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 "object-store.h"
  17 #include "userdiff.h"
  18 #include "submodule-config.h"
  19 #include "submodule.h"
  20 #include "hashmap.h"
  21 #include "ll-merge.h"
  22 #include "string-list.h"
  23 #include "argv-array.h"
  24 #include "graph.h"
  25 #include "packfile.h"
  26 #include "help.h"
  27
  28 #ifdef NO_FAST_WORKING_DIRECTORY
  29 #define FAST_WORKING_DIRECTORY 0
  30 #else
  31 #define FAST_WORKING_DIRECTORY 1
  32 #endif
  33
  34 static int diff_detect_rename_default;
  35 static int diff_indent_heuristic = 1;
  36 static int diff_rename_limit_default = 400;
  37 static int diff_suppress_blank_empty;
  38 static int diff_use_color_default = -1;
  39 static int diff_color_moved_default;
  40 static int diff_color_moved_ws_default;
  41 static int diff_context_default = 3;
  42 static int diff_interhunk_context_default;
  43 static const char *diff_word_regex_cfg;
  44 static const char *external_diff_cmd_cfg;
  45 static const char *diff_order_file_cfg;
  46 int diff_auto_refresh_index = 1;
  47 static int diff_mnemonic_prefix;
  48 static int diff_no_prefix;
  49 static int diff_stat_graph_width;
  50 static int diff_dirstat_permille_default = 30;
  51 static struct diff_options default_diff_options;
  52 static long diff_algorithm;
  53 static unsigned ws_error_highlight_default = WSEH_NEW;
  54
  55 static char diff_colors[][COLOR_MAXLEN] = {
  56         GIT_COLOR_RESET,
  57         GIT_COLOR_NORMAL,       /* CONTEXT */
  58         GIT_COLOR_BOLD,         /* METAINFO */
  59         GIT_COLOR_CYAN,         /* FRAGINFO */
  60         GIT_COLOR_RED,          /* OLD */
  61         GIT_COLOR_GREEN,        /* NEW */
  62         GIT_COLOR_YELLOW,       /* COMMIT */
  63         GIT_COLOR_BG_RED,       /* WHITESPACE */
  64         GIT_COLOR_NORMAL,       /* FUNCINFO */
  65         GIT_COLOR_BOLD_MAGENTA, /* OLD_MOVED */
  66         GIT_COLOR_BOLD_BLUE,    /* OLD_MOVED ALTERNATIVE */
  67         GIT_COLOR_FAINT,        /* OLD_MOVED_DIM */
  68         GIT_COLOR_FAINT_ITALIC, /* OLD_MOVED_ALTERNATIVE_DIM */
  69         GIT_COLOR_BOLD_CYAN,    /* NEW_MOVED */
  70         GIT_COLOR_BOLD_YELLOW,  /* NEW_MOVED ALTERNATIVE */
  71         GIT_COLOR_FAINT,        /* NEW_MOVED_DIM */
  72         GIT_COLOR_FAINT_ITALIC, /* NEW_MOVED_ALTERNATIVE_DIM */
  73         GIT_COLOR_FAINT,        /* CONTEXT_DIM */
  74         GIT_COLOR_FAINT_RED,    /* OLD_DIM */
  75         GIT_COLOR_FAINT_GREEN,  /* NEW_DIM */
  76         GIT_COLOR_BOLD,         /* CONTEXT_BOLD */
  77         GIT_COLOR_BOLD_RED,     /* OLD_BOLD */
  78         GIT_COLOR_BOLD_GREEN,   /* NEW_BOLD */
  79 };
  80
  81 static const char *color_diff_slots[] = {
  82         [DIFF_CONTEXT]                = "context",
  83         [DIFF_METAINFO]               = "meta",
  84         [DIFF_FRAGINFO]               = "frag",
  85         [DIFF_FILE_OLD]               = "old",
  86         [DIFF_FILE_NEW]               = "new",
  87         [DIFF_COMMIT]                 = "commit",
  88         [DIFF_WHITESPACE]             = "whitespace",
  89         [DIFF_FUNCINFO]               = "func",
  90         [DIFF_FILE_OLD_MOVED]         = "oldMoved",
  91         [DIFF_FILE_OLD_MOVED_ALT]     = "oldMovedAlternative",
  92         [DIFF_FILE_OLD_MOVED_DIM]     = "oldMovedDimmed",
  93         [DIFF_FILE_OLD_MOVED_ALT_DIM] = "oldMovedAlternativeDimmed",
  94         [DIFF_FILE_NEW_MOVED]         = "newMoved",
  95         [DIFF_FILE_NEW_MOVED_ALT]     = "newMovedAlternative",
  96         [DIFF_FILE_NEW_MOVED_DIM]     = "newMovedDimmed",
  97         [DIFF_FILE_NEW_MOVED_ALT_DIM] = "newMovedAlternativeDimmed",
  98         [DIFF_CONTEXT_DIM]            = "contextDimmed",
  99         [DIFF_FILE_OLD_DIM]           = "oldDimmed",
 100         [DIFF_FILE_NEW_DIM]           = "newDimmed",
 101         [DIFF_CONTEXT_BOLD]           = "contextBold",
 102         [DIFF_FILE_OLD_BOLD]          = "oldBold",
 103         [DIFF_FILE_NEW_BOLD]          = "newBold",
 104 };
 105
 106 static NORETURN void die_want_option(const char *option_name)
 107 {
 108         die(_("option '%s' requires a value"), option_name);
 109 }
 110
 111 define_list_config_array_extra(color_diff_slots, {"plain"});
 112
 113 static int parse_diff_color_slot(const char *var)
 114 {
 115         if (!strcasecmp(var, "plain"))
 116                 return DIFF_CONTEXT;
 117         return LOOKUP_CONFIG(color_diff_slots, var);
 118 }
 119
 120 static int parse_dirstat_params(struct diff_options *options, const char *params_string,
 121                                 struct strbuf *errmsg)
 122 {
 123         char *params_copy = xstrdup(params_string);
 124         struct string_list params = STRING_LIST_INIT_NODUP;
 125         int ret = 0;
 126         int i;
 127
 128         if (*params_copy)
 129                 string_list_split_in_place(&params, params_copy, ',', -1);
 130         for (i = 0; i < params.nr; i++) {
 131                 const char *p = params.items[i].string;
 132                 if (!strcmp(p, "changes")) {
 133                         options->flags.dirstat_by_line = 0;
 134                         options->flags.dirstat_by_file = 0;
 135                 } else if (!strcmp(p, "lines")) {
 136                         options->flags.dirstat_by_line = 1;
 137                         options->flags.dirstat_by_file = 0;
 138                 } else if (!strcmp(p, "files")) {
 139                         options->flags.dirstat_by_line = 0;
 140                         options->flags.dirstat_by_file = 1;
 141                 } else if (!strcmp(p, "noncumulative")) {
 142                         options->flags.dirstat_cumulative = 0;
 143                 } else if (!strcmp(p, "cumulative")) {
 144                         options->flags.dirstat_cumulative = 1;
 145                 } else if (isdigit(*p)) {
 146                         char *end;
 147                         int permille = strtoul(p, &end, 10) * 10;
 148                         if (*end == '.' && isdigit(*++end)) {
 149                                 /* only use first digit */
 150                                 permille += *end - '0';
 151                                 /* .. and ignore any further digits */
 152                                 while (isdigit(*++end))
 153                                         ; /* nothing */
 154                         }
 155                         if (!*end)
 156                                 options->dirstat_permille = permille;
 157                         else {
 158                                 strbuf_addf(errmsg, _("  Failed to parse dirstat cut-off percentage '%s'\n"),
 159                                             p);
 160                                 ret++;
 161                         }
 162                 } else {
 163                         strbuf_addf(errmsg, _("  Unknown dirstat parameter '%s'\n"), p);
 164                         ret++;
 165                 }
 166
 167         }
 168         string_list_clear(&params, 0);
 169         free(params_copy);
 170         return ret;
 171 }
 172
 173 static int parse_submodule_params(struct diff_options *options, const char *value)
 174 {
 175         if (!strcmp(value, "log"))
 176                 options->submodule_format = DIFF_SUBMODULE_LOG;
 177         else if (!strcmp(value, "short"))
 178                 options->submodule_format = DIFF_SUBMODULE_SHORT;
 179         else if (!strcmp(value, "diff"))
 180                 options->submodule_format = DIFF_SUBMODULE_INLINE_DIFF;
 181         else
 182                 return -1;
 183         return 0;
 184 }
 185
 186 int git_config_rename(const char *var, const char *value)
 187 {
 188         if (!value)
 189                 return DIFF_DETECT_RENAME;
 190         if (!strcasecmp(value, "copies") || !strcasecmp(value, "copy"))
 191                 return  DIFF_DETECT_COPY;
 192         return git_config_bool(var,value) ? DIFF_DETECT_RENAME : 0;
 193 }
 194
 195 long parse_algorithm_value(const char *value)
 196 {
 197         if (!value)
 198                 return -1;
 199         else if (!strcasecmp(value, "myers") || !strcasecmp(value, "default"))
 200                 return 0;
 201         else if (!strcasecmp(value, "minimal"))
 202                 return XDF_NEED_MINIMAL;
 203         else if (!strcasecmp(value, "patience"))
 204                 return XDF_PATIENCE_DIFF;
 205         else if (!strcasecmp(value, "histogram"))
 206                 return XDF_HISTOGRAM_DIFF;
 207         return -1;
 208 }
 209
 210 static int parse_one_token(const char **arg, const char *token)
 211 {
 212         const char *rest;
 213         if (skip_prefix(*arg, token, &rest) && (!*rest || *rest == ',')) {
 214                 *arg = rest;
 215                 return 1;
 216         }
 217         return 0;
 218 }
 219
 220 static int parse_ws_error_highlight(const char *arg)
 221 {
 222         const char *orig_arg = arg;
 223         unsigned val = 0;
 224
 225         while (*arg) {
 226                 if (parse_one_token(&arg, "none"))
 227                         val = 0;
 228                 else if (parse_one_token(&arg, "default"))
 229                         val = WSEH_NEW;
 230                 else if (parse_one_token(&arg, "all"))
 231                         val = WSEH_NEW | WSEH_OLD | WSEH_CONTEXT;
 232                 else if (parse_one_token(&arg, "new"))
 233                         val |= WSEH_NEW;
 234                 else if (parse_one_token(&arg, "old"))
 235                         val |= WSEH_OLD;
 236                 else if (parse_one_token(&arg, "context"))
 237                         val |= WSEH_CONTEXT;
 238                 else {
 239                         return -1 - (int)(arg - orig_arg);
 240                 }
 241                 if (*arg)
 242                         arg++;
 243         }
 244         return val;
 245 }
 246
 247 /*
 248  * These are to give UI layer defaults.
 249  * The core-level commands such as git-diff-files should
 250  * never be affected by the setting of diff.renames
 251  * the user happens to have in the configuration file.
 252  */
 253 void init_diff_ui_defaults(void)
 254 {
 255         diff_detect_rename_default = DIFF_DETECT_RENAME;
 256 }
 257
 258 int git_diff_heuristic_config(const char *var, const char *value, void *cb)
 259 {
 260         if (!strcmp(var, "diff.indentheuristic"))
 261                 diff_indent_heuristic = git_config_bool(var, value);
 262         return 0;
 263 }
 264
 265 static int parse_color_moved(const char *arg)
 266 {
 267         switch (git_parse_maybe_bool(arg)) {
 268         case 0:
 269                 return COLOR_MOVED_NO;
 270         case 1:
 271                 return COLOR_MOVED_DEFAULT;
 272         default:
 273                 break;
 274         }
 275
 276         if (!strcmp(arg, "no"))
 277                 return COLOR_MOVED_NO;
 278         else if (!strcmp(arg, "plain"))
 279                 return COLOR_MOVED_PLAIN;
 280         else if (!strcmp(arg, "blocks"))
 281                 return COLOR_MOVED_BLOCKS;
 282         else if (!strcmp(arg, "zebra"))
 283                 return COLOR_MOVED_ZEBRA;
 284         else if (!strcmp(arg, "default"))
 285                 return COLOR_MOVED_DEFAULT;
 286         else if (!strcmp(arg, "dimmed-zebra"))
 287                 return COLOR_MOVED_ZEBRA_DIM;
 288         else if (!strcmp(arg, "dimmed_zebra"))
 289                 return COLOR_MOVED_ZEBRA_DIM;
 290         else
 291                 return error(_("color moved setting must be one of 'no', 'default', 'blocks', 'zebra', 'dimmed-zebra', 'plain'"));
 292 }
 293
 294 static int parse_color_moved_ws(const char *arg)
 295 {
 296         int ret = 0;
 297         struct string_list l = STRING_LIST_INIT_DUP;
 298         struct string_list_item *i;
 299
 300         string_list_split(&l, arg, ',', -1);
 301
 302         for_each_string_list_item(i, &l) {
 303                 struct strbuf sb = STRBUF_INIT;
 304                 strbuf_addstr(&sb, i->string);
 305                 strbuf_trim(&sb);
 306
 307                 if (!strcmp(sb.buf, "no"))
 308                         ret = 0;
 309                 else if (!strcmp(sb.buf, "ignore-space-change"))
 310                         ret |= XDF_IGNORE_WHITESPACE_CHANGE;
 311                 else if (!strcmp(sb.buf, "ignore-space-at-eol"))
 312                         ret |= XDF_IGNORE_WHITESPACE_AT_EOL;
 313                 else if (!strcmp(sb.buf, "ignore-all-space"))
 314                         ret |= XDF_IGNORE_WHITESPACE;
 315                 else if (!strcmp(sb.buf, "allow-indentation-change"))
 316                         ret |= COLOR_MOVED_WS_ALLOW_INDENTATION_CHANGE;
 317                 else
 318                         error(_("ignoring unknown color-moved-ws mode '%s'"), sb.buf);
 319
 320                 strbuf_release(&sb);
 321         }
 322
 323         if ((ret & COLOR_MOVED_WS_ALLOW_INDENTATION_CHANGE) &&
 324             (ret & XDF_WHITESPACE_FLAGS))
 325                 die(_("color-moved-ws: allow-indentation-change cannot be combined with other whitespace modes"));
 326
 327         string_list_clear(&l, 0);
 328
 329         return ret;
 330 }
 331
 332 int git_diff_ui_config(const char *var, const char *value, void *cb)
 333 {
 334         if (!strcmp(var, "diff.color") || !strcmp(var, "color.diff")) {
 335                 diff_use_color_default = git_config_colorbool(var, value);
 336                 return 0;
 337         }
 338         if (!strcmp(var, "diff.colormoved")) {
 339                 int cm = parse_color_moved(value);
 340                 if (cm < 0)
 341                         return -1;
 342                 diff_color_moved_default = cm;
 343                 return 0;
 344         }
 345         if (!strcmp(var, "diff.colormovedws")) {
 346                 int cm = parse_color_moved_ws(value);
 347                 if (cm < 0)
 348                         return -1;
 349                 diff_color_moved_ws_default = cm;
 350                 return 0;
 351         }
 352         if (!strcmp(var, "diff.context")) {
 353                 diff_context_default = git_config_int(var, value);
 354                 if (diff_context_default < 0)
 355                         return -1;
 356                 return 0;
 357         }
 358         if (!strcmp(var, "diff.interhunkcontext")) {
 359                 diff_interhunk_context_default = git_config_int(var, value);
 360                 if (diff_interhunk_context_default < 0)
 361                         return -1;
 362                 return 0;
 363         }
 364         if (!strcmp(var, "diff.renames")) {
 365                 diff_detect_rename_default = git_config_rename(var, value);
 366                 return 0;
 367         }
 368         if (!strcmp(var, "diff.autorefreshindex")) {
 369                 diff_auto_refresh_index = git_config_bool(var, value);
 370                 return 0;
 371         }
 372         if (!strcmp(var, "diff.mnemonicprefix")) {
 373                 diff_mnemonic_prefix = git_config_bool(var, value);
 374                 return 0;
 375         }
 376         if (!strcmp(var, "diff.noprefix")) {
 377                 diff_no_prefix = git_config_bool(var, value);
 378                 return 0;
 379         }
 380         if (!strcmp(var, "diff.statgraphwidth")) {
 381                 diff_stat_graph_width = git_config_int(var, value);
 382                 return 0;
 383         }
 384         if (!strcmp(var, "diff.external"))
 385                 return git_config_string(&external_diff_cmd_cfg, var, value);
 386         if (!strcmp(var, "diff.wordregex"))
 387                 return git_config_string(&diff_word_regex_cfg, var, value);
 388         if (!strcmp(var, "diff.orderfile"))
 389                 return git_config_pathname(&diff_order_file_cfg, var, value);
 390
 391         if (!strcmp(var, "diff.ignoresubmodules"))
 392                 handle_ignore_submodules_arg(&default_diff_options, value);
 393
 394         if (!strcmp(var, "diff.submodule")) {
 395                 if (parse_submodule_params(&default_diff_options, value))
 396                         warning(_("Unknown value for 'diff.submodule' config variable: '%s'"),
 397                                 value);
 398                 return 0;
 399         }
 400
 401         if (!strcmp(var, "diff.algorithm")) {
 402                 diff_algorithm = parse_algorithm_value(value);
 403                 if (diff_algorithm < 0)
 404                         return -1;
 405                 return 0;
 406         }
 407
 408         if (!strcmp(var, "diff.wserrorhighlight")) {
 409                 int val = parse_ws_error_highlight(value);
 410                 if (val < 0)
 411                         return -1;
 412                 ws_error_highlight_default = val;
 413                 return 0;
 414         }
 415
 416         if (git_color_config(var, value, cb) < 0)
 417                 return -1;
 418
 419         return git_diff_basic_config(var, value, cb);
 420 }
 421
 422 int git_diff_basic_config(const char *var, const char *value, void *cb)
 423 {
 424         const char *name;
 425
 426         if (!strcmp(var, "diff.renamelimit")) {
 427                 diff_rename_limit_default = git_config_int(var, value);
 428                 return 0;
 429         }
 430
 431         if (userdiff_config(var, value) < 0)
 432                 return -1;
 433
 434         if (skip_prefix(var, "diff.color.", &name) ||
 435             skip_prefix(var, "color.diff.", &name)) {
 436                 int slot = parse_diff_color_slot(name);
 437                 if (slot < 0)
 438                         return 0;
 439                 if (!value)
 440                         return config_error_nonbool(var);
 441                 return color_parse(value, diff_colors[slot]);
 442         }
 443
 444         /* like GNU diff's --suppress-blank-empty option  */
 445         if (!strcmp(var, "diff.suppressblankempty") ||
 446                         /* for backwards compatibility */
 447                         !strcmp(var, "diff.suppress-blank-empty")) {
 448                 diff_suppress_blank_empty = git_config_bool(var, value);
 449                 return 0;
 450         }
 451
 452         if (!strcmp(var, "diff.dirstat")) {
 453                 struct strbuf errmsg = STRBUF_INIT;
 454                 default_diff_options.dirstat_permille = diff_dirstat_permille_default;
 455                 if (parse_dirstat_params(&default_diff_options, value, &errmsg))
 456                         warning(_("Found errors in 'diff.dirstat' config variable:\n%s"),
 457                                 errmsg.buf);
 458                 strbuf_release(&errmsg);
 459                 diff_dirstat_permille_default = default_diff_options.dirstat_permille;
 460                 return 0;
 461         }
 462
 463         if (git_diff_heuristic_config(var, value, cb) < 0)
 464                 return -1;
 465
 466         return git_default_config(var, value, cb);
 467 }
 468
 469 static char *quote_two(const char *one, const char *two)
 470 {
 471         int need_one = quote_c_style(one, NULL, NULL, 1);
 472         int need_two = quote_c_style(two, NULL, NULL, 1);
 473         struct strbuf res = STRBUF_INIT;
 474
 475         if (need_one + need_two) {
 476                 strbuf_addch(&res, '"');
 477                 quote_c_style(one, &res, NULL, 1);
 478                 quote_c_style(two, &res, NULL, 1);
 479                 strbuf_addch(&res, '"');
 480         } else {
 481                 strbuf_addstr(&res, one);
 482                 strbuf_addstr(&res, two);
 483         }
 484         return strbuf_detach(&res, NULL);
 485 }
 486
 487 static const char *external_diff(void)
 488 {
 489         static const char *external_diff_cmd = NULL;
 490         static int done_preparing = 0;
 491
 492         if (done_preparing)
 493                 return external_diff_cmd;
 494         external_diff_cmd = getenv("GIT_EXTERNAL_DIFF");
 495         if (!external_diff_cmd)
 496                 external_diff_cmd = external_diff_cmd_cfg;
 497         done_preparing = 1;
 498         return external_diff_cmd;
 499 }
 500
 501 /*
 502  * Keep track of files used for diffing. Sometimes such an entry
 503  * refers to a temporary file, sometimes to an existing file, and
 504  * sometimes to "/dev/null".
 505  */
 506 static struct diff_tempfile {
 507         /*
 508          * filename external diff should read from, or NULL if this
 509          * entry is currently not in use:
 510          */
 511         const char *name;
 512
 513         char hex[GIT_MAX_HEXSZ + 1];
 514         char mode[10];
 515
 516         /*
 517          * If this diff_tempfile instance refers to a temporary file,
 518          * this tempfile object is used to manage its lifetime.
 519          */
 520         struct tempfile *tempfile;
 521 } diff_temp[2];
 522
 523 struct emit_callback {
 524         int color_diff;
 525         unsigned ws_rule;
 526         int blank_at_eof_in_preimage;
 527         int blank_at_eof_in_postimage;
 528         int lno_in_preimage;
 529         int lno_in_postimage;
 530         const char **label_path;
 531         struct diff_words_data *diff_words;
 532         struct diff_options *opt;
 533         struct strbuf *header;
 534 };
 535
 536 static int count_lines(const char *data, int size)
 537 {
 538         int count, ch, completely_empty = 1, nl_just_seen = 0;
 539         count = 0;
 540         while (0 < size--) {
 541                 ch = *data++;
 542                 if (ch == '\n') {
 543                         count++;
 544                         nl_just_seen = 1;
 545                         completely_empty = 0;
 546                 }
 547                 else {
 548                         nl_just_seen = 0;
 549                         completely_empty = 0;
 550                 }
 551         }
 552         if (completely_empty)
 553                 return 0;
 554         if (!nl_just_seen)
 555                 count++; /* no trailing newline */
 556         return count;
 557 }
 558
 559 static int fill_mmfile(struct repository *r, mmfile_t *mf,
 560                        struct diff_filespec *one)
 561 {
 562         if (!DIFF_FILE_VALID(one)) {
 563                 mf->ptr = (char *)""; /* does not matter */
 564                 mf->size = 0;
 565                 return 0;
 566         }
 567         else if (diff_populate_filespec(r, one, 0))
 568                 return -1;
 569
 570         mf->ptr = one->data;
 571         mf->size = one->size;
 572         return 0;
 573 }
 574
 575 /* like fill_mmfile, but only for size, so we can avoid retrieving blob */
 576 static unsigned long diff_filespec_size(struct repository *r,
 577                                         struct diff_filespec *one)
 578 {
 579         if (!DIFF_FILE_VALID(one))
 580                 return 0;
 581         diff_populate_filespec(r, one, CHECK_SIZE_ONLY);
 582         return one->size;
 583 }
 584
 585 static int count_trailing_blank(mmfile_t *mf, unsigned ws_rule)
 586 {
 587         char *ptr = mf->ptr;
 588         long size = mf->size;
 589         int cnt = 0;
 590
 591         if (!size)
 592                 return cnt;
 593         ptr += size - 1; /* pointing at the very end */
 594         if (*ptr != '\n')
 595                 ; /* incomplete line */
 596         else
 597                 ptr--; /* skip the last LF */
 598         while (mf->ptr < ptr) {
 599                 char *prev_eol;
 600                 for (prev_eol = ptr; mf->ptr <= prev_eol; prev_eol--)
 601                         if (*prev_eol == '\n')
 602                                 break;
 603                 if (!ws_blank_line(prev_eol + 1, ptr - prev_eol, ws_rule))
 604                         break;
 605                 cnt++;
 606                 ptr = prev_eol - 1;
 607         }
 608         return cnt;
 609 }
 610
 611 static void check_blank_at_eof(mmfile_t *mf1, mmfile_t *mf2,
 612                                struct emit_callback *ecbdata)
 613 {
 614         int l1, l2, at;
 615         unsigned ws_rule = ecbdata->ws_rule;
 616         l1 = count_trailing_blank(mf1, ws_rule);
 617         l2 = count_trailing_blank(mf2, ws_rule);
 618         if (l2 <= l1) {
 619                 ecbdata->blank_at_eof_in_preimage = 0;
 620                 ecbdata->blank_at_eof_in_postimage = 0;
 621                 return;
 622         }
 623         at = count_lines(mf1->ptr, mf1->size);
 624         ecbdata->blank_at_eof_in_preimage = (at - l1) + 1;
 625
 626         at = count_lines(mf2->ptr, mf2->size);
 627         ecbdata->blank_at_eof_in_postimage = (at - l2) + 1;
 628 }
 629
 630 static void emit_line_0(struct diff_options *o,
 631                         const char *set_sign, const char *set, unsigned reverse, const char *reset,
 632                         int first, const char *line, int len)
 633 {
 634         int has_trailing_newline, has_trailing_carriage_return;
 635         int needs_reset = 0; /* at the end of the line */
 636         FILE *file = o->file;
 637
 638         fputs(diff_line_prefix(o), file);
 639
 640         has_trailing_newline = (len > 0 && line[len-1] == '\n');
 641         if (has_trailing_newline)
 642                 len--;
 643
 644         has_trailing_carriage_return = (len > 0 && line[len-1] == '\r');
 645         if (has_trailing_carriage_return)
 646                 len--;
 647
 648         if (!len && !first)
 649                 goto end_of_line;
 650
 651         if (reverse && want_color(o->use_color)) {
 652                 fputs(GIT_COLOR_REVERSE, file);
 653                 needs_reset = 1;
 654         }
 655
 656         if (set_sign) {
 657                 fputs(set_sign, file);
 658                 needs_reset = 1;
 659         }
 660
 661         if (first)
 662                 fputc(first, file);
 663
 664         if (!len)
 665                 goto end_of_line;
 666
 667         if (set) {
 668                 if (set_sign && set != set_sign)
 669                         fputs(reset, file);
 670                 fputs(set, file);
 671                 needs_reset = 1;
 672         }
 673         fwrite(line, len, 1, file);
 674         needs_reset = 1; /* 'line' may contain color codes. */
 675
 676 end_of_line:
 677         if (needs_reset)
 678                 fputs(reset, file);
 679         if (has_trailing_carriage_return)
 680                 fputc('\r', file);
 681         if (has_trailing_newline)
 682                 fputc('\n', file);
 683 }
 684
 685 static void emit_line(struct diff_options *o, const char *set, const char *reset,
 686                       const char *line, int len)
 687 {
 688         emit_line_0(o, set, NULL, 0, reset, 0, line, len);
 689 }
 690
 691 enum diff_symbol {
 692         DIFF_SYMBOL_BINARY_DIFF_HEADER,
 693         DIFF_SYMBOL_BINARY_DIFF_HEADER_DELTA,
 694         DIFF_SYMBOL_BINARY_DIFF_HEADER_LITERAL,
 695         DIFF_SYMBOL_BINARY_DIFF_BODY,
 696         DIFF_SYMBOL_BINARY_DIFF_FOOTER,
 697         DIFF_SYMBOL_STATS_SUMMARY_NO_FILES,
 698         DIFF_SYMBOL_STATS_SUMMARY_ABBREV,
 699         DIFF_SYMBOL_STATS_SUMMARY_INSERTS_DELETES,
 700         DIFF_SYMBOL_STATS_LINE,
 701         DIFF_SYMBOL_WORD_DIFF,
 702         DIFF_SYMBOL_STAT_SEP,
 703         DIFF_SYMBOL_SUMMARY,
 704         DIFF_SYMBOL_SUBMODULE_ADD,
 705         DIFF_SYMBOL_SUBMODULE_DEL,
 706         DIFF_SYMBOL_SUBMODULE_UNTRACKED,
 707         DIFF_SYMBOL_SUBMODULE_MODIFIED,
 708         DIFF_SYMBOL_SUBMODULE_HEADER,
 709         DIFF_SYMBOL_SUBMODULE_ERROR,
 710         DIFF_SYMBOL_SUBMODULE_PIPETHROUGH,
 711         DIFF_SYMBOL_REWRITE_DIFF,
 712         DIFF_SYMBOL_BINARY_FILES,
 713         DIFF_SYMBOL_HEADER,
 714         DIFF_SYMBOL_FILEPAIR_PLUS,
 715         DIFF_SYMBOL_FILEPAIR_MINUS,
 716         DIFF_SYMBOL_WORDS_PORCELAIN,
 717         DIFF_SYMBOL_WORDS,
 718         DIFF_SYMBOL_CONTEXT,
 719         DIFF_SYMBOL_CONTEXT_INCOMPLETE,
 720         DIFF_SYMBOL_PLUS,
 721         DIFF_SYMBOL_MINUS,
 722         DIFF_SYMBOL_NO_LF_EOF,
 723         DIFF_SYMBOL_CONTEXT_FRAGINFO,
 724         DIFF_SYMBOL_CONTEXT_MARKER,
 725         DIFF_SYMBOL_SEPARATOR
 726 };
 727 /*
 728  * Flags for content lines:
 729  * 0..12 are whitespace rules
 730  * 13-15 are WSEH_NEW | WSEH_OLD | WSEH_CONTEXT
 731  * 16 is marking if the line is blank at EOF
 732  */
 733 #define DIFF_SYMBOL_CONTENT_BLANK_LINE_EOF      (1<<16)
 734 #define DIFF_SYMBOL_MOVED_LINE                  (1<<17)
 735 #define DIFF_SYMBOL_MOVED_LINE_ALT              (1<<18)
 736 #define DIFF_SYMBOL_MOVED_LINE_UNINTERESTING    (1<<19)
 737 #define DIFF_SYMBOL_CONTENT_WS_MASK (WSEH_NEW | WSEH_OLD | WSEH_CONTEXT | WS_RULE_MASK)
 738
 739 /*
 740  * This struct is used when we need to buffer the output of the diff output.
