
3587                  * through convert_to_git().  This may not strictly be
3588                  * correct, but the whole point of big_file_threshold
3589                  * and is_binary check being that we want to avoid
3590                  * opening the file and inspecting the contents, this
3591                  * is probably fine.
3592                  */
3593                 if ((flags & CHECK_BINARY) &&
3594                     s->size > big_file_threshold && s->is_binary == -1) {
3595                         s->is_binary = 1;
3596                         return 0;
3597                 }
3598                 fd = open(s->path, O_RDONLY);
3599                 if (fd < 0)
3600                         goto err_empty;
3601                 s->data = xmmap(NULL, s->size, PROT_READ, MAP_PRIVATE, fd, 0);
3602                 close(fd);
3603                 s->should_munmap = 1;
3604
3605                 /*
3606                  * Convert from working tree format to canonical git format
3607                  */
3608                 if (convert_to_git(&the_index, s->path, s->data, s->size, &buf, crlf_warn)) {
3609                         size_t size = 0;
3610                         munmap(s->data, s->size);
3611                         s->should_munmap = 0;
3612                         s->data = strbuf_detach(&buf, &size);
3613                         s->size = size;
3614                         s->should_free = 1;
3615                 }
3616         }
3617         else {
3618                 enum object_type type;
3619                 if (size_only || (flags & CHECK_BINARY)) {
3620                         type = sha1_object_info(s->oid.hash, &s->size);
3621                         if (type < 0)
3622                                 die("unable to read %s",
3623                                     oid_to_hex(&s->oid));
3624                         if (size_only)
3625                                 return 0;
3626                         if (s->size > big_file_threshold && s->is_binary == -1) {
3627                                 s->is_binary = 1;
3628                                 return 0;
3629                         }
3630                 }
3631                 s->data = read_sha1_file(s->oid.hash, &type, &s->size);
3632                 if (!s->data)
3633                         die("unable to read %s", oid_to_hex(&s->oid));
3634                 s->should_free = 1;
3635         }
3636         return 0;
3637 }
3638
3639 void diff_free_filespec_blob(struct diff_filespec *s)
3640 {
3641         if (s->should_free)
3642                 free(s->data);
3643         else if (s->should_munmap)
3644                 munmap(s->data, s->size);
3645
3646         if (s->should_free || s->should_munmap) {
3647                 s->should_free = s->should_munmap = 0;
3648                 s->data = NULL;
3649         }
3650 }
3651
3652 void diff_free_filespec_data(struct diff_filespec *s)
3653 {
3654         diff_free_filespec_blob(s);
3655         FREE_AND_NULL(s->cnt_data);
3656 }
3657
3658 static void prep_temp_blob(const char *path, struct diff_tempfile *temp,
3659                            void *blob,
3660                            unsigned long size,
3661                            const struct object_id *oid,
3662                            int mode)
3663 {
3664         struct strbuf buf = STRBUF_INIT;
3665         struct strbuf template = STRBUF_INIT;
3666         char *path_dup = xstrdup(path);
3667         const char *base = basename(path_dup);
3668
3669         /* Generate "XXXXXX_basename.ext" */
3670         strbuf_addstr(&template, "XXXXXX_");
3671         strbuf_addstr(&template, base);
3672
3673         temp->tempfile = mks_tempfile_ts(template.buf, strlen(base) + 1);
3674         if (!temp->tempfile)
3675                 die_errno("unable to create temp-file");
3676         if (convert_to_working_tree(path,
3677                         (const char *)blob, (size_t)size, &buf)) {
3678                 blob = buf.buf;
3679                 size = buf.len;
3680         }
3681         if (write_in_full(temp->tempfile->fd, blob, size) < 0 ||
3682             close_tempfile_gently(temp->tempfile))
3683                 die_errno("unable to write temp-file");
3684         temp->name = get_tempfile_path(temp->tempfile);
3685         oid_to_hex_r(temp->hex, oid);
3686         xsnprintf(temp->mode, sizeof(temp->mode), "%06o", mode);
3687         strbuf_release(&buf);
3688         strbuf_release(&template);
3689         free(path_dup);
3690 }
3691
3692 static struct diff_tempfile *prepare_temp_file(const char *name,
3693                 struct diff_filespec *one)
3694 {
3695         struct diff_tempfile *temp = claim_diff_tempfile();
3696
3697         if (!DIFF_FILE_VALID(one)) {
3698         not_a_valid_file:
3699                 /* A '-' entry produces this for file-2, and
3700                  * a '+' entry produces this for file-1.
3701                  */
3702                 temp->name = "/dev/null";
3703                 xsnprintf(temp->hex, sizeof(temp->hex), ".");
3704                 xsnprintf(temp->mode, sizeof(temp->mode), ".");
3705                 return temp;
3706         }
3707
3708         if (!S_ISGITLINK(one->mode) &&
3709             (!one->oid_valid ||
3710              reuse_worktree_file(name, &one->oid, 1))) {
3711                 struct stat st;
3712                 if (lstat(name, &st) < 0) {
3713                         if (errno == ENOENT)
3714                                 goto not_a_valid_file;
3715                         die_errno("stat(%s)", name);
3716                 }
3717                 if (S_ISLNK(st.st_mode)) {
3718                         struct strbuf sb = STRBUF_INIT;
3719                         if (strbuf_readlink(&sb, name, st.st_size) < 0)
3720                                 die_errno("readlink(%s)", name);
3721                         prep_temp_blob(name, temp, sb.buf, sb.len,
3722                                        (one->oid_valid ?
3723                                         &one->oid : &null_oid),
3724                                        (one->oid_valid ?
3725                                         one->mode : S_IFLNK));
3726                         strbuf_release(&sb);
3727                 }
3728                 else {
3729                         /* we can borrow from the file in the work tree */
3730                         temp->name = name;
3731                         if (!one->oid_valid)
3732                                 oid_to_hex_r(temp->hex, &null_oid);
3733                         else
3734                                 oid_to_hex_r(temp->hex, &one->oid);
3735                         /* Even though we may sometimes borrow the
3736                          * contents from the work tree, we always want
3737                          * one->mode.  mode is trustworthy even when
3738                          * !(one->oid_valid), as long as
3739                          * DIFF_FILE_VALID(one).
3740                          */
3741                         xsnprintf(temp->mode, sizeof(temp->mode), "%06o", one->mode);
3742                 }
3743                 return temp;
3744         }
3745         else {
3746                 if (diff_populate_filespec(one, 0))
3747                         die("cannot read data blob for %s", one->path);
3748                 prep_temp_blob(name, temp, one->data, one->size,
3749                                &one->oid, one->mode);
3750         }
3751         return temp;
3752 }
3753
3754 static void add_external_diff_name(struct argv_array *argv,
3755                                    const char *name,
3756                                    struct diff_filespec *df)
3757 {
3758         struct diff_tempfile *temp = prepare_temp_file(name, df);
3759         argv_array_push(argv, temp->name);
3760         argv_array_push(argv, temp->hex);
3761         argv_array_push(argv, temp->mode);
3762 }
3763
3764 /* An external diff command takes:
3765  *
3766  * diff-cmd name infile1 infile1-sha1 infile1-mode \
3767  *               infile2 infile2-sha1 infile2-mode [ rename-to ]
3768  *
3769  */
3770 static void run_external_diff(const char *pgm,
3771                               const char *name,
3772                               const char *other,
3773                               struct diff_filespec *one,
3774                               struct diff_filespec *two,
3775                               const char *xfrm_msg,
3776                               int complete_rewrite,
3777                               struct diff_options *o)
3778 {
3779         struct argv_array argv = ARGV_ARRAY_INIT;
3780         struct argv_array env = ARGV_ARRAY_INIT;
3781         struct diff_queue_struct *q = &diff_queued_diff;
3782
3783         argv_array_push(&argv, pgm);
3784         argv_array_push(&argv, name);
3785
3786         if (one && two) {
3787                 add_external_diff_name(&argv, name, one);
3788                 if (!other)
3789                         add_external_diff_name(&argv, name, two);
3790                 else {
3791                         add_external_diff_name(&argv, other, two);
3792                         argv_array_push(&argv, other);
3793                         argv_array_push(&argv, xfrm_msg);
3794                 }
3795         }
3796
3797         argv_array_pushf(&env, "GIT_DIFF_PATH_COUNTER=%d", ++o->diff_path_counter);
3798         argv_array_pushf(&env, "GIT_DIFF_PATH_TOTAL=%d", q->nr);
3799
3800         if (run_command_v_opt_cd_env(argv.argv, RUN_USING_SHELL, NULL, env.argv))
3801                 die(_("external diff died, stopping at %s"), name);
3802
3803         remove_tempfile();
3804         argv_array_clear(&argv);
3805         argv_array_clear(&env);
3806 }
3807
3808 static int similarity_index(struct diff_filepair *p)
3809 {
3810         return p->score * 100 / MAX_SCORE;
3811 }
3812
3813 static const char *diff_abbrev_oid(const struct object_id *oid, int abbrev)
3814 {
3815         if (startup_info->have_repository)
3816                 return find_unique_abbrev(oid->hash, abbrev);
3817         else {
3818                 char *hex = oid_to_hex(oid);
3819                 if (abbrev < 0)
3820                         abbrev = FALLBACK_DEFAULT_ABBREV;
3821                 if (abbrev > GIT_SHA1_HEXSZ)
3822                         die("BUG: oid abbreviation out of range: %d", abbrev);
3823                 if (abbrev)
3824                         hex[abbrev] = '\0';
3825                 return hex;
3826         }
3827 }
3828
3829 static void fill_metainfo(struct strbuf *msg,
3830                           const char *name,
3831                           const char *other,
3832                           struct diff_filespec *one,
3833                           struct diff_filespec *two,
3834                           struct diff_options *o,
3835                           struct diff_filepair *p,
3836                           int *must_show_header,
3837                           int use_color)
3838 {
3839         const char *set = diff_get_color(use_color, DIFF_METAINFO);
3840         const char *reset = diff_get_color(use_color, DIFF_RESET);
3841         const char *line_prefix = diff_line_prefix(o);
3842
3843         *must_show_header = 1;
3844         strbuf_init(msg, PATH_MAX * 2 + 300);
3845         switch (p->status) {
3846         case DIFF_STATUS_COPIED:
3847                 strbuf_addf(msg, "%s%ssimilarity index %d%%",
3848                             line_prefix, set, similarity_index(p));
3849                 strbuf_addf(msg, "%s\n%s%scopy from ",
3850                             reset,  line_prefix, set);
3851                 quote_c_style(name, msg, NULL, 0);
3852                 strbuf_addf(msg, "%s\n%s%scopy to ", reset, line_prefix, set);
3853                 quote_c_style(other, msg, NULL, 0);
3854                 strbuf_addf(msg, "%s\n", reset);
3855                 break;
3856         case DIFF_STATUS_RENAMED:
3857                 strbuf_addf(msg, "%s%ssimilarity index %d%%",
3858                             line_prefix, set, similarity_index(p));
3859                 strbuf_addf(msg, "%s\n%s%srename from ",
3860                             reset, line_prefix, set);
3861                 quote_c_style(name, msg, NULL, 0);
3862                 strbuf_addf(msg, "%s\n%s%srename to ",
3863                             reset, line_prefix, set);
3864                 quote_c_style(other, msg, NULL, 0);
3865                 strbuf_addf(msg, "%s\n", reset);
3866                 break;
3867         case DIFF_STATUS_MODIFIED:
3868                 if (p->score) {
3869                         strbuf_addf(msg, "%s%sdissimilarity index %d%%%s\n",
3870                                     line_prefix,
3871                                     set, similarity_index(p), reset);
3872                         break;
3873                 }
3874                 /* fallthru */
3875         default:
3876                 *must_show_header = 0;
3877         }
3878         if (one && two && oidcmp(&one->oid, &two->oid)) {
3879                 int abbrev = o->flags.full_index ? 40 : DEFAULT_ABBREV;
3880
3881                 if (o->flags.binary) {
3882                         mmfile_t mf;
3883                         if ((!fill_mmfile(&mf, one) && diff_filespec_is_binary(one)) ||
3884                             (!fill_mmfile(&mf, two) && diff_filespec_is_binary(two)))