 741  *
 742  * NEEDSWORK: Instead of storing a copy of the line, add an offset pointer
 743  * into the pre/post image file. This pointer could be a union with the
 744  * line pointer. By storing an offset into the file instead of the literal line,
 745  * we can decrease the memory footprint for the buffered output. At first we
 746  * may want to only have indirection for the content lines, but we could also
 747  * enhance the state for emitting prefabricated lines, e.g. the similarity
 748  * score line or hunk/file headers would only need to store a number or path
 749  * and then the output can be constructed later on depending on state.
 750  */
 751 struct emitted_diff_symbol {
 752         const char *line;
 753         int len;
 754         int flags;
 755         enum diff_symbol s;
 756 };
 757 #define EMITTED_DIFF_SYMBOL_INIT {NULL}
 758
 759 struct emitted_diff_symbols {
 760         struct emitted_diff_symbol *buf;
 761         int nr, alloc;
 762 };
 763 #define EMITTED_DIFF_SYMBOLS_INIT {NULL, 0, 0}
 764
 765 static void append_emitted_diff_symbol(struct diff_options *o,
 766                                        struct emitted_diff_symbol *e)
 767 {
 768         struct emitted_diff_symbol *f;
 769
 770         ALLOC_GROW(o->emitted_symbols->buf,
 771                    o->emitted_symbols->nr + 1,
 772                    o->emitted_symbols->alloc);
 773         f = &o->emitted_symbols->buf[o->emitted_symbols->nr++];
 774
 775         memcpy(f, e, sizeof(struct emitted_diff_symbol));
 776         f->line = e->line ? xmemdupz(e->line, e->len) : NULL;
 777 }
 778
 779 struct moved_entry {
 780         struct hashmap_entry ent;
 781         const struct emitted_diff_symbol *es;
 782         struct moved_entry *next_line;
 783 };
 784
 785 /**
 786  * The struct ws_delta holds white space differences between moved lines, i.e.
 787  * between '+' and '-' lines that have been detected to be a move.
 788  * The string contains the difference in leading white spaces, before the
 789  * rest of the line is compared using the white space config for move
 790  * coloring. The current_longer indicates if the first string in the
 791  * comparision is longer than the second.
 792  */
 793 struct ws_delta {
 794         char *string;
 795         unsigned int current_longer : 1;
 796 };
 797 #define WS_DELTA_INIT { NULL, 0 }
 798
 799 struct moved_block {
 800         struct moved_entry *match;
 801         struct ws_delta wsd;
 802 };
 803
 804 static void moved_block_clear(struct moved_block *b)
 805 {
 806         FREE_AND_NULL(b->wsd.string);
 807         b->match = NULL;
 808 }
 809
 810 static int compute_ws_delta(const struct emitted_diff_symbol *a,
 811                              const struct emitted_diff_symbol *b,
 812                              struct ws_delta *out)
 813 {
 814         const struct emitted_diff_symbol *longer =  a->len > b->len ? a : b;
 815         const struct emitted_diff_symbol *shorter = a->len > b->len ? b : a;
 816         int d = longer->len - shorter->len;
 817
 818         if (strncmp(longer->line + d, shorter->line, shorter->len))
 819                 return 0;
 820
 821         out->string = xmemdupz(longer->line, d);
 822         out->current_longer = (a == longer);
 823
 824         return 1;
 825 }
 826
 827 static int cmp_in_block_with_wsd(const struct diff_options *o,
 828                                  const struct moved_entry *cur,
 829                                  const struct moved_entry *match,
 830                                  struct moved_block *pmb,
 831                                  int n)
 832 {
 833         struct emitted_diff_symbol *l = &o->emitted_symbols->buf[n];
 834         int al = cur->es->len, cl = l->len;
 835         const char *a = cur->es->line,
 836                    *b = match->es->line,
 837                    *c = l->line;
 838
 839         int wslen;
 840
 841         /*
 842          * We need to check if 'cur' is equal to 'match'.
 843          * As those are from the same (+/-) side, we do not need to adjust for
 844          * indent changes. However these were found using fuzzy matching
 845          * so we do have to check if they are equal.
 846          */
 847         if (strcmp(a, b))
 848                 return 1;
 849
 850         if (!pmb->wsd.string)
 851                 /*
 852                  * The white space delta is not active? This can happen
 853                  * when we exit early in this function.
 854                  */
 855                 return 1;
 856
 857         /*
 858          * The indent changes of the block are known and stored in
 859          * pmb->wsd; however we need to check if the indent changes of the
 860          * current line are still the same as before.
 861          *
 862          * To do so we need to compare 'l' to 'cur', adjusting the
 863          * one of them for the white spaces, depending which was longer.
 864          */
 865
 866         wslen = strlen(pmb->wsd.string);
 867         if (pmb->wsd.current_longer) {
 868                 c += wslen;
 869                 cl -= wslen;
 870         } else {
 871                 a += wslen;
 872                 al -= wslen;
 873         }
 874
 875         if (al != cl || memcmp(a, c, al))
 876                 return 1;
 877
 878         return 0;
 879 }
 880
 881 static int moved_entry_cmp(const void *hashmap_cmp_fn_data,
 882                            const void *entry,
 883                            const void *entry_or_key,
 884                            const void *keydata)
 885 {
 886         const struct diff_options *diffopt = hashmap_cmp_fn_data;
 887         const struct moved_entry *a = entry;
 888         const struct moved_entry *b = entry_or_key;
 889         unsigned flags = diffopt->color_moved_ws_handling
 890                          & XDF_WHITESPACE_FLAGS;
 891
 892         if (diffopt->color_moved_ws_handling &
 893             COLOR_MOVED_WS_ALLOW_INDENTATION_CHANGE)
 894                 /*
 895                  * As there is not specific white space config given,
 896                  * we'd need to check for a new block, so ignore all
 897                  * white space. The setup of the white space
 898                  * configuration for the next block is done else where
 899                  */
 900                 flags |= XDF_IGNORE_WHITESPACE;
 901
 902         return !xdiff_compare_lines(a->es->line, a->es->len,
 903                                     b->es->line, b->es->len,
 904                                     flags);
 905 }
 906
 907 static struct moved_entry *prepare_entry(struct diff_options *o,
 908                                          int line_no)
 909 {
 910         struct moved_entry *ret = xmalloc(sizeof(*ret));
 911         struct emitted_diff_symbol *l = &o->emitted_symbols->buf[line_no];
 912         unsigned flags = o->color_moved_ws_handling & XDF_WHITESPACE_FLAGS;
 913
 914         ret->ent.hash = xdiff_hash_string(l->line, l->len, flags);
 915         ret->es = l;
 916         ret->next_line = NULL;
 917
 918         return ret;
 919 }
 920
 921 static void add_lines_to_move_detection(struct diff_options *o,
 922                                         struct hashmap *add_lines,
 923                                         struct hashmap *del_lines)
 924 {
 925         struct moved_entry *prev_line = NULL;
 926
 927         int n;
 928         for (n = 0; n < o->emitted_symbols->nr; n++) {
 929                 struct hashmap *hm;
 930                 struct moved_entry *key;
 931
 932                 switch (o->emitted_symbols->buf[n].s) {
 933                 case DIFF_SYMBOL_PLUS:
 934                         hm = add_lines;
 935                         break;
 936                 case DIFF_SYMBOL_MINUS:
 937                         hm = del_lines;
 938                         break;
 939                 default:
 940                         prev_line = NULL;
 941                         continue;
 942                 }
 943
 944                 key = prepare_entry(o, n);
 945                 if (prev_line && prev_line->es->s == o->emitted_symbols->buf[n].s)
 946                         prev_line->next_line = key;
 947
 948                 hashmap_add(hm, key);
 949                 prev_line = key;
 950         }
 951 }
 952
 953 static void pmb_advance_or_null(struct diff_options *o,
 954                                 struct moved_entry *match,
 955                                 struct hashmap *hm,
 956                                 struct moved_block *pmb,
 957                                 int pmb_nr)
 958 {
 959         int i;
 960         for (i = 0; i < pmb_nr; i++) {
 961                 struct moved_entry *prev = pmb[i].match;
 962                 struct moved_entry *cur = (prev && prev->next_line) ?
 963                                 prev->next_line : NULL;
 964                 if (cur && !hm->cmpfn(o, cur, match, NULL)) {
 965                         pmb[i].match = cur;
 966                 } else {
 967                         pmb[i].match = NULL;
 968                 }
 969         }
 970 }
 971
 972 static void pmb_advance_or_null_multi_match(struct diff_options *o,
 973                                             struct moved_entry *match,
 974                                             struct hashmap *hm,
 975                                             struct moved_block *pmb,
 976                                             int pmb_nr, int n)
 977 {
 978         int i;
 979         char *got_match = xcalloc(1, pmb_nr);
 980
 981         for (; match; match = hashmap_get_next(hm, match)) {
 982                 for (i = 0; i < pmb_nr; i++) {
 983                         struct moved_entry *prev = pmb[i].match;
 984                         struct moved_entry *cur = (prev && prev->next_line) ?
 985                                         prev->next_line : NULL;
 986                         if (!cur)
 987                                 continue;
 988                         if (!cmp_in_block_with_wsd(o, cur, match, &pmb[i], n))
 989                                 got_match[i] |= 1;
 990                 }
 991         }
 992
 993         for (i = 0; i < pmb_nr; i++) {
 994                 if (got_match[i]) {
 995                         /* Advance to the next line */
 996                         pmb[i].match = pmb[i].match->next_line;
 997                 } else {
 998                         moved_block_clear(&pmb[i]);
 999                 }
1000         }
1001
1002         free(got_match);
1003 }
1004
1005 static int shrink_potential_moved_blocks(struct moved_block *pmb,
1006                                          int pmb_nr)
1007 {
1008         int lp, rp;
1009
1010         /* Shrink the set of potential block to the remaining running */
1011         for (lp = 0, rp = pmb_nr - 1; lp <= rp;) {
1012                 while (lp < pmb_nr && pmb[lp].match)
1013                         lp++;
1014                 /* lp points at the first NULL now */
1015
1016                 while (rp > -1 && !pmb[rp].match)
1017                         rp--;
1018                 /* rp points at the last non-NULL */
1019
1020                 if (lp < pmb_nr && rp > -1 && lp < rp) {
1021                         pmb[lp] = pmb[rp];
1022                         pmb[rp].match = NULL;
1023                         pmb[rp].wsd.string = NULL;
1024                         rp--;
1025                         lp++;
1026                 }
1027         }
1028
1029         /* Remember the number of running sets */
1030         return rp + 1;
1031 }
1032
1033 /*
1034  * If o->color_moved is COLOR_MOVED_PLAIN, this function does nothing.
1035  *
1036  * Otherwise, if the last block has fewer alphanumeric characters than
1037  * COLOR_MOVED_MIN_ALNUM_COUNT, unset DIFF_SYMBOL_MOVED_LINE on all lines in
1038  * that block.
1039  *
1040  * The last block consists of the (n - block_length)'th line up to but not
1041  * including the nth line.
1042  *
1043  * NEEDSWORK: This uses the same heuristic as blame_entry_score() in blame.c.
1044  * Think of a way to unify them.
1045  */
1046 static void adjust_last_block(struct diff_options *o, int n, int block_length)
1047 {
1048         int i, alnum_count = 0;
1049         if (o->color_moved == COLOR_MOVED_PLAIN)
1050                 return;
1051         for (i = 1; i < block_length + 1; i++) {
1052                 const char *c = o->emitted_symbols->buf[n - i].line;
1053                 for (; *c; c++) {
1054                         if (!isalnum(*c))
1055                                 continue;
1056                         alnum_count++;
1057                         if (alnum_count >= COLOR_MOVED_MIN_ALNUM_COUNT)
1058                                 return;
1059                 }
1060         }
1061         for (i = 1; i < block_length + 1; i++)
1062                 o->emitted_symbols->buf[n - i].flags &= ~DIFF_SYMBOL_MOVED_LINE;
1063 }
1064
1065 /* Find blocks of moved code, delegate actual coloring decision to helper */
1066 static void mark_color_as_moved(struct diff_options *o,
1067                                 struct hashmap *add_lines,
1068                                 struct hashmap *del_lines)
1069 {
1070         struct moved_block *pmb = NULL; /* potentially moved blocks */
1071         int pmb_nr = 0, pmb_alloc = 0;
1072         int n, flipped_block = 1, block_length = 0;
1073
1074
1075         for (n = 0; n < o->emitted_symbols->nr; n++) {
1076                 struct hashmap *hm = NULL;
1077                 struct moved_entry *key;
1078                 struct moved_entry *match = NULL;
1079                 struct emitted_diff_symbol *l = &o->emitted_symbols->buf[n];
1080
1081                 switch (l->s) {
1082                 case DIFF_SYMBOL_PLUS:
1083                         hm = del_lines;
1084                         key = prepare_entry(o, n);
1085                         match = hashmap_get(hm, key, NULL);
1086                         free(key);
1087                         break;
1088                 case DIFF_SYMBOL_MINUS:
1089                         hm = add_lines;
1090                         key = prepare_entry(o, n);
1091                         match = hashmap_get(hm, key, NULL);
1092                         free(key);
1093                         break;
1094                 default:
1095                         flipped_block = 1;
1096                 }
1097
1098                 if (!match) {
1099                         int i;
1100
1101                         adjust_last_block(o, n, block_length);
1102                         for(i = 0; i < pmb_nr; i++)
1103                                 moved_block_clear(&pmb[i]);
1104                         pmb_nr = 0;
1105                         block_length = 0;
1106                         continue;
1107                 }
1108
1109                 l->flags |= DIFF_SYMBOL_MOVED_LINE;
1110
1111                 if (o->color_moved == COLOR_MOVED_PLAIN)
1112                         continue;
1113
1114                 if (o->color_moved_ws_handling &
1115                     COLOR_MOVED_WS_ALLOW_INDENTATION_CHANGE)
1116                         pmb_advance_or_null_multi_match(o, match, hm, pmb, pmb_nr, n);
1117                 else
1118                         pmb_advance_or_null(o, match, hm, pmb, pmb_nr);
1119
1120                 pmb_nr = shrink_potential_moved_blocks(pmb, pmb_nr);
1121
1122                 if (pmb_nr == 0) {
1123                         /*
1124                          * The current line is the start of a new block.
1125                          * Setup the set of potential blocks.
1126                          */
1127                         for (; match; match = hashmap_get_next(hm, match)) {
1128                                 ALLOC_GROW(pmb, pmb_nr + 1, pmb_alloc);
1129                                 if (o->color_moved_ws_handling &
1130                                     COLOR_MOVED_WS_ALLOW_INDENTATION_CHANGE) {
1131                                         if (compute_ws_delta(l, match->es,
1132                                                              &pmb[pmb_nr].wsd))
1133                                                 pmb[pmb_nr++].match = match;
1134                                 } else {
1135                                         pmb[pmb_nr].wsd.string = NULL;
1136                                         pmb[pmb_nr++].match = match;
1137                                 }
1138                         }
1139
1140                         flipped_block = (flipped_block + 1) % 2;
1141
1142                         adjust_last_block(o, n, block_length);
1143                         block_length = 0;
1144                 }
1145
1146                 block_length++;
1147
1148                 if (flipped_block && o->color_moved != COLOR_MOVED_BLOCKS)
1149                         l->flags |= DIFF_SYMBOL_MOVED_LINE_ALT;
1150         }
1151         adjust_last_block(o, n, block_length);
1152
1153         for(n = 0; n < pmb_nr; n++)
1154                 moved_block_clear(&pmb[n]);
1155         free(pmb);
1156 }
1157
1158 #define DIFF_SYMBOL_MOVED_LINE_ZEBRA_MASK \
1159   (DIFF_SYMBOL_MOVED_LINE | DIFF_SYMBOL_MOVED_LINE_ALT)
1160 static void dim_moved_lines(struct diff_options *o)
1161 {
1162         int n;
1163         for (n = 0; n < o->emitted_symbols->nr; n++) {
1164                 struct emitted_diff_symbol *prev = (n != 0) ?
1165                                 &o->emitted_symbols->buf[n - 1] : NULL;
1166                 struct emitted_diff_symbol *l = &o->emitted_symbols->buf[n];
1167                 struct emitted_diff_symbol *next =
1168                                 (n < o->emitted_symbols->nr - 1) ?
1169                                 &o->emitted_symbols->buf[n + 1] : NULL;
1170
1171                 /* Not a plus or minus line? */
1172                 if (l->s != DIFF_SYMBOL_PLUS && l->s != DIFF_SYMBOL_MINUS)
1173                         continue;
1174
1175                 /* Not a moved line? */
1176                 if (!(l->flags & DIFF_SYMBOL_MOVED_LINE))
1177                         continue;
1178
1179                 /*
1180                  * If prev or next are not a plus or minus line,
1181                  * pretend they don't exist
1182                  */
1183                 if (prev && prev->s != DIFF_SYMBOL_PLUS &&
1184                             prev->s != DIFF_SYMBOL_MINUS)
1185                         prev = NULL;
1186                 if (next && next->s != DIFF_SYMBOL_PLUS &&
1187                             next->s != DIFF_SYMBOL_MINUS)
1188                         next = NULL;
1189
1190                 /* Inside a block? */
1191                 if ((prev &&
1192                     (prev->flags & DIFF_SYMBOL_MOVED_LINE_ZEBRA_MASK) ==
1193                     (l->flags & DIFF_SYMBOL_MOVED_LINE_ZEBRA_MASK)) &&
1194                     (next &&
1195                     (next->flags & DIFF_SYMBOL_MOVED_LINE_ZEBRA_MASK) ==
1196                     (l->flags & DIFF_SYMBOL_MOVED_LINE_ZEBRA_MASK))) {
1197                         l->flags |= DIFF_SYMBOL_MOVED_LINE_UNINTERESTING;
1198                         continue;
1199                 }
1200
1201                 /* Check if we are at an interesting bound: */
1202                 if (prev && (prev->flags & DIFF_SYMBOL_MOVED_LINE) &&
1203                     (prev->flags & DIFF_SYMBOL_MOVED_LINE_ALT) !=
1204                        (l->flags & DIFF_SYMBOL_MOVED_LINE_ALT))
1205                         continue;
1206                 if (next && (next->flags & DIFF_SYMBOL_MOVED_LINE) &&
1207                     (next->flags & DIFF_SYMBOL_MOVED_LINE_ALT) !=
1208                        (l->flags & DIFF_SYMBOL_MOVED_LINE_ALT))
1209                         continue;
1210
1211                 /*
1212                  * The boundary to prev and next are not interesting,
1213                  * so this line is not interesting as a whole
1214                  */
1215                 l->flags |= DIFF_SYMBOL_MOVED_LINE_UNINTERESTING;
1216         }
1217 }
1218
1219 static void emit_line_ws_markup(struct diff_options *o,
1220                                 const char *set_sign, const char *set,
1221                                 const char *reset,
1222                                 int sign_index, const char *line, int len,
1223                                 unsigned ws_rule, int blank_at_eof)
1224 {
1225         const char *ws = NULL;
1226         int sign = o->output_indicators[sign_index];
1227
1228         if (o->ws_error_highlight & ws_rule) {
1229                 ws = diff_get_color_opt(o, DIFF_WHITESPACE);
1230                 if (!*ws)
1231                         ws = NULL;
1232         }
1233
1234         if (!ws && !set_sign)
1235                 emit_line_0(o, set, NULL, 0, reset, sign, line, len);
1236         else if (!ws) {
1237                 emit_line_0(o, set_sign, set, !!set_sign, reset, sign, line, len);
1238         } else if (blank_at_eof)
1239                 /* Blank line at EOF - paint '+' as well */
1240                 emit_line_0(o, ws, NULL, 0, reset, sign, line, len);
1241         else {
1242                 /* Emit just the prefix, then the rest. */
1243                 emit_line_0(o, set_sign ? set_sign : set, NULL, !!set_sign, reset,
1244                             sign, "", 0);
1245                 ws_check_emit(line, len, ws_rule,
1246                               o->file, set, reset, ws);
1247         }
1248 }
1249
1250 static void emit_diff_symbol_from_struct(struct diff_options *o,
1251                                          struct emitted_diff_symbol *eds)
1252 {
1253         static const char *nneof = " No newline at end of file\n";
1254         const char *context, *reset, *set, *set_sign, *meta, *fraginfo;
1255         struct strbuf sb = STRBUF_INIT;
1256
1257         enum diff_symbol s = eds->s;
1258         const char *line = eds->line;
1259         int len = eds->len;
1260         unsigned flags = eds->flags;
1261
1262         switch (s) {
1263         case DIFF_SYMBOL_NO_LF_EOF:
1264                 context = diff_get_color_opt(o, DIFF_CONTEXT);
1265                 reset = diff_get_color_opt(o, DIFF_RESET);
1266                 putc('\n', o->file);
1267                 emit_line_0(o, context, NULL, 0, reset, '\\',
1268                             nneof, strlen(nneof));
1269                 break;
1270         case DIFF_SYMBOL_SUBMODULE_HEADER:
1271         case DIFF_SYMBOL_SUBMODULE_ERROR:
1272         case DIFF_SYMBOL_SUBMODULE_PIPETHROUGH:
1273         case DIFF_SYMBOL_STATS_SUMMARY_INSERTS_DELETES:
1274         case DIFF_SYMBOL_SUMMARY:
1275         case DIFF_SYMBOL_STATS_LINE:
1276         case DIFF_SYMBOL_BINARY_DIFF_BODY:
1277         case DIFF_SYMBOL_CONTEXT_FRAGINFO:
1278                 emit_line(o, "", "", line, len);
1279                 break;
1280         case DIFF_SYMBOL_CONTEXT_INCOMPLETE:
1281         case DIFF_SYMBOL_CONTEXT_MARKER:
1282                 context = diff_get_color_opt(o, DIFF_CONTEXT);
1283                 reset = diff_get_color_opt(o, DIFF_RESET);
1284                 emit_line(o, context, reset, line, len);
1285                 break;
1286         case DIFF_SYMBOL_SEPARATOR:
1287                 fprintf(o->file, "%s%c",
1288                         diff_line_prefix(o),
1289                         o->line_termination);
1290                 break;
1291         case DIFF_SYMBOL_CONTEXT:
1292                 set = diff_get_color_opt(o, DIFF_CONTEXT);
1293                 reset = diff_get_color_opt(o, DIFF_RESET);
1294                 set_sign = NULL;
1295                 if (o->flags.dual_color_diffed_diffs) {
1296                         char c = !len ? 0 : line[0];
1297
1298                         if (c == '+')
1299                                 set = diff_get_color_opt(o, DIFF_FILE_NEW);
1300                         else if (c == '@')
1301                                 set = diff_get_color_opt(o, DIFF_FRAGINFO);
1302                         else if (c == '-')
1303                                 set = diff_get_color_opt(o, DIFF_FILE_OLD);
1304                 }
1305                 emit_line_ws_markup(o, set_sign, set, reset,
1306                                     OUTPUT_INDICATOR_CONTEXT, line, len,
1307                                     flags & (DIFF_SYMBOL_CONTENT_WS_MASK), 0);
1308                 break;
1309         case DIFF_SYMBOL_PLUS:
1310                 switch (flags & (DIFF_SYMBOL_MOVED_LINE |
1311                                  DIFF_SYMBOL_MOVED_LINE_ALT |
1312                                  DIFF_SYMBOL_MOVED_LINE_UNINTERESTING)) {
1313                 case DIFF_SYMBOL_MOVED_LINE |
1314                      DIFF_SYMBOL_MOVED_LINE_ALT |
1315                      DIFF_SYMBOL_MOVED_LINE_UNINTERESTING:
1316                         set = diff_get_color_opt(o, DIFF_FILE_NEW_MOVED_ALT_DIM);
1317                         break;
1318                 case DIFF_SYMBOL_MOVED_LINE |
1319                      DIFF_SYMBOL_MOVED_LINE_ALT:
1320                         set = diff_get_color_opt(o, DIFF_FILE_NEW_MOVED_ALT);
1321                         break;
1322                 case DIFF_SYMBOL_MOVED_LINE |
1323                      DIFF_SYMBOL_MOVED_LINE_UNINTERESTING:
1324                         set = diff_get_color_opt(o, DIFF_FILE_NEW_MOVED_DIM);
1325                         break;
1326                 case DIFF_SYMBOL_MOVED_LINE:
1327                         set = diff_get_color_opt(o, DIFF_FILE_NEW_MOVED);
1328                         break;
1329                 default:
1330                         set = diff_get_color_opt(o, DIFF_FILE_NEW);
1331                 }
1332                 reset = diff_get_color_opt(o, DIFF_RESET);
1333                 if (!o->flags.dual_color_diffed_diffs)
1334                         set_sign = NULL;
1335                 else {
1336                         char c = !len ? 0 : line[0];
1337
1338                         set_sign = set;
1339                         if (c == '-')
1340                                 set = diff_get_color_opt(o, DIFF_FILE_OLD_BOLD);
1341                         else if (c == '@')
1342                                 set = diff_get_color_opt(o, DIFF_FRAGINFO);
1343                         else if (c == '+')
1344                                 set = diff_get_color_opt(o, DIFF_FILE_NEW_BOLD);
1345                         else
1346                                 set = diff_get_color_opt(o, DIFF_CONTEXT_BOLD);
1347                         flags &= ~DIFF_SYMBOL_CONTENT_WS_MASK;
1348                 }
1349                 emit_line_ws_markup(o, set_sign, set, reset,
1350                                     OUTPUT_INDICATOR_NEW, line, len,
1351                                     flags & DIFF_SYMBOL_CONTENT_WS_MASK,
1352                                     flags & DIFF_SYMBOL_CONTENT_BLANK_LINE_EOF);
1353                 break;
1354         case DIFF_SYMBOL_MINUS:
1355                 switch (flags & (DIFF_SYMBOL_MOVED_LINE |
1356                                  DIFF_SYMBOL_MOVED_LINE_ALT |
1357                                  DIFF_SYMBOL_MOVED_LINE_UNINTERESTING)) {
1358                 case DIFF_SYMBOL_MOVED_LINE |
1359                      DIFF_SYMBOL_MOVED_LINE_ALT |
1360                      DIFF_SYMBOL_MOVED_LINE_UNINTERESTING:
1361                         set = diff_get_color_opt(o, DIFF_FILE_OLD_MOVED_ALT_DIM);
1362                         break;
1363                 case DIFF_SYMBOL_MOVED_LINE |
1364                      DIFF_SYMBOL_MOVED_LINE_ALT:
1365                         set = diff_get_color_opt(o, DIFF_FILE_OLD_MOVED_ALT);
1366                         break;
1367                 case DIFF_SYMBOL_MOVED_LINE |
1368                      DIFF_SYMBOL_MOVED_LINE_UNINTERESTING:
1369                         set = diff_get_color_opt(o, DIFF_FILE_OLD_MOVED_DIM);
1370                         break;
1371                 case DIFF_SYMBOL_MOVED_LINE:
1372                         set = diff_get_color_opt(o, DIFF_FILE_OLD_MOVED);
1373                         break;
1374                 default:
1375                         set = diff_get_color_opt(o, DIFF_FILE_OLD);
1376                 }
1377                 reset = diff_get_color_opt(o, DIFF_RESET);
1378                 if (!o->flags.dual_color_diffed_diffs)
1379                         set_sign = NULL;
1380                 else {
1381                         char c = !len ? 0 : line[0];
1382
1383                         set_sign = set;
1384                         if (c == '+')
1385                                 set = diff_get_color_opt(o, DIFF_FILE_NEW_DIM);
1386                         else if (c == '@')
1387                                 set = diff_get_color_opt(o, DIFF_FRAGINFO);
1388                         else if (c == '-')
1389                                 set = diff_get_color_opt(o, DIFF_FILE_OLD_DIM);
1390                         else
1391                                 set = diff_get_color_opt(o, DIFF_CONTEXT_DIM);
1392                 }
1393                 emit_line_ws_markup(o, set_sign, set, reset,
1394                                     OUTPUT_INDICATOR_OLD, line, len,
1395                                     flags & DIFF_SYMBOL_CONTENT_WS_MASK, 0);
1396                 break;
1397         case DIFF_SYMBOL_WORDS_PORCELAIN:
1398                 context = diff_get_color_opt(o, DIFF_CONTEXT);
1399                 reset = diff_get_color_opt(o, DIFF_RESET);
1400                 emit_line(o, context, reset, line, len);
1401                 fputs("~\n", o->file);
1402                 break;
1403         case DIFF_SYMBOL_WORDS:
1404                 context = diff_get_color_opt(o, DIFF_CONTEXT);
1405                 reset = diff_get_color_opt(o, DIFF_RESET);
1406                 /*
1407                  * Skip the prefix character, if any.  With
1408                  * diff_suppress_blank_empty, there may be
1409                  * none.
1410                  */
1411                 if (line[0] != '\n') {
1412                         line++;
1413                         len--;
1414                 }
1415                 emit_line(o, context, reset, line, len);
1416                 break;
1417         case DIFF_SYMBOL_FILEPAIR_PLUS:
1418                 meta = diff_get_color_opt(o, DIFF_METAINFO);
1419                 reset = diff_get_color_opt(o, DIFF_RESET);
1420                 fprintf(o->file, "%s%s+++ %s%s%s\n", diff_line_prefix(o), meta,
1421                         line, reset,
1422                         strchr(line, ' ') ? "\t" : "");
1423                 break;
1424         case DIFF_SYMBOL_FILEPAIR_MINUS:
1425                 meta = diff_get_color_opt(o, DIFF_METAINFO);
1426                 reset = diff_get_color_opt(o, DIFF_RESET);
1427                 fprintf(o->file, "%s%s--- %s%s%s\n", diff_line_prefix(o), meta,
1428                         line, reset,
1429                         strchr(line, ' ') ? "\t" : "");
1430                 break;
1431         case DIFF_SYMBOL_BINARY_FILES:
1432         case DIFF_SYMBOL_HEADER:
1433                 fprintf(o->file, "%s", line);
1434                 break;
1435         case DIFF_SYMBOL_BINARY_DIFF_HEADER:
1436                 fprintf(o->file, "%sGIT binary patch\n", diff_line_prefix(o));
1437                 break;
1438         case DIFF_SYMBOL_BINARY_DIFF_HEADER_DELTA:
1439                 fprintf(o->file, "%sdelta %s\n", diff_line_prefix(o), line);
1440                 break;
1441         case DIFF_SYMBOL_BINARY_DIFF_HEADER_LITERAL:
1442                 fprintf(o->file, "%sliteral %s\n", diff_line_prefix(o), line);
1443                 break;
1444         case DIFF_SYMBOL_BINARY_DIFF_FOOTER:
1445                 fputs(diff_line_prefix(o), o->file);
1446                 fputc('\n', o->file);
1447                 break;
1448         case DIFF_SYMBOL_REWRITE_DIFF:
1449                 fraginfo = diff_get_color(o->use_color, DIFF_FRAGINFO);
1450                 reset = diff_get_color_opt(o, DIFF_RESET);
1451                 emit_line(o, fraginfo, reset, line, len);
1452                 break;
1453         case DIFF_SYMBOL_SUBMODULE_ADD:
1454                 set = diff_get_color_opt(o, DIFF_FILE_NEW);
1455                 reset = diff_get_color_opt(o, DIFF_RESET);
1456                 emit_line(o, set, reset, line, len);
1457                 break;
1458         case DIFF_SYMBOL_SUBMODULE_DEL:
1459                 set = diff_get_color_opt(o, DIFF_FILE_OLD);
1460                 reset = diff_get_color_opt(o, DIFF_RESET);
1461                 emit_line(o, set, reset, line, len);
1462                 break;
1463         case DIFF_SYMBOL_SUBMODULE_UNTRACKED:
1464                 fprintf(o->file, "%sSubmodule %s contains untracked content\n",
1465                         diff_line_prefix(o), line);
1466                 break;
1467         case DIFF_SYMBOL_SUBMODULE_MODIFIED:
1468                 fprintf(o->file, "%sSubmodule %s contains modified content\n",
1469                         diff_line_prefix(o), line);
1470                 break;
1471         case DIFF_SYMBOL_STATS_SUMMARY_NO_FILES:
1472                 emit_line(o, "", "", " 0 files changed\n",
1473                           strlen(" 0 files changed\n"));
1474                 break;
1475         case DIFF_SYMBOL_STATS_SUMMARY_ABBREV:
1476                 emit_line(o, "", "", " ...\n", strlen(" ...\n"));
1477                 break;
1478         case DIFF_SYMBOL_WORD_DIFF:
1479                 fprintf(o->file, "%.*s", len, line);
1480                 break;
1481         case DIFF_SYMBOL_STAT_SEP:
1482                 fputs(o->stat_sep, o->file);
1483                 break;
1484         default:
1485                 BUG("unknown diff symbol");
1486         }
1487         strbuf_release(&sb);
1488 }
1489
1490 static void emit_diff_symbol(struct diff_options *o, enum diff_symbol s,
1491                              const char *line, int len, unsigned flags)
1492 {
1493         struct emitted_diff_symbol e = {line, len, flags, s};
1494
1495         if (o->emitted_symbols)
1496                 append_emitted_diff_symbol(o, &e);
1497         else
1498                 emit_diff_symbol_from_struct(o, &e);
1499 }
1500
1501 void diff_emit_submodule_del(struct diff_options *o, const char *line)
1502 {
1503         emit_diff_symbol(o, DIFF_SYMBOL_SUBMODULE_DEL, line, strlen(line), 0);
1504 }
1505
1506 void diff_emit_submodule_add(struct diff_options *o, const char *line)
1507 {
1508         emit_diff_symbol(o, DIFF_SYMBOL_SUBMODULE_ADD, line, strlen(line), 0);
1509 }
1510
1511 void diff_emit_submodule_untracked(struct diff_options *o, const char *path)
1512 {
1513         emit_diff_symbol(o, DIFF_SYMBOL_SUBMODULE_UNTRACKED,
1514                          path, strlen(path), 0);
1515 }
1516
1517 void diff_emit_submodule_modified(struct diff_options *o, const char *path)
1518 {
1519         emit_diff_symbol(o, DIFF_SYMBOL_SUBMODULE_MODIFIED,
1520                          path, strlen(path), 0);
1521 }
1522
1523 void diff_emit_submodule_header(struct diff_options *o, const char *header)
1524 {
1525         emit_diff_symbol(o, DIFF_SYMBOL_SUBMODULE_HEADER,
1526                          header, strlen(header), 0);
1527 }
1528
1529 void diff_emit_submodule_error(struct diff_options *o, const char *err)
1530 {
1531         emit_diff_symbol(o, DIFF_SYMBOL_SUBMODULE_ERROR, err, strlen(err), 0);
1532 }
1533
1534 void diff_emit_submodule_pipethrough(struct diff_options *o,
1535                                      const char *line, int len)
1536 {
1537         emit_diff_symbol(o, DIFF_SYMBOL_SUBMODULE_PIPETHROUGH, line, len, 0);
1538 }
1539
1540 static int new_blank_line_at_eof(struct emit_callback *ecbdata, const char *line, int len)
1541 {
1542         if (!((ecbdata->ws_rule & WS_BLANK_AT_EOF) &&
1543               ecbdata->blank_at_eof_in_preimage &&
1544               ecbdata->blank_at_eof_in_postimage &&
1545               ecbdata->blank_at_eof_in_preimage <= ecbdata->lno_in_preimage &&
1546               ecbdata->blank_at_eof_in_postimage <= ecbdata->lno_in_postimage))
1547                 return 0;
1548         return ws_blank_line(line, len, ecbdata->ws_rule);
1549 }
1550
1551 static void emit_add_line(const char *reset,
1552                           struct emit_callback *ecbdata,
1553                           const char *line, int len)
1554 {
1555         unsigned flags = WSEH_NEW | ecbdata->ws_rule;
1556         if (new_blank_line_at_eof(ecbdata, line, len))
1557                 flags |= DIFF_SYMBOL_CONTENT_BLANK_LINE_EOF;
1558
1559         emit_diff_symbol(ecbdata->opt, DIFF_SYMBOL_PLUS, line, len, flags);
1560 }
1561
1562 static void emit_del_line(const char *reset,
1563                           struct emit_callback *ecbdata,
1564                           const char *line, int len)
1565 {
1566         unsigned flags = WSEH_OLD | ecbdata->ws_rule;
1567         emit_diff_symbol(ecbdata->opt, DIFF_SYMBOL_MINUS, line, len, flags);
1568 }
1569
1570 static void emit_context_line(const char *reset,
1571                               struct emit_callback *ecbdata,
1572                               const char *line, int len)
1573 {
1574         unsigned flags = WSEH_CONTEXT | ecbdata->ws_rule;
1575         emit_diff_symbol(ecbdata->opt, DIFF_SYMBOL_CONTEXT, line, len, flags);
1576 }
1577
1578 static void emit_hunk_header(struct emit_callback *ecbdata,
1579                              const char *line, int len)
1580 {
1581         const char *context = diff_get_color(ecbdata->color_diff, DIFF_CONTEXT);
1582         const char *frag = diff_get_color(ecbdata->color_diff, DIFF_FRAGINFO);
1583         const char *func = diff_get_color(ecbdata->color_diff, DIFF_FUNCINFO);
1584         const char *reset = diff_get_color(ecbdata->color_diff, DIFF_RESET);
1585         const char *reverse = ecbdata->color_diff ? GIT_COLOR_REVERSE : "";
1586         static const char atat[2] = { '@', '@' };
1587         const char *cp, *ep;
1588         struct strbuf msgbuf = STRBUF_INIT;
1589         int org_len = len;
1590         int i = 1;
1591
1592         /*
1593          * As a hunk header must begin with "@@ -<old>, +<new> @@",
1594          * it always is at least 10 bytes long.
1595          */
1596         if (len < 10 ||
1597             memcmp(line, atat, 2) ||
1598             !(ep = memmem(line + 2, len - 2, atat, 2))) {
1599                 emit_diff_symbol(ecbdata->opt,
1600                                  DIFF_SYMBOL_CONTEXT_MARKER, line, len, 0);
1601                 return;
1602         }
1603         ep += 2; /* skip over @@ */
1604
1605         /* The hunk header in fraginfo color */
1606         if (ecbdata->opt->flags.dual_color_diffed_diffs)
1607                 strbuf_addstr(&msgbuf, reverse);
1608         strbuf_addstr(&msgbuf, frag);
1609         strbuf_add(&msgbuf, line, ep - line);
1610         strbuf_addstr(&msgbuf, reset);
1611
1612         /*
1613          * trailing "\r\n"
1614          */
1615         for ( ; i < 3; i++)
1616                 if (line[len - i] == '\r' || line[len - i] == '\n')
1617                         len--;
1618
1619         /* blank before the func header */
1620         for (cp = ep; ep - line < len; ep++)
1621                 if (*ep != ' ' && *ep != '\t')
1622                         break;
1623         if (ep != cp) {
1624                 strbuf_addstr(&msgbuf, context);
1625                 strbuf_add(&msgbuf, cp, ep - cp);
1626                 strbuf_addstr(&msgbuf, reset);
1627         }
1628
1629         if (ep < line + len) {
1630                 strbuf_addstr(&msgbuf, func);
1631                 strbuf_add(&msgbuf, ep, line + len - ep);
1632                 strbuf_addstr(&msgbuf, reset);
1633         }
1634
1635         strbuf_add(&msgbuf, line + len, org_len - len);
1636         strbuf_complete_line(&msgbuf);
1637         emit_diff_symbol(ecbdata->opt,
1638                          DIFF_SYMBOL_CONTEXT_FRAGINFO, msgbuf.buf, msgbuf.len, 0);
1639         strbuf_release(&msgbuf);
1640 }
1641
1642 static struct diff_tempfile *claim_diff_tempfile(void) {
1643         int i;
1644         for (i = 0; i < ARRAY_SIZE(diff_temp); i++)
1645                 if (!diff_temp[i].name)
1646                         return diff_temp + i;
1647         BUG("diff is failing to clean up its tempfiles");
1648 }
1649
1650 static void remove_tempfile(void)
1651 {
1652         int i;
1653         for (i = 0; i < ARRAY_SIZE(diff_temp); i++) {
1654                 if (is_tempfile_active(diff_temp[i].tempfile))
1655                         delete_tempfile(&diff_temp[i].tempfile);
1656                 diff_temp[i].name = NULL;
1657         }
1658 }
1659
1660 static void add_line_count(struct strbuf *out, int count)
1661 {
1662         switch (count) {
1663         case 0:
1664                 strbuf_addstr(out, "0,0");
1665                 break;
1666         case 1:
1667                 strbuf_addstr(out, "1");
1668                 break;
1669         default:
1670                 strbuf_addf(out, "1,%d", count);
1671                 break;
1672         }
1673 }
1674
1675 static void emit_rewrite_lines(struct emit_callback *ecb,
1676                                int prefix, const char *data, int size)
1677 {
1678         const char *endp = NULL;
1679         const char *reset = diff_get_color(ecb->color_diff, DIFF_RESET);
1680
1681         while (0 < size) {
1682                 int len;
1683
1684                 endp = memchr(data, '\n', size);
1685                 len = endp ? (endp - data + 1) : size;
1686                 if (prefix != '+') {
1687                         ecb->lno_in_preimage++;
1688                         emit_del_line(reset, ecb, data, len);
1689                 } else {
1690                         ecb->lno_in_postimage++;
1691                         emit_add_line(reset, ecb, data, len);
1692                 }
1693                 size -= len;
1694                 data += len;
1695         }
1696         if (!endp)
1697                 emit_diff_symbol(ecb->opt, DIFF_SYMBOL_NO_LF_EOF, NULL, 0, 0);
1698 }
1699
1700 static void emit_rewrite_diff(const char *name_a,
1701                               const char *name_b,
1702                               struct diff_filespec *one,
1703                               struct diff_filespec *two,
1704                               struct userdiff_driver *textconv_one,
1705                               struct userdiff_driver *textconv_two,
1706                               struct diff_options *o)
1707 {
1708         int lc_a, lc_b;
1709         static struct strbuf a_name = STRBUF_INIT, b_name = STRBUF_INIT;
1710         const char *a_prefix, *b_prefix;
1711         char *data_one, *data_two;
1712         size_t size_one, size_two;
1713         struct emit_callback ecbdata;
1714         struct strbuf out = STRBUF_INIT;
1715
1716         if (diff_mnemonic_prefix && o->flags.reverse_diff) {
1717                 a_prefix = o->b_prefix;
1718                 b_prefix = o->a_prefix;
1719         } else {
1720                 a_prefix = o->a_prefix;
1721                 b_prefix = o->b_prefix;
1722         }
1723
1724         name_a += (*name_a == '/');
1725         name_b += (*name_b == '/');
1726
1727         strbuf_reset(&a_name);
1728         strbuf_reset(&b_name);
1729         quote_two_c_style(&a_name, a_prefix, name_a, 0);
1730         quote_two_c_style(&b_name, b_prefix, name_b, 0);
1731
1732         size_one = fill_textconv(o->repo, textconv_one, one, &data_one);
1733         size_two = fill_textconv(o->repo, textconv_two, two, &data_two);
1734
1735         memset(&ecbdata, 0, sizeof(ecbdata));
1736         ecbdata.color_diff = want_color(o->use_color);
1737         ecbdata.ws_rule = whitespace_rule(o->repo->index, name_b);
1738         ecbdata.opt = o;
1739         if (ecbdata.ws_rule & WS_BLANK_AT_EOF) {
1740                 mmfile_t mf1, mf2;
1741                 mf1.ptr = (char *)data_one;
1742                 mf2.ptr = (char *)data_two;
1743                 mf1.size = size_one;
1744                 mf2.size = size_two;
1745                 check_blank_at_eof(&mf1, &mf2, &ecbdata);
1746         }
1747         ecbdata.lno_in_preimage = 1;
1748         ecbdata.lno_in_postimage = 1;
1749
1750         lc_a = count_lines(data_one, size_one);
1751         lc_b = count_lines(data_two, size_two);
1752
1753         emit_diff_symbol(o, DIFF_SYMBOL_FILEPAIR_MINUS,
1754                          a_name.buf, a_name.len, 0);
1755         emit_diff_symbol(o, DIFF_SYMBOL_FILEPAIR_PLUS,
1756                          b_name.buf, b_name.len, 0);
1757
1758         strbuf_addstr(&out, "@@ -");
1759         if (!o->irreversible_delete)
1760                 add_line_count(&out, lc_a);
1761         else
1762                 strbuf_addstr(&out, "?,?");
1763         strbuf_addstr(&out, " +");
1764         add_line_count(&out, lc_b);
1765         strbuf_addstr(&out, " @@\n");
1766         emit_diff_symbol(o, DIFF_SYMBOL_REWRITE_DIFF, out.buf, out.len, 0);
1767         strbuf_release(&out);
1768
1769         if (lc_a && !o->irreversible_delete)
1770                 emit_rewrite_lines(&ecbdata, '-', data_one, size_one);
1771         if (lc_b)
1772                 emit_rewrite_lines(&ecbdata, '+', data_two, size_two);
1773         if (textconv_one)
1774                 free((char *)data_one);
1775         if (textconv_two)
1776                 free((char *)data_two);
1777 }
1778
1779 struct diff_words_buffer {
1780         mmfile_t text;
1781         unsigned long alloc;
1782         struct diff_words_orig {
1783                 const char *begin, *end;
1784         } *orig;
1785         int orig_nr, orig_alloc;
1786 };
1787
1788 static void diff_words_append(char *line, unsigned long len,
1789                 struct diff_words_buffer *buffer)
1790 {
1791         ALLOC_GROW(buffer->text.ptr, buffer->text.size + len, buffer->alloc);
1792         line++;
1793         len--;
1794         memcpy(buffer->text.ptr + buffer->text.size, line, len);
1795         buffer->text.size += len;
1796         buffer->text.ptr[buffer->text.size] = '\0';
1797 }
1798
1799 struct diff_words_style_elem {
1800         const char *prefix;
1801         const char *suffix;
1802         const char *color; /* NULL; filled in by the setup code if
1803                             * color is enabled */
1804 };
1805
1806 struct diff_words_style {
1807         enum diff_words_type type;
1808         struct diff_words_style_elem new_word, old_word, ctx;
1809         const char *newline;
1810 };
1811
1812 static struct diff_words_style diff_words_styles[] = {
1813         { DIFF_WORDS_PORCELAIN, {"+", "\n"}, {"-", "\n"}, {" ", "\n"}, "~\n" },
1814         { DIFF_WORDS_PLAIN, {"{+", "+}"}, {"[-", "-]"}, {"", ""}, "\n" },
1815         { DIFF_WORDS_COLOR, {"", ""}, {"", ""}, {"", ""}, "\n" }
1816 };
1817
1818 struct diff_words_data {
1819         struct diff_words_buffer minus, plus;
1820         const char *current_plus;
1821         int last_minus;
1822         struct diff_options *opt;
1823         regex_t *word_regex;
1824         enum diff_words_type type;
1825         struct diff_words_style *style;
1826 };
1827
1828 static int fn_out_diff_words_write_helper(struct diff_options *o,
1829                                           struct diff_words_style_elem *st_el,
1830                                           const char *newline,
1831                                           size_t count, const char *buf)
1832 {
1833         int print = 0;
1834         struct strbuf sb = STRBUF_INIT;
1835
1836         while (count) {
1837                 char *p = memchr(buf, '\n', count);
1838                 if (print)
1839                         strbuf_addstr(&sb, diff_line_prefix(o));
1840
1841                 if (p != buf) {
1842                         const char *reset = st_el->color && *st_el->color ?
1843                                             GIT_COLOR_RESET : NULL;
1844                         if (st_el->color && *st_el->color)
1845                                 strbuf_addstr(&sb, st_el->color);
1846                         strbuf_addstr(&sb, st_el->prefix);
1847                         strbuf_add(&sb, buf, p ? p - buf : count);
1848                         strbuf_addstr(&sb, st_el->suffix);
1849                         if (reset)
1850                                 strbuf_addstr(&sb, reset);
1851                 }
1852                 if (!p)
1853                         goto out;
1854
1855                 strbuf_addstr(&sb, newline);
1856                 count -= p + 1 - buf;
1857                 buf = p + 1;
1858                 print = 1;
1859                 if (count) {
1860                         emit_diff_symbol(o, DIFF_SYMBOL_WORD_DIFF,
1861                                          sb.buf, sb.len, 0);
1862                         strbuf_reset(&sb);
1863                 }
1864         }
1865
1866 out:
1867         if (sb.len)
1868                 emit_diff_symbol(o, DIFF_SYMBOL_WORD_DIFF,
1869                                  sb.buf, sb.len, 0);
1870         strbuf_release(&sb);
1871         return 0;
1872 }
1873
1874 /*
1875  * '--color-words' algorithm can be described as:
1876  *
1877  *   1. collect the minus/plus lines of a diff hunk, divided into
1878  *      minus-lines and plus-lines;
1879  *
1880  *   2. break both minus-lines and plus-lines into words and
1881  *      place them into two mmfile_t with one word for each line;
1882  *
1883  *   3. use xdiff to run diff on the two mmfile_t to get the words level diff;
1884  *
1885  * And for the common parts of the both file, we output the plus side text.
1886  * diff_words->current_plus is used to trace the current position of the plus file
1887  * which printed. diff_words->last_minus is used to trace the last minus word
1888  * printed.
1889  *
1890  * For '--graph' to work with '--color-words', we need to output the graph prefix
1891  * on each line of color words output. Generally, there are two conditions on
1892  * which we should output the prefix.
1893  *
1894  *   1. diff_words->last_minus == 0 &&
1895  *      diff_words->current_plus == diff_words->plus.text.ptr
1896  *
1897  *      that is: the plus text must start as a new line, and if there is no minus
1898  *      word printed, a graph prefix must be printed.
1899  *
1900  *   2. diff_words->current_plus > diff_words->plus.text.ptr &&
1901  *      *(diff_words->current_plus - 1) == '\n'
1902  *
1903  *      that is: a graph prefix must be printed following a '\n'
1904  */
1905 static int color_words_output_graph_prefix(struct diff_words_data *diff_words)
1906 {
1907         if ((diff_words->last_minus == 0 &&
1908                 diff_words->current_plus == diff_words->plus.text.ptr) ||
1909                 (diff_words->current_plus > diff_words->plus.text.ptr &&
1910                 *(diff_words->current_plus - 1) == '\n')) {
1911                 return 1;
1912         } else {
1913                 return 0;
1914         }
1915 }
1916
1917 static void fn_out_diff_words_aux(void *priv,
1918                                   long minus_first, long minus_len,
1919                                   long plus_first, long plus_len,
1920                                   const char *func, long funclen)
1921 {
1922         struct diff_words_data *diff_words = priv;
1923         struct diff_words_style *style = diff_words->style;
1924         const char *minus_begin, *minus_end, *plus_begin, *plus_end;
1925         struct diff_options *opt = diff_words->opt;
1926         const char *line_prefix;
1927
1928         assert(opt);
1929         line_prefix = diff_line_prefix(opt);
1930
1931         /* POSIX requires that first be decremented by one if len == 0... */
1932         if (minus_len) {
1933                 minus_begin = diff_words->minus.orig[minus_first].begin;
1934                 minus_end =
1935                         diff_words->minus.orig[minus_first + minus_len - 1].end;
1936         } else
1937                 minus_begin = minus_end =
1938                         diff_words->minus.orig[minus_first].end;
1939
1940         if (plus_len) {
1941                 plus_begin = diff_words->plus.orig[plus_first].begin;
1942                 plus_end = diff_words->plus.orig[plus_first + plus_len - 1].end;
1943         } else
1944                 plus_begin = plus_end = diff_words->plus.orig[plus_first].end;
1945
1946         if (color_words_output_graph_prefix(diff_words)) {
1947                 fputs(line_prefix, diff_words->opt->file);
1948         }
1949         if (diff_words->current_plus != plus_begin) {
1950                 fn_out_diff_words_write_helper(diff_words->opt,
1951                                 &style->ctx, style->newline,
1952                                 plus_begin - diff_words->current_plus,
1953                                 diff_words->current_plus);
1954         }
1955         if (minus_begin != minus_end) {
1956                 fn_out_diff_words_write_helper(diff_words->opt,
1957                                 &style->old_word, style->newline,
1958                                 minus_end - minus_begin, minus_begin);
1959         }
1960         if (plus_begin != plus_end) {
1961                 fn_out_diff_words_write_helper(diff_words->opt,
1962                                 &style->new_word, style->newline,
1963                                 plus_end - plus_begin, plus_begin);
1964         }
1965
1966         diff_words->current_plus = plus_end;
1967         diff_words->last_minus = minus_first;
1968 }
1969
1970 /* This function starts looking at *begin, and returns 0 iff a word was found. */
1971 static int find_word_boundaries(mmfile_t *buffer, regex_t *word_regex,
1972                 int *begin, int *end)
1973 {
1974         if (word_regex && *begin < buffer->size) {
1975                 regmatch_t match[1];
1976                 if (!regexec_buf(word_regex, buffer->ptr + *begin,
1977                                  buffer->size - *begin, 1, match, 0)) {
1978                         char *p = memchr(buffer->ptr + *begin + match[0].rm_so,
1979                                         '\n', match[0].rm_eo - match[0].rm_so);
1980                         *end = p ? p - buffer->ptr : match[0].rm_eo + *begin;
1981                         *begin += match[0].rm_so;
1982                         return *begin >= *end;
1983                 }
1984                 return -1;
1985         }
1986
1987         /* find the next word */
1988         while (*begin < buffer->size && isspace(buffer->ptr[*begin]))
1989                 (*begin)++;
1990         if (*begin >= buffer->size)
1991                 return -1;
1992
1993         /* find the end of the word */
1994         *end = *begin + 1;
1995         while (*end < buffer->size && !isspace(buffer->ptr[*end]))
1996                 (*end)++;
1997
1998         return 0;
1999 }
2000
2001 /*
2002  * This function splits the words in buffer->text, stores the list with
2003  * newline separator into out, and saves the offsets of the original words
2004  * in buffer->orig.