3885                                 abbrev = 40;
3886                 }
3887                 strbuf_addf(msg, "%s%sindex %s..%s", line_prefix, set,
3888                             diff_abbrev_oid(&one->oid, abbrev),
3889                             diff_abbrev_oid(&two->oid, abbrev));
3890                 if (one->mode == two->mode)
3891                         strbuf_addf(msg, " %06o", one->mode);
3892                 strbuf_addf(msg, "%s\n", reset);
3893         }
3894 }
3895
3896 static void run_diff_cmd(const char *pgm,
3897                          const char *name,
3898                          const char *other,
3899                          const char *attr_path,
3900                          struct diff_filespec *one,
3901                          struct diff_filespec *two,
3902                          struct strbuf *msg,
3903                          struct diff_options *o,
3904                          struct diff_filepair *p)
3905 {
3906         const char *xfrm_msg = NULL;
3907         int complete_rewrite = (p->status == DIFF_STATUS_MODIFIED) && p->score;
3908         int must_show_header = 0;
3909
3910
3911         if (o->flags.allow_external) {
3912                 struct userdiff_driver *drv = userdiff_find_by_path(attr_path);
3913                 if (drv && drv->external)
3914                         pgm = drv->external;
3915         }
3916
3917         if (msg) {
3918                 /*
3919                  * don't use colors when the header is intended for an
3920                  * external diff driver
3921                  */
3922                 fill_metainfo(msg, name, other, one, two, o, p,
3923                               &must_show_header,
3924                               want_color(o->use_color) && !pgm);
3925                 xfrm_msg = msg->len ? msg->buf : NULL;
3926         }
3927
3928         if (pgm) {
3929                 run_external_diff(pgm, name, other, one, two, xfrm_msg,
3930                                   complete_rewrite, o);
3931                 return;
3932         }
3933         if (one && two)
3934                 builtin_diff(name, other ? other : name,
3935                              one, two, xfrm_msg, must_show_header,
3936                              o, complete_rewrite);
3937         else
3938                 fprintf(o->file, "* Unmerged path %s\n", name);
3939 }
3940
3941 static void diff_fill_oid_info(struct diff_filespec *one)
3942 {
3943         if (DIFF_FILE_VALID(one)) {
3944                 if (!one->oid_valid) {
3945                         struct stat st;
3946                         if (one->is_stdin) {
3947                                 oidclr(&one->oid);
3948                                 return;
3949                         }
3950                         if (lstat(one->path, &st) < 0)
3951                                 die_errno("stat '%s'", one->path);
3952                         if (index_path(&one->oid, one->path, &st, 0))
3953                                 die("cannot hash %s", one->path);
3954                 }
3955         }
3956         else
3957                 oidclr(&one->oid);
3958 }
3959
3960 static void strip_prefix(int prefix_length, const char **namep, const char **otherp)
3961 {
3962         /* Strip the prefix but do not molest /dev/null and absolute paths */
3963         if (*namep && **namep != '/') {
3964                 *namep += prefix_length;
3965                 if (**namep == '/')
3966                         ++*namep;
3967         }
3968         if (*otherp && **otherp != '/') {
3969                 *otherp += prefix_length;
3970                 if (**otherp == '/')
3971                         ++*otherp;
3972         }
3973 }
3974
3975 static void run_diff(struct diff_filepair *p, struct diff_options *o)
3976 {
3977         const char *pgm = external_diff();
3978         struct strbuf msg;
3979         struct diff_filespec *one = p->one;
3980         struct diff_filespec *two = p->two;
3981         const char *name;
3982         const char *other;
3983         const char *attr_path;
3984
3985         name  = one->path;
3986         other = (strcmp(name, two->path) ? two->path : NULL);
3987         attr_path = name;
3988         if (o->prefix_length)
3989                 strip_prefix(o->prefix_length, &name, &other);
3990
3991         if (!o->flags.allow_external)
3992                 pgm = NULL;
3993
3994         if (DIFF_PAIR_UNMERGED(p)) {
3995                 run_diff_cmd(pgm, name, NULL, attr_path,
3996                              NULL, NULL, NULL, o, p);
3997                 return;
3998         }
3999
4000         diff_fill_oid_info(one);
4001         diff_fill_oid_info(two);
4002
4003         if (!pgm &&
4004             DIFF_FILE_VALID(one) && DIFF_FILE_VALID(two) &&
4005             (S_IFMT & one->mode) != (S_IFMT & two->mode)) {
4006                 /*
4007                  * a filepair that changes between file and symlink
4008                  * needs to be split into deletion and creation.
4009                  */
4010                 struct diff_filespec *null = alloc_filespec(two->path);
4011                 run_diff_cmd(NULL, name, other, attr_path,
4012                              one, null, &msg, o, p);
4013                 free(null);
4014                 strbuf_release(&msg);
4015
4016                 null = alloc_filespec(one->path);
4017                 run_diff_cmd(NULL, name, other, attr_path,
4018                              null, two, &msg, o, p);
4019                 free(null);
4020         }
4021         else
4022                 run_diff_cmd(pgm, name, other, attr_path,
4023                              one, two, &msg, o, p);
4024
4025         strbuf_release(&msg);
4026 }
4027
4028 static void run_diffstat(struct diff_filepair *p, struct diff_options *o,
4029                          struct diffstat_t *diffstat)
4030 {
4031         const char *name;
4032         const char *other;
4033
4034         if (DIFF_PAIR_UNMERGED(p)) {
4035                 /* unmerged */
4036                 builtin_diffstat(p->one->path, NULL, NULL, NULL, diffstat, o, p);
4037                 return;
4038         }
4039
4040         name = p->one->path;
4041         other = (strcmp(name, p->two->path) ? p->two->path : NULL);
4042
4043         if (o->prefix_length)
4044                 strip_prefix(o->prefix_length, &name, &other);
4045
4046         diff_fill_oid_info(p->one);
4047         diff_fill_oid_info(p->two);
4048
4049         builtin_diffstat(name, other, p->one, p->two, diffstat, o, p);
4050 }
4051
4052 static void run_checkdiff(struct diff_filepair *p, struct diff_options *o)
4053 {
4054         const char *name;
4055         const char *other;
4056         const char *attr_path;
4057
4058         if (DIFF_PAIR_UNMERGED(p)) {
4059                 /* unmerged */
4060                 return;
4061         }
4062
4063         name = p->one->path;
4064         other = (strcmp(name, p->two->path) ? p->two->path : NULL);
4065         attr_path = other ? other : name;
4066
4067         if (o->prefix_length)
4068                 strip_prefix(o->prefix_length, &name, &other);
4069
4070         diff_fill_oid_info(p->one);
4071         diff_fill_oid_info(p->two);
4072
4073         builtin_checkdiff(name, other, attr_path, p->one, p->two, o);
4074 }
4075
4076 void diff_setup(struct diff_options *options)
4077 {
4078         memcpy(options, &default_diff_options, sizeof(*options));
4079
4080         options->file = stdout;
4081
4082         options->abbrev = DEFAULT_ABBREV;
4083         options->line_termination = '\n';
4084         options->break_opt = -1;
4085         options->rename_limit = -1;
4086         options->dirstat_permille = diff_dirstat_permille_default;
4087         options->context = diff_context_default;
4088         options->interhunkcontext = diff_interhunk_context_default;
4089         options->ws_error_highlight = ws_error_highlight_default;
4090         options->flags.rename_empty = 1;
4091
4092         /* pathchange left =NULL by default */
4093         options->change = diff_change;
4094         options->add_remove = diff_addremove;
4095         options->use_color = diff_use_color_default;
4096         options->detect_rename = diff_detect_rename_default;
4097         options->xdl_opts |= diff_algorithm;
4098         if (diff_indent_heuristic)
4099                 DIFF_XDL_SET(options, INDENT_HEURISTIC);
4100
4101         options->orderfile = diff_order_file_cfg;
4102
4103         if (diff_no_prefix) {
4104                 options->a_prefix = options->b_prefix = "";
4105         } else if (!diff_mnemonic_prefix) {
4106                 options->a_prefix = "a/";
4107                 options->b_prefix = "b/";
4108         }
4109
4110         options->color_moved = diff_color_moved_default;
4111 }
4112
4113 void diff_setup_done(struct diff_options *options)
4114 {
4115         int count = 0;
4116
4117         if (options->set_default)
4118                 options->set_default(options);
4119
4120         if (options->output_format & DIFF_FORMAT_NAME)
4121                 count++;
4122         if (options->output_format & DIFF_FORMAT_NAME_STATUS)
4123                 count++;
4124         if (options->output_format & DIFF_FORMAT_CHECKDIFF)
4125                 count++;
4126         if (options->output_format & DIFF_FORMAT_NO_OUTPUT)
4127                 count++;
4128         if (count > 1)
4129                 die(_("--name-only, --name-status, --check and -s are mutually exclusive"));
4130
4131         /*
4132          * Most of the time we can say "there are changes"
4133          * only by checking if there are changed paths, but
4134          * --ignore-whitespace* options force us to look
4135          * inside contents.
4136          */
4137
4138         if ((options->xdl_opts & XDF_WHITESPACE_FLAGS))
4139                 options->flags.diff_from_contents = 1;
4140         else
4141                 options->flags.diff_from_contents = 0;
4142
4143         if (options->flags.find_copies_harder)
4144                 options->detect_rename = DIFF_DETECT_COPY;
4145
4146         if (!options->flags.relative_name)
4147                 options->prefix = NULL;
4148         if (options->prefix)
4149                 options->prefix_length = strlen(options->prefix);
4150         else
4151                 options->prefix_length = 0;
4152
4153         if (options->output_format & (DIFF_FORMAT_NAME |
4154                                       DIFF_FORMAT_NAME_STATUS |
4155                                       DIFF_FORMAT_CHECKDIFF |
4156                                       DIFF_FORMAT_NO_OUTPUT))
4157                 options->output_format &= ~(DIFF_FORMAT_RAW |
4158                                             DIFF_FORMAT_NUMSTAT |
4159                                             DIFF_FORMAT_DIFFSTAT |
4160                                             DIFF_FORMAT_SHORTSTAT |
4161                                             DIFF_FORMAT_DIRSTAT |
4162                                             DIFF_FORMAT_SUMMARY |
4163                                             DIFF_FORMAT_PATCH);
4164
4165         /*
4166          * These cases always need recursive; we do not drop caller-supplied
4167          * recursive bits for other formats here.
4168          */
4169         if (options->output_format & (DIFF_FORMAT_PATCH |
4170                                       DIFF_FORMAT_NUMSTAT |
4171                                       DIFF_FORMAT_DIFFSTAT |
4172                                       DIFF_FORMAT_SHORTSTAT |
4173                                       DIFF_FORMAT_DIRSTAT |
4174                                       DIFF_FORMAT_SUMMARY |
4175                                       DIFF_FORMAT_CHECKDIFF))
4176                 options->flags.recursive = 1;
4177         /*
4178          * Also pickaxe would not work very well if you do not say recursive
4179          */
4180         if (options->pickaxe)
4181                 options->flags.recursive = 1;
4182         /*
4183          * When patches are generated, submodules diffed against the work tree
4184          * must be checked for dirtiness too so it can be shown in the output
4185          */
4186         if (options->output_format & DIFF_FORMAT_PATCH)
4187                 options->flags.dirty_submodules = 1;
4188
4189         if (options->detect_rename && options->rename_limit < 0)
4190                 options->rename_limit = diff_rename_limit_default;
4191         if (options->setup & DIFF_SETUP_USE_CACHE) {
4192                 if (!active_cache)
4193                         /* read-cache does not die even when it fails
4194                          * so it is safe for us to do this here.  Also
4195                          * it does not smudge active_cache or active_nr
4196                          * when it fails, so we do not have to worry about
4197                          * cleaning it up ourselves either.
4198                          */
4199                         read_cache();
4200         }
4201         if (40 < options->abbrev)
4202                 options->abbrev = 40; /* full */
4203
4204         /*
4205          * It does not make sense to show the first hit we happened
4206          * to have found.  It does not make sense not to return with
4207          * exit code in such a case either.