2005  */
2006 static void diff_words_fill(struct diff_words_buffer *buffer, mmfile_t *out,
2007                 regex_t *word_regex)
2008 {
2009         int i, j;
2010         long alloc = 0;
2011
2012         out->size = 0;
2013         out->ptr = NULL;
2014
2015         /* fake an empty "0th" word */
2016         ALLOC_GROW(buffer->orig, 1, buffer->orig_alloc);
2017         buffer->orig[0].begin = buffer->orig[0].end = buffer->text.ptr;
2018         buffer->orig_nr = 1;
2019
2020         for (i = 0; i < buffer->text.size; i++) {
2021                 if (find_word_boundaries(&buffer->text, word_regex, &i, &j))
2022                         return;
2023
2024                 /* store original boundaries */
2025                 ALLOC_GROW(buffer->orig, buffer->orig_nr + 1,
2026                                 buffer->orig_alloc);
2027                 buffer->orig[buffer->orig_nr].begin = buffer->text.ptr + i;
2028                 buffer->orig[buffer->orig_nr].end = buffer->text.ptr + j;
2029                 buffer->orig_nr++;
2030
2031                 /* store one word */
2032                 ALLOC_GROW(out->ptr, out->size + j - i + 1, alloc);
2033                 memcpy(out->ptr + out->size, buffer->text.ptr + i, j - i);
2034                 out->ptr[out->size + j - i] = '\n';
2035                 out->size += j - i + 1;
2036
2037                 i = j - 1;
2038         }
2039 }
2040
2041 /* this executes the word diff on the accumulated buffers */
2042 static void diff_words_show(struct diff_words_data *diff_words)
2043 {
2044         xpparam_t xpp;
2045         xdemitconf_t xecfg;
2046         mmfile_t minus, plus;
2047         struct diff_words_style *style = diff_words->style;
2048
2049         struct diff_options *opt = diff_words->opt;
2050         const char *line_prefix;
2051
2052         assert(opt);
2053         line_prefix = diff_line_prefix(opt);
2054
2055         /* special case: only removal */
2056         if (!diff_words->plus.text.size) {
2057                 emit_diff_symbol(diff_words->opt, DIFF_SYMBOL_WORD_DIFF,
2058                                  line_prefix, strlen(line_prefix), 0);
2059                 fn_out_diff_words_write_helper(diff_words->opt,
2060                         &style->old_word, style->newline,
2061                         diff_words->minus.text.size,
2062                         diff_words->minus.text.ptr);
2063                 diff_words->minus.text.size = 0;
2064                 return;
2065         }
2066
2067         diff_words->current_plus = diff_words->plus.text.ptr;
2068         diff_words->last_minus = 0;
2069
2070         memset(&xpp, 0, sizeof(xpp));
2071         memset(&xecfg, 0, sizeof(xecfg));
2072         diff_words_fill(&diff_words->minus, &minus, diff_words->word_regex);
2073         diff_words_fill(&diff_words->plus, &plus, diff_words->word_regex);
2074         xpp.flags = 0;
2075         /* as only the hunk header will be parsed, we need a 0-context */
2076         xecfg.ctxlen = 0;
2077         if (xdi_diff_outf(&minus, &plus, fn_out_diff_words_aux, NULL,
2078                           diff_words, &xpp, &xecfg))
2079                 die("unable to generate word diff");
2080         free(minus.ptr);
2081         free(plus.ptr);
2082         if (diff_words->current_plus != diff_words->plus.text.ptr +
2083                         diff_words->plus.text.size) {
2084                 if (color_words_output_graph_prefix(diff_words))
2085                         emit_diff_symbol(diff_words->opt, DIFF_SYMBOL_WORD_DIFF,
2086                                          line_prefix, strlen(line_prefix), 0);
2087                 fn_out_diff_words_write_helper(diff_words->opt,
2088                         &style->ctx, style->newline,
2089                         diff_words->plus.text.ptr + diff_words->plus.text.size
2090                         - diff_words->current_plus, diff_words->current_plus);
2091         }
2092         diff_words->minus.text.size = diff_words->plus.text.size = 0;
2093 }
2094
2095 /* In "color-words" mode, show word-diff of words accumulated in the buffer */
2096 static void diff_words_flush(struct emit_callback *ecbdata)
2097 {
2098         struct diff_options *wo = ecbdata->diff_words->opt;
2099
2100         if (ecbdata->diff_words->minus.text.size ||
2101             ecbdata->diff_words->plus.text.size)
2102                 diff_words_show(ecbdata->diff_words);
2103
2104         if (wo->emitted_symbols) {
2105                 struct diff_options *o = ecbdata->opt;
2106                 struct emitted_diff_symbols *wol = wo->emitted_symbols;
2107                 int i;
2108
2109                 /*
2110                  * NEEDSWORK:
2111                  * Instead of appending each, concat all words to a line?
2112                  */
2113                 for (i = 0; i < wol->nr; i++)
2114                         append_emitted_diff_symbol(o, &wol->buf[i]);
2115
2116                 for (i = 0; i < wol->nr; i++)
2117                         free((void *)wol->buf[i].line);
2118
2119                 wol->nr = 0;
2120         }
2121 }
2122
2123 static void diff_filespec_load_driver(struct diff_filespec *one,
2124                                       struct index_state *istate)
2125 {
2126         /* Use already-loaded driver */
2127         if (one->driver)
2128                 return;
2129
2130         if (S_ISREG(one->mode))
2131                 one->driver = userdiff_find_by_path(istate, one->path);
2132
2133         /* Fallback to default settings */
2134         if (!one->driver)
2135                 one->driver = userdiff_find_by_name("default");
2136 }
2137
2138 static const char *userdiff_word_regex(struct diff_filespec *one,
2139                                        struct index_state *istate)
2140 {
2141         diff_filespec_load_driver(one, istate);
2142         return one->driver->word_regex;
2143 }
2144
2145 static void init_diff_words_data(struct emit_callback *ecbdata,
2146                                  struct diff_options *orig_opts,
2147                                  struct diff_filespec *one,
2148                                  struct diff_filespec *two)
2149 {
2150         int i;
2151         struct diff_options *o = xmalloc(sizeof(struct diff_options));
2152         memcpy(o, orig_opts, sizeof(struct diff_options));
2153
2154         ecbdata->diff_words =
2155                 xcalloc(1, sizeof(struct diff_words_data));
2156         ecbdata->diff_words->type = o->word_diff;
2157         ecbdata->diff_words->opt = o;
2158
2159         if (orig_opts->emitted_symbols)
2160                 o->emitted_symbols =
2161                         xcalloc(1, sizeof(struct emitted_diff_symbols));
2162
2163         if (!o->word_regex)
2164                 o->word_regex = userdiff_word_regex(one, o->repo->index);
2165         if (!o->word_regex)
2166                 o->word_regex = userdiff_word_regex(two, o->repo->index);
2167         if (!o->word_regex)
2168                 o->word_regex = diff_word_regex_cfg;
2169         if (o->word_regex) {
2170                 ecbdata->diff_words->word_regex = (regex_t *)
2171                         xmalloc(sizeof(regex_t));
2172                 if (regcomp(ecbdata->diff_words->word_regex,
2173                             o->word_regex,
2174                             REG_EXTENDED | REG_NEWLINE))
2175                         die("invalid regular expression: %s",
2176                             o->word_regex);
2177         }
2178         for (i = 0; i < ARRAY_SIZE(diff_words_styles); i++) {
2179                 if (o->word_diff == diff_words_styles[i].type) {
2180                         ecbdata->diff_words->style =
2181                                 &diff_words_styles[i];
2182                         break;
2183                 }
2184         }
2185         if (want_color(o->use_color)) {
2186                 struct diff_words_style *st = ecbdata->diff_words->style;
2187                 st->old_word.color = diff_get_color_opt(o, DIFF_FILE_OLD);
2188                 st->new_word.color = diff_get_color_opt(o, DIFF_FILE_NEW);
2189                 st->ctx.color = diff_get_color_opt(o, DIFF_CONTEXT);
2190         }
2191 }
2192
2193 static void free_diff_words_data(struct emit_callback *ecbdata)
2194 {
2195         if (ecbdata->diff_words) {
2196                 diff_words_flush(ecbdata);
2197                 free (ecbdata->diff_words->opt->emitted_symbols);
2198                 free (ecbdata->diff_words->opt);
2199                 free (ecbdata->diff_words->minus.text.ptr);
2200                 free (ecbdata->diff_words->minus.orig);
2201                 free (ecbdata->diff_words->plus.text.ptr);
2202                 free (ecbdata->diff_words->plus.orig);
2203                 if (ecbdata->diff_words->word_regex) {
2204                         regfree(ecbdata->diff_words->word_regex);
2205                         free(ecbdata->diff_words->word_regex);
2206                 }
2207                 FREE_AND_NULL(ecbdata->diff_words);
2208         }
2209 }
2210
2211 const char *diff_get_color(int diff_use_color, enum color_diff ix)
2212 {
2213         if (want_color(diff_use_color))
2214                 return diff_colors[ix];
2215         return "";
2216 }
2217
2218 const char *diff_line_prefix(struct diff_options *opt)
2219 {
2220         struct strbuf *msgbuf;
2221         if (!opt->output_prefix)
2222                 return "";
2223
2224         msgbuf = opt->output_prefix(opt, opt->output_prefix_data);
2225         return msgbuf->buf;
2226 }
2227
2228 static unsigned long sane_truncate_line(struct emit_callback *ecb, char *line, unsigned long len)
2229 {
2230         const char *cp;
2231         unsigned long allot;
2232         size_t l = len;
2233
2234         cp = line;
2235         allot = l;
2236         while (0 < l) {
2237                 (void) utf8_width(&cp, &l);
2238                 if (!cp)
2239                         break; /* truncated in the middle? */
2240         }
2241         return allot - l;
2242 }
2243
2244 static void find_lno(const char *line, struct emit_callback *ecbdata)
2245 {
2246         const char *p;
2247         ecbdata->lno_in_preimage = 0;
2248         ecbdata->lno_in_postimage = 0;
2249         p = strchr(line, '-');
2250         if (!p)
2251                 return; /* cannot happen */
2252         ecbdata->lno_in_preimage = strtol(p + 1, NULL, 10);
2253         p = strchr(p, '+');
2254         if (!p)
2255                 return; /* cannot happen */
2256         ecbdata->lno_in_postimage = strtol(p + 1, NULL, 10);
2257 }
2258
2259 static void fn_out_consume(void *priv, char *line, unsigned long len)
2260 {
2261         struct emit_callback *ecbdata = priv;
2262         const char *reset = diff_get_color(ecbdata->color_diff, DIFF_RESET);
2263         struct diff_options *o = ecbdata->opt;
2264
2265         o->found_changes = 1;
2266
2267         if (ecbdata->header) {
2268                 emit_diff_symbol(o, DIFF_SYMBOL_HEADER,
2269                                  ecbdata->header->buf, ecbdata->header->len, 0);
2270                 strbuf_reset(ecbdata->header);
2271                 ecbdata->header = NULL;
2272         }
2273
2274         if (ecbdata->label_path[0]) {
2275                 emit_diff_symbol(o, DIFF_SYMBOL_FILEPAIR_MINUS,
2276                                  ecbdata->label_path[0],
2277                                  strlen(ecbdata->label_path[0]), 0);
2278                 emit_diff_symbol(o, DIFF_SYMBOL_FILEPAIR_PLUS,
2279                                  ecbdata->label_path[1],
2280                                  strlen(ecbdata->label_path[1]), 0);
2281                 ecbdata->label_path[0] = ecbdata->label_path[1] = NULL;
2282         }
2283
2284         if (diff_suppress_blank_empty
2285             && len == 2 && line[0] == ' ' && line[1] == '\n') {
2286                 line[0] = '\n';
2287                 len = 1;
2288         }
2289
2290         if (line[0] == '@') {
2291                 if (ecbdata->diff_words)
2292                         diff_words_flush(ecbdata);
2293                 len = sane_truncate_line(ecbdata, line, len);
2294                 find_lno(line, ecbdata);
2295                 emit_hunk_header(ecbdata, line, len);
2296                 return;
2297         }
2298
2299         if (ecbdata->diff_words) {
2300                 enum diff_symbol s =
2301                         ecbdata->diff_words->type == DIFF_WORDS_PORCELAIN ?
2302                         DIFF_SYMBOL_WORDS_PORCELAIN : DIFF_SYMBOL_WORDS;
2303                 if (line[0] == '-') {
2304                         diff_words_append(line, len,
2305                                           &ecbdata->diff_words->minus);
2306                         return;
2307                 } else if (line[0] == '+') {
2308                         diff_words_append(line, len,
2309                                           &ecbdata->diff_words->plus);
2310                         return;
2311                 } else if (starts_with(line, "\\ ")) {
2312                         /*
2313                          * Eat the "no newline at eof" marker as if we
2314                          * saw a "+" or "-" line with nothing on it,
2315                          * and return without diff_words_flush() to
2316                          * defer processing. If this is the end of
2317                          * preimage, more "+" lines may come after it.
2318                          */
2319                         return;
2320                 }
2321                 diff_words_flush(ecbdata);
2322                 emit_diff_symbol(o, s, line, len, 0);
2323                 return;
2324         }
2325
2326         switch (line[0]) {
2327         case '+':
2328                 ecbdata->lno_in_postimage++;
2329                 emit_add_line(reset, ecbdata, line + 1, len - 1);
2330                 break;
2331         case '-':
2332                 ecbdata->lno_in_preimage++;
2333                 emit_del_line(reset, ecbdata, line + 1, len - 1);
2334                 break;
2335         case ' ':
2336                 ecbdata->lno_in_postimage++;
2337                 ecbdata->lno_in_preimage++;
2338                 emit_context_line(reset, ecbdata, line + 1, len - 1);
2339                 break;
2340         default:
2341                 /* incomplete line at the end */
2342                 ecbdata->lno_in_preimage++;
2343                 emit_diff_symbol(o, DIFF_SYMBOL_CONTEXT_INCOMPLETE,
2344                                  line, len, 0);
2345                 break;
2346         }
2347 }
2348
2349 static void pprint_rename(struct strbuf *name, const char *a, const char *b)
2350 {
2351         const char *old_name = a;
2352         const char *new_name = b;
2353         int pfx_length, sfx_length;
2354         int pfx_adjust_for_slash;
2355         int len_a = strlen(a);
2356         int len_b = strlen(b);
2357         int a_midlen, b_midlen;
2358         int qlen_a = quote_c_style(a, NULL, NULL, 0);
2359         int qlen_b = quote_c_style(b, NULL, NULL, 0);
2360
2361         if (qlen_a || qlen_b) {
2362                 quote_c_style(a, name, NULL, 0);
2363                 strbuf_addstr(name, " => ");
2364                 quote_c_style(b, name, NULL, 0);
2365                 return;
2366         }
2367
2368         /* Find common prefix */
2369         pfx_length = 0;
2370         while (*old_name && *new_name && *old_name == *new_name) {
2371                 if (*old_name == '/')
2372                         pfx_length = old_name - a + 1;
2373                 old_name++;
2374                 new_name++;
2375         }
2376
2377         /* Find common suffix */
2378         old_name = a + len_a;
2379         new_name = b + len_b;
2380         sfx_length = 0;
2381         /*
2382          * If there is a common prefix, it must end in a slash.  In
2383          * that case we let this loop run 1 into the prefix to see the
2384          * same slash.
2385          *
2386          * If there is no common prefix, we cannot do this as it would
2387          * underrun the input strings.
2388          */
2389         pfx_adjust_for_slash = (pfx_length ? 1 : 0);
2390         while (a + pfx_length - pfx_adjust_for_slash <= old_name &&
2391                b + pfx_length - pfx_adjust_for_slash <= new_name &&
2392                *old_name == *new_name) {
2393                 if (*old_name == '/')
2394                         sfx_length = len_a - (old_name - a);
2395                 old_name--;
2396                 new_name--;
2397         }
2398
2399         /*
2400          * pfx{mid-a => mid-b}sfx
2401          * {pfx-a => pfx-b}sfx
2402          * pfx{sfx-a => sfx-b}
2403          * name-a => name-b
2404          */
2405         a_midlen = len_a - pfx_length - sfx_length;
2406         b_midlen = len_b - pfx_length - sfx_length;
2407         if (a_midlen < 0)
2408                 a_midlen = 0;
2409         if (b_midlen < 0)
2410                 b_midlen = 0;
2411
2412         strbuf_grow(name, pfx_length + a_midlen + b_midlen + sfx_length + 7);
2413         if (pfx_length + sfx_length) {
2414                 strbuf_add(name, a, pfx_length);
2415                 strbuf_addch(name, '{');
2416         }
2417         strbuf_add(name, a + pfx_length, a_midlen);
2418         strbuf_addstr(name, " => ");
2419         strbuf_add(name, b + pfx_length, b_midlen);
2420         if (pfx_length + sfx_length) {
2421                 strbuf_addch(name, '}');
2422                 strbuf_add(name, a + len_a - sfx_length, sfx_length);
2423         }
2424 }
2425
2426 struct diffstat_t {
2427         int nr;
2428         int alloc;
2429         struct diffstat_file {
2430                 char *from_name;
2431                 char *name;
2432                 char *print_name;
2433                 const char *comments;
2434                 unsigned is_unmerged:1;
2435                 unsigned is_binary:1;
2436                 unsigned is_renamed:1;
2437                 unsigned is_interesting:1;
2438                 uintmax_t added, deleted;
2439         } **files;
2440 };
2441
2442 static struct diffstat_file *diffstat_add(struct diffstat_t *diffstat,
2443                                           const char *name_a,
2444                                           const char *name_b)
2445 {
2446         struct diffstat_file *x;
2447         x = xcalloc(1, sizeof(*x));
2448         ALLOC_GROW(diffstat->files, diffstat->nr + 1, diffstat->alloc);
2449         diffstat->files[diffstat->nr++] = x;
2450         if (name_b) {
2451                 x->from_name = xstrdup(name_a);
2452                 x->name = xstrdup(name_b);
2453                 x->is_renamed = 1;
2454         }
2455         else {
2456                 x->from_name = NULL;
2457                 x->name = xstrdup(name_a);
2458         }
2459         return x;
2460 }
2461
2462 static void diffstat_consume(void *priv, char *line, unsigned long len)
2463 {
2464         struct diffstat_t *diffstat = priv;
2465         struct diffstat_file *x = diffstat->files[diffstat->nr - 1];
2466
2467         if (line[0] == '+')
2468                 x->added++;
2469         else if (line[0] == '-')
2470                 x->deleted++;
2471 }
2472
2473 const char mime_boundary_leader[] = "------------";
2474
2475 static int scale_linear(int it, int width, int max_change)
2476 {
2477         if (!it)
2478                 return 0;
2479         /*
2480          * make sure that at least one '-' or '+' is printed if
2481          * there is any change to this path. The easiest way is to
2482          * scale linearly as if the alloted width is one column shorter
2483          * than it is, and then add 1 to the result.
2484          */
2485         return 1 + (it * (width - 1) / max_change);
2486 }
2487
2488 static void show_graph(struct strbuf *out, char ch, int cnt,
2489                        const char *set, const char *reset)
2490 {
2491         if (cnt <= 0)
2492                 return;
2493         strbuf_addstr(out, set);
2494         strbuf_addchars(out, ch, cnt);
2495         strbuf_addstr(out, reset);
2496 }
2497
2498 static void fill_print_name(struct diffstat_file *file)
2499 {
2500         struct strbuf pname = STRBUF_INIT;
2501
2502         if (file->print_name)
2503                 return;
2504
2505         if (file->is_renamed)
2506                 pprint_rename(&pname, file->from_name, file->name);
2507         else
2508                 quote_c_style(file->name, &pname, NULL, 0);
2509
2510         if (file->comments)
2511                 strbuf_addf(&pname, " (%s)", file->comments);
2512
2513         file->print_name = strbuf_detach(&pname, NULL);
2514 }
2515
2516 static void print_stat_summary_inserts_deletes(struct diff_options *options,
2517                 int files, int insertions, int deletions)
2518 {
2519         struct strbuf sb = STRBUF_INIT;
2520
2521         if (!files) {
2522                 assert(insertions == 0 && deletions == 0);
2523                 emit_diff_symbol(options, DIFF_SYMBOL_STATS_SUMMARY_NO_FILES,
2524                                  NULL, 0, 0);
2525                 return;
2526         }
2527
2528         strbuf_addf(&sb,
2529                     (files == 1) ? " %d file changed" : " %d files changed",
2530                     files);
2531
2532         /*
2533          * For binary diff, the caller may want to print "x files
2534          * changed" with insertions == 0 && deletions == 0.
2535          *
2536          * Not omitting "0 insertions(+), 0 deletions(-)" in this case
2537          * is probably less confusing (i.e skip over "2 files changed
2538          * but nothing about added/removed lines? Is this a bug in Git?").
2539          */
2540         if (insertions || deletions == 0) {
2541                 strbuf_addf(&sb,
2542                             (insertions == 1) ? ", %d insertion(+)" : ", %d insertions(+)",
2543                             insertions);
2544         }
2545
2546         if (deletions || insertions == 0) {
2547                 strbuf_addf(&sb,
2548                             (deletions == 1) ? ", %d deletion(-)" : ", %d deletions(-)",
2549                             deletions);
2550         }
2551         strbuf_addch(&sb, '\n');
2552         emit_diff_symbol(options, DIFF_SYMBOL_STATS_SUMMARY_INSERTS_DELETES,
2553                          sb.buf, sb.len, 0);
2554         strbuf_release(&sb);
2555 }
2556
2557 void print_stat_summary(FILE *fp, int files,
2558                         int insertions, int deletions)
2559 {
2560         struct diff_options o;
2561         memset(&o, 0, sizeof(o));
2562         o.file = fp;
2563
2564         print_stat_summary_inserts_deletes(&o, files, insertions, deletions);
2565 }
2566
2567 static void show_stats(struct diffstat_t *data, struct diff_options *options)
2568 {
2569         int i, len, add, del, adds = 0, dels = 0;
2570         uintmax_t max_change = 0, max_len = 0;
2571         int total_files = data->nr, count;
2572         int width, name_width, graph_width, number_width = 0, bin_width = 0;
2573         const char *reset, *add_c, *del_c;
2574         int extra_shown = 0;
2575         const char *line_prefix = diff_line_prefix(options);
2576         struct strbuf out = STRBUF_INIT;
2577
2578         if (data->nr == 0)
2579                 return;
2580
2581         count = options->stat_count ? options->stat_count : data->nr;
2582
2583         reset = diff_get_color_opt(options, DIFF_RESET);
2584         add_c = diff_get_color_opt(options, DIFF_FILE_NEW);
2585         del_c = diff_get_color_opt(options, DIFF_FILE_OLD);
2586
2587         /*
2588          * Find the longest filename and max number of changes
2589          */
2590         for (i = 0; (i < count) && (i < data->nr); i++) {
2591                 struct diffstat_file *file = data->files[i];
2592                 uintmax_t change = file->added + file->deleted;
2593
2594                 if (!file->is_interesting && (change == 0)) {
2595                         count++; /* not shown == room for one more */
2596                         continue;
2597                 }
2598                 fill_print_name(file);
2599                 len = strlen(file->print_name);
2600                 if (max_len < len)
2601                         max_len = len;
2602
2603                 if (file->is_unmerged) {
2604                         /* "Unmerged" is 8 characters */
2605                         bin_width = bin_width < 8 ? 8 : bin_width;
2606                         continue;
2607                 }
2608                 if (file->is_binary) {
2609                         /* "Bin XXX -> YYY bytes" */
2610                         int w = 14 + decimal_width(file->added)
2611                                 + decimal_width(file->deleted);
2612                         bin_width = bin_width < w ? w : bin_width;
2613                         /* Display change counts aligned with "Bin" */
2614                         number_width = 3;
2615                         continue;
2616                 }
2617
2618                 if (max_change < change)
2619                         max_change = change;
2620         }
2621         count = i; /* where we can stop scanning in data->files[] */
2622
2623         /*
2624          * We have width = stat_width or term_columns() columns total.
2625          * We want a maximum of min(max_len, stat_name_width) for the name part.
2626          * We want a maximum of min(max_change, stat_graph_width) for the +- part.
2627          * We also need 1 for " " and 4 + decimal_width(max_change)
2628          * for " | NNNN " and one the empty column at the end, altogether
2629          * 6 + decimal_width(max_change).
2630          *
2631          * If there's not enough space, we will use the smaller of
2632          * stat_name_width (if set) and 5/8*width for the filename,
2633          * and the rest for constant elements + graph part, but no more
2634          * than stat_graph_width for the graph part.
2635          * (5/8 gives 50 for filename and 30 for the constant parts + graph
2636          * for the standard terminal size).
2637          *
2638          * In other words: stat_width limits the maximum width, and
2639          * stat_name_width fixes the maximum width of the filename,
2640          * and is also used to divide available columns if there
2641          * aren't enough.
2642          *
2643          * Binary files are displayed with "Bin XXX -> YYY bytes"
2644          * instead of the change count and graph. This part is treated
2645          * similarly to the graph part, except that it is not
2646          * "scaled". If total width is too small to accommodate the
2647          * guaranteed minimum width of the filename part and the
2648          * separators and this message, this message will "overflow"
2649          * making the line longer than the maximum width.
2650          */
2651
2652         if (options->stat_width == -1)
2653                 width = term_columns() - strlen(line_prefix);
2654         else
2655                 width = options->stat_width ? options->stat_width : 80;
2656         number_width = decimal_width(max_change) > number_width ?
2657                 decimal_width(max_change) : number_width;
2658
2659         if (options->stat_graph_width == -1)
2660                 options->stat_graph_width = diff_stat_graph_width;
2661
2662         /*
2663          * Guarantee 3/8*16==6 for the graph part
2664          * and 5/8*16==10 for the filename part
2665          */
2666         if (width < 16 + 6 + number_width)
2667                 width = 16 + 6 + number_width;
2668
2669         /*
2670          * First assign sizes that are wanted, ignoring available width.
2671          * strlen("Bin XXX -> YYY bytes") == bin_width, and the part
2672          * starting from "XXX" should fit in graph_width.
2673          */
2674         graph_width = max_change + 4 > bin_width ? max_change : bin_width - 4;
2675         if (options->stat_graph_width &&
2676             options->stat_graph_width < graph_width)
2677                 graph_width = options->stat_graph_width;
2678
2679         name_width = (options->stat_name_width > 0 &&
2680                       options->stat_name_width < max_len) ?
2681                 options->stat_name_width : max_len;
2682
2683         /*
2684          * Adjust adjustable widths not to exceed maximum width
2685          */
2686         if (name_width + number_width + 6 + graph_width > width) {
2687                 if (graph_width > width * 3/8 - number_width - 6) {
2688                         graph_width = width * 3/8 - number_width - 6;
2689                         if (graph_width < 6)
2690                                 graph_width = 6;
2691                 }
2692
2693                 if (options->stat_graph_width &&
2694                     graph_width > options->stat_graph_width)
2695                         graph_width = options->stat_graph_width;
2696                 if (name_width > width - number_width - 6 - graph_width)
2697                         name_width = width - number_width - 6 - graph_width;
2698                 else
2699                         graph_width = width - number_width - 6 - name_width;
2700         }
2701
2702         /*
2703          * From here name_width is the width of the name area,
2704          * and graph_width is the width of the graph area.
2705          * max_change is used to scale graph properly.
2706          */
2707         for (i = 0; i < count; i++) {
2708                 const char *prefix = "";
2709                 struct diffstat_file *file = data->files[i];
2710                 char *name = file->print_name;
2711                 uintmax_t added = file->added;
2712                 uintmax_t deleted = file->deleted;
2713                 int name_len;
2714
2715                 if (!file->is_interesting && (added + deleted == 0))
2716                         continue;
2717
2718                 /*
2719                  * "scale" the filename
2720                  */
2721                 len = name_width;
2722                 name_len = strlen(name);
2723                 if (name_width < name_len) {
2724                         char *slash;
2725                         prefix = "...";
2726                         len -= 3;
2727                         name += name_len - len;
2728                         slash = strchr(name, '/');
2729                         if (slash)
2730                                 name = slash;
2731                 }
2732
2733                 if (file->is_binary) {
2734                         strbuf_addf(&out, " %s%-*s |", prefix, len, name);
2735                         strbuf_addf(&out, " %*s", number_width, "Bin");
2736                         if (!added && !deleted) {
2737                                 strbuf_addch(&out, '\n');
2738                                 emit_diff_symbol(options, DIFF_SYMBOL_STATS_LINE,
2739                                                  out.buf, out.len, 0);
2740                                 strbuf_reset(&out);
2741                                 continue;
2742                         }
2743                         strbuf_addf(&out, " %s%"PRIuMAX"%s",
2744                                 del_c, deleted, reset);
2745                         strbuf_addstr(&out, " -> ");
2746                         strbuf_addf(&out, "%s%"PRIuMAX"%s",
2747                                 add_c, added, reset);
2748                         strbuf_addstr(&out, " bytes\n");
2749                         emit_diff_symbol(options, DIFF_SYMBOL_STATS_LINE,
2750                                          out.buf, out.len, 0);
2751                         strbuf_reset(&out);
2752                         continue;
2753                 }
2754                 else if (file->is_unmerged) {
2755                         strbuf_addf(&out, " %s%-*s |", prefix, len, name);
2756                         strbuf_addstr(&out, " Unmerged\n");
2757                         emit_diff_symbol(options, DIFF_SYMBOL_STATS_LINE,
2758                                          out.buf, out.len, 0);
2759                         strbuf_reset(&out);
2760                         continue;
2761                 }
2762
2763                 /*
2764                  * scale the add/delete
2765                  */
2766                 add = added;
2767                 del = deleted;
2768
2769                 if (graph_width <= max_change) {
2770                         int total = scale_linear(add + del, graph_width, max_change);
2771                         if (total < 2 && add && del)
2772                                 /* width >= 2 due to the sanity check */
2773                                 total = 2;
2774                         if (add < del) {
2775                                 add = scale_linear(add, graph_width, max_change);
2776                                 del = total - add;
2777                         } else {
2778                                 del = scale_linear(del, graph_width, max_change);
2779                                 add = total - del;
2780                         }
2781                 }
2782                 strbuf_addf(&out, " %s%-*s |", prefix, len, name);
2783                 strbuf_addf(&out, " %*"PRIuMAX"%s",
2784                         number_width, added + deleted,
2785                         added + deleted ? " " : "");
2786                 show_graph(&out, '+', add, add_c, reset);
2787                 show_graph(&out, '-', del, del_c, reset);
2788                 strbuf_addch(&out, '\n');
2789                 emit_diff_symbol(options, DIFF_SYMBOL_STATS_LINE,
2790                                  out.buf, out.len, 0);
2791                 strbuf_reset(&out);
2792         }
2793
2794         for (i = 0; i < data->nr; i++) {
2795                 struct diffstat_file *file = data->files[i];
2796                 uintmax_t added = file->added;
2797                 uintmax_t deleted = file->deleted;
2798
2799                 if (file->is_unmerged ||
2800                     (!file->is_interesting && (added + deleted == 0))) {
2801                         total_files--;
2802                         continue;
2803                 }
2804
2805                 if (!file->is_binary) {
2806                         adds += added;
2807                         dels += deleted;
2808                 }
2809                 if (i < count)
2810                         continue;
2811                 if (!extra_shown)
2812                         emit_diff_symbol(options,
2813                                          DIFF_SYMBOL_STATS_SUMMARY_ABBREV,
2814                                          NULL, 0, 0);
2815                 extra_shown = 1;
2816         }
2817
2818         print_stat_summary_inserts_deletes(options, total_files, adds, dels);
2819         strbuf_release(&out);
2820 }
2821
2822 static void show_shortstats(struct diffstat_t *data, struct diff_options *options)
2823 {
2824         int i, adds = 0, dels = 0, total_files = data->nr;
2825
2826         if (data->nr == 0)
2827                 return;
2828
2829         for (i = 0; i < data->nr; i++) {
2830                 int added = data->files[i]->added;
2831                 int deleted = data->files[i]->deleted;
2832
2833                 if (data->files[i]->is_unmerged ||
2834                     (!data->files[i]->is_interesting && (added + deleted == 0))) {
2835                         total_files--;
2836                 } else if (!data->files[i]->is_binary) { /* don't count bytes */
2837                         adds += added;
2838                         dels += deleted;
2839                 }
2840         }
2841         print_stat_summary_inserts_deletes(options, total_files, adds, dels);
2842 }
2843
2844 static void show_numstat(struct diffstat_t *data, struct diff_options *options)
2845 {
2846         int i;
2847
2848         if (data->nr == 0)
2849                 return;
2850
2851         for (i = 0; i < data->nr; i++) {
2852                 struct diffstat_file *file = data->files[i];
2853
2854                 fprintf(options->file, "%s", diff_line_prefix(options));
2855
2856                 if (file->is_binary)
2857                         fprintf(options->file, "-\t-\t");
2858                 else
2859                         fprintf(options->file,
2860                                 "%"PRIuMAX"\t%"PRIuMAX"\t",
2861                                 file->added, file->deleted);
2862                 if (options->line_termination) {
2863                         fill_print_name(file);
2864                         if (!file->is_renamed)
2865                                 write_name_quoted(file->name, options->file,
2866                                                   options->line_termination);
2867                         else {
2868                                 fputs(file->print_name, options->file);
2869                                 putc(options->line_termination, options->file);
2870                         }
2871                 } else {
2872                         if (file->is_renamed) {
2873                                 putc('\0', options->file);
2874                                 write_name_quoted(file->from_name, options->file, '\0');
2875                         }
2876                         write_name_quoted(file->name, options->file, '\0');
2877                 }
2878         }
2879 }
2880
2881 struct dirstat_file {
2882         const char *name;
2883         unsigned long changed;
2884 };
2885
2886 struct dirstat_dir {
2887         struct dirstat_file *files;
2888         int alloc, nr, permille, cumulative;
2889 };
2890
2891 static long gather_dirstat(struct diff_options *opt, struct dirstat_dir *dir,
2892                 unsigned long changed, const char *base, int baselen)
2893 {
2894         unsigned long sum_changes = 0;
2895         unsigned int sources = 0;
2896         const char *line_prefix = diff_line_prefix(opt);
2897
2898         while (dir->nr) {
2899                 struct dirstat_file *f = dir->files;
2900                 int namelen = strlen(f->name);
2901                 unsigned long changes;
2902                 char *slash;
2903
2904                 if (namelen < baselen)
2905                         break;
2906                 if (memcmp(f->name, base, baselen))
2907                         break;
2908                 slash = strchr(f->name + baselen, '/');
2909                 if (slash) {
2910                         int newbaselen = slash + 1 - f->name;
2911                         changes = gather_dirstat(opt, dir, changed, f->name, newbaselen);
2912                         sources++;
2913                 } else {
2914                         changes = f->changed;
2915                         dir->files++;
2916                         dir->nr--;
2917                         sources += 2;
2918                 }
2919                 sum_changes += changes;
2920         }
2921
2922         /*
2923          * We don't report dirstat's for
2924          *  - the top level
2925          *  - or cases where everything came from a single directory
2926          *    under this directory (sources == 1).
2927          */
2928         if (baselen && sources != 1) {
2929                 if (sum_changes) {
2930                         int permille = sum_changes * 1000 / changed;
2931                         if (permille >= dir->permille) {
2932                                 fprintf(opt->file, "%s%4d.%01d%% %.*s\n", line_prefix,
2933                                         permille / 10, permille % 10, baselen, base);
2934                                 if (!dir->cumulative)
2935                                         return 0;
2936                         }
2937                 }
2938         }
2939         return sum_changes;
2940 }
2941
2942 static int dirstat_compare(const void *_a, const void *_b)
2943 {
2944         const struct dirstat_file *a = _a;
2945         const struct dirstat_file *b = _b;
2946         return strcmp(a->name, b->name);
2947 }
2948
2949 static void show_dirstat(struct diff_options *options)
2950 {
2951         int i;
2952         unsigned long changed;
2953         struct dirstat_dir dir;
2954         struct diff_queue_struct *q = &diff_queued_diff;
2955
2956         dir.files = NULL;
2957         dir.alloc = 0;
2958         dir.nr = 0;
2959         dir.permille = options->dirstat_permille;
2960         dir.cumulative = options->flags.dirstat_cumulative;
2961
2962         changed = 0;
2963         for (i = 0; i < q->nr; i++) {
2964                 struct diff_filepair *p = q->queue[i];
2965                 const char *name;
2966                 unsigned long copied, added, damage;
2967
2968                 name = p->two->path ? p->two->path : p->one->path;
2969
2970                 if (p->one->oid_valid && p->two->oid_valid &&
2971                     oideq(&p->one->oid, &p->two->oid)) {
2972                         /*
2973                          * The SHA1 has not changed, so pre-/post-content is
2974                          * identical. We can therefore skip looking at the
2975                          * file contents altogether.
2976                          */
2977                         damage = 0;
2978                         goto found_damage;
2979                 }
2980
2981                 if (options->flags.dirstat_by_file) {
2982                         /*
2983                          * In --dirstat-by-file mode, we don't really need to
2984                          * look at the actual file contents at all.
2985                          * The fact that the SHA1 changed is enough for us to
2986                          * add this file to the list of results
2987                          * (with each file contributing equal damage).
2988                          */
2989                         damage = 1;
2990                         goto found_damage;
2991                 }
2992
2993                 if (DIFF_FILE_VALID(p->one) && DIFF_FILE_VALID(p->two)) {
2994                         diff_populate_filespec(options->repo, p->one, 0);
2995                         diff_populate_filespec(options->repo, p->two, 0);
2996                         diffcore_count_changes(options->repo,
2997                                                p->one, p->two, NULL, NULL,
2998                                                &copied, &added);
2999                         diff_free_filespec_data(p->one);
3000                         diff_free_filespec_data(p->two);
3001                 } else if (DIFF_FILE_VALID(p->one)) {
3002                         diff_populate_filespec(options->repo, p->one, CHECK_SIZE_ONLY);
3003                         copied = added = 0;
3004                         diff_free_filespec_data(p->one);
3005                 } else if (DIFF_FILE_VALID(p->two)) {
3006                         diff_populate_filespec(options->repo, p->two, CHECK_SIZE_ONLY);
3007                         copied = 0;
3008                         added = p->two->size;
3009                         diff_free_filespec_data(p->two);
3010                 } else
3011                         continue;
3012
3013                 /*
3014                  * Original minus copied is the removed material,
3015                  * added is the new material.  They are both damages
3016                  * made to the preimage.
3017                  * If the resulting damage is zero, we know that
3018                  * diffcore_count_changes() considers the two entries to
3019                  * be identical, but since the oid changed, we
3020                  * know that there must have been _some_ kind of change,
3021                  * so we force all entries to have damage > 0.
3022                  */
3023                 damage = (p->one->size - copied) + added;
3024                 if (!damage)
3025                         damage = 1;
3026
3027 found_damage:
3028                 ALLOC_GROW(dir.files, dir.nr + 1, dir.alloc);
3029                 dir.files[dir.nr].name = name;
3030                 dir.files[dir.nr].changed = damage;
3031                 changed += damage;
3032                 dir.nr++;
3033         }
3034
3035         /* This can happen even with many files, if everything was renames */
3036         if (!changed)
3037                 return;
3038
3039         /* Show all directories with more than x% of the changes */
3040         QSORT(dir.files, dir.nr, dirstat_compare);
3041         gather_dirstat(options, &dir, changed, "", 0);
3042 }
3043
3044 static void show_dirstat_by_line(struct diffstat_t *data, struct diff_options *options)
3045 {
3046         int i;
3047         unsigned long changed;
3048         struct dirstat_dir dir;
3049
3050         if (data->nr == 0)
3051                 return;
3052
3053         dir.files = NULL;
3054         dir.alloc = 0;
3055         dir.nr = 0;
3056         dir.permille = options->dirstat_permille;
3057         dir.cumulative = options->flags.dirstat_cumulative;
3058
3059         changed = 0;
3060         for (i = 0; i < data->nr; i++) {
3061                 struct diffstat_file *file = data->files[i];
3062                 unsigned long damage = file->added + file->deleted;
3063                 if (file->is_binary)
3064                         /*
3065                          * binary files counts bytes, not lines. Must find some
3066                          * way to normalize binary bytes vs. textual lines.
3067                          * The following heuristic assumes that there are 64
3068                          * bytes per "line".
3069                          * This is stupid and ugly, but very cheap...
3070                          */
3071                         damage = DIV_ROUND_UP(damage, 64);
3072                 ALLOC_GROW(dir.files, dir.nr + 1, dir.alloc);
3073                 dir.files[dir.nr].name = file->name;
3074                 dir.files[dir.nr].changed = damage;
3075                 changed += damage;
3076                 dir.nr++;
3077         }
3078
3079         /* This can happen even with many files, if everything was renames */
3080         if (!changed)
3081                 return;
3082
3083         /* Show all directories with more than x% of the changes */
3084         QSORT(dir.files, dir.nr, dirstat_compare);
3085         gather_dirstat(options, &dir, changed, "", 0);
3086 }
3087
3088 static void free_diffstat_info(struct diffstat_t *diffstat)
3089 {
3090         int i;
3091         for (i = 0; i < diffstat->nr; i++) {
3092                 struct diffstat_file *f = diffstat->files[i];
3093                 free(f->print_name);
3094                 free(f->name);
3095                 free(f->from_name);
3096                 free(f);
3097         }
3098         free(diffstat->files);
3099 }
3100
3101 struct checkdiff_t {
3102         const char *filename;
3103         int lineno;
3104         int conflict_marker_size;
3105         struct diff_options *o;
3106         unsigned ws_rule;
3107         unsigned status;
3108 };
3109
3110 static int is_conflict_marker(const char *line, int marker_size, unsigned long len)
3111 {
3112         char firstchar;
3113         int cnt;
3114
3115         if (len < marker_size + 1)
3116                 return 0;
3117         firstchar = line[0];
3118         switch (firstchar) {
3119         case '=': case '>': case '<': case '|':
3120                 break;
3121         default:
3122                 return 0;
3123         }
3124         for (cnt = 1; cnt < marker_size; cnt++)
3125                 if (line[cnt] != firstchar)
3126                         return 0;
3127         /* line[1] thru line[marker_size-1] are same as firstchar */
3128         if (len < marker_size + 1 || !isspace(line[marker_size]))
3129                 return 0;
3130         return 1;
3131 }
3132
3133 static void checkdiff_consume_hunk(void *priv,
3134                                    long ob, long on, long nb, long nn,
3135                                    const char *func, long funclen)
3136
3137 {
3138         struct checkdiff_t *data = priv;
3139         data->lineno = nb - 1;
3140 }
3141
3142 static void checkdiff_consume(void *priv, char *line, unsigned long len)
3143 {
3144         struct checkdiff_t *data = priv;
3145         int marker_size = data->conflict_marker_size;
3146         const char *ws = diff_get_color(data->o->use_color, DIFF_WHITESPACE);
3147         const char *reset = diff_get_color(data->o->use_color, DIFF_RESET);
3148         const char *set = diff_get_color(data->o->use_color, DIFF_FILE_NEW);
3149         char *err;
3150         const char *line_prefix;
3151
3152         assert(data->o);
3153         line_prefix = diff_line_prefix(data->o);
3154
3155         if (line[0] == '+') {
3156                 unsigned bad;
3157                 data->lineno++;
3158                 if (is_conflict_marker(line + 1, marker_size, len - 1)) {
3159                         data->status |= 1;
3160                         fprintf(data->o->file,
3161                                 "%s%s:%d: leftover conflict marker\n",
3162                                 line_prefix, data->filename, data->lineno);
3163                 }
3164                 bad = ws_check(line + 1, len - 1, data->ws_rule);
3165                 if (!bad)
3166                         return;
3167                 data->status |= bad;
3168                 err = whitespace_error_string(bad);
3169                 fprintf(data->o->file, "%s%s:%d: %s.\n",
3170                         line_prefix, data->filename, data->lineno, err);
3171                 free(err);
3172                 emit_line(data->o, set, reset, line, 1);
3173                 ws_check_emit(line + 1, len - 1, data->ws_rule,
3174                               data->o->file, set, reset, ws);
3175         } else if (line[0] == ' ') {
3176                 data->lineno++;
3177         }
3178 }
3179
3180 static unsigned char *deflate_it(char *data,
3181                                  unsigned long size,
3182                                  unsigned long *result_size)
3183 {
3184         int bound;
3185         unsigned char *deflated;
3186         git_zstream stream;
3187
3188         git_deflate_init(&stream, zlib_compression_level);
3189         bound = git_deflate_bound(&stream, size);
3190         deflated = xmalloc(bound);
3191         stream.next_out = deflated;
3192         stream.avail_out = bound;
3193
3194         stream.next_in = (unsigned char *)data;
3195         stream.avail_in = size;
3196         while (git_deflate(&stream, Z_FINISH) == Z_OK)
3197                 ; /* nothing */
3198         git_deflate_end(&stream);
3199         *result_size = stream.total_out;
3200         return deflated;
3201 }
3202
3203 static void emit_binary_diff_body(struct diff_options *o,
3204                                   mmfile_t *one, mmfile_t *two)
3205 {
3206         void *cp;
3207         void *delta;
3208         void *deflated;
3209         void *data;
3210         unsigned long orig_size;
3211         unsigned long delta_size;
3212         unsigned long deflate_size;
3213         unsigned long data_size;
3214
3215         /* We could do deflated delta, or we could do just deflated two,
3216          * whichever is smaller.
3217          */
3218         delta = NULL;
3219         deflated = deflate_it(two->ptr, two->size, &deflate_size);
3220         if (one->size && two->size) {
3221                 delta = diff_delta(one->ptr, one->size,
3222                                    two->ptr, two->size,
3223                                    &delta_size, deflate_size);
3224                 if (delta) {
3225                         void *to_free = delta;
3226                         orig_size = delta_size;
3227                         delta = deflate_it(delta, delta_size, &delta_size);
3228                         free(to_free);
3229                 }
3230         }
3231
3232         if (delta && delta_size < deflate_size) {
3233                 char *s = xstrfmt("%"PRIuMAX , (uintmax_t)orig_size);
3234                 emit_diff_symbol(o, DIFF_SYMBOL_BINARY_DIFF_HEADER_DELTA,
3235                                  s, strlen(s), 0);
3236                 free(s);
3237                 free(deflated);
3238                 data = delta;
3239                 data_size = delta_size;
3240         } else {
3241                 char *s = xstrfmt("%lu", two->size);
3242                 emit_diff_symbol(o, DIFF_SYMBOL_BINARY_DIFF_HEADER_LITERAL,
3243                                  s, strlen(s), 0);
3244                 free(s);
3245                 free(delta);
3246                 data = deflated;
3247                 data_size = deflate_size;
3248         }
3249
3250         /* emit data encoded in base85 */
3251         cp = data;
3252         while (data_size) {
3253                 int len;
3254                 int bytes = (52 < data_size) ? 52 : data_size;
3255                 char line[71];
3256                 data_size -= bytes;
3257                 if (bytes <= 26)
3258                         line[0] = bytes + 'A' - 1;
3259                 else
3260                         line[0] = bytes - 26 + 'a' - 1;
3261                 encode_85(line + 1, cp, bytes);
3262                 cp = (char *) cp + bytes;
3263
3264                 len = strlen(line);
3265                 line[len++] = '\n';
3266                 line[len] = '\0';
3267
3268                 emit_diff_symbol(o, DIFF_SYMBOL_BINARY_DIFF_BODY,
3269                                  line, len, 0);
3270         }
3271         emit_diff_symbol(o, DIFF_SYMBOL_BINARY_DIFF_FOOTER, NULL, 0, 0);
3272         free(data);
3273 }
3274
3275 static void emit_binary_diff(struct diff_options *o,
3276                              mmfile_t *one, mmfile_t *two)
3277 {
3278         emit_diff_symbol(o, DIFF_SYMBOL_BINARY_DIFF_HEADER, NULL, 0, 0);
3279         emit_binary_diff_body(o, one, two);
3280         emit_binary_diff_body(o, two, one);
3281 }
3282
3283 int diff_filespec_is_binary(struct repository *r,
3284                             struct diff_filespec *one)
3285 {
3286         if (one->is_binary == -1) {
3287                 diff_filespec_load_driver(one, r->index);
3288                 if (one->driver->binary != -1)
3289                         one->is_binary = one->driver->binary;
3290                 else {
3291                         if (!one->data && DIFF_FILE_VALID(one))
3292                                 diff_populate_filespec(r, one, CHECK_BINARY);
3293                         if (one->is_binary == -1 && one->data)
3294                                 one->is_binary = buffer_is_binary(one->data,
3295                                                 one->size);
3296                         if (one->is_binary == -1)
3297                                 one->is_binary = 0;
3298                 }
3299         }
3300         return one->is_binary;
3301 }
3302
3303 static const struct userdiff_funcname *
3304 diff_funcname_pattern(struct diff_options *o, struct diff_filespec *one)
3305 {
3306         diff_filespec_load_driver(one, o->repo->index);
3307         return one->driver->funcname.pattern ? &one->driver->funcname : NULL;
3308 }
3309
3310 void diff_set_mnemonic_prefix(struct diff_options *options, const char *a, const char *b)
3311 {
3312         if (!options->a_prefix)
3313                 options->a_prefix = a;
3314         if (!options->b_prefix)
3315                 options->b_prefix = b;
3316 }
3317
3318 struct userdiff_driver *get_textconv(struct index_state *istate,
3319                                      struct diff_filespec *one)
3320 {
3321         if (!DIFF_FILE_VALID(one))
3322                 return NULL;
3323
3324         diff_filespec_load_driver(one, istate);
3325         return userdiff_get_textconv(one->driver);
3326 }
3327
3328 static void builtin_diff(const char *name_a,
3329                          const char *name_b,
3330                          struct diff_filespec *one,
3331                          struct diff_filespec *two,
3332                          const char *xfrm_msg,
3333                          int must_show_header,
3334                          struct diff_options *o,
3335                          int complete_rewrite)
3336 {
3337         mmfile_t mf1, mf2;
3338         const char *lbl[2];
3339         char *a_one, *b_two;
3340         const char *meta = diff_get_color_opt(o, DIFF_METAINFO);
3341         const char *reset = diff_get_color_opt(o, DIFF_RESET);
3342         const char *a_prefix, *b_prefix;
3343         struct userdiff_driver *textconv_one = NULL;
3344         struct userdiff_driver *textconv_two = NULL;
3345         struct strbuf header = STRBUF_INIT;
3346         const char *line_prefix = diff_line_prefix(o);
3347
3348         diff_set_mnemonic_prefix(o, "a/", "b/");
3349         if (o->flags.reverse_diff) {
3350                 a_prefix = o->b_prefix;
3351                 b_prefix = o->a_prefix;
3352         } else {
3353                 a_prefix = o->a_prefix;
3354                 b_prefix = o->b_prefix;
3355         }
3356
3357         if (o->submodule_format == DIFF_SUBMODULE_LOG &&
3358             (!one->mode || S_ISGITLINK(one->mode)) &&
3359             (!two->mode || S_ISGITLINK(two->mode))) {
3360                 show_submodule_summary(o, one->path ? one->path : two->path,
3361                                 &one->oid, &two->oid,
3362                                 two->dirty_submodule);
3363                 return;
3364         } else if (o->submodule_format == DIFF_SUBMODULE_INLINE_DIFF &&
3365                    (!one->mode || S_ISGITLINK(one->mode)) &&
3366                    (!two->mode || S_ISGITLINK(two->mode))) {
3367                 show_submodule_inline_diff(o, one->path ? one->path : two->path,
3368                                 &one->oid, &two->oid,
3369                                 two->dirty_submodule);
3370                 return;
3371         }
3372
3373         if (o->flags.allow_textconv) {
3374                 textconv_one = get_textconv(o->repo->index, one);
3375                 textconv_two = get_textconv(o->repo->index, two);
3376         }
3377
3378         /* Never use a non-valid filename anywhere if at all possible */
3379         name_a = DIFF_FILE_VALID(one) ? name_a : name_b;
3380         name_b = DIFF_FILE_VALID(two) ? name_b : name_a;
3381
3382         a_one = quote_two(a_prefix, name_a + (*name_a == '/'));
3383         b_two = quote_two(b_prefix, name_b + (*name_b == '/'));
3384         lbl[0] = DIFF_FILE_VALID(one) ? a_one : "/dev/null";
3385         lbl[1] = DIFF_FILE_VALID(two) ? b_two : "/dev/null";
3386         strbuf_addf(&header, "%s%sdiff --git %s %s%s\n", line_prefix, meta, a_one, b_two, reset);
3387         if (lbl[0][0] == '/') {
3388                 /* /dev/null */
3389                 strbuf_addf(&header, "%s%snew file mode %06o%s\n", line_prefix, meta, two->mode, reset);
3390                 if (xfrm_msg)
3391                         strbuf_addstr(&header, xfrm_msg);
3392                 must_show_header = 1;
3393         }
3394         else if (lbl[1][0] == '/') {
3395                 strbuf_addf(&header, "%s%sdeleted file mode %06o%s\n", line_prefix, meta, one->mode, reset);
3396                 if (xfrm_msg)
3397                         strbuf_addstr(&header, xfrm_msg);
3398                 must_show_header = 1;
3399         }
3400         else {
3401                 if (one->mode != two->mode) {
3402                         strbuf_addf(&header, "%s%sold mode %06o%s\n", line_prefix, meta, one->mode, reset);
3403                         strbuf_addf(&header, "%s%snew mode %06o%s\n", line_prefix, meta, two->mode, reset);
3404                         must_show_header = 1;
3405                 }
3406                 if (xfrm_msg)
3407                         strbuf_addstr(&header, xfrm_msg);
3408
3409                 /*
3410                  * we do not run diff between different kind
3411                  * of objects.