4208          */
4209         if (options->flags.quick) {
4210                 options->output_format = DIFF_FORMAT_NO_OUTPUT;
4211                 options->flags.exit_with_status = 1;
4212         }
4213
4214         options->diff_path_counter = 0;
4215
4216         if (options->flags.follow_renames && options->pathspec.nr != 1)
4217                 die(_("--follow requires exactly one pathspec"));
4218
4219         if (!options->use_color || external_diff())
4220                 options->color_moved = 0;
4221 }
4222
4223 static int opt_arg(const char *arg, int arg_short, const char *arg_long, int *val)
4224 {
4225         char c, *eq;
4226         int len;
4227
4228         if (*arg != '-')
4229                 return 0;
4230         c = *++arg;
4231         if (!c)
4232                 return 0;
4233         if (c == arg_short) {
4234                 c = *++arg;
4235                 if (!c)
4236                         return 1;
4237                 if (val && isdigit(c)) {
4238                         char *end;
4239                         int n = strtoul(arg, &end, 10);
4240                         if (*end)
4241                                 return 0;
4242                         *val = n;
4243                         return 1;
4244                 }
4245                 return 0;
4246         }
4247         if (c != '-')
4248                 return 0;
4249         arg++;
4250         eq = strchrnul(arg, '=');
4251         len = eq - arg;
4252         if (!len || strncmp(arg, arg_long, len))
4253                 return 0;
4254         if (*eq) {
4255                 int n;
4256                 char *end;
4257                 if (!isdigit(*++eq))
4258                         return 0;
4259                 n = strtoul(eq, &end, 10);
4260                 if (*end)
4261                         return 0;
4262                 *val = n;
4263         }
4264         return 1;
4265 }
4266
4267 static int diff_scoreopt_parse(const char *opt);
4268
4269 static inline int short_opt(char opt, const char **argv,
4270                             const char **optarg)
4271 {
4272         const char *arg = argv[0];
4273         if (arg[0] != '-' || arg[1] != opt)
4274                 return 0;
4275         if (arg[2] != '\0') {
4276                 *optarg = arg + 2;
4277                 return 1;
4278         }
4279         if (!argv[1])
4280                 die("Option '%c' requires a value", opt);
4281         *optarg = argv[1];
4282         return 2;
4283 }
4284
4285 int parse_long_opt(const char *opt, const char **argv,
4286                    const char **optarg)
4287 {
4288         const char *arg = argv[0];
4289         if (!skip_prefix(arg, "--", &arg))
4290                 return 0;
4291         if (!skip_prefix(arg, opt, &arg))
4292                 return 0;
4293         if (*arg == '=') { /* stuck form: --option=value */
4294                 *optarg = arg + 1;
4295                 return 1;
4296         }
4297         if (*arg != '\0')
4298                 return 0;
4299         /* separate form: --option value */
4300         if (!argv[1])
4301                 die("Option '--%s' requires a value", opt);
4302         *optarg = argv[1];
4303         return 2;
4304 }
4305
4306 static int stat_opt(struct diff_options *options, const char **av)
4307 {
4308         const char *arg = av[0];
4309         char *end;
4310         int width = options->stat_width;
4311         int name_width = options->stat_name_width;
4312         int graph_width = options->stat_graph_width;
4313         int count = options->stat_count;
4314         int argcount = 1;
4315
4316         if (!skip_prefix(arg, "--stat", &arg))
4317                 die("BUG: stat option does not begin with --stat: %s", arg);
4318         end = (char *)arg;
4319
4320         switch (*arg) {
4321         case '-':
4322                 if (skip_prefix(arg, "-width", &arg)) {
4323                         if (*arg == '=')
4324                                 width = strtoul(arg + 1, &end, 10);
4325                         else if (!*arg && !av[1])
4326                                 die_want_option("--stat-width");
4327                         else if (!*arg) {
4328                                 width = strtoul(av[1], &end, 10);
4329                                 argcount = 2;
4330                         }
4331                 } else if (skip_prefix(arg, "-name-width", &arg)) {
4332                         if (*arg == '=')
4333                                 name_width = strtoul(arg + 1, &end, 10);
4334                         else if (!*arg && !av[1])
4335                                 die_want_option("--stat-name-width");
4336                         else if (!*arg) {
4337                                 name_width = strtoul(av[1], &end, 10);
4338                                 argcount = 2;
4339                         }
4340                 } else if (skip_prefix(arg, "-graph-width", &arg)) {
4341                         if (*arg == '=')
4342                                 graph_width = strtoul(arg + 1, &end, 10);
4343                         else if (!*arg && !av[1])
4344                                 die_want_option("--stat-graph-width");
4345                         else if (!*arg) {
4346                                 graph_width = strtoul(av[1], &end, 10);
4347                                 argcount = 2;
4348                         }
4349                 } else if (skip_prefix(arg, "-count", &arg)) {
4350                         if (*arg == '=')
4351                                 count = strtoul(arg + 1, &end, 10);
4352                         else if (!*arg && !av[1])
4353                                 die_want_option("--stat-count");
4354                         else if (!*arg) {
4355                                 count = strtoul(av[1], &end, 10);
4356                                 argcount = 2;
4357                         }
4358                 }
4359                 break;
4360         case '=':
4361                 width = strtoul(arg+1, &end, 10);
4362                 if (*end == ',')
4363                         name_width = strtoul(end+1, &end, 10);
4364                 if (*end == ',')
4365                         count = strtoul(end+1, &end, 10);
4366         }
4367
4368         /* Important! This checks all the error cases! */
4369         if (*end)
4370                 return 0;
4371         options->output_format |= DIFF_FORMAT_DIFFSTAT;
4372         options->stat_name_width = name_width;
4373         options->stat_graph_width = graph_width;
4374         options->stat_width = width;
4375         options->stat_count = count;
4376         return argcount;
4377 }
4378
4379 static int parse_dirstat_opt(struct diff_options *options, const char *params)
4380 {
4381         struct strbuf errmsg = STRBUF_INIT;
4382         if (parse_dirstat_params(options, params, &errmsg))
4383                 die(_("Failed to parse --dirstat/-X option parameter:\n%s"),
4384                     errmsg.buf);
4385         strbuf_release(&errmsg);
4386         /*
4387          * The caller knows a dirstat-related option is given from the command
4388          * line; allow it to say "return this_function();"
4389          */
4390         options->output_format |= DIFF_FORMAT_DIRSTAT;
4391         return 1;
4392 }
4393
4394 static int parse_submodule_opt(struct diff_options *options, const char *value)
4395 {
4396         if (parse_submodule_params(options, value))
4397                 die(_("Failed to parse --submodule option parameter: '%s'"),
4398                         value);
4399         return 1;
4400 }
4401
4402 static const char diff_status_letters[] = {
4403         DIFF_STATUS_ADDED,
4404         DIFF_STATUS_COPIED,
4405         DIFF_STATUS_DELETED,
4406         DIFF_STATUS_MODIFIED,
4407         DIFF_STATUS_RENAMED,
4408         DIFF_STATUS_TYPE_CHANGED,
4409         DIFF_STATUS_UNKNOWN,
4410         DIFF_STATUS_UNMERGED,
4411         DIFF_STATUS_FILTER_AON,
4412         DIFF_STATUS_FILTER_BROKEN,
4413         '\0',
4414 };
4415
4416 static unsigned int filter_bit['Z' + 1];
4417
4418 static void prepare_filter_bits(void)
4419 {
4420         int i;
4421
4422         if (!filter_bit[DIFF_STATUS_ADDED]) {
4423                 for (i = 0; diff_status_letters[i]; i++)
4424                         filter_bit[(int) diff_status_letters[i]] = (1 << i);
4425         }
4426 }
4427
4428 static unsigned filter_bit_tst(char status, const struct diff_options *opt)
4429 {
4430         return opt->filter & filter_bit[(int) status];
4431 }
4432
4433 static int parse_diff_filter_opt(const char *optarg, struct diff_options *opt)
4434 {
4435         int i, optch;
4436
4437         prepare_filter_bits();
4438
4439         /*
4440          * If there is a negation e.g. 'd' in the input, and we haven't
4441          * initialized the filter field with another --diff-filter, start
4442          * from full set of bits, except for AON.
4443          */
4444         if (!opt->filter) {
4445                 for (i = 0; (optch = optarg[i]) != '\0'; i++) {
4446                         if (optch < 'a' || 'z' < optch)
4447                                 continue;
4448                         opt->filter = (1 << (ARRAY_SIZE(diff_status_letters) - 1)) - 1;
4449                         opt->filter &= ~filter_bit[DIFF_STATUS_FILTER_AON];
4450                         break;
4451                 }
4452         }
4453
4454         for (i = 0; (optch = optarg[i]) != '\0'; i++) {
4455                 unsigned int bit;
4456                 int negate;
4457
4458                 if ('a' <= optch && optch <= 'z') {
4459                         negate = 1;
4460                         optch = toupper(optch);
4461                 } else {
4462                         negate = 0;
4463                 }
4464
4465                 bit = (0 <= optch && optch <= 'Z') ? filter_bit[optch] : 0;
4466                 if (!bit)
4467                         return optarg[i];
4468                 if (negate)
4469                         opt->filter &= ~bit;
4470                 else
4471                         opt->filter |= bit;
4472         }
4473         return 0;
4474 }
4475
4476 static void enable_patch_output(int *fmt) {
4477         *fmt &= ~DIFF_FORMAT_NO_OUTPUT;
4478         *fmt |= DIFF_FORMAT_PATCH;
4479 }
4480
4481 static int parse_ws_error_highlight_opt(struct diff_options *opt, const char *arg)
4482 {
4483         int val = parse_ws_error_highlight(arg);
4484
4485         if (val < 0) {
4486                 error("unknown value after ws-error-highlight=%.*s",
4487                       -1 - val, arg);
4488                 return 0;
4489         }
4490         opt->ws_error_highlight = val;
4491         return 1;
4492 }
4493
4494 int diff_opt_parse(struct diff_options *options,
4495                    const char **av, int ac, const char *prefix)
4496 {
4497         const char *arg = av[0];
4498         const char *optarg;
4499         int argcount;
4500
4501         if (!prefix)
4502                 prefix = "";
4503
4504         /* Output format options */
4505         if (!strcmp(arg, "-p") || !strcmp(arg, "-u") || !strcmp(arg, "--patch")
4506             || opt_arg(arg, 'U', "unified", &options->context))
4507                 enable_patch_output(&options->output_format);
4508         else if (!strcmp(arg, "--raw"))
4509                 options->output_format |= DIFF_FORMAT_RAW;
4510         else if (!strcmp(arg, "--patch-with-raw")) {
4511                 enable_patch_output(&options->output_format);
4512                 options->output_format |= DIFF_FORMAT_RAW;
4513         } else if (!strcmp(arg, "--numstat"))
4514                 options->output_format |= DIFF_FORMAT_NUMSTAT;
4515         else if (!strcmp(arg, "--shortstat"))
4516                 options->output_format |= DIFF_FORMAT_SHORTSTAT;
4517         else if (skip_prefix(arg, "-X", &arg) ||
4518                  skip_to_optional_arg(arg, "--dirstat", &arg))
4519                 return parse_dirstat_opt(options, arg);
4520         else if (!strcmp(arg, "--cumulative"))
4521                 return parse_dirstat_opt(options, "cumulative");
4522         else if (skip_to_optional_arg(arg, "--dirstat-by-file", &arg)) {
4523                 parse_dirstat_opt(options, "files");
4524                 return parse_dirstat_opt(options, arg);
4525         }
4526         else if (!strcmp(arg, "--check"))
4527                 options->output_format |= DIFF_FORMAT_CHECKDIFF;
4528         else if (!strcmp(arg, "--summary"))
4529                 options->output_format |= DIFF_FORMAT_SUMMARY;
4530         else if (!strcmp(arg, "--patch-with-stat")) {
4531                 enable_patch_output(&options->output_format);
4532                 options->output_format |= DIFF_FORMAT_DIFFSTAT;
4533         } else if (!strcmp(arg, "--name-only"))
4534                 options->output_format |= DIFF_FORMAT_NAME;
4535         else if (!strcmp(arg, "--name-status"))
4536                 options->output_format |= DIFF_FORMAT_NAME_STATUS;
4537         else if (!strcmp(arg, "-s") || !strcmp(arg, "--no-patch"))
4538                 options->output_format |= DIFF_FORMAT_NO_OUTPUT;
4539         else if (starts_with(arg, "--stat"))
4540                 /* --stat, --stat-width, --stat-name-width, or --stat-count */
4541                 return stat_opt(options, av);
4542
4543         /* renames options */
4544         else if (starts_with(arg, "-B") ||
4545                  skip_to_optional_arg(arg, "--break-rewrites", NULL)) {
4546                 if ((options->break_opt = diff_scoreopt_parse(arg)) == -1)
4547                         return error("invalid argument to -B: %s", arg+2);
4548         }
4549         else if (starts_with(arg, "-M") ||
4550                  skip_to_optional_arg(arg, "--find-renames", NULL)) {
4551                 if ((options->rename_score = diff_scoreopt_parse(arg)) == -1)
4552                         return error("invalid argument to -M: %s", arg+2);
4553                 options->detect_rename = DIFF_DETECT_RENAME;
4554         }
4555         else if (!strcmp(arg, "-D") || !strcmp(arg, "--irreversible-delete")) {
4556                 options->irreversible_delete = 1;
4557         }
4558         else if (starts_with(arg, "-C") ||
4559                  skip_to_optional_arg(arg, "--find-copies", NULL)) {
4560                 if (options->detect_rename == DIFF_DETECT_COPY)
4561                         options->flags.find_copies_harder = 1;
4562                 if ((options->rename_score = diff_scoreopt_parse(arg)) == -1)
4563                         return error("invalid argument to -C: %s", arg+2);
4564                 options->detect_rename = DIFF_DETECT_COPY;
4565         }
4566         else if (!strcmp(arg, "--no-renames"))
4567                 options->detect_rename = 0;
4568         else if (!strcmp(arg, "--rename-empty"))
4569                 options->flags.rename_empty = 1;
4570         else if (!strcmp(arg, "--no-rename-empty"))
4571                 options->flags.rename_empty = 0;
4572         else if (skip_to_optional_arg_default(arg, "--relative", &arg, NULL)) {
4573                 options->flags.relative_name = 1;
4574                 if (arg)
4575                         options->prefix = arg;
4576         }
4577
4578         /* xdiff options */
4579         else if (!strcmp(arg, "--minimal"))
4580                 DIFF_XDL_SET(options, NEED_MINIMAL);
4581         else if (!strcmp(arg, "--no-minimal"))
4582                 DIFF_XDL_CLR(options, NEED_MINIMAL);
4583         else if (!strcmp(arg, "-w") || !strcmp(arg, "--ignore-all-space"))
4584                 DIFF_XDL_SET(options, IGNORE_WHITESPACE);
4585         else if (!strcmp(arg, "-b") || !strcmp(arg, "--ignore-space-change"))
4586                 DIFF_XDL_SET(options, IGNORE_WHITESPACE_CHANGE);
4587         else if (!strcmp(arg, "--ignore-space-at-eol"))
4588                 DIFF_XDL_SET(options, IGNORE_WHITESPACE_AT_EOL);
4589         else if (!strcmp(arg, "--ignore-cr-at-eol"))
4590                 DIFF_XDL_SET(options, IGNORE_CR_AT_EOL);
4591         else if (!strcmp(arg, "--ignore-blank-lines"))
4592                 DIFF_XDL_SET(options, IGNORE_BLANK_LINES);
4593         else if (!strcmp(arg, "--indent-heuristic"))
4594                 DIFF_XDL_SET(options, INDENT_HEURISTIC);
4595         else if (!strcmp(arg, "--no-indent-heuristic"))
4596                 DIFF_XDL_CLR(options, INDENT_HEURISTIC);
4597         else if (!strcmp(arg, "--patience")) {
4598                 int i;
4599                 options->xdl_opts = DIFF_WITH_ALG(options, PATIENCE_DIFF);
4600                 /*
4601                  * Both --patience and --anchored use PATIENCE_DIFF
4602                  * internally, so remove any anchors previously
4603                  * specified.