3412                  */
3413                 if ((one->mode ^ two->mode) & S_IFMT)
3414                         goto free_ab_and_return;
3415                 if (complete_rewrite &&
3416                     (textconv_one || !diff_filespec_is_binary(o->repo, one)) &&
3417                     (textconv_two || !diff_filespec_is_binary(o->repo, two))) {
3418                         emit_diff_symbol(o, DIFF_SYMBOL_HEADER,
3419                                          header.buf, header.len, 0);
3420                         strbuf_reset(&header);
3421                         emit_rewrite_diff(name_a, name_b, one, two,
3422                                           textconv_one, textconv_two, o);
3423                         o->found_changes = 1;
3424                         goto free_ab_and_return;
3425                 }
3426         }
3427
3428         if (o->irreversible_delete && lbl[1][0] == '/') {
3429                 emit_diff_symbol(o, DIFF_SYMBOL_HEADER, header.buf,
3430                                  header.len, 0);
3431                 strbuf_reset(&header);
3432                 goto free_ab_and_return;
3433         } else if (!o->flags.text &&
3434                    ( (!textconv_one && diff_filespec_is_binary(o->repo, one)) ||
3435                      (!textconv_two && diff_filespec_is_binary(o->repo, two)) )) {
3436                 struct strbuf sb = STRBUF_INIT;
3437                 if (!one->data && !two->data &&
3438                     S_ISREG(one->mode) && S_ISREG(two->mode) &&
3439                     !o->flags.binary) {
3440                         if (oideq(&one->oid, &two->oid)) {
3441                                 if (must_show_header)
3442                                         emit_diff_symbol(o, DIFF_SYMBOL_HEADER,
3443                                                          header.buf, header.len,
3444                                                          0);
3445                                 goto free_ab_and_return;
3446                         }
3447                         emit_diff_symbol(o, DIFF_SYMBOL_HEADER,
3448                                          header.buf, header.len, 0);
3449                         strbuf_addf(&sb, "%sBinary files %s and %s differ\n",
3450                                     diff_line_prefix(o), lbl[0], lbl[1]);
3451                         emit_diff_symbol(o, DIFF_SYMBOL_BINARY_FILES,
3452                                          sb.buf, sb.len, 0);
3453                         strbuf_release(&sb);
3454                         goto free_ab_and_return;
3455                 }
3456                 if (fill_mmfile(o->repo, &mf1, one) < 0 ||
3457                     fill_mmfile(o->repo, &mf2, two) < 0)
3458                         die("unable to read files to diff");
3459                 /* Quite common confusing case */
3460                 if (mf1.size == mf2.size &&
3461                     !memcmp(mf1.ptr, mf2.ptr, mf1.size)) {
3462                         if (must_show_header)
3463                                 emit_diff_symbol(o, DIFF_SYMBOL_HEADER,
3464                                                  header.buf, header.len, 0);
3465                         goto free_ab_and_return;
3466                 }
3467                 emit_diff_symbol(o, DIFF_SYMBOL_HEADER, header.buf, header.len, 0);
3468                 strbuf_reset(&header);
3469                 if (o->flags.binary)
3470                         emit_binary_diff(o, &mf1, &mf2);
3471                 else {
3472                         strbuf_addf(&sb, "%sBinary files %s and %s differ\n",
3473                                     diff_line_prefix(o), lbl[0], lbl[1]);
3474                         emit_diff_symbol(o, DIFF_SYMBOL_BINARY_FILES,
3475                                          sb.buf, sb.len, 0);
3476                         strbuf_release(&sb);
3477                 }
3478                 o->found_changes = 1;
3479         } else {
3480                 /* Crazy xdl interfaces.. */
3481                 const char *diffopts = getenv("GIT_DIFF_OPTS");
3482                 const char *v;
3483                 xpparam_t xpp;
3484                 xdemitconf_t xecfg;
3485                 struct emit_callback ecbdata;
3486                 const struct userdiff_funcname *pe;
3487
3488                 if (must_show_header) {
3489                         emit_diff_symbol(o, DIFF_SYMBOL_HEADER,
3490                                          header.buf, header.len, 0);
3491                         strbuf_reset(&header);
3492                 }
3493
3494                 mf1.size = fill_textconv(o->repo, textconv_one, one, &mf1.ptr);
3495                 mf2.size = fill_textconv(o->repo, textconv_two, two, &mf2.ptr);
3496
3497                 pe = diff_funcname_pattern(o, one);
3498                 if (!pe)
3499                         pe = diff_funcname_pattern(o, two);
3500
3501                 memset(&xpp, 0, sizeof(xpp));
3502                 memset(&xecfg, 0, sizeof(xecfg));
3503                 memset(&ecbdata, 0, sizeof(ecbdata));
3504                 if (o->flags.suppress_diff_headers)
3505                         lbl[0] = NULL;
3506                 ecbdata.label_path = lbl;
3507                 ecbdata.color_diff = want_color(o->use_color);
3508                 ecbdata.ws_rule = whitespace_rule(o->repo->index, name_b);
3509                 if (ecbdata.ws_rule & WS_BLANK_AT_EOF)
3510                         check_blank_at_eof(&mf1, &mf2, &ecbdata);
3511                 ecbdata.opt = o;
3512                 if (header.len && !o->flags.suppress_diff_headers)
3513                         ecbdata.header = &header;
3514                 xpp.flags = o->xdl_opts;
3515                 xpp.anchors = o->anchors;
3516                 xpp.anchors_nr = o->anchors_nr;
3517                 xecfg.ctxlen = o->context;
3518                 xecfg.interhunkctxlen = o->interhunkcontext;
3519                 xecfg.flags = XDL_EMIT_FUNCNAMES;
3520                 if (o->flags.funccontext)
3521                         xecfg.flags |= XDL_EMIT_FUNCCONTEXT;
3522                 if (pe)
3523                         xdiff_set_find_func(&xecfg, pe->pattern, pe->cflags);
3524                 if (!diffopts)
3525                         ;
3526                 else if (skip_prefix(diffopts, "--unified=", &v))
3527                         xecfg.ctxlen = strtoul(v, NULL, 10);
3528                 else if (skip_prefix(diffopts, "-u", &v))
3529                         xecfg.ctxlen = strtoul(v, NULL, 10);
3530                 if (o->word_diff)
3531                         init_diff_words_data(&ecbdata, o, one, two);
3532                 if (xdi_diff_outf(&mf1, &mf2, NULL, fn_out_consume,
3533                                   &ecbdata, &xpp, &xecfg))
3534                         die("unable to generate diff for %s", one->path);
3535                 if (o->word_diff)
3536                         free_diff_words_data(&ecbdata);
3537                 if (textconv_one)
3538                         free(mf1.ptr);
3539                 if (textconv_two)
3540                         free(mf2.ptr);
3541                 xdiff_clear_find_func(&xecfg);
3542         }
3543
3544  free_ab_and_return:
3545         strbuf_release(&header);
3546         diff_free_filespec_data(one);
3547         diff_free_filespec_data(two);
3548         free(a_one);
3549         free(b_two);
3550         return;
3551 }
3552
3553 static char *get_compact_summary(const struct diff_filepair *p, int is_renamed)
3554 {
3555         if (!is_renamed) {
3556                 if (p->status == DIFF_STATUS_ADDED) {
3557                         if (S_ISLNK(p->two->mode))
3558                                 return "new +l";
3559                         else if ((p->two->mode & 0777) == 0755)
3560                                 return "new +x";
3561                         else
3562                                 return "new";
3563                 } else if (p->status == DIFF_STATUS_DELETED)
3564                         return "gone";
3565         }
3566         if (S_ISLNK(p->one->mode) && !S_ISLNK(p->two->mode))
3567                 return "mode -l";
3568         else if (!S_ISLNK(p->one->mode) && S_ISLNK(p->two->mode))
3569                 return "mode +l";
3570         else if ((p->one->mode & 0777) == 0644 &&
3571                  (p->two->mode & 0777) == 0755)
3572                 return "mode +x";
3573         else if ((p->one->mode & 0777) == 0755 &&
3574                  (p->two->mode & 0777) == 0644)
3575                 return "mode -x";
3576         return NULL;
3577 }
3578
3579 static void builtin_diffstat(const char *name_a, const char *name_b,
3580                              struct diff_filespec *one,
3581                              struct diff_filespec *two,
3582                              struct diffstat_t *diffstat,
3583                              struct diff_options *o,
3584                              struct diff_filepair *p)
3585 {
3586         mmfile_t mf1, mf2;
3587         struct diffstat_file *data;
3588         int same_contents;
3589         int complete_rewrite = 0;
3590
3591         if (!DIFF_PAIR_UNMERGED(p)) {
3592                 if (p->status == DIFF_STATUS_MODIFIED && p->score)
3593                         complete_rewrite = 1;
3594         }
3595
3596         data = diffstat_add(diffstat, name_a, name_b);
3597         data->is_interesting = p->status != DIFF_STATUS_UNKNOWN;
3598         if (o->flags.stat_with_summary)
3599                 data->comments = get_compact_summary(p, data->is_renamed);
3600
3601         if (!one || !two) {
3602                 data->is_unmerged = 1;
3603                 return;
3604         }
3605
3606         same_contents = oideq(&one->oid, &two->oid);
3607
3608         if (diff_filespec_is_binary(o->repo, one) ||
3609             diff_filespec_is_binary(o->repo, two)) {
3610                 data->is_binary = 1;
3611                 if (same_contents) {
3612                         data->added = 0;
3613                         data->deleted = 0;
3614                 } else {
3615                         data->added = diff_filespec_size(o->repo, two);
3616                         data->deleted = diff_filespec_size(o->repo, one);
3617                 }
3618         }
3619
3620         else if (complete_rewrite) {
3621                 diff_populate_filespec(o->repo, one, 0);
3622                 diff_populate_filespec(o->repo, two, 0);
3623                 data->deleted = count_lines(one->data, one->size);
3624                 data->added = count_lines(two->data, two->size);
3625         }
3626
3627         else if (!same_contents) {
3628                 /* Crazy xdl interfaces.. */
3629                 xpparam_t xpp;
3630                 xdemitconf_t xecfg;
3631
3632                 if (fill_mmfile(o->repo, &mf1, one) < 0 ||
3633                     fill_mmfile(o->repo, &mf2, two) < 0)
3634                         die("unable to read files to diff");
3635
3636                 memset(&xpp, 0, sizeof(xpp));
3637                 memset(&xecfg, 0, sizeof(xecfg));
3638                 xpp.flags = o->xdl_opts;
3639                 xpp.anchors = o->anchors;
3640                 xpp.anchors_nr = o->anchors_nr;
3641                 xecfg.ctxlen = o->context;
3642                 xecfg.interhunkctxlen = o->interhunkcontext;
3643                 if (xdi_diff_outf(&mf1, &mf2, discard_hunk_line,
3644                                   diffstat_consume, diffstat, &xpp, &xecfg))
3645                         die("unable to generate diffstat for %s", one->path);
3646         }
3647
3648         diff_free_filespec_data(one);
3649         diff_free_filespec_data(two);
3650 }
3651
3652 static void builtin_checkdiff(const char *name_a, const char *name_b,
3653                               const char *attr_path,
3654                               struct diff_filespec *one,
3655                               struct diff_filespec *two,
3656                               struct diff_options *o)
3657 {
3658         mmfile_t mf1, mf2;
3659         struct checkdiff_t data;
3660
3661         if (!two)
3662                 return;
3663
3664         memset(&data, 0, sizeof(data));
3665         data.filename = name_b ? name_b : name_a;
3666         data.lineno = 0;
3667         data.o = o;
3668         data.ws_rule = whitespace_rule(o->repo->index, attr_path);
3669         data.conflict_marker_size = ll_merge_marker_size(o->repo->index, attr_path);
3670
3671         if (fill_mmfile(o->repo, &mf1, one) < 0 ||
3672             fill_mmfile(o->repo, &mf2, two) < 0)
3673                 die("unable to read files to diff");
3674
3675         /*
3676          * All the other codepaths check both sides, but not checking
3677          * the "old" side here is deliberate.  We are checking the newly
3678          * introduced changes, and as long as the "new" side is text, we
3679          * can and should check what it introduces.
3680          */
3681         if (diff_filespec_is_binary(o->repo, two))
3682                 goto free_and_return;
3683         else {
3684                 /* Crazy xdl interfaces.. */
3685                 xpparam_t xpp;
3686                 xdemitconf_t xecfg;
3687
3688                 memset(&xpp, 0, sizeof(xpp));
3689                 memset(&xecfg, 0, sizeof(xecfg));
3690                 xecfg.ctxlen = 1; /* at least one context line */
3691                 xpp.flags = 0;
3692                 if (xdi_diff_outf(&mf1, &mf2, checkdiff_consume_hunk,
3693                                   checkdiff_consume, &data,
3694                                   &xpp, &xecfg))
3695                         die("unable to generate checkdiff for %s", one->path);
3696
3697                 if (data.ws_rule & WS_BLANK_AT_EOF) {
3698                         struct emit_callback ecbdata;
3699                         int blank_at_eof;
3700
3701                         ecbdata.ws_rule = data.ws_rule;
3702                         check_blank_at_eof(&mf1, &mf2, &ecbdata);
3703                         blank_at_eof = ecbdata.blank_at_eof_in_postimage;
3704
3705                         if (blank_at_eof) {
3706                                 static char *err;
3707                                 if (!err)
3708                                         err = whitespace_error_string(WS_BLANK_AT_EOF);
3709                                 fprintf(o->file, "%s:%d: %s.\n",
3710                                         data.filename, blank_at_eof, err);
3711                                 data.status = 1; /* report errors */
3712                         }
3713                 }
3714         }
3715  free_and_return:
3716         diff_free_filespec_data(one);
3717         diff_free_filespec_data(two);
3718         if (data.status)
3719                 o->flags.check_failed = 1;
3720 }
3721
3722 struct diff_filespec *alloc_filespec(const char *path)
3723 {
3724         struct diff_filespec *spec;
3725
3726         FLEXPTR_ALLOC_STR(spec, path, path);
3727         spec->count = 1;
3728         spec->is_binary = -1;
3729         return spec;
3730 }
3731
3732 void free_filespec(struct diff_filespec *spec)
3733 {
3734         if (!--spec->count) {
3735                 diff_free_filespec_data(spec);
3736                 free(spec);
3737         }
3738 }
3739
3740 void fill_filespec(struct diff_filespec *spec, const struct object_id *oid,
3741                    int oid_valid, unsigned short mode)
3742 {
3743         if (mode) {
3744                 spec->mode = canon_mode(mode);
3745                 oidcpy(&spec->oid, oid);
3746                 spec->oid_valid = oid_valid;
3747         }
3748 }
3749
3750 /*
3751  * Given a name and sha1 pair, if the index tells us the file in
3752  * the work tree has that object contents, return true, so that
3753  * prepare_temp_file() does not have to inflate and extract.
3754  */
3755 static int reuse_worktree_file(struct index_state *istate,
3756                                const char *name,
3757                                const struct object_id *oid,
3758                                int want_file)
3759 {
3760         const struct cache_entry *ce;
3761         struct stat st;
3762         int pos, len;
3763
3764         /*
3765          * We do not read the cache ourselves here, because the
3766          * benchmark with my previous version that always reads cache
3767          * shows that it makes things worse for diff-tree comparing
3768          * two linux-2.6 kernel trees in an already checked out work
3769          * tree.  This is because most diff-tree comparisons deal with
3770          * only a small number of files, while reading the cache is
3771          * expensive for a large project, and its cost outweighs the
3772          * savings we get by not inflating the object to a temporary
3773          * file.  Practically, this code only helps when we are used
3774          * by diff-cache --cached, which does read the cache before
3775          * calling us.
3776          */
3777         if (!istate->cache)
3778                 return 0;
3779
3780         /* We want to avoid the working directory if our caller
3781          * doesn't need the data in a normal file, this system
3782          * is rather slow with its stat/open/mmap/close syscalls,
3783          * and the object is contained in a pack file.  The pack
3784          * is probably already open and will be faster to obtain
3785          * the data through than the working directory.  Loose
3786          * objects however would tend to be slower as they need
3787          * to be individually opened and inflated.
3788          */
3789         if (!FAST_WORKING_DIRECTORY && !want_file && has_object_pack(oid))
3790                 return 0;
3791
3792         /*
3793          * Similarly, if we'd have to convert the file contents anyway, that
3794          * makes the optimization not worthwhile.
3795          */
3796         if (!want_file && would_convert_to_git(istate, name))
3797                 return 0;
3798
3799         len = strlen(name);
3800         pos = index_name_pos(istate, name, len);
3801         if (pos < 0)
3802                 return 0;
3803         ce = istate->cache[pos];
3804
3805         /*
3806          * This is not the sha1 we are looking for, or
3807          * unreusable because it is not a regular file.
3808          */
3809         if (!oideq(oid, &ce->oid) || !S_ISREG(ce->ce_mode))
3810                 return 0;
3811
3812         /*
3813          * If ce is marked as "assume unchanged", there is no
3814          * guarantee that work tree matches what we are looking for.
3815          */
3816         if ((ce->ce_flags & CE_VALID) || ce_skip_worktree(ce))
3817                 return 0;
3818
3819         /*
3820          * If ce matches the file in the work tree, we can reuse it.
3821          */
3822         if (ce_uptodate(ce) ||
3823             (!lstat(name, &st) && !ie_match_stat(istate, ce, &st, 0)))
3824                 return 1;
3825
3826         return 0;
3827 }
3828
3829 static int diff_populate_gitlink(struct diff_filespec *s, int size_only)
3830 {
3831         struct strbuf buf = STRBUF_INIT;
3832         char *dirty = "";
3833
3834         /* Are we looking at the work tree? */
3835         if (s->dirty_submodule)
3836                 dirty = "-dirty";
3837
3838         strbuf_addf(&buf, "Subproject commit %s%s\n",
3839                     oid_to_hex(&s->oid), dirty);
3840         s->size = buf.len;
3841         if (size_only) {
3842                 s->data = NULL;
3843                 strbuf_release(&buf);
3844         } else {
3845                 s->data = strbuf_detach(&buf, NULL);
3846                 s->should_free = 1;
3847         }
3848         return 0;
3849 }
3850
3851 /*
3852  * While doing rename detection and pickaxe operation, we may need to
3853  * grab the data for the blob (or file) for our own in-core comparison.
3854  * diff_filespec has data and size fields for this purpose.
3855  */
3856 int diff_populate_filespec(struct repository *r,
3857                            struct diff_filespec *s,
3858                            unsigned int flags)
3859 {
3860         int size_only = flags & CHECK_SIZE_ONLY;
3861         int err = 0;
3862         int conv_flags = global_conv_flags_eol;
3863         /*
3864          * demote FAIL to WARN to allow inspecting the situation
3865          * instead of refusing.
3866          */
3867         if (conv_flags & CONV_EOL_RNDTRP_DIE)
3868                 conv_flags = CONV_EOL_RNDTRP_WARN;
3869
3870         if (!DIFF_FILE_VALID(s))
3871                 die("internal error: asking to populate invalid file.");
3872         if (S_ISDIR(s->mode))
3873                 return -1;
3874
3875         if (s->data)
3876                 return 0;
3877
3878         if (size_only && 0 < s->size)
3879                 return 0;
3880
3881         if (S_ISGITLINK(s->mode))
3882                 return diff_populate_gitlink(s, size_only);
3883
3884         if (!s->oid_valid ||
3885             reuse_worktree_file(r->index, s->path, &s->oid, 0)) {
3886                 struct strbuf buf = STRBUF_INIT;
3887                 struct stat st;
3888                 int fd;
3889
3890                 if (lstat(s->path, &st) < 0) {
3891                 err_empty:
3892                         err = -1;
3893                 empty:
3894                         s->data = (char *)"";
3895                         s->size = 0;
3896                         return err;
3897                 }
3898                 s->size = xsize_t(st.st_size);
3899                 if (!s->size)
3900                         goto empty;
3901                 if (S_ISLNK(st.st_mode)) {
3902                         struct strbuf sb = STRBUF_INIT;
3903
3904                         if (strbuf_readlink(&sb, s->path, s->size))
3905                                 goto err_empty;
3906                         s->size = sb.len;
3907                         s->data = strbuf_detach(&sb, NULL);
3908                         s->should_free = 1;
3909                         return 0;
3910                 }
3911
3912                 /*
3913                  * Even if the caller would be happy with getting
3914                  * only the size, we cannot return early at this
3915                  * point if the path requires us to run the content
3916                  * conversion.
3917                  */
3918                 if (size_only && !would_convert_to_git(r->index, s->path))
3919                         return 0;
3920
3921                 /*
3922                  * Note: this check uses xsize_t(st.st_size) that may
3923                  * not be the true size of the blob after it goes
3924                  * through convert_to_git().  This may not strictly be
3925                  * correct, but the whole point of big_file_threshold
3926                  * and is_binary check being that we want to avoid
3927                  * opening the file and inspecting the contents, this
3928                  * is probably fine.
3929                  */
3930                 if ((flags & CHECK_BINARY) &&
3931                     s->size > big_file_threshold && s->is_binary == -1) {
3932                         s->is_binary = 1;
3933                         return 0;
3934                 }
3935                 fd = open(s->path, O_RDONLY);
3936                 if (fd < 0)
3937                         goto err_empty;
3938                 s->data = xmmap(NULL, s->size, PROT_READ, MAP_PRIVATE, fd, 0);
3939                 close(fd);
3940                 s->should_munmap = 1;
3941
3942                 /*
3943                  * Convert from working tree format to canonical git format
3944                  */
3945                 if (convert_to_git(r->index, s->path, s->data, s->size, &buf, conv_flags)) {
3946                         size_t size = 0;
3947                         munmap(s->data, s->size);
3948                         s->should_munmap = 0;
3949                         s->data = strbuf_detach(&buf, &size);
3950                         s->size = size;
3951                         s->should_free = 1;
3952                 }
3953         }
3954         else {
3955                 enum object_type type;
3956                 if (size_only || (flags & CHECK_BINARY)) {
3957                         type = oid_object_info(r, &s->oid, &s->size);
3958                         if (type < 0)
3959                                 die("unable to read %s",
3960                                     oid_to_hex(&s->oid));
3961                         if (size_only)
3962                                 return 0;
3963                         if (s->size > big_file_threshold && s->is_binary == -1) {
3964                                 s->is_binary = 1;
3965                                 return 0;
3966                         }
3967                 }
3968                 s->data = read_object_file(&s->oid, &type, &s->size);
3969                 if (!s->data)
3970                         die("unable to read %s", oid_to_hex(&s->oid));
3971                 s->should_free = 1;
3972         }
3973         return 0;
3974 }
3975
3976 void diff_free_filespec_blob(struct diff_filespec *s)
3977 {
3978         if (s->should_free)
3979                 free(s->data);
3980         else if (s->should_munmap)
3981                 munmap(s->data, s->size);
3982
3983         if (s->should_free || s->should_munmap) {
3984                 s->should_free = s->should_munmap = 0;
3985                 s->data = NULL;
3986         }
3987 }
3988
3989 void diff_free_filespec_data(struct diff_filespec *s)
3990 {
3991         diff_free_filespec_blob(s);
3992         FREE_AND_NULL(s->cnt_data);
3993 }
3994
3995 static void prep_temp_blob(struct index_state *istate,
3996                            const char *path, struct diff_tempfile *temp,
3997                            void *blob,
3998                            unsigned long size,
3999                            const struct object_id *oid,
4000                            int mode)
4001 {
4002         struct strbuf buf = STRBUF_INIT;
4003         struct strbuf tempfile = STRBUF_INIT;
4004         char *path_dup = xstrdup(path);
4005         const char *base = basename(path_dup);
4006
4007         /* Generate "XXXXXX_basename.ext" */
4008         strbuf_addstr(&tempfile, "XXXXXX_");
4009         strbuf_addstr(&tempfile, base);
4010
4011         temp->tempfile = mks_tempfile_ts(tempfile.buf, strlen(base) + 1);
4012         if (!temp->tempfile)
4013                 die_errno("unable to create temp-file");
4014         if (convert_to_working_tree(istate, path,
4015                         (const char *)blob, (size_t)size, &buf)) {
4016                 blob = buf.buf;
4017                 size = buf.len;
4018         }
4019         if (write_in_full(temp->tempfile->fd, blob, size) < 0 ||
4020             close_tempfile_gently(temp->tempfile))
4021                 die_errno("unable to write temp-file");
4022         temp->name = get_tempfile_path(temp->tempfile);
4023         oid_to_hex_r(temp->hex, oid);
4024         xsnprintf(temp->mode, sizeof(temp->mode), "%06o", mode);
4025         strbuf_release(&buf);
4026         strbuf_release(&tempfile);
4027         free(path_dup);
4028 }
4029
4030 static struct diff_tempfile *prepare_temp_file(struct repository *r,
4031                                                const char *name,
4032                                                struct diff_filespec *one)
4033 {
4034         struct diff_tempfile *temp = claim_diff_tempfile();
4035
4036         if (!DIFF_FILE_VALID(one)) {
4037         not_a_valid_file:
4038                 /* A '-' entry produces this for file-2, and
4039                  * a '+' entry produces this for file-1.
4040                  */
4041                 temp->name = "/dev/null";
4042                 xsnprintf(temp->hex, sizeof(temp->hex), ".");
4043                 xsnprintf(temp->mode, sizeof(temp->mode), ".");
4044                 return temp;
4045         }
4046
4047         if (!S_ISGITLINK(one->mode) &&
4048             (!one->oid_valid ||
4049              reuse_worktree_file(r->index, name, &one->oid, 1))) {
4050                 struct stat st;
4051                 if (lstat(name, &st) < 0) {
4052                         if (errno == ENOENT)
4053                                 goto not_a_valid_file;
4054                         die_errno("stat(%s)", name);
4055                 }
4056                 if (S_ISLNK(st.st_mode)) {
4057                         struct strbuf sb = STRBUF_INIT;
4058                         if (strbuf_readlink(&sb, name, st.st_size) < 0)
4059                                 die_errno("readlink(%s)", name);
4060                         prep_temp_blob(r->index, name, temp, sb.buf, sb.len,
4061                                        (one->oid_valid ?
4062                                         &one->oid : &null_oid),
4063                                        (one->oid_valid ?