4604                  */
4605                 for (i = 0; i < options->anchors_nr; i++)
4606                         free(options->anchors[i]);
4607                 options->anchors_nr = 0;
4608         } else if (!strcmp(arg, "--histogram"))
4609                 options->xdl_opts = DIFF_WITH_ALG(options, HISTOGRAM_DIFF);
4610         else if ((argcount = parse_long_opt("diff-algorithm", av, &optarg))) {
4611                 long value = parse_algorithm_value(optarg);
4612                 if (value < 0)
4613                         return error("option diff-algorithm accepts \"myers\", "
4614                                      "\"minimal\", \"patience\" and \"histogram\"");
4615                 /* clear out previous settings */
4616                 DIFF_XDL_CLR(options, NEED_MINIMAL);
4617                 options->xdl_opts &= ~XDF_DIFF_ALGORITHM_MASK;
4618                 options->xdl_opts |= value;
4619                 return argcount;
4620         } else if (skip_prefix(arg, "--anchored=", &arg)) {
4621                 options->xdl_opts = DIFF_WITH_ALG(options, PATIENCE_DIFF);
4622                 ALLOC_GROW(options->anchors, options->anchors_nr + 1,
4623                            options->anchors_alloc);
4624                 options->anchors[options->anchors_nr++] = xstrdup(arg);
4625         }
4626
4627         /* flags options */
4628         else if (!strcmp(arg, "--binary")) {
4629                 enable_patch_output(&options->output_format);
4630                 options->flags.binary = 1;
4631         }
4632         else if (!strcmp(arg, "--full-index"))
4633                 options->flags.full_index = 1;
4634         else if (!strcmp(arg, "-a") || !strcmp(arg, "--text"))
4635                 options->flags.text = 1;
4636         else if (!strcmp(arg, "-R"))
4637                 options->flags.reverse_diff = 1;
4638         else if (!strcmp(arg, "--find-copies-harder"))
4639                 options->flags.find_copies_harder = 1;
4640         else if (!strcmp(arg, "--follow"))
4641                 options->flags.follow_renames = 1;
4642         else if (!strcmp(arg, "--no-follow")) {
4643                 options->flags.follow_renames = 0;
4644                 options->flags.default_follow_renames = 0;
4645         } else if (skip_to_optional_arg_default(arg, "--color", &arg, "always")) {
4646                 int value = git_config_colorbool(NULL, arg);
4647                 if (value < 0)
4648                         return error("option `color' expects \"always\", \"auto\", or \"never\"");
4649                 options->use_color = value;
4650         }
4651         else if (!strcmp(arg, "--no-color"))
4652                 options->use_color = 0;
4653         else if (!strcmp(arg, "--color-moved")) {
4654                 if (diff_color_moved_default)
4655                         options->color_moved = diff_color_moved_default;
4656                 if (options->color_moved == COLOR_MOVED_NO)
4657                         options->color_moved = COLOR_MOVED_DEFAULT;
4658         } else if (!strcmp(arg, "--no-color-moved"))
4659                 options->color_moved = COLOR_MOVED_NO;
4660         else if (skip_prefix(arg, "--color-moved=", &arg)) {
4661                 int cm = parse_color_moved(arg);
4662                 if (cm < 0)
4663                         die("bad --color-moved argument: %s", arg);
4664                 options->color_moved = cm;
4665         } else if (skip_to_optional_arg_default(arg, "--color-words", &options->word_regex, NULL)) {
4666                 options->use_color = 1;
4667                 options->word_diff = DIFF_WORDS_COLOR;
4668         }
4669         else if (!strcmp(arg, "--word-diff")) {
4670                 if (options->word_diff == DIFF_WORDS_NONE)
4671                         options->word_diff = DIFF_WORDS_PLAIN;
4672         }
4673         else if (skip_prefix(arg, "--word-diff=", &arg)) {
4674                 if (!strcmp(arg, "plain"))
4675                         options->word_diff = DIFF_WORDS_PLAIN;
4676                 else if (!strcmp(arg, "color")) {
4677                         options->use_color = 1;
4678                         options->word_diff = DIFF_WORDS_COLOR;
4679                 }
4680                 else if (!strcmp(arg, "porcelain"))
4681                         options->word_diff = DIFF_WORDS_PORCELAIN;
4682                 else if (!strcmp(arg, "none"))
4683                         options->word_diff = DIFF_WORDS_NONE;
4684                 else
4685                         die("bad --word-diff argument: %s", arg);
4686         }
4687         else if ((argcount = parse_long_opt("word-diff-regex", av, &optarg))) {
4688                 if (options->word_diff == DIFF_WORDS_NONE)
4689                         options->word_diff = DIFF_WORDS_PLAIN;
4690                 options->word_regex = optarg;
4691                 return argcount;
4692         }
4693         else if (!strcmp(arg, "--exit-code"))
4694                 options->flags.exit_with_status = 1;
4695         else if (!strcmp(arg, "--quiet"))
4696                 options->flags.quick = 1;
4697         else if (!strcmp(arg, "--ext-diff"))
4698                 options->flags.allow_external = 1;
4699         else if (!strcmp(arg, "--no-ext-diff"))
4700                 options->flags.allow_external = 0;
4701         else if (!strcmp(arg, "--textconv")) {
4702                 options->flags.allow_textconv = 1;
4703                 options->flags.textconv_set_via_cmdline = 1;
4704         } else if (!strcmp(arg, "--no-textconv"))
4705                 options->flags.allow_textconv = 0;
4706         else if (skip_to_optional_arg_default(arg, "--ignore-submodules", &arg, "all")) {
4707                 options->flags.override_submodule_config = 1;
4708                 handle_ignore_submodules_arg(options, arg);
4709         } else if (skip_to_optional_arg_default(arg, "--submodule", &arg, "log"))
4710                 return parse_submodule_opt(options, arg);
4711         else if (skip_prefix(arg, "--ws-error-highlight=", &arg))
4712                 return parse_ws_error_highlight_opt(options, arg);
4713         else if (!strcmp(arg, "--ita-invisible-in-index"))
4714                 options->ita_invisible_in_index = 1;
4715         else if (!strcmp(arg, "--ita-visible-in-index"))
4716                 options->ita_invisible_in_index = 0;
4717
4718         /* misc options */
4719         else if (!strcmp(arg, "-z"))
4720                 options->line_termination = 0;
4721         else if ((argcount = short_opt('l', av, &optarg))) {
4722                 options->rename_limit = strtoul(optarg, NULL, 10);
4723                 return argcount;
4724         }
4725         else if ((argcount = short_opt('S', av, &optarg))) {
4726                 options->pickaxe = optarg;
4727                 options->pickaxe_opts |= DIFF_PICKAXE_KIND_S;
4728                 return argcount;
4729         } else if ((argcount = short_opt('G', av, &optarg))) {
4730                 options->pickaxe = optarg;
4731                 options->pickaxe_opts |= DIFF_PICKAXE_KIND_G;
4732                 return argcount;
4733         }
4734         else if (!strcmp(arg, "--pickaxe-all"))
4735                 options->pickaxe_opts |= DIFF_PICKAXE_ALL;
4736         else if (!strcmp(arg, "--pickaxe-regex"))
4737                 options->pickaxe_opts |= DIFF_PICKAXE_REGEX;
4738         else if ((argcount = short_opt('O', av, &optarg))) {
4739                 options->orderfile = prefix_filename(prefix, optarg);
4740                 return argcount;
4741         }
4742         else if ((argcount = parse_long_opt("diff-filter", av, &optarg))) {
4743                 int offending = parse_diff_filter_opt(optarg, options);
4744                 if (offending)
4745                         die("unknown change class '%c' in --diff-filter=%s",
4746                             offending, optarg);
4747                 return argcount;
4748         }
4749         else if (!strcmp(arg, "--no-abbrev"))
4750                 options->abbrev = 0;
4751         else if (!strcmp(arg, "--abbrev"))
4752                 options->abbrev = DEFAULT_ABBREV;
4753         else if (skip_prefix(arg, "--abbrev=", &arg)) {
4754                 options->abbrev = strtoul(arg, NULL, 10);
4755                 if (options->abbrev < MINIMUM_ABBREV)
4756                         options->abbrev = MINIMUM_ABBREV;
4757                 else if (40 < options->abbrev)
4758                         options->abbrev = 40;
4759         }
4760         else if ((argcount = parse_long_opt("src-prefix", av, &optarg))) {
4761                 options->a_prefix = optarg;
4762                 return argcount;
4763         }
4764         else if ((argcount = parse_long_opt("line-prefix", av, &optarg))) {
4765                 options->line_prefix = optarg;
4766                 options->line_prefix_length = strlen(options->line_prefix);
4767                 graph_setup_line_prefix(options);
4768                 return argcount;
4769         }
4770         else if ((argcount = parse_long_opt("dst-prefix", av, &optarg))) {
4771                 options->b_prefix = optarg;
4772                 return argcount;
4773         }
4774         else if (!strcmp(arg, "--no-prefix"))
4775                 options->a_prefix = options->b_prefix = "";
4776         else if (opt_arg(arg, '\0', "inter-hunk-context",
4777                          &options->interhunkcontext))
4778                 ;
4779         else if (!strcmp(arg, "-W"))
4780                 options->flags.funccontext = 1;
4781         else if (!strcmp(arg, "--function-context"))
4782                 options->flags.funccontext = 1;
4783         else if (!strcmp(arg, "--no-function-context"))
4784                 options->flags.funccontext = 0;
4785         else if ((argcount = parse_long_opt("output", av, &optarg))) {
4786                 char *path = prefix_filename(prefix, optarg);
4787                 options->file = xfopen(path, "w");
4788                 options->close_file = 1;
4789                 if (options->use_color != GIT_COLOR_ALWAYS)
4790                         options->use_color = GIT_COLOR_NEVER;
4791                 free(path);
4792                 return argcount;
4793         } else
4794                 return 0;
4795         return 1;
4796 }
4797
4798 int parse_rename_score(const char **cp_p)
4799 {
4800         unsigned long num, scale;
4801         int ch, dot;
4802         const char *cp = *cp_p;
4803
4804         num = 0;
4805         scale = 1;
4806         dot = 0;
4807         for (;;) {
4808                 ch = *cp;
4809                 if ( !dot && ch == '.' ) {
4810                         scale = 1;
4811                         dot = 1;
4812                 } else if ( ch == '%' ) {
4813                         scale = dot ? scale*100 : 100;
4814                         cp++;   /* % is always at the end */
4815                         break;
4816                 } else if ( ch >= '0' && ch <= '9' ) {
4817                         if ( scale < 100000 ) {
4818                                 scale *= 10;
4819                                 num = (num*10) + (ch-'0');
4820                         }
4821                 } else {
4822                         break;
4823                 }
4824                 cp++;
4825         }
4826         *cp_p = cp;
4827
4828         /* user says num divided by scale and we say internally that
4829          * is MAX_SCORE * num / scale.
4830          */
4831         return (int)((num >= scale) ? MAX_SCORE : (MAX_SCORE * num / scale));
4832 }
4833
4834 static int diff_scoreopt_parse(const char *opt)
4835 {
4836         int opt1, opt2, cmd;
4837
4838         if (*opt++ != '-')
4839                 return -1;
4840         cmd = *opt++;
4841         if (cmd == '-') {
4842                 /* convert the long-form arguments into short-form versions */
4843                 if (skip_prefix(opt, "break-rewrites", &opt)) {
4844                         if (*opt == 0 || *opt++ == '=')
4845                                 cmd = 'B';
4846                 } else if (skip_prefix(opt, "find-copies", &opt)) {
4847                         if (*opt == 0 || *opt++ == '=')
4848                                 cmd = 'C';
4849                 } else if (skip_prefix(opt, "find-renames", &opt)) {
4850                         if (*opt == 0 || *opt++ == '=')
4851                                 cmd = 'M';
4852                 }
4853         }
4854         if (cmd != 'M' && cmd != 'C' && cmd != 'B')
4855                 return -1; /* that is not a -M, -C, or -B option */
4856
4857         opt1 = parse_rename_score(&opt);
4858         if (cmd != 'B')
4859                 opt2 = 0;
4860         else {
4861                 if (*opt == 0)
4862                         opt2 = 0;
4863                 else if (*opt != '/')
4864                         return -1; /* we expect -B80/99 or -B80 */
4865                 else {
4866                         opt++;
4867                         opt2 = parse_rename_score(&opt);
4868                 }
4869         }
4870         if (*opt != 0)
4871                 return -1;
4872         return opt1 | (opt2 << 16);
4873 }
4874
4875 struct diff_queue_struct diff_queued_diff;
4876
4877 void diff_q(struct diff_queue_struct *queue, struct diff_filepair *dp)
4878 {
4879         ALLOC_GROW(queue->queue, queue->nr + 1, queue->alloc);
4880         queue->queue[queue->nr++] = dp;
4881 }
4882
4883 struct diff_filepair *diff_queue(struct diff_queue_struct *queue,
4884                                  struct diff_filespec *one,
4885                                  struct diff_filespec *two)
4886 {
4887         struct diff_filepair *dp = xcalloc(1, sizeof(*dp));
4888         dp->one = one;
4889         dp->two = two;
4890         if (queue)
4891                 diff_q(queue, dp);
4892         return dp;
4893 }
4894
4895 void diff_free_filepair(struct diff_filepair *p)
4896 {
4897         free_filespec(p->one);
4898         free_filespec(p->two);
4899         free(p);
4900 }
4901
4902 const char *diff_aligned_abbrev(const struct object_id *oid, int len)
4903 {
4904         int abblen;
4905         const char *abbrev;
4906
4907         /* Do we want all 40 hex characters? */
4908         if (len == GIT_SHA1_HEXSZ)
4909                 return oid_to_hex(oid);
4910
4911         /* An abbreviated value is fine, possibly followed by an ellipsis. */
4912         abbrev = diff_abbrev_oid(oid, len);
4913
4914         if (!print_sha1_ellipsis())
4915                 return abbrev;
4916
4917         abblen = strlen(abbrev);
4918
4919         /*
4920          * In well-behaved cases, where the abbreviated result is the
4921          * same as the requested length, append three dots after the
4922          * abbreviation (hence the whole logic is limited to the case
4923          * where abblen < 37); when the actual abbreviated result is a
4924          * bit longer than the requested length, we reduce the number
4925          * of dots so that they match the well-behaved ones.  However,
4926          * if the actual abbreviation is longer than the requested
4927          * length by more than three, we give up on aligning, and add
4928          * three dots anyway, to indicate that the output is not the
4929          * full object name.  Yes, this may be suboptimal, but this
4930          * appears only in "diff --raw --abbrev" output and it is not
4931          * worth the effort to change it now.  Note that this would
4932          * likely to work fine when the automatic sizing of default
4933          * abbreviation length is used--we would be fed -1 in "len" in
4934          * that case, and will end up always appending three-dots, but
4935          * the automatic sizing is supposed to give abblen that ensures
4936          * uniqueness across all objects (statistically speaking).
4937          */
4938         if (abblen < GIT_SHA1_HEXSZ - 3) {
4939                 static char hex[GIT_MAX_HEXSZ + 1];
4940                 if (len < abblen && abblen <= len + 2)
4941                         xsnprintf(hex, sizeof(hex), "%s%.*s", abbrev, len+3-abblen, "..");
4942                 else
4943                         xsnprintf(hex, sizeof(hex), "%s...", abbrev);
4944                 return hex;
4945         }
4946
4947         return oid_to_hex(oid);
4948 }
4949
4950 static void diff_flush_raw(struct diff_filepair *p, struct diff_options *opt)
4951 {
4952         int line_termination = opt->line_termination;
4953         int inter_name_termination = line_termination ? '\t' : '\0';
4954
4955         fprintf(opt->file, "%s", diff_line_prefix(opt));
4956         if (!(opt->output_format & DIFF_FORMAT_NAME_STATUS)) {
4957                 fprintf(opt->file, ":%06o %06o %s ", p->one->mode, p->two->mode,
4958                         diff_aligned_abbrev(&p->one->oid, opt->abbrev));
4959                 fprintf(opt->file, "%s ",
4960                         diff_aligned_abbrev(&p->two->oid, opt->abbrev));
4961         }
4962         if (p->score) {
4963                 fprintf(opt->file, "%c%03d%c", p->status, similarity_index(p),
4964                         inter_name_termination);
4965         } else {
4966                 fprintf(opt->file, "%c%c", p->status, inter_name_termination);
4967         }
4968
4969         if (p->status == DIFF_STATUS_COPIED ||
4970             p->status == DIFF_STATUS_RENAMED) {
4971                 const char *name_a, *name_b;
4972                 name_a = p->one->path;
4973                 name_b = p->two->path;
4974                 strip_prefix(opt->prefix_length, &name_a, &name_b);
4975                 write_name_quoted(name_a, opt->file, inter_name_termination);
4976                 write_name_quoted(name_b, opt->file, line_termination);
4977         } else {
4978                 const char *name_a, *name_b;
4979                 name_a = p->one->mode ? p->one->path : p->two->path;
4980                 name_b = NULL;
4981                 strip_prefix(opt->prefix_length, &name_a, &name_b);
4982                 write_name_quoted(name_a, opt->file, line_termination);
4983         }
4984 }
4985
4986 int diff_unmodified_pair(struct diff_filepair *p)
4987 {
4988         /* This function is written stricter than necessary to support
4989          * the currently implemented transformers, but the idea is to
4990          * let transformers to produce diff_filepairs any way they want,
4991          * and filter and clean them up here before producing the output.
4992          */
4993         struct diff_filespec *one = p->one, *two = p->two;
4994
4995         if (DIFF_PAIR_UNMERGED(p))
4996                 return 0; /* unmerged is interesting */
4997
4998         /* deletion, addition, mode or type change
4999          * and rename are all interesting.
5000          */
5001         if (DIFF_FILE_VALID(one) != DIFF_FILE_VALID(two) ||
5002             DIFF_PAIR_MODE_CHANGED(p) ||
5003             strcmp(one->path, two->path))
5004                 return 0;
5005
5006         /* both are valid and point at the same path.  that is, we are
5007          * dealing with a change.
5008          */
5009         if (one->oid_valid && two->oid_valid &&
5010             !oidcmp(&one->oid, &two->oid) &&
5011             !one->dirty_submodule && !two->dirty_submodule)
5012                 return 1; /* no change */
5013         if (!one->oid_valid && !two->oid_valid)
5014                 return 1; /* both look at the same file on the filesystem. */
5015         return 0;
5016 }
5017
5018 static void diff_flush_patch(struct diff_filepair *p, struct diff_options *o)
5019 {
5020         if (diff_unmodified_pair(p))
5021                 return;
5022
5023         if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
5024             (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
5025                 return; /* no tree diffs in patch format */
5026
5027         run_diff(p, o);
5028 }
5029
5030 static void diff_flush_stat(struct diff_filepair *p, struct diff_options *o,
5031                             struct diffstat_t *diffstat)
5032 {
5033         if (diff_unmodified_pair(p))
5034                 return;
5035
5036         if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
5037             (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
5038                 return; /* no useful stat for tree diffs */
5039
5040         run_diffstat(p, o, diffstat);
5041 }
5042
5043 static void diff_flush_checkdiff(struct diff_filepair *p,
5044                 struct diff_options *o)
5045 {
5046         if (diff_unmodified_pair(p))
5047                 return;
5048
5049         if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
5050             (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
5051                 return; /* nothing to check in tree diffs */
5052
5053         run_checkdiff(p, o);
5054 }
5055
5056 int diff_queue_is_empty(void)
5057 {
5058         struct diff_queue_struct *q = &diff_queued_diff;
5059         int i;
5060         for (i = 0; i < q->nr; i++)
5061                 if (!diff_unmodified_pair(q->queue[i]))
5062                         return 0;
5063         return 1;
5064 }
5065
5066 #if DIFF_DEBUG
5067 void diff_debug_filespec(struct diff_filespec *s, int x, const char *one)
5068 {
5069         fprintf(stderr, "queue[%d] %s (%s) %s %06o %s\n",
5070                 x, one ? one : "",
5071                 s->path,
5072                 DIFF_FILE_VALID(s) ? "valid" : "invalid",
5073                 s->mode,
5074                 s->oid_valid ? oid_to_hex(&s->oid) : "");
5075         fprintf(stderr, "queue[%d] %s size %lu\n",
5076                 x, one ? one : "",
5077                 s->size);
5078 }
5079
5080 void diff_debug_filepair(const struct diff_filepair *p, int i)
5081 {
5082         diff_debug_filespec(p->one, i, "one");
5083         diff_debug_filespec(p->two, i, "two");
5084         fprintf(stderr, "score %d, status %c rename_used %d broken %d\n",
5085                 p->score, p->status ? p->status : '?',
5086                 p->one->rename_used, p->broken_pair);
5087 }
5088
5089 void diff_debug_queue(const char *msg, struct diff_queue_struct *q)
5090 {
5091         int i;
5092         if (msg)
5093                 fprintf(stderr, "%s\n", msg);
5094         fprintf(stderr, "q->nr = %d\n", q->nr);
5095         for (i = 0; i < q->nr; i++) {
5096                 struct diff_filepair *p = q->queue[i];
5097                 diff_debug_filepair(p, i);
5098         }
5099 }
5100 #endif
5101
5102 static void diff_resolve_rename_copy(void)
5103 {
5104         int i;
5105         struct diff_filepair *p;
5106         struct diff_queue_struct *q = &diff_queued_diff;
5107
5108         diff_debug_queue("resolve-rename-copy", q);
5109
5110         for (i = 0; i < q->nr; i++) {
5111                 p = q->queue[i];
5112                 p->status = 0; /* undecided */
5113                 if (DIFF_PAIR_UNMERGED(p))
5114                         p->status = DIFF_STATUS_UNMERGED;
5115                 else if (!DIFF_FILE_VALID(p->one))
5116                         p->status = DIFF_STATUS_ADDED;
5117                 else if (!DIFF_FILE_VALID(p->two))
5118                         p->status = DIFF_STATUS_DELETED;
5119                 else if (DIFF_PAIR_TYPE_CHANGED(p))
5120                         p->status = DIFF_STATUS_TYPE_CHANGED;
5121
5122                 /* from this point on, we are dealing with a pair
5123                  * whose both sides are valid and of the same type, i.e.