4064                                         one->mode : S_IFLNK));
4065                         strbuf_release(&sb);
4066                 }
4067                 else {
4068                         /* we can borrow from the file in the work tree */
4069                         temp->name = name;
4070                         if (!one->oid_valid)
4071                                 oid_to_hex_r(temp->hex, &null_oid);
4072                         else
4073                                 oid_to_hex_r(temp->hex, &one->oid);
4074                         /* Even though we may sometimes borrow the
4075                          * contents from the work tree, we always want
4076                          * one->mode.  mode is trustworthy even when
4077                          * !(one->oid_valid), as long as
4078                          * DIFF_FILE_VALID(one).
4079                          */
4080                         xsnprintf(temp->mode, sizeof(temp->mode), "%06o", one->mode);
4081                 }
4082                 return temp;
4083         }
4084         else {
4085                 if (diff_populate_filespec(r, one, 0))
4086                         die("cannot read data blob for %s", one->path);
4087                 prep_temp_blob(r->index, name, temp,
4088                                one->data, one->size,
4089                                &one->oid, one->mode);
4090         }
4091         return temp;
4092 }
4093
4094 static void add_external_diff_name(struct repository *r,
4095                                    struct argv_array *argv,
4096                                    const char *name,
4097                                    struct diff_filespec *df)
4098 {
4099         struct diff_tempfile *temp = prepare_temp_file(r, name, df);
4100         argv_array_push(argv, temp->name);
4101         argv_array_push(argv, temp->hex);
4102         argv_array_push(argv, temp->mode);
4103 }
4104
4105 /* An external diff command takes:
4106  *
4107  * diff-cmd name infile1 infile1-sha1 infile1-mode \
4108  *               infile2 infile2-sha1 infile2-mode [ rename-to ]
4109  *
4110  */
4111 static void run_external_diff(const char *pgm,
4112                               const char *name,
4113                               const char *other,
4114                               struct diff_filespec *one,
4115                               struct diff_filespec *two,
4116                               const char *xfrm_msg,
4117                               int complete_rewrite,
4118                               struct diff_options *o)
4119 {
4120         struct argv_array argv = ARGV_ARRAY_INIT;
4121         struct argv_array env = ARGV_ARRAY_INIT;
4122         struct diff_queue_struct *q = &diff_queued_diff;
4123
4124         argv_array_push(&argv, pgm);
4125         argv_array_push(&argv, name);
4126
4127         if (one && two) {
4128                 add_external_diff_name(o->repo, &argv, name, one);
4129                 if (!other)
4130                         add_external_diff_name(o->repo, &argv, name, two);
4131                 else {
4132                         add_external_diff_name(o->repo, &argv, other, two);
4133                         argv_array_push(&argv, other);
4134                         argv_array_push(&argv, xfrm_msg);
4135                 }
4136         }
4137
4138         argv_array_pushf(&env, "GIT_DIFF_PATH_COUNTER=%d", ++o->diff_path_counter);
4139         argv_array_pushf(&env, "GIT_DIFF_PATH_TOTAL=%d", q->nr);
4140
4141         if (run_command_v_opt_cd_env(argv.argv, RUN_USING_SHELL, NULL, env.argv))
4142                 die(_("external diff died, stopping at %s"), name);
4143
4144         remove_tempfile();
4145         argv_array_clear(&argv);
4146         argv_array_clear(&env);
4147 }
4148
4149 static int similarity_index(struct diff_filepair *p)
4150 {
4151         return p->score * 100 / MAX_SCORE;
4152 }
4153
4154 static const char *diff_abbrev_oid(const struct object_id *oid, int abbrev)
4155 {
4156         if (startup_info->have_repository)
4157                 return find_unique_abbrev(oid, abbrev);
4158         else {
4159                 char *hex = oid_to_hex(oid);
4160                 if (abbrev < 0)
4161                         abbrev = FALLBACK_DEFAULT_ABBREV;
4162                 if (abbrev > the_hash_algo->hexsz)
4163                         BUG("oid abbreviation out of range: %d", abbrev);
4164                 if (abbrev)
4165                         hex[abbrev] = '\0';
4166                 return hex;
4167         }
4168 }
4169
4170 static void fill_metainfo(struct strbuf *msg,
4171                           const char *name,
4172                           const char *other,
4173                           struct diff_filespec *one,
4174                           struct diff_filespec *two,
4175                           struct diff_options *o,
4176                           struct diff_filepair *p,
4177                           int *must_show_header,
4178                           int use_color)
4179 {
4180         const char *set = diff_get_color(use_color, DIFF_METAINFO);
4181         const char *reset = diff_get_color(use_color, DIFF_RESET);
4182         const char *line_prefix = diff_line_prefix(o);
4183
4184         *must_show_header = 1;
4185         strbuf_init(msg, PATH_MAX * 2 + 300);
4186         switch (p->status) {
4187         case DIFF_STATUS_COPIED:
4188                 strbuf_addf(msg, "%s%ssimilarity index %d%%",
4189                             line_prefix, set, similarity_index(p));
4190                 strbuf_addf(msg, "%s\n%s%scopy from ",
4191                             reset,  line_prefix, set);
4192                 quote_c_style(name, msg, NULL, 0);
4193                 strbuf_addf(msg, "%s\n%s%scopy to ", reset, line_prefix, set);
4194                 quote_c_style(other, msg, NULL, 0);
4195                 strbuf_addf(msg, "%s\n", reset);
4196                 break;
4197         case DIFF_STATUS_RENAMED:
4198                 strbuf_addf(msg, "%s%ssimilarity index %d%%",
4199                             line_prefix, set, similarity_index(p));
4200                 strbuf_addf(msg, "%s\n%s%srename from ",
4201                             reset, line_prefix, set);
4202                 quote_c_style(name, msg, NULL, 0);
4203                 strbuf_addf(msg, "%s\n%s%srename to ",
4204                             reset, line_prefix, set);
4205                 quote_c_style(other, msg, NULL, 0);
4206                 strbuf_addf(msg, "%s\n", reset);
4207                 break;
4208         case DIFF_STATUS_MODIFIED:
4209                 if (p->score) {
4210                         strbuf_addf(msg, "%s%sdissimilarity index %d%%%s\n",
4211                                     line_prefix,
4212                                     set, similarity_index(p), reset);
4213                         break;
4214                 }
4215                 /* fallthru */
4216         default:
4217                 *must_show_header = 0;
4218         }
4219         if (one && two && !oideq(&one->oid, &two->oid)) {
4220                 const unsigned hexsz = the_hash_algo->hexsz;
4221                 int abbrev = o->flags.full_index ? hexsz : DEFAULT_ABBREV;
4222
4223                 if (o->flags.binary) {
4224                         mmfile_t mf;
4225                         if ((!fill_mmfile(o->repo, &mf, one) &&
4226                              diff_filespec_is_binary(o->repo, one)) ||
4227                             (!fill_mmfile(o->repo, &mf, two) &&
4228                              diff_filespec_is_binary(o->repo, two)))
4229                                 abbrev = hexsz;
4230                 }
4231                 strbuf_addf(msg, "%s%sindex %s..%s", line_prefix, set,
4232                             diff_abbrev_oid(&one->oid, abbrev),
4233                             diff_abbrev_oid(&two->oid, abbrev));
4234                 if (one->mode == two->mode)
4235                         strbuf_addf(msg, " %06o", one->mode);
4236                 strbuf_addf(msg, "%s\n", reset);
4237         }
4238 }
4239
4240 static void run_diff_cmd(const char *pgm,
4241                          const char *name,
4242                          const char *other,
4243                          const char *attr_path,
4244                          struct diff_filespec *one,
4245                          struct diff_filespec *two,
4246                          struct strbuf *msg,
4247                          struct diff_options *o,
4248                          struct diff_filepair *p)
4249 {
4250         const char *xfrm_msg = NULL;
4251         int complete_rewrite = (p->status == DIFF_STATUS_MODIFIED) && p->score;
4252         int must_show_header = 0;
4253
4254
4255         if (o->flags.allow_external) {
4256                 struct userdiff_driver *drv;
4257
4258                 drv = userdiff_find_by_path(o->repo->index, attr_path);
4259                 if (drv && drv->external)
4260                         pgm = drv->external;
4261         }
4262
4263         if (msg) {
4264                 /*
4265                  * don't use colors when the header is intended for an
4266                  * external diff driver
4267                  */
4268                 fill_metainfo(msg, name, other, one, two, o, p,
4269                               &must_show_header,
4270                               want_color(o->use_color) && !pgm);
4271                 xfrm_msg = msg->len ? msg->buf : NULL;
4272         }
4273
4274         if (pgm) {
4275                 run_external_diff(pgm, name, other, one, two, xfrm_msg,
4276                                   complete_rewrite, o);
4277                 return;
4278         }
4279         if (one && two)
4280                 builtin_diff(name, other ? other : name,
4281                              one, two, xfrm_msg, must_show_header,
4282                              o, complete_rewrite);
4283         else
4284                 fprintf(o->file, "* Unmerged path %s\n", name);
4285 }
4286
4287 static void diff_fill_oid_info(struct diff_filespec *one, struct index_state *istate)
4288 {
4289         if (DIFF_FILE_VALID(one)) {
4290                 if (!one->oid_valid) {
4291                         struct stat st;
4292                         if (one->is_stdin) {
4293                                 oidclr(&one->oid);
4294                                 return;
4295                         }
4296                         if (lstat(one->path, &st) < 0)
4297                                 die_errno("stat '%s'", one->path);
4298                         if (index_path(istate, &one->oid, one->path, &st, 0))
4299                                 die("cannot hash %s", one->path);
4300                 }
4301         }
4302         else
4303                 oidclr(&one->oid);
4304 }
4305
4306 static void strip_prefix(int prefix_length, const char **namep, const char **otherp)
4307 {
4308         /* Strip the prefix but do not molest /dev/null and absolute paths */
4309         if (*namep && !is_absolute_path(*namep)) {
4310                 *namep += prefix_length;
4311                 if (**namep == '/')
4312                         ++*namep;
4313         }
4314         if (*otherp && !is_absolute_path(*otherp)) {
4315                 *otherp += prefix_length;
4316                 if (**otherp == '/')
4317                         ++*otherp;
4318         }
4319 }
4320
4321 static void run_diff(struct diff_filepair *p, struct diff_options *o)
4322 {
4323         const char *pgm = external_diff();
4324         struct strbuf msg;
4325         struct diff_filespec *one = p->one;
4326         struct diff_filespec *two = p->two;
4327         const char *name;
4328         const char *other;
4329         const char *attr_path;
4330
4331         name  = one->path;
4332         other = (strcmp(name, two->path) ? two->path : NULL);
4333         attr_path = name;
4334         if (o->prefix_length)
4335                 strip_prefix(o->prefix_length, &name, &other);
4336
4337         if (!o->flags.allow_external)
4338                 pgm = NULL;
4339
4340         if (DIFF_PAIR_UNMERGED(p)) {
4341                 run_diff_cmd(pgm, name, NULL, attr_path,
4342                              NULL, NULL, NULL, o, p);
4343                 return;
4344         }
4345
4346         diff_fill_oid_info(one, o->repo->index);
4347         diff_fill_oid_info(two, o->repo->index);
4348
4349         if (!pgm &&
4350             DIFF_FILE_VALID(one) && DIFF_FILE_VALID(two) &&
4351             (S_IFMT & one->mode) != (S_IFMT & two->mode)) {
4352                 /*
4353                  * a filepair that changes between file and symlink
4354                  * needs to be split into deletion and creation.
4355                  */
4356                 struct diff_filespec *null = alloc_filespec(two->path);
4357                 run_diff_cmd(NULL, name, other, attr_path,
4358                              one, null, &msg,
4359                              o, p);
4360                 free(null);
4361                 strbuf_release(&msg);
4362
4363                 null = alloc_filespec(one->path);
4364                 run_diff_cmd(NULL, name, other, attr_path,
4365                              null, two, &msg, o, p);
4366                 free(null);
4367         }
4368         else
4369                 run_diff_cmd(pgm, name, other, attr_path,
4370                              one, two, &msg, o, p);
4371
4372         strbuf_release(&msg);
4373 }
4374
4375 static void run_diffstat(struct diff_filepair *p, struct diff_options *o,
4376                          struct diffstat_t *diffstat)
4377 {
4378         const char *name;
4379         const char *other;
4380
4381         if (DIFF_PAIR_UNMERGED(p)) {
4382                 /* unmerged */
4383                 builtin_diffstat(p->one->path, NULL, NULL, NULL,
4384                                  diffstat, o, p);
4385                 return;
4386         }
4387
4388         name = p->one->path;
4389         other = (strcmp(name, p->two->path) ? p->two->path : NULL);
4390
4391         if (o->prefix_length)
4392                 strip_prefix(o->prefix_length, &name, &other);
4393
4394         diff_fill_oid_info(p->one, o->repo->index);
4395         diff_fill_oid_info(p->two, o->repo->index);
4396
4397         builtin_diffstat(name, other, p->one, p->two,
4398                          diffstat, o, p);
4399 }
4400
4401 static void run_checkdiff(struct diff_filepair *p, struct diff_options *o)
4402 {
4403         const char *name;
4404         const char *other;
4405         const char *attr_path;
4406
4407         if (DIFF_PAIR_UNMERGED(p)) {
4408                 /* unmerged */
4409                 return;
4410         }
4411
4412         name = p->one->path;
4413         other = (strcmp(name, p->two->path) ? p->two->path : NULL);
4414         attr_path = other ? other : name;
4415
4416         if (o->prefix_length)
4417                 strip_prefix(o->prefix_length, &name, &other);
4418
4419         diff_fill_oid_info(p->one, o->repo->index);
4420         diff_fill_oid_info(p->two, o->repo->index);
4421
4422         builtin_checkdiff(name, other, attr_path, p->one, p->two, o);
4423 }
4424
4425 void repo_diff_setup(struct repository *r, struct diff_options *options)
4426 {
4427         memcpy(options, &default_diff_options, sizeof(*options));
4428
4429         options->file = stdout;
4430         options->repo = r;
4431
4432         options->output_indicators[OUTPUT_INDICATOR_NEW] = '+';
4433         options->output_indicators[OUTPUT_INDICATOR_OLD] = '-';
4434         options->output_indicators[OUTPUT_INDICATOR_CONTEXT] = ' ';
4435         options->abbrev = DEFAULT_ABBREV;
4436         options->line_termination = '\n';
4437         options->break_opt = -1;
4438         options->rename_limit = -1;
4439         options->dirstat_permille = diff_dirstat_permille_default;
4440         options->context = diff_context_default;
4441         options->interhunkcontext = diff_interhunk_context_default;
4442         options->ws_error_highlight = ws_error_highlight_default;
4443         options->flags.rename_empty = 1;
4444         options->objfind = NULL;
4445
4446         /* pathchange left =NULL by default */
4447         options->change = diff_change;
4448         options->add_remove = diff_addremove;
4449         options->use_color = diff_use_color_default;
4450         options->detect_rename = diff_detect_rename_default;
4451         options->xdl_opts |= diff_algorithm;
4452         if (diff_indent_heuristic)
4453                 DIFF_XDL_SET(options, INDENT_HEURISTIC);
4454
4455         options->orderfile = diff_order_file_cfg;
4456
4457         if (diff_no_prefix) {
4458                 options->a_prefix = options->b_prefix = "";
4459         } else if (!diff_mnemonic_prefix) {
4460                 options->a_prefix = "a/";
4461                 options->b_prefix = "b/";
4462         }
4463
4464         options->color_moved = diff_color_moved_default;
4465         options->color_moved_ws_handling = diff_color_moved_ws_default;
4466 }
4467
4468 void diff_setup_done(struct diff_options *options)
4469 {
4470         unsigned check_mask = DIFF_FORMAT_NAME |
4471                               DIFF_FORMAT_NAME_STATUS |
4472                               DIFF_FORMAT_CHECKDIFF |
4473                               DIFF_FORMAT_NO_OUTPUT;
4474         /*
4475          * This must be signed because we're comparing against a potentially
4476          * negative value.
4477          */
4478         const int hexsz = the_hash_algo->hexsz;
4479
4480         if (options->set_default)
4481                 options->set_default(options);
4482
4483         if (HAS_MULTI_BITS(options->output_format & check_mask))
4484                 die(_("--name-only, --name-status, --check and -s are mutually exclusive"));
4485
4486         if (HAS_MULTI_BITS(options->pickaxe_opts & DIFF_PICKAXE_KINDS_MASK))
4487                 die(_("-G, -S and --find-object are mutually exclusive"));
4488
4489         /*
4490          * Most of the time we can say "there are changes"
4491          * only by checking if there are changed paths, but
4492          * --ignore-whitespace* options force us to look
4493          * inside contents.
4494          */
4495
4496         if ((options->xdl_opts & XDF_WHITESPACE_FLAGS))
4497                 options->flags.diff_from_contents = 1;
4498         else
4499                 options->flags.diff_from_contents = 0;
4500
4501         if (options->flags.find_copies_harder)
4502                 options->detect_rename = DIFF_DETECT_COPY;
4503
4504         if (!options->flags.relative_name)
4505                 options->prefix = NULL;
4506         if (options->prefix)
4507                 options->prefix_length = strlen(options->prefix);
4508         else
4509                 options->prefix_length = 0;
4510
4511         if (options->output_format & (DIFF_FORMAT_NAME |
4512                                       DIFF_FORMAT_NAME_STATUS |
4513                                       DIFF_FORMAT_CHECKDIFF |
4514                                       DIFF_FORMAT_NO_OUTPUT))
4515                 options->output_format &= ~(DIFF_FORMAT_RAW |
4516                                             DIFF_FORMAT_NUMSTAT |
4517                                             DIFF_FORMAT_DIFFSTAT |
4518                                             DIFF_FORMAT_SHORTSTAT |
4519                                             DIFF_FORMAT_DIRSTAT |
4520                                             DIFF_FORMAT_SUMMARY |
4521                                             DIFF_FORMAT_PATCH);
4522
4523         /*
4524          * These cases always need recursive; we do not drop caller-supplied
4525          * recursive bits for other formats here.
4526          */
4527         if (options->output_format & (DIFF_FORMAT_PATCH |
4528                                       DIFF_FORMAT_NUMSTAT |
4529                                       DIFF_FORMAT_DIFFSTAT |
4530                                       DIFF_FORMAT_SHORTSTAT |
4531                                       DIFF_FORMAT_DIRSTAT |
4532                                       DIFF_FORMAT_SUMMARY |
4533                                       DIFF_FORMAT_CHECKDIFF))
4534                 options->flags.recursive = 1;
4535         /*
4536          * Also pickaxe would not work very well if you do not say recursive
4537          */
4538         if (options->pickaxe_opts & DIFF_PICKAXE_KINDS_MASK)
4539                 options->flags.recursive = 1;
4540         /*
4541          * When patches are generated, submodules diffed against the work tree
4542          * must be checked for dirtiness too so it can be shown in the output
4543          */
4544         if (options->output_format & DIFF_FORMAT_PATCH)
4545                 options->flags.dirty_submodules = 1;
4546
4547         if (options->detect_rename && options->rename_limit < 0)
4548                 options->rename_limit = diff_rename_limit_default;
4549         if (hexsz < options->abbrev)
4550                 options->abbrev = hexsz; /* full */
4551
4552         /*
4553          * It does not make sense to show the first hit we happened
4554          * to have found.  It does not make sense not to return with
4555          * exit code in such a case either.
4556          */
4557         if (options->flags.quick) {
4558                 options->output_format = DIFF_FORMAT_NO_OUTPUT;
4559                 options->flags.exit_with_status = 1;
4560         }
4561
4562         options->diff_path_counter = 0;
4563
4564         if (options->flags.follow_renames && options->pathspec.nr != 1)
4565                 die(_("--follow requires exactly one pathspec"));
4566
4567         if (!options->use_color || external_diff())
4568                 options->color_moved = 0;
4569 }
4570
4571 static int opt_arg(const char *arg, int arg_short, const char *arg_long, int *val)
4572 {
4573         char c, *eq;
4574         int len;
4575
4576         if (*arg != '-')
4577                 return 0;
4578         c = *++arg;
4579         if (!c)
4580                 return 0;
4581         if (c == arg_short) {
4582                 c = *++arg;
4583                 if (!c)
4584                         return 1;
4585                 if (val && isdigit(c)) {
4586                         char *end;
4587                         int n = strtoul(arg, &end, 10);
4588                         if (*end)
4589                                 return 0;
4590                         *val = n;
4591                         return 1;
4592                 }
4593                 return 0;
4594         }
4595         if (c != '-')
4596                 return 0;
4597         arg++;
4598         eq = strchrnul(arg, '=');
4599         len = eq - arg;
4600         if (!len || strncmp(arg, arg_long, len))
4601                 return 0;
4602         if (*eq) {
4603                 int n;
4604                 char *end;
4605                 if (!isdigit(*++eq))
4606                         return 0;
4607                 n = strtoul(eq, &end, 10);
4608                 if (*end)
4609                         return 0;
4610                 *val = n;
4611         }
4612         return 1;
4613 }
4614
4615 static int diff_scoreopt_parse(const char *opt);
4616
4617 static inline int short_opt(char opt, const char **argv,
4618                             const char **optarg)
4619 {
4620         const char *arg = argv[0];
4621         if (arg[0] != '-' || arg[1] != opt)
4622                 return 0;
4623         if (arg[2] != '\0') {
4624                 *optarg = arg + 2;
4625                 return 1;
4626         }
4627         if (!argv[1])
4628                 die("Option '%c' requires a value", opt);
4629         *optarg = argv[1];
4630         return 2;
4631 }
4632
4633 int parse_long_opt(const char *opt, const char **argv,
4634                    const char **optarg)
4635 {
4636         const char *arg = argv[0];
4637         if (!skip_prefix(arg, "--", &arg))
4638                 return 0;
4639         if (!skip_prefix(arg, opt, &arg))
4640                 return 0;
4641         if (*arg == '=') { /* stuck form: --option=value */
4642                 *optarg = arg + 1;
4643                 return 1;
4644         }
4645         if (*arg != '\0')
4646                 return 0;
4647         /* separate form: --option value */
4648         if (!argv[1])
4649                 die("Option '--%s' requires a value", opt);
4650         *optarg = argv[1];
4651         return 2;
4652 }
4653
4654 static int stat_opt(struct diff_options *options, const char **av)
4655 {
4656         const char *arg = av[0];
4657         char *end;
4658         int width = options->stat_width;
4659         int name_width = options->stat_name_width;
4660         int graph_width = options->stat_graph_width;
4661         int count = options->stat_count;
4662         int argcount = 1;
4663
4664         if (!skip_prefix(arg, "--stat", &arg))
4665                 BUG("stat option does not begin with --stat: %s", arg);
4666         end = (char *)arg;
4667
4668         switch (*arg) {
4669         case '-':
4670                 if (skip_prefix(arg, "-width", &arg)) {
4671                         if (*arg == '=')
4672                                 width = strtoul(arg + 1, &end, 10);
4673                         else if (!*arg && !av[1])
4674                                 die_want_option("--stat-width");
4675                         else if (!*arg) {
4676                                 width = strtoul(av[1], &end, 10);
4677                                 argcount = 2;
4678                         }
4679                 } else if (skip_prefix(arg, "-name-width", &arg)) {
4680                         if (*arg == '=')
4681                                 name_width = strtoul(arg + 1, &end, 10);
4682                         else if (!*arg && !av[1])
4683                                 die_want_option("--stat-name-width");
4684                         else if (!*arg) {
4685                                 name_width = strtoul(av[1], &end, 10);
4686                                 argcount = 2;
4687                         }
4688                 } else if (skip_prefix(arg, "-graph-width", &arg)) {
4689                         if (*arg == '=')
4690                                 graph_width = strtoul(arg + 1, &end, 10);
4691                         else if (!*arg && !av[1])
4692                                 die_want_option("--stat-graph-width");
4693                         else if (!*arg) {
4694                                 graph_width = strtoul(av[1], &end, 10);
4695                                 argcount = 2;
4696                         }
4697                 } else if (skip_prefix(arg, "-count", &arg)) {
4698                         if (*arg == '=')
4699                                 count = strtoul(arg + 1, &end, 10);
4700                         else if (!*arg && !av[1])
4701                                 die_want_option("--stat-count");
4702                         else if (!*arg) {
4703                                 count = strtoul(av[1], &end, 10);
4704                                 argcount = 2;
4705                         }
4706                 }
4707                 break;
4708         case '=':
4709                 width = strtoul(arg+1, &end, 10);
4710                 if (*end == ',')
4711                         name_width = strtoul(end+1, &end, 10);
4712                 if (*end == ',')
4713                         count = strtoul(end+1, &end, 10);
4714         }
4715
4716         /* Important! This checks all the error cases! */
4717         if (*end)
4718                 return 0;
4719         options->output_format |= DIFF_FORMAT_DIFFSTAT;
4720         options->stat_name_width = name_width;
4721         options->stat_graph_width = graph_width;
4722         options->stat_width = width;
4723         options->stat_count = count;
4724         return argcount;
4725 }
4726
4727 static int parse_dirstat_opt(struct diff_options *options, const char *params)
4728 {
4729         struct strbuf errmsg = STRBUF_INIT;
4730         if (parse_dirstat_params(options, params, &errmsg))
4731                 die(_("Failed to parse --dirstat/-X option parameter:\n%s"),
4732                     errmsg.buf);
4733         strbuf_release(&errmsg);
4734         /*
4735          * The caller knows a dirstat-related option is given from the command
4736          * line; allow it to say "return this_function();"
4737          */
4738         options->output_format |= DIFF_FORMAT_DIRSTAT;
4739         return 1;
4740 }
4741
4742 static int parse_submodule_opt(struct diff_options *options, const char *value)
4743 {
4744         if (parse_submodule_params(options, value))
4745                 die(_("Failed to parse --submodule option parameter: '%s'"),
4746                         value);
4747         return 1;
4748 }
4749
4750 static const char diff_status_letters[] = {
4751         DIFF_STATUS_ADDED,
4752         DIFF_STATUS_COPIED,
4753         DIFF_STATUS_DELETED,
4754         DIFF_STATUS_MODIFIED,
4755         DIFF_STATUS_RENAMED,
4756         DIFF_STATUS_TYPE_CHANGED,
4757         DIFF_STATUS_UNKNOWN,
4758         DIFF_STATUS_UNMERGED,
4759         DIFF_STATUS_FILTER_AON,
4760         DIFF_STATUS_FILTER_BROKEN,
4761         '\0',
4762 };
4763
4764 static unsigned int filter_bit['Z' + 1];
4765
4766 static void prepare_filter_bits(void)
4767 {
4768         int i;
4769
4770         if (!filter_bit[DIFF_STATUS_ADDED]) {
4771                 for (i = 0; diff_status_letters[i]; i++)
4772                         filter_bit[(int) diff_status_letters[i]] = (1 << i);
4773         }
4774 }
4775
4776 static unsigned filter_bit_tst(char status, const struct diff_options *opt)
4777 {
4778         return opt->filter & filter_bit[(int) status];
4779 }
4780
4781 static int parse_diff_filter_opt(const char *optarg, struct diff_options *opt)
4782 {
4783         int i, optch;
4784
4785         prepare_filter_bits();
4786
4787         /*
4788          * If there is a negation e.g. 'd' in the input, and we haven't
4789          * initialized the filter field with another --diff-filter, start
4790          * from full set of bits, except for AON.