5124                  * either in-place edit or rename/copy edit.
5125                  */
5126                 else if (DIFF_PAIR_RENAME(p)) {
5127                         /*
5128                          * A rename might have re-connected a broken
5129                          * pair up, causing the pathnames to be the
5130                          * same again. If so, that's not a rename at
5131                          * all, just a modification..
5132                          *
5133                          * Otherwise, see if this source was used for
5134                          * multiple renames, in which case we decrement
5135                          * the count, and call it a copy.
5136                          */
5137                         if (!strcmp(p->one->path, p->two->path))
5138                                 p->status = DIFF_STATUS_MODIFIED;
5139                         else if (--p->one->rename_used > 0)
5140                                 p->status = DIFF_STATUS_COPIED;
5141                         else
5142                                 p->status = DIFF_STATUS_RENAMED;
5143                 }
5144                 else if (oidcmp(&p->one->oid, &p->two->oid) ||
5145                          p->one->mode != p->two->mode ||
5146                          p->one->dirty_submodule ||
5147                          p->two->dirty_submodule ||
5148                          is_null_oid(&p->one->oid))
5149                         p->status = DIFF_STATUS_MODIFIED;
5150                 else {
5151                         /* This is a "no-change" entry and should not
5152                          * happen anymore, but prepare for broken callers.
5153                          */
5154                         error("feeding unmodified %s to diffcore",
5155                               p->one->path);
5156                         p->status = DIFF_STATUS_UNKNOWN;
5157                 }
5158         }
5159         diff_debug_queue("resolve-rename-copy done", q);
5160 }
5161
5162 static int check_pair_status(struct diff_filepair *p)
5163 {
5164         switch (p->status) {
5165         case DIFF_STATUS_UNKNOWN:
5166                 return 0;
5167         case 0:
5168                 die("internal error in diff-resolve-rename-copy");
5169         default:
5170                 return 1;
5171         }
5172 }
5173
5174 static void flush_one_pair(struct diff_filepair *p, struct diff_options *opt)
5175 {
5176         int fmt = opt->output_format;
5177
5178         if (fmt & DIFF_FORMAT_CHECKDIFF)
5179                 diff_flush_checkdiff(p, opt);
5180         else if (fmt & (DIFF_FORMAT_RAW | DIFF_FORMAT_NAME_STATUS))
5181                 diff_flush_raw(p, opt);
5182         else if (fmt & DIFF_FORMAT_NAME) {
5183                 const char *name_a, *name_b;
5184                 name_a = p->two->path;
5185                 name_b = NULL;
5186                 strip_prefix(opt->prefix_length, &name_a, &name_b);
5187                 fprintf(opt->file, "%s", diff_line_prefix(opt));
5188                 write_name_quoted(name_a, opt->file, opt->line_termination);
5189         }
5190 }
5191
5192 static void show_file_mode_name(struct diff_options *opt, const char *newdelete, struct diff_filespec *fs)
5193 {
5194         struct strbuf sb = STRBUF_INIT;
5195         if (fs->mode)
5196                 strbuf_addf(&sb, " %s mode %06o ", newdelete, fs->mode);
5197         else
5198                 strbuf_addf(&sb, " %s ", newdelete);
5199
5200         quote_c_style(fs->path, &sb, NULL, 0);
5201         strbuf_addch(&sb, '\n');
5202         emit_diff_symbol(opt, DIFF_SYMBOL_SUMMARY,
5203                          sb.buf, sb.len, 0);
5204         strbuf_release(&sb);
5205 }
5206
5207 static void show_mode_change(struct diff_options *opt, struct diff_filepair *p,
5208                 int show_name)
5209 {
5210         if (p->one->mode && p->two->mode && p->one->mode != p->two->mode) {
5211                 struct strbuf sb = STRBUF_INIT;
5212                 strbuf_addf(&sb, " mode change %06o => %06o",
5213                             p->one->mode, p->two->mode);
5214                 if (show_name) {
5215                         strbuf_addch(&sb, ' ');
5216                         quote_c_style(p->two->path, &sb, NULL, 0);
5217                 }
5218                 strbuf_addch(&sb, '\n');
5219                 emit_diff_symbol(opt, DIFF_SYMBOL_SUMMARY,
5220                                  sb.buf, sb.len, 0);
5221                 strbuf_release(&sb);
5222         }
5223 }
5224
5225 static void show_rename_copy(struct diff_options *opt, const char *renamecopy,
5226                 struct diff_filepair *p)
5227 {
5228         struct strbuf sb = STRBUF_INIT;
5229         char *names = pprint_rename(p->one->path, p->two->path);
5230         strbuf_addf(&sb, " %s %s (%d%%)\n",
5231                         renamecopy, names, similarity_index(p));
5232         free(names);
5233         emit_diff_symbol(opt, DIFF_SYMBOL_SUMMARY,
5234                                  sb.buf, sb.len, 0);
5235         show_mode_change(opt, p, 0);
5236         strbuf_release(&sb);
5237 }
5238
5239 static void diff_summary(struct diff_options *opt, struct diff_filepair *p)
5240 {
5241         switch(p->status) {
5242         case DIFF_STATUS_DELETED:
5243                 show_file_mode_name(opt, "delete", p->one);
5244                 break;
5245         case DIFF_STATUS_ADDED:
5246                 show_file_mode_name(opt, "create", p->two);
5247                 break;
5248         case DIFF_STATUS_COPIED:
5249                 show_rename_copy(opt, "copy", p);
5250                 break;
5251         case DIFF_STATUS_RENAMED:
5252                 show_rename_copy(opt, "rename", p);
5253                 break;
5254         default:
5255                 if (p->score) {
5256                         struct strbuf sb = STRBUF_INIT;
5257                         strbuf_addstr(&sb, " rewrite ");
5258                         quote_c_style(p->two->path, &sb, NULL, 0);
5259                         strbuf_addf(&sb, " (%d%%)\n", similarity_index(p));
5260                         emit_diff_symbol(opt, DIFF_SYMBOL_SUMMARY,
5261                                          sb.buf, sb.len, 0);
5262                         strbuf_release(&sb);
5263                 }
5264                 show_mode_change(opt, p, !p->score);
5265                 break;
5266         }
5267 }
5268
5269 struct patch_id_t {
5270         git_SHA_CTX *ctx;
5271         int patchlen;
5272 };
5273
5274 static int remove_space(char *line, int len)
5275 {
5276         int i;
5277         char *dst = line;
5278         unsigned char c;
5279
5280         for (i = 0; i < len; i++)
5281                 if (!isspace((c = line[i])))
5282                         *dst++ = c;
5283
5284         return dst - line;
5285 }
5286
5287 static void patch_id_consume(void *priv, char *line, unsigned long len)
5288 {
5289         struct patch_id_t *data = priv;
5290         int new_len;
5291
5292         /* Ignore line numbers when computing the SHA1 of the patch */
5293         if (starts_with(line, "@@ -"))
5294                 return;
5295
5296         new_len = remove_space(line, len);
5297
5298         git_SHA1_Update(data->ctx, line, new_len);
5299         data->patchlen += new_len;
5300 }
5301
5302 static void patch_id_add_string(git_SHA_CTX *ctx, const char *str)
5303 {
5304         git_SHA1_Update(ctx, str, strlen(str));
5305 }
5306
5307 static void patch_id_add_mode(git_SHA_CTX *ctx, unsigned mode)
5308 {
5309         /* large enough for 2^32 in octal */
5310         char buf[12];
5311         int len = xsnprintf(buf, sizeof(buf), "%06o", mode);
5312         git_SHA1_Update(ctx, buf, len);
5313 }
5314
5315 /* returns 0 upon success, and writes result into sha1 */
5316 static int diff_get_patch_id(struct diff_options *options, struct object_id *oid, int diff_header_only)
5317 {
5318         struct diff_queue_struct *q = &diff_queued_diff;
5319         int i;
5320         git_SHA_CTX ctx;
5321         struct patch_id_t data;
5322
5323         git_SHA1_Init(&ctx);
5324         memset(&data, 0, sizeof(struct patch_id_t));
5325         data.ctx = &ctx;
5326
5327         for (i = 0; i < q->nr; i++) {
5328                 xpparam_t xpp;
5329                 xdemitconf_t xecfg;
5330                 mmfile_t mf1, mf2;
5331                 struct diff_filepair *p = q->queue[i];
5332                 int len1, len2;
5333
5334                 memset(&xpp, 0, sizeof(xpp));
5335                 memset(&xecfg, 0, sizeof(xecfg));
5336                 if (p->status == 0)
5337                         return error("internal diff status error");
5338                 if (p->status == DIFF_STATUS_UNKNOWN)
5339                         continue;
5340                 if (diff_unmodified_pair(p))
5341                         continue;
5342                 if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
5343                     (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
5344                         continue;
5345                 if (DIFF_PAIR_UNMERGED(p))
5346                         continue;
5347
5348                 diff_fill_oid_info(p->one);
5349                 diff_fill_oid_info(p->two);
5350
5351                 len1 = remove_space(p->one->path, strlen(p->one->path));
5352                 len2 = remove_space(p->two->path, strlen(p->two->path));
5353                 patch_id_add_string(&ctx, "diff--git");
5354                 patch_id_add_string(&ctx, "a/");
5355                 git_SHA1_Update(&ctx, p->one->path, len1);
5356                 patch_id_add_string(&ctx, "b/");
5357                 git_SHA1_Update(&ctx, p->two->path, len2);
5358
5359                 if (p->one->mode == 0) {
5360                         patch_id_add_string(&ctx, "newfilemode");
5361                         patch_id_add_mode(&ctx, p->two->mode);
5362                         patch_id_add_string(&ctx, "---/dev/null");
5363                         patch_id_add_string(&ctx, "+++b/");
5364                         git_SHA1_Update(&ctx, p->two->path, len2);
5365                 } else if (p->two->mode == 0) {
5366                         patch_id_add_string(&ctx, "deletedfilemode");
5367                         patch_id_add_mode(&ctx, p->one->mode);
5368                         patch_id_add_string(&ctx, "---a/");
5369                         git_SHA1_Update(&ctx, p->one->path, len1);
5370                         patch_id_add_string(&ctx, "+++/dev/null");
5371                 } else {
5372                         patch_id_add_string(&ctx, "---a/");
5373                         git_SHA1_Update(&ctx, p->one->path, len1);
5374                         patch_id_add_string(&ctx, "+++b/");
5375                         git_SHA1_Update(&ctx, p->two->path, len2);
5376                 }
5377
5378                 if (diff_header_only)
5379                         continue;
5380
5381                 if (fill_mmfile(&mf1, p->one) < 0 ||
5382                     fill_mmfile(&mf2, p->two) < 0)
5383                         return error("unable to read files to diff");
5384
5385                 if (diff_filespec_is_binary(p->one) ||
5386                     diff_filespec_is_binary(p->two)) {
5387                         git_SHA1_Update(&ctx, oid_to_hex(&p->one->oid),
5388                                         GIT_SHA1_HEXSZ);
5389                         git_SHA1_Update(&ctx, oid_to_hex(&p->two->oid),
5390                                         GIT_SHA1_HEXSZ);
5391                         continue;
5392                 }
5393
5394                 xpp.flags = 0;
5395                 xecfg.ctxlen = 3;
5396                 xecfg.flags = 0;
5397                 if (xdi_diff_outf(&mf1, &mf2, patch_id_consume, &data,
5398                                   &xpp, &xecfg))
5399                         return error("unable to generate patch-id diff for %s",
5400                                      p->one->path);
5401         }
5402
5403         git_SHA1_Final(oid->hash, &ctx);
5404         return 0;
5405 }
5406
5407 int diff_flush_patch_id(struct diff_options *options, struct object_id *oid, int diff_header_only)
5408 {
5409         struct diff_queue_struct *q = &diff_queued_diff;
5410         int i;
5411         int result = diff_get_patch_id(options, oid, diff_header_only);
5412
5413         for (i = 0; i < q->nr; i++)
5414                 diff_free_filepair(q->queue[i]);
5415
5416         free(q->queue);
5417         DIFF_QUEUE_CLEAR(q);
5418
5419         return result;
5420 }
5421
5422 static int is_summary_empty(const struct diff_queue_struct *q)
5423 {
5424         int i;
5425
5426         for (i = 0; i < q->nr; i++) {
5427                 const struct diff_filepair *p = q->queue[i];
5428
5429                 switch (p->status) {
5430                 case DIFF_STATUS_DELETED:
5431                 case DIFF_STATUS_ADDED:
5432                 case DIFF_STATUS_COPIED:
5433                 case DIFF_STATUS_RENAMED:
5434                         return 0;
5435                 default:
5436                         if (p->score)
5437                                 return 0;
5438                         if (p->one->mode && p->two->mode &&
5439                             p->one->mode != p->two->mode)
5440                                 return 0;
5441                         break;
5442                 }
5443         }
5444         return 1;
5445 }
5446
5447 static const char rename_limit_warning[] =
5448 N_("inexact rename detection was skipped due to too many files.");
5449
5450 static const char degrade_cc_to_c_warning[] =
5451 N_("only found copies from modified paths due to too many files.");
5452
5453 static const char rename_limit_advice[] =
5454 N_("you may want to set your %s variable to at least "
5455    "%d and retry the command.");
5456
5457 void diff_warn_rename_limit(const char *varname, int needed, int degraded_cc)
5458 {
5459         fflush(stdout);
5460         if (degraded_cc)
5461                 warning(_(degrade_cc_to_c_warning));
5462         else if (needed)
5463                 warning(_(rename_limit_warning));
5464         else
5465                 return;
5466         if (0 < needed)
5467                 warning(_(rename_limit_advice), varname, needed);
5468 }
5469
5470 static void diff_flush_patch_all_file_pairs(struct diff_options *o)
5471 {
5472         int i;
5473         static struct emitted_diff_symbols esm = EMITTED_DIFF_SYMBOLS_INIT;
5474         struct diff_queue_struct *q = &diff_queued_diff;
5475
5476         if (WSEH_NEW & WS_RULE_MASK)
5477                 die("BUG: WS rules bit mask overlaps with diff symbol flags");
5478
5479         if (o->color_moved)
5480                 o->emitted_symbols = &esm;
5481
5482         for (i = 0; i < q->nr; i++) {
5483                 struct diff_filepair *p = q->queue[i];
5484                 if (check_pair_status(p))
5485                         diff_flush_patch(p, o);
5486         }
5487
5488         if (o->emitted_symbols) {
5489                 if (o->color_moved) {
5490                         struct hashmap add_lines, del_lines;
5491
5492                         hashmap_init(&del_lines,
5493                                      (hashmap_cmp_fn)moved_entry_cmp, o, 0);
5494                         hashmap_init(&add_lines,
5495                                      (hashmap_cmp_fn)moved_entry_cmp, o, 0);
5496
5497                         add_lines_to_move_detection(o, &add_lines, &del_lines);
5498                         mark_color_as_moved(o, &add_lines, &del_lines);
5499                         if (o->color_moved == COLOR_MOVED_ZEBRA_DIM)
5500                                 dim_moved_lines(o);
5501
5502                         hashmap_free(&add_lines, 0);
5503                         hashmap_free(&del_lines, 0);
5504                 }
5505
5506                 for (i = 0; i < esm.nr; i++)
5507                         emit_diff_symbol_from_struct(o, &esm.buf[i]);
5508
5509                 for (i = 0; i < esm.nr; i++)
5510                         free((void *)esm.buf[i].line);
5511         }
5512         esm.nr = 0;
5513 }
5514
5515 void diff_flush(struct diff_options *options)
5516 {
5517         struct diff_queue_struct *q = &diff_queued_diff;
5518         int i, output_format = options->output_format;
5519         int separator = 0;
5520         int dirstat_by_line = 0;
5521
5522         /*
5523          * Order: raw, stat, summary, patch
5524          * or:    name/name-status/checkdiff (other bits clear)
5525          */
5526         if (!q->nr)
5527                 goto free_queue;
5528
5529         if (output_format & (DIFF_FORMAT_RAW |
5530                              DIFF_FORMAT_NAME |
5531                              DIFF_FORMAT_NAME_STATUS |
5532                              DIFF_FORMAT_CHECKDIFF)) {
5533                 for (i = 0; i < q->nr; i++) {
5534                         struct diff_filepair *p = q->queue[i];
5535                         if (check_pair_status(p))
5536                                 flush_one_pair(p, options);
5537                 }
5538                 separator++;
5539         }
5540
5541         if (output_format & DIFF_FORMAT_DIRSTAT && options->flags.dirstat_by_line)
5542                 dirstat_by_line = 1;
5543
5544         if (output_format & (DIFF_FORMAT_DIFFSTAT|DIFF_FORMAT_SHORTSTAT|DIFF_FORMAT_NUMSTAT) ||
5545             dirstat_by_line) {
5546                 struct diffstat_t diffstat;
5547
5548                 memset(&diffstat, 0, sizeof(struct diffstat_t));
5549                 for (i = 0; i < q->nr; i++) {
5550                         struct diff_filepair *p = q->queue[i];
5551                         if (check_pair_status(p))
5552                                 diff_flush_stat(p, options, &diffstat);
5553                 }
5554                 if (output_format & DIFF_FORMAT_NUMSTAT)
5555                         show_numstat(&diffstat, options);
5556                 if (output_format & DIFF_FORMAT_DIFFSTAT)
5557                         show_stats(&diffstat, options);
5558                 if (output_format & DIFF_FORMAT_SHORTSTAT)
5559                         show_shortstats(&diffstat, options);
5560                 if (output_format & DIFF_FORMAT_DIRSTAT && dirstat_by_line)
5561                         show_dirstat_by_line(&diffstat, options);
5562                 free_diffstat_info(&diffstat);
5563                 separator++;
5564         }
5565         if ((output_format & DIFF_FORMAT_DIRSTAT) && !dirstat_by_line)
5566                 show_dirstat(options);
5567
5568         if (output_format & DIFF_FORMAT_SUMMARY && !is_summary_empty(q)) {
5569                 for (i = 0; i < q->nr; i++) {
5570                         diff_summary(options, q->queue[i]);
5571                 }
5572                 separator++;
5573         }
5574
5575         if (output_format & DIFF_FORMAT_NO_OUTPUT &&
5576             options->flags.exit_with_status &&
5577             options->flags.diff_from_contents) {
5578                 /*
5579                  * run diff_flush_patch for the exit status. setting
5580                  * options->file to /dev/null should be safe, because we
5581                  * aren't supposed to produce any output anyway.
5582                  */
5583                 if (options->close_file)
5584                         fclose(options->file);
5585                 options->file = xfopen("/dev/null", "w");
5586                 options->close_file = 1;
5587                 options->color_moved = 0;
5588                 for (i = 0; i < q->nr; i++) {
5589                         struct diff_filepair *p = q->queue[i];
5590                         if (check_pair_status(p))
5591                                 diff_flush_patch(p, options);
5592                         if (options->found_changes)
5593                                 break;
5594                 }
5595         }
5596
5597         if (output_format & DIFF_FORMAT_PATCH) {
5598                 if (separator) {
5599                         emit_diff_symbol(options, DIFF_SYMBOL_SEPARATOR, NULL, 0, 0);
5600                         if (options->stat_sep)
5601                                 /* attach patch instead of inline */
5602                                 emit_diff_symbol(options, DIFF_SYMBOL_STAT_SEP,
5603                                                  NULL, 0, 0);
5604                 }
5605
5606                 diff_flush_patch_all_file_pairs(options);
5607         }
5608
5609         if (output_format & DIFF_FORMAT_CALLBACK)
5610                 options->format_callback(q, options, options->format_callback_data);
5611
5612         for (i = 0; i < q->nr; i++)
5613                 diff_free_filepair(q->queue[i]);
5614 free_queue:
5615         free(q->queue);
5616         DIFF_QUEUE_CLEAR(q);
5617         if (options->close_file)
5618                 fclose(options->file);
5619
5620         /*
5621          * Report the content-level differences with HAS_CHANGES;
5622          * diff_addremove/diff_change does not set the bit when
5623          * DIFF_FROM_CONTENTS is in effect (e.g. with -w).
5624          */
5625         if (options->flags.diff_from_contents) {
5626                 if (options->found_changes)
5627                         options->flags.has_changes = 1;
5628                 else
5629                         options->flags.has_changes = 0;
5630         }
5631 }
5632
5633 static int match_filter(const struct diff_options *options, const struct diff_filepair *p)
5634 {
5635         return (((p->status == DIFF_STATUS_MODIFIED) &&
5636                  ((p->score &&
5637                    filter_bit_tst(DIFF_STATUS_FILTER_BROKEN, options)) ||
5638                   (!p->score &&
5639                    filter_bit_tst(DIFF_STATUS_MODIFIED, options)))) ||
5640                 ((p->status != DIFF_STATUS_MODIFIED) &&
5641                  filter_bit_tst(p->status, options)));
5642 }
5643
5644 static void diffcore_apply_filter(struct diff_options *options)
5645 {
5646         int i;
5647         struct diff_queue_struct *q = &diff_queued_diff;
5648         struct diff_queue_struct outq;
5649
5650         DIFF_QUEUE_CLEAR(&outq);
5651
5652         if (!options->filter)
5653                 return;
5654
5655         if (filter_bit_tst(DIFF_STATUS_FILTER_AON, options)) {
5656                 int found;
5657                 for (i = found = 0; !found && i < q->nr; i++) {
5658                         if (match_filter(options, q->queue[i]))
5659                                 found++;
5660                 }
5661                 if (found)
5662                         return;
5663
5664                 /* otherwise we will clear the whole queue
5665                  * by copying the empty outq at the end of this
5666                  * function, but first clear the current entries
5667                  * in the queue.
5668                  */
5669                 for (i = 0; i < q->nr; i++)
5670                         diff_free_filepair(q->queue[i]);
5671         }
5672         else {
5673                 /* Only the matching ones */
5674                 for (i = 0; i < q->nr; i++) {
5675                         struct diff_filepair *p = q->queue[i];
5676                         if (match_filter(options, p))
5677                                 diff_q(&outq, p);
5678                         else
5679                                 diff_free_filepair(p);
5680                 }
5681         }
5682         free(q->queue);
5683         *q = outq;
5684 }
5685
5686 /* Check whether two filespecs with the same mode and size are identical */
5687 static int diff_filespec_is_identical(struct diff_filespec *one,
5688                                       struct diff_filespec *two)
5689 {
5690         if (S_ISGITLINK(one->mode))
5691                 return 0;
5692         if (diff_populate_filespec(one, 0))
5693                 return 0;
5694         if (diff_populate_filespec(two, 0))
5695                 return 0;
5696         return !memcmp(one->data, two->data, one->size);
5697 }
5698
5699 static int diff_filespec_check_stat_unmatch(struct diff_filepair *p)
5700 {
5701         if (p->done_skip_stat_unmatch)
5702                 return p->skip_stat_unmatch_result;
5703
5704         p->done_skip_stat_unmatch = 1;
5705         p->skip_stat_unmatch_result = 0;
5706         /*
5707          * 1. Entries that come from stat info dirtiness
5708          *    always have both sides (iow, not create/delete),
5709          *    one side of the object name is unknown, with
5710          *    the same mode and size.  Keep the ones that
5711          *    do not match these criteria.  They have real
5712          *    differences.
5713          *
5714          * 2. At this point, the file is known to be modified,
5715          *    with the same mode and size, and the object
5716          *    name of one side is unknown.  Need to inspect
5717          *    the identical contents.
5718          */
5719         if (!DIFF_FILE_VALID(p->one) || /* (1) */
5720             !DIFF_FILE_VALID(p->two) ||
5721             (p->one->oid_valid && p->two->oid_valid) ||
5722             (p->one->mode != p->two->mode) ||
5723             diff_populate_filespec(p->one, CHECK_SIZE_ONLY) ||
5724             diff_populate_filespec(p->two, CHECK_SIZE_ONLY) ||
5725             (p->one->size != p->two->size) ||
5726             !diff_filespec_is_identical(p->one, p->two)) /* (2) */
5727                 p->skip_stat_unmatch_result = 1;
5728         return p->skip_stat_unmatch_result;
5729 }
5730
5731 static void diffcore_skip_stat_unmatch(struct diff_options *diffopt)
5732 {
5733         int i;
5734         struct diff_queue_struct *q = &diff_queued_diff;
5735         struct diff_queue_struct outq;
5736         DIFF_QUEUE_CLEAR(&outq);
5737
5738         for (i = 0; i < q->nr; i++) {
5739                 struct diff_filepair *p = q->queue[i];
5740
5741                 if (diff_filespec_check_stat_unmatch(p))
5742                         diff_q(&outq, p);
5743                 else {
5744                         /*
5745                          * The caller can subtract 1 from skip_stat_unmatch
5746                          * to determine how many paths were dirty only
5747                          * due to stat info mismatch.
5748                          */
5749                         if (!diffopt->flags.no_index)
5750                                 diffopt->skip_stat_unmatch++;
5751                         diff_free_filepair(p);
5752                 }
5753         }
5754         free(q->queue);
5755         *q = outq;
5756 }
5757
5758 static int diffnamecmp(const void *a_, const void *b_)
5759 {
5760         const struct diff_filepair *a = *((const struct diff_filepair **)a_);
5761         const struct diff_filepair *b = *((const struct diff_filepair **)b_);
5762         const char *name_a, *name_b;
5763
5764         name_a = a->one ? a->one->path : a->two->path;
5765         name_b = b->one ? b->one->path : b->two->path;
5766         return strcmp(name_a, name_b);
5767 }
5768
5769 void diffcore_fix_diff_index(struct diff_options *options)
5770 {
5771         struct diff_queue_struct *q = &diff_queued_diff;
5772         QSORT(q->queue, q->nr, diffnamecmp);
5773 }
5774
5775 void diffcore_std(struct diff_options *options)
5776 {
5777         /* NOTE please keep the following in sync with diff_tree_combined() */
5778         if (options->skip_stat_unmatch)
5779                 diffcore_skip_stat_unmatch(options);
5780         if (!options->found_follow) {
5781                 /* See try_to_follow_renames() in tree-diff.c */
5782                 if (options->break_opt != -1)
5783                         diffcore_break(options->break_opt);
5784                 if (options->detect_rename)
5785                         diffcore_rename(options);
5786                 if (options->break_opt != -1)
5787                         diffcore_merge_broken();
5788         }
5789         if (options->pickaxe)
5790                 diffcore_pickaxe(options);
5791         if (options->orderfile)
5792                 diffcore_order(options->orderfile);
5793         if (!options->found_follow)
5794                 /* See try_to_follow_renames() in tree-diff.c */
5795                 diff_resolve_rename_copy();
5796         diffcore_apply_filter(options);
5797
5798         if (diff_queued_diff.nr && !options->flags.diff_from_contents)
5799                 options->flags.has_changes = 1;
5800         else
5801                 options->flags.has_changes = 0;
5802
5803         options->found_follow = 0;
5804 }
5805
5806 int diff_result_code(struct diff_options *opt, int status)
5807 {
5808         int result = 0;
5809
5810         diff_warn_rename_limit("diff.renameLimit",
5811                                opt->needed_rename_limit,
5812                                opt->degraded_cc_to_c);
5813         if (!opt->flags.exit_with_status &&
5814             !(opt->output_format & DIFF_FORMAT_CHECKDIFF))
5815                 return status;
5816         if (opt->flags.exit_with_status &&
5817             opt->flags.has_changes)
5818                 result |= 01;
5819         if ((opt->output_format & DIFF_FORMAT_CHECKDIFF) &&
5820             opt->flags.check_failed)
5821                 result |= 02;
5822         return result;
5823 }
5824
5825 int diff_can_quit_early(struct diff_options *opt)
5826 {
5827         return (opt->flags.quick &&
5828                 !opt->filter &&
5829                 opt->flags.has_changes);
5830 }
5831
5832 /*
5833  * Shall changes to this submodule be ignored?
5834  *
5835  * Submodule changes can be configured to be ignored separately for each path,
5836  * but that configuration can be overridden from the command line.
5837  */
5838 static int is_submodule_ignored(const char *path, struct diff_options *options)
5839 {
5840         int ignored = 0;
5841         struct diff_flags orig_flags = options->flags;
5842         if (!options->flags.override_submodule_config)
5843                 set_diffopt_flags_from_submodule_config(options, path);
5844         if (options->flags.ignore_submodules)
5845                 ignored = 1;
5846         options->flags = orig_flags;
5847         return ignored;
5848 }
5849
5850 void diff_addremove(struct diff_options *options,
5851                     int addremove, unsigned mode,
5852                     const struct object_id *oid,
5853                     int oid_valid,
5854                     const char *concatpath, unsigned dirty_submodule)
5855 {
5856         struct diff_filespec *one, *two;
5857
5858         if (S_ISGITLINK(mode) && is_submodule_ignored(concatpath, options))
5859                 return;
5860
5861         /* This may look odd, but it is a preparation for
5862          * feeding "there are unchanged files which should
5863          * not produce diffs, but when you are doing copy
5864          * detection you would need them, so here they are"
5865          * entries to the diff-core.  They will be prefixed
5866          * with something like '=' or '*' (I haven't decided
5867          * which but should not make any difference).
5868          * Feeding the same new and old to diff_change()
5869          * also has the same effect.
5870          * Before the final output happens, they are pruned after
5871          * merged into rename/copy pairs as appropriate.
5872          */
5873         if (options->flags.reverse_diff)
5874                 addremove = (addremove == '+' ? '-' :
5875                              addremove == '-' ? '+' : addremove);
5876
5877         if (options->prefix &&
5878             strncmp(concatpath, options->prefix, options->prefix_length))
5879                 return;
5880
5881         one = alloc_filespec(concatpath);
5882         two = alloc_filespec(concatpath);
5883
5884         if (addremove != '+')
5885                 fill_filespec(one, oid, oid_valid, mode);
5886         if (addremove != '-') {
5887                 fill_filespec(two, oid, oid_valid, mode);
5888                 two->dirty_submodule = dirty_submodule;
5889         }
5890
5891         diff_queue(&diff_queued_diff, one, two);
5892         if (!options->flags.diff_from_contents)
5893                 options->flags.has_changes = 1;
5894 }
5895
5896 void diff_change(struct diff_options *options,
5897                  unsigned old_mode, unsigned new_mode,
5898                  const struct object_id *old_oid,
5899                  const struct object_id *new_oid,
5900                  int old_oid_valid, int new_oid_valid,
5901                  const char *concatpath,
5902                  unsigned old_dirty_submodule, unsigned new_dirty_submodule)
5903 {
5904         struct diff_filespec *one, *two;
5905         struct diff_filepair *p;
5906
5907         if (S_ISGITLINK(old_mode) && S_ISGITLINK(new_mode) &&
5908             is_submodule_ignored(concatpath, options))
5909                 return;
5910
5911         if (options->flags.reverse_diff) {
5912                 SWAP(old_mode, new_mode);
5913                 SWAP(old_oid, new_oid);
5914                 SWAP(old_oid_valid, new_oid_valid);
5915                 SWAP(old_dirty_submodule, new_dirty_submodule);
5916         }
5917
5918         if (options->prefix &&
5919             strncmp(concatpath, options->prefix, options->prefix_length))
5920                 return;
5921
5922         one = alloc_filespec(concatpath);
5923         two = alloc_filespec(concatpath);
5924         fill_filespec(one, old_oid, old_oid_valid, old_mode);
5925         fill_filespec(two, new_oid, new_oid_valid, new_mode);
5926         one->dirty_submodule = old_dirty_submodule;
5927         two->dirty_submodule = new_dirty_submodule;
5928         p = diff_queue(&diff_queued_diff, one, two);
5929
5930         if (options->flags.diff_from_contents)
5931                 return;
5932
5933         if (options->flags.quick && options->skip_stat_unmatch &&
5934             !diff_filespec_check_stat_unmatch(p))
5935                 return;
5936
5937         options->flags.has_changes = 1;
5938 }
5939
5940 struct diff_filepair *diff_unmerge(struct diff_options *options, const char *path)
5941 {
5942         struct diff_filepair *pair;
5943         struct diff_filespec *one, *two;
5944
5945         if (options->prefix &&
5946             strncmp(path, options->prefix, options->prefix_length))
5947                 return NULL;
5948
5949         one = alloc_filespec(path);
5950         two = alloc_filespec(path);
5951         pair = diff_queue(&diff_queued_diff, one, two);
5952         pair->is_unmerged = 1;
5953         return pair;
5954 }
5955
5956 static char *run_textconv(const char *pgm, struct diff_filespec *spec,
5957                 size_t *outsize)
5958 {
5959         struct diff_tempfile *temp;
5960         const char *argv[3];
5961         const char **arg = argv;
5962         struct child_process child = CHILD_PROCESS_INIT;
5963         struct strbuf buf = STRBUF_INIT;
5964         int err = 0;
5965
5966         temp = prepare_temp_file(spec->path, spec);
5967         *arg++ = pgm;
5968         *arg++ = temp->name;
5969         *arg = NULL;
5970
5971         child.use_shell = 1;
5972         child.argv = argv;
5973         child.out = -1;
5974         if (start_command(&child)) {
5975                 remove_tempfile();
5976                 return NULL;
5977         }
5978
5979         if (strbuf_read(&buf, child.out, 0) < 0)
5980                 err = error("error reading from textconv command '%s'", pgm);
5981         close(child.out);
5982
5983         if (finish_command(&child) || err) {
5984                 strbuf_release(&buf);
5985                 remove_tempfile();
5986                 return NULL;
5987         }
5988         remove_tempfile();
5989
5990         return strbuf_detach(&buf, outsize);
5991 }
5992
5993 size_t fill_textconv(struct userdiff_driver *driver,
5994                      struct diff_filespec *df,
5995                      char **outbuf)
5996 {
5997         size_t size;
5998
5999         if (!driver) {
6000                 if (!DIFF_FILE_VALID(df)) {
6001                         *outbuf = "";
6002                         return 0;
6003                 }
6004                 if (diff_populate_filespec(df, 0))
6005                         die("unable to read files to diff");
6006                 *outbuf = df->data;
6007                 return df->size;
6008         }
6009
6010         if (!driver->textconv)
6011                 die("BUG: fill_textconv called with non-textconv driver");
6012
6013         if (driver->textconv_cache && df->oid_valid) {
6014                 *outbuf = notes_cache_get(driver->textconv_cache,
6015                                           &df->oid,
6016                                           &size);
6017                 if (*outbuf)
6018                         return size;
6019         }
6020
6021         *outbuf = run_textconv(driver->textconv, df, &size);
6022         if (!*outbuf)
6023                 die("unable to read files to diff");
6024
6025         if (driver->textconv_cache && df->oid_valid) {
6026                 /* ignore errors, as we might be in a readonly repository */
6027                 notes_cache_put(driver->textconv_cache, &df->oid, *outbuf,
6028                                 size);
6029                 /*
6030                  * we could save up changes and flush them all at the end,
6031                  * but we would need an extra call after all diffing is done.
6032                  * Since generating a cache entry is the slow path anyway,
6033                  * this extra overhead probably isn't a big deal.
6034                  */
6035                 notes_cache_write(driver->textconv_cache);
6036         }
6037
6038         return size;
6039 }
6040
6041 int textconv_object(const char *path,
6042                     unsigned mode,
6043                     const struct object_id *oid,
6044                     int oid_valid,
6045                     char **buf,
6046                     unsigned long *buf_size)
6047 {
6048         struct diff_filespec *df;
6049         struct userdiff_driver *textconv;
6050
6051         df = alloc_filespec(path);
6052         fill_filespec(df, oid, oid_valid, mode);
6053         textconv = get_textconv(df);
6054         if (!textconv) {
6055                 free_filespec(df);
6056                 return 0;
6057         }
6058
6059         *buf_size = fill_textconv(textconv, df, buf);
6060         free_filespec(df);
6061         return 1;
6062 }
6063
6064 void setup_diff_pager(struct diff_options *opt)
6065 {
6066         /*
6067          * If the user asked for our exit code, then either they want --quiet
6068          * or --exit-code. We should definitely not bother with a pager in the
6069          * former case, as we will generate no output. Since we still properly
6070          * report our exit code even when a pager is run, we _could_ run a
6071          * pager with --exit-code. But since we have not done so historically,
6072          * and because it is easy to find people oneline advising "git diff
6073          * --exit-code" in hooks and other scripts, we do not do so.
6074          */
6075         if (!opt->flags.exit_with_status &&
6076             check_pager_config("diff") != 0)
6077                 setup_pager();
6078 }