4791          */
4792         if (!opt->filter) {
4793                 for (i = 0; (optch = optarg[i]) != '\0'; i++) {
4794                         if (optch < 'a' || 'z' < optch)
4795                                 continue;
4796                         opt->filter = (1 << (ARRAY_SIZE(diff_status_letters) - 1)) - 1;
4797                         opt->filter &= ~filter_bit[DIFF_STATUS_FILTER_AON];
4798                         break;
4799                 }
4800         }
4801
4802         for (i = 0; (optch = optarg[i]) != '\0'; i++) {
4803                 unsigned int bit;
4804                 int negate;
4805
4806                 if ('a' <= optch && optch <= 'z') {
4807                         negate = 1;
4808                         optch = toupper(optch);
4809                 } else {
4810                         negate = 0;
4811                 }
4812
4813                 bit = (0 <= optch && optch <= 'Z') ? filter_bit[optch] : 0;
4814                 if (!bit)
4815                         return optarg[i];
4816                 if (negate)
4817                         opt->filter &= ~bit;
4818                 else
4819                         opt->filter |= bit;
4820         }
4821         return 0;
4822 }
4823
4824 static void enable_patch_output(int *fmt) {
4825         *fmt &= ~DIFF_FORMAT_NO_OUTPUT;
4826         *fmt |= DIFF_FORMAT_PATCH;
4827 }
4828
4829 static int parse_ws_error_highlight_opt(struct diff_options *opt, const char *arg)
4830 {
4831         int val = parse_ws_error_highlight(arg);
4832
4833         if (val < 0) {
4834                 error("unknown value after ws-error-highlight=%.*s",
4835                       -1 - val, arg);
4836                 return 0;
4837         }
4838         opt->ws_error_highlight = val;
4839         return 1;
4840 }
4841
4842 static int parse_objfind_opt(struct diff_options *opt, const char *arg)
4843 {
4844         struct object_id oid;
4845
4846         if (get_oid(arg, &oid))
4847                 return error("unable to resolve '%s'", arg);
4848
4849         if (!opt->objfind)
4850                 opt->objfind = xcalloc(1, sizeof(*opt->objfind));
4851
4852         opt->pickaxe_opts |= DIFF_PICKAXE_KIND_OBJFIND;
4853         opt->flags.recursive = 1;
4854         opt->flags.tree_in_recursive = 1;
4855         oidset_insert(opt->objfind, &oid);
4856         return 1;
4857 }
4858
4859 int diff_opt_parse(struct diff_options *options,
4860                    const char **av, int ac, const char *prefix)
4861 {
4862         const char *arg = av[0];
4863         const char *optarg;
4864         int argcount;
4865
4866         if (!prefix)
4867                 prefix = "";
4868
4869         /* Output format options */
4870         if (!strcmp(arg, "-p") || !strcmp(arg, "-u") || !strcmp(arg, "--patch")
4871             || opt_arg(arg, 'U', "unified", &options->context))
4872                 enable_patch_output(&options->output_format);
4873         else if (!strcmp(arg, "--raw"))
4874                 options->output_format |= DIFF_FORMAT_RAW;
4875         else if (!strcmp(arg, "--patch-with-raw")) {
4876                 enable_patch_output(&options->output_format);
4877                 options->output_format |= DIFF_FORMAT_RAW;
4878         } else if (!strcmp(arg, "--numstat"))
4879                 options->output_format |= DIFF_FORMAT_NUMSTAT;
4880         else if (!strcmp(arg, "--shortstat"))
4881                 options->output_format |= DIFF_FORMAT_SHORTSTAT;
4882         else if (skip_prefix(arg, "-X", &arg) ||
4883                  skip_to_optional_arg(arg, "--dirstat", &arg))
4884                 return parse_dirstat_opt(options, arg);
4885         else if (!strcmp(arg, "--cumulative"))
4886                 return parse_dirstat_opt(options, "cumulative");
4887         else if (skip_to_optional_arg(arg, "--dirstat-by-file", &arg)) {
4888                 parse_dirstat_opt(options, "files");
4889                 return parse_dirstat_opt(options, arg);
4890         }
4891         else if (!strcmp(arg, "--check"))
4892                 options->output_format |= DIFF_FORMAT_CHECKDIFF;
4893         else if (!strcmp(arg, "--summary"))
4894                 options->output_format |= DIFF_FORMAT_SUMMARY;
4895         else if (!strcmp(arg, "--patch-with-stat")) {
4896                 enable_patch_output(&options->output_format);
4897                 options->output_format |= DIFF_FORMAT_DIFFSTAT;
4898         } else if (!strcmp(arg, "--name-only"))
4899                 options->output_format |= DIFF_FORMAT_NAME;
4900         else if (!strcmp(arg, "--name-status"))
4901                 options->output_format |= DIFF_FORMAT_NAME_STATUS;
4902         else if (!strcmp(arg, "-s") || !strcmp(arg, "--no-patch"))
4903                 options->output_format |= DIFF_FORMAT_NO_OUTPUT;
4904         else if (starts_with(arg, "--stat"))
4905                 /* --stat, --stat-width, --stat-name-width, or --stat-count */
4906                 return stat_opt(options, av);
4907         else if (!strcmp(arg, "--compact-summary")) {
4908                  options->flags.stat_with_summary = 1;
4909                  options->output_format |= DIFF_FORMAT_DIFFSTAT;
4910         } else if (!strcmp(arg, "--no-compact-summary"))
4911                  options->flags.stat_with_summary = 0;
4912         else if (skip_prefix(arg, "--output-indicator-new=", &arg))
4913                 options->output_indicators[OUTPUT_INDICATOR_NEW] = arg[0];
4914         else if (skip_prefix(arg, "--output-indicator-old=", &arg))
4915                 options->output_indicators[OUTPUT_INDICATOR_OLD] = arg[0];
4916         else if (skip_prefix(arg, "--output-indicator-context=", &arg))
4917                 options->output_indicators[OUTPUT_INDICATOR_CONTEXT] = arg[0];
4918
4919         /* renames options */
4920         else if (starts_with(arg, "-B") ||
4921                  skip_to_optional_arg(arg, "--break-rewrites", NULL)) {
4922                 if ((options->break_opt = diff_scoreopt_parse(arg)) == -1)
4923                         return error("invalid argument to -B: %s", arg+2);
4924         }
4925         else if (starts_with(arg, "-M") ||
4926                  skip_to_optional_arg(arg, "--find-renames", NULL)) {
4927                 if ((options->rename_score = diff_scoreopt_parse(arg)) == -1)
4928                         return error("invalid argument to -M: %s", arg+2);
4929                 options->detect_rename = DIFF_DETECT_RENAME;
4930         }
4931         else if (!strcmp(arg, "-D") || !strcmp(arg, "--irreversible-delete")) {
4932                 options->irreversible_delete = 1;
4933         }
4934         else if (starts_with(arg, "-C") ||
4935                  skip_to_optional_arg(arg, "--find-copies", NULL)) {
4936                 if (options->detect_rename == DIFF_DETECT_COPY)
4937                         options->flags.find_copies_harder = 1;
4938                 if ((options->rename_score = diff_scoreopt_parse(arg)) == -1)
4939                         return error("invalid argument to -C: %s", arg+2);
4940                 options->detect_rename = DIFF_DETECT_COPY;
4941         }
4942         else if (!strcmp(arg, "--no-renames"))
4943                 options->detect_rename = 0;
4944         else if (!strcmp(arg, "--rename-empty"))
4945                 options->flags.rename_empty = 1;
4946         else if (!strcmp(arg, "--no-rename-empty"))
4947                 options->flags.rename_empty = 0;
4948         else if (skip_to_optional_arg_default(arg, "--relative", &arg, NULL)) {
4949                 options->flags.relative_name = 1;
4950                 if (arg)
4951                         options->prefix = arg;
4952         }
4953
4954         /* xdiff options */
4955         else if (!strcmp(arg, "--minimal"))
4956                 DIFF_XDL_SET(options, NEED_MINIMAL);
4957         else if (!strcmp(arg, "--no-minimal"))
4958                 DIFF_XDL_CLR(options, NEED_MINIMAL);
4959         else if (!strcmp(arg, "-w") || !strcmp(arg, "--ignore-all-space"))
4960                 DIFF_XDL_SET(options, IGNORE_WHITESPACE);
4961         else if (!strcmp(arg, "-b") || !strcmp(arg, "--ignore-space-change"))
4962                 DIFF_XDL_SET(options, IGNORE_WHITESPACE_CHANGE);
4963         else if (!strcmp(arg, "--ignore-space-at-eol"))
4964                 DIFF_XDL_SET(options, IGNORE_WHITESPACE_AT_EOL);
4965         else if (!strcmp(arg, "--ignore-cr-at-eol"))
4966                 DIFF_XDL_SET(options, IGNORE_CR_AT_EOL);
4967         else if (!strcmp(arg, "--ignore-blank-lines"))
4968                 DIFF_XDL_SET(options, IGNORE_BLANK_LINES);
4969         else if (!strcmp(arg, "--indent-heuristic"))
4970                 DIFF_XDL_SET(options, INDENT_HEURISTIC);
4971         else if (!strcmp(arg, "--no-indent-heuristic"))
4972                 DIFF_XDL_CLR(options, INDENT_HEURISTIC);
4973         else if (!strcmp(arg, "--patience")) {
4974                 int i;
4975                 options->xdl_opts = DIFF_WITH_ALG(options, PATIENCE_DIFF);
4976                 /*
4977                  * Both --patience and --anchored use PATIENCE_DIFF
4978                  * internally, so remove any anchors previously
4979                  * specified.
4980                  */
4981                 for (i = 0; i < options->anchors_nr; i++)
4982                         free(options->anchors[i]);
4983                 options->anchors_nr = 0;
4984         } else if (!strcmp(arg, "--histogram"))
4985                 options->xdl_opts = DIFF_WITH_ALG(options, HISTOGRAM_DIFF);
4986         else if ((argcount = parse_long_opt("diff-algorithm", av, &optarg))) {
4987                 long value = parse_algorithm_value(optarg);
4988                 if (value < 0)
4989                         return error("option diff-algorithm accepts \"myers\", "
4990                                      "\"minimal\", \"patience\" and \"histogram\"");
4991                 /* clear out previous settings */
4992                 DIFF_XDL_CLR(options, NEED_MINIMAL);
4993                 options->xdl_opts &= ~XDF_DIFF_ALGORITHM_MASK;
4994                 options->xdl_opts |= value;
4995                 return argcount;
4996         } else if (skip_prefix(arg, "--anchored=", &arg)) {
4997                 options->xdl_opts = DIFF_WITH_ALG(options, PATIENCE_DIFF);
4998                 ALLOC_GROW(options->anchors, options->anchors_nr + 1,
4999                            options->anchors_alloc);
5000                 options->anchors[options->anchors_nr++] = xstrdup(arg);
5001         }
5002
5003         /* flags options */
5004         else if (!strcmp(arg, "--binary")) {
5005                 enable_patch_output(&options->output_format);
5006                 options->flags.binary = 1;
5007         }
5008         else if (!strcmp(arg, "--full-index"))
5009                 options->flags.full_index = 1;
5010         else if (!strcmp(arg, "-a") || !strcmp(arg, "--text"))
5011                 options->flags.text = 1;
5012         else if (!strcmp(arg, "-R"))
5013                 options->flags.reverse_diff = 1;
5014         else if (!strcmp(arg, "--find-copies-harder"))
5015                 options->flags.find_copies_harder = 1;
5016         else if (!strcmp(arg, "--follow"))
5017                 options->flags.follow_renames = 1;
5018         else if (!strcmp(arg, "--no-follow")) {
5019                 options->flags.follow_renames = 0;
5020                 options->flags.default_follow_renames = 0;
5021         } else if (skip_to_optional_arg_default(arg, "--color", &arg, "always")) {
5022                 int value = git_config_colorbool(NULL, arg);
5023                 if (value < 0)
5024                         return error("option `color' expects \"always\", \"auto\", or \"never\"");
5025                 options->use_color = value;
5026         }
5027         else if (!strcmp(arg, "--no-color"))
5028                 options->use_color = 0;
5029         else if (!strcmp(arg, "--color-moved")) {
5030                 if (diff_color_moved_default)
5031                         options->color_moved = diff_color_moved_default;
5032                 if (options->color_moved == COLOR_MOVED_NO)
5033                         options->color_moved = COLOR_MOVED_DEFAULT;
5034         } else if (!strcmp(arg, "--no-color-moved"))
5035                 options->color_moved = COLOR_MOVED_NO;
5036         else if (skip_prefix(arg, "--color-moved=", &arg)) {
5037                 int cm = parse_color_moved(arg);
5038                 if (cm < 0)
5039                         die("bad --color-moved argument: %s", arg);
5040                 options->color_moved = cm;
5041         } else if (!strcmp(arg, "--no-color-moved-ws")) {
5042                 options->color_moved_ws_handling = 0;
5043         } else if (skip_prefix(arg, "--color-moved-ws=", &arg)) {
5044                 options->color_moved_ws_handling = parse_color_moved_ws(arg);
5045         } else if (skip_to_optional_arg_default(arg, "--color-words", &options->word_regex, NULL)) {
5046                 options->use_color = 1;
5047                 options->word_diff = DIFF_WORDS_COLOR;
5048         }
5049         else if (!strcmp(arg, "--word-diff")) {
5050                 if (options->word_diff == DIFF_WORDS_NONE)
5051                         options->word_diff = DIFF_WORDS_PLAIN;
5052         }
5053         else if (skip_prefix(arg, "--word-diff=", &arg)) {
5054                 if (!strcmp(arg, "plain"))
5055                         options->word_diff = DIFF_WORDS_PLAIN;
5056                 else if (!strcmp(arg, "color")) {
5057                         options->use_color = 1;
5058                         options->word_diff = DIFF_WORDS_COLOR;
5059                 }
5060                 else if (!strcmp(arg, "porcelain"))
5061                         options->word_diff = DIFF_WORDS_PORCELAIN;
5062                 else if (!strcmp(arg, "none"))
5063                         options->word_diff = DIFF_WORDS_NONE;
5064                 else
5065                         die("bad --word-diff argument: %s", arg);
5066         }
5067         else if ((argcount = parse_long_opt("word-diff-regex", av, &optarg))) {
5068                 if (options->word_diff == DIFF_WORDS_NONE)
5069                         options->word_diff = DIFF_WORDS_PLAIN;
5070                 options->word_regex = optarg;
5071                 return argcount;
5072         }
5073         else if (!strcmp(arg, "--exit-code"))
5074                 options->flags.exit_with_status = 1;
5075         else if (!strcmp(arg, "--quiet"))
5076                 options->flags.quick = 1;
5077         else if (!strcmp(arg, "--ext-diff"))
5078                 options->flags.allow_external = 1;
5079         else if (!strcmp(arg, "--no-ext-diff"))
5080                 options->flags.allow_external = 0;
5081         else if (!strcmp(arg, "--textconv")) {
5082                 options->flags.allow_textconv = 1;
5083                 options->flags.textconv_set_via_cmdline = 1;
5084         } else if (!strcmp(arg, "--no-textconv"))
5085                 options->flags.allow_textconv = 0;
5086         else if (skip_to_optional_arg_default(arg, "--ignore-submodules", &arg, "all")) {
5087                 options->flags.override_submodule_config = 1;
5088                 handle_ignore_submodules_arg(options, arg);
5089         } else if (skip_to_optional_arg_default(arg, "--submodule", &arg, "log"))
5090                 return parse_submodule_opt(options, arg);
5091         else if (skip_prefix(arg, "--ws-error-highlight=", &arg))
5092                 return parse_ws_error_highlight_opt(options, arg);
5093         else if (!strcmp(arg, "--ita-invisible-in-index"))
5094                 options->ita_invisible_in_index = 1;
5095         else if (!strcmp(arg, "--ita-visible-in-index"))
5096                 options->ita_invisible_in_index = 0;
5097
5098         /* misc options */
5099         else if (!strcmp(arg, "-z"))
5100                 options->line_termination = 0;
5101         else if ((argcount = short_opt('l', av, &optarg))) {
5102                 options->rename_limit = strtoul(optarg, NULL, 10);
5103                 return argcount;
5104         }
5105         else if ((argcount = short_opt('S', av, &optarg))) {
5106                 options->pickaxe = optarg;
5107                 options->pickaxe_opts |= DIFF_PICKAXE_KIND_S;
5108                 return argcount;
5109         } else if ((argcount = short_opt('G', av, &optarg))) {
5110                 options->pickaxe = optarg;
5111                 options->pickaxe_opts |= DIFF_PICKAXE_KIND_G;
5112                 return argcount;
5113         }
5114         else if (!strcmp(arg, "--pickaxe-all"))
5115                 options->pickaxe_opts |= DIFF_PICKAXE_ALL;
5116         else if (!strcmp(arg, "--pickaxe-regex"))
5117                 options->pickaxe_opts |= DIFF_PICKAXE_REGEX;
5118         else if ((argcount = short_opt('O', av, &optarg))) {
5119                 options->orderfile = prefix_filename(prefix, optarg);
5120                 return argcount;
5121         } else if (skip_prefix(arg, "--find-object=", &arg))
5122                 return parse_objfind_opt(options, arg);
5123         else if ((argcount = parse_long_opt("diff-filter", av, &optarg))) {
5124                 int offending = parse_diff_filter_opt(optarg, options);
5125                 if (offending)
5126                         die("unknown change class '%c' in --diff-filter=%s",
5127                             offending, optarg);
5128                 return argcount;
5129         }
5130         else if (!strcmp(arg, "--no-abbrev"))
5131                 options->abbrev = 0;
5132         else if (!strcmp(arg, "--abbrev"))
5133                 options->abbrev = DEFAULT_ABBREV;
5134         else if (skip_prefix(arg, "--abbrev=", &arg)) {
5135                 options->abbrev = strtoul(arg, NULL, 10);
5136                 if (options->abbrev < MINIMUM_ABBREV)
5137                         options->abbrev = MINIMUM_ABBREV;
5138                 else if (the_hash_algo->hexsz < options->abbrev)
5139                         options->abbrev = the_hash_algo->hexsz;
5140         }
5141         else if ((argcount = parse_long_opt("src-prefix", av, &optarg))) {
5142                 options->a_prefix = optarg;
5143                 return argcount;
5144         }
5145         else if ((argcount = parse_long_opt("line-prefix", av, &optarg))) {
5146                 options->line_prefix = optarg;
5147                 options->line_prefix_length = strlen(options->line_prefix);
5148                 graph_setup_line_prefix(options);
5149                 return argcount;
5150         }
5151         else if ((argcount = parse_long_opt("dst-prefix", av, &optarg))) {
5152                 options->b_prefix = optarg;
5153                 return argcount;
5154         }
5155         else if (!strcmp(arg, "--no-prefix"))
5156                 options->a_prefix = options->b_prefix = "";
5157         else if (opt_arg(arg, '\0', "inter-hunk-context",
5158                          &options->interhunkcontext))
5159                 ;
5160         else if (!strcmp(arg, "-W"))
5161                 options->flags.funccontext = 1;
5162         else if (!strcmp(arg, "--function-context"))
5163                 options->flags.funccontext = 1;
5164         else if (!strcmp(arg, "--no-function-context"))
5165                 options->flags.funccontext = 0;
5166         else if ((argcount = parse_long_opt("output", av, &optarg))) {
5167                 char *path = prefix_filename(prefix, optarg);
5168                 options->file = xfopen(path, "w");
5169                 options->close_file = 1;
5170                 if (options->use_color != GIT_COLOR_ALWAYS)
5171                         options->use_color = GIT_COLOR_NEVER;
5172                 free(path);
5173                 return argcount;
5174         } else
5175                 return 0;
5176         return 1;
5177 }
5178
5179 int parse_rename_score(const char **cp_p)
5180 {
5181         unsigned long num, scale;
5182         int ch, dot;
5183         const char *cp = *cp_p;
5184
5185         num = 0;
5186         scale = 1;
5187         dot = 0;
5188         for (;;) {
5189                 ch = *cp;
5190                 if ( !dot && ch == '.' ) {
5191                         scale = 1;
5192                         dot = 1;
5193                 } else if ( ch == '%' ) {
5194                         scale = dot ? scale*100 : 100;
5195                         cp++;   /* % is always at the end */
5196                         break;
5197                 } else if ( ch >= '0' && ch <= '9' ) {
5198                         if ( scale < 100000 ) {
5199                                 scale *= 10;
5200                                 num = (num*10) + (ch-'0');
5201                         }
5202                 } else {
5203                         break;
5204                 }
5205                 cp++;
5206         }
5207         *cp_p = cp;
5208
5209         /* user says num divided by scale and we say internally that
5210          * is MAX_SCORE * num / scale.
5211          */
5212         return (int)((num >= scale) ? MAX_SCORE : (MAX_SCORE * num / scale));
5213 }
5214
5215 static int diff_scoreopt_parse(const char *opt)
5216 {
5217         int opt1, opt2, cmd;
5218
5219         if (*opt++ != '-')
5220                 return -1;
5221         cmd = *opt++;
5222         if (cmd == '-') {
5223                 /* convert the long-form arguments into short-form versions */
5224                 if (skip_prefix(opt, "break-rewrites", &opt)) {
5225                         if (*opt == 0 || *opt++ == '=')
5226                                 cmd = 'B';
5227                 } else if (skip_prefix(opt, "find-copies", &opt)) {
5228                         if (*opt == 0 || *opt++ == '=')
5229                                 cmd = 'C';
5230                 } else if (skip_prefix(opt, "find-renames", &opt)) {
5231                         if (*opt == 0 || *opt++ == '=')
5232                                 cmd = 'M';
5233                 }
5234         }
5235         if (cmd != 'M' && cmd != 'C' && cmd != 'B')
5236                 return -1; /* that is not a -M, -C, or -B option */
5237
5238         opt1 = parse_rename_score(&opt);
5239         if (cmd != 'B')
5240                 opt2 = 0;
5241         else {
5242                 if (*opt == 0)
5243                         opt2 = 0;
5244                 else if (*opt != '/')
5245                         return -1; /* we expect -B80/99 or -B80 */
5246                 else {
5247                         opt++;
5248                         opt2 = parse_rename_score(&opt);
5249                 }
5250         }
5251         if (*opt != 0)
5252                 return -1;
5253         return opt1 | (opt2 << 16);
5254 }
5255
5256 struct diff_queue_struct diff_queued_diff;
5257
5258 void diff_q(struct diff_queue_struct *queue, struct diff_filepair *dp)
5259 {
5260         ALLOC_GROW(queue->queue, queue->nr + 1, queue->alloc);
5261         queue->queue[queue->nr++] = dp;
5262 }
5263
5264 struct diff_filepair *diff_queue(struct diff_queue_struct *queue,
5265                                  struct diff_filespec *one,
5266                                  struct diff_filespec *two)
5267 {
5268         struct diff_filepair *dp = xcalloc(1, sizeof(*dp));
5269         dp->one = one;
5270         dp->two = two;
5271         if (queue)
5272                 diff_q(queue, dp);
5273         return dp;
5274 }
5275
5276 void diff_free_filepair(struct diff_filepair *p)
5277 {
5278         free_filespec(p->one);
5279         free_filespec(p->two);
5280         free(p);
5281 }
5282
5283 const char *diff_aligned_abbrev(const struct object_id *oid, int len)
5284 {
5285         int abblen;
5286         const char *abbrev;
5287
5288         /* Do we want all 40 hex characters? */
5289         if (len == the_hash_algo->hexsz)
5290                 return oid_to_hex(oid);
5291
5292         /* An abbreviated value is fine, possibly followed by an ellipsis. */
5293         abbrev = diff_abbrev_oid(oid, len);
5294
5295         if (!print_sha1_ellipsis())
5296                 return abbrev;
5297
5298         abblen = strlen(abbrev);
5299
5300         /*
5301          * In well-behaved cases, where the abbreviated result is the
5302          * same as the requested length, append three dots after the
5303          * abbreviation (hence the whole logic is limited to the case
5304          * where abblen < 37); when the actual abbreviated result is a
5305          * bit longer than the requested length, we reduce the number
5306          * of dots so that they match the well-behaved ones.  However,
5307          * if the actual abbreviation is longer than the requested
5308          * length by more than three, we give up on aligning, and add
5309          * three dots anyway, to indicate that the output is not the
5310          * full object name.  Yes, this may be suboptimal, but this
5311          * appears only in "diff --raw --abbrev" output and it is not
5312          * worth the effort to change it now.  Note that this would
5313          * likely to work fine when the automatic sizing of default
5314          * abbreviation length is used--we would be fed -1 in "len" in
5315          * that case, and will end up always appending three-dots, but
5316          * the automatic sizing is supposed to give abblen that ensures
5317          * uniqueness across all objects (statistically speaking).
5318          */
5319         if (abblen < the_hash_algo->hexsz - 3) {
5320                 static char hex[GIT_MAX_HEXSZ + 1];
5321                 if (len < abblen && abblen <= len + 2)
5322                         xsnprintf(hex, sizeof(hex), "%s%.*s", abbrev, len+3-abblen, "..");
5323                 else
5324                         xsnprintf(hex, sizeof(hex), "%s...", abbrev);
5325                 return hex;
5326         }
5327
5328         return oid_to_hex(oid);
5329 }
5330
5331 static void diff_flush_raw(struct diff_filepair *p, struct diff_options *opt)
5332 {
5333         int line_termination = opt->line_termination;
5334         int inter_name_termination = line_termination ? '\t' : '\0';
5335
5336         fprintf(opt->file, "%s", diff_line_prefix(opt));
5337         if (!(opt->output_format & DIFF_FORMAT_NAME_STATUS)) {
5338                 fprintf(opt->file, ":%06o %06o %s ", p->one->mode, p->two->mode,
5339                         diff_aligned_abbrev(&p->one->oid, opt->abbrev));
5340                 fprintf(opt->file, "%s ",
5341                         diff_aligned_abbrev(&p->two->oid, opt->abbrev));
5342         }
5343         if (p->score) {
5344                 fprintf(opt->file, "%c%03d%c", p->status, similarity_index(p),
5345                         inter_name_termination);
5346         } else {
5347                 fprintf(opt->file, "%c%c", p->status, inter_name_termination);
5348         }
5349
5350         if (p->status == DIFF_STATUS_COPIED ||
5351             p->status == DIFF_STATUS_RENAMED) {
5352                 const char *name_a, *name_b;
5353                 name_a = p->one->path;
5354                 name_b = p->two->path;
5355                 strip_prefix(opt->prefix_length, &name_a, &name_b);
5356                 write_name_quoted(name_a, opt->file, inter_name_termination);
5357                 write_name_quoted(name_b, opt->file, line_termination);
5358         } else {
5359                 const char *name_a, *name_b;
5360                 name_a = p->one->mode ? p->one->path : p->two->path;
5361                 name_b = NULL;
5362                 strip_prefix(opt->prefix_length, &name_a, &name_b);
5363                 write_name_quoted(name_a, opt->file, line_termination);
5364         }
5365 }
5366
5367 int diff_unmodified_pair(struct diff_filepair *p)
5368 {
5369         /* This function is written stricter than necessary to support
5370          * the currently implemented transformers, but the idea is to
5371          * let transformers to produce diff_filepairs any way they want,
5372          * and filter and clean them up here before producing the output.
5373          */
5374         struct diff_filespec *one = p->one, *two = p->two;
5375
5376         if (DIFF_PAIR_UNMERGED(p))
5377                 return 0; /* unmerged is interesting */
5378
5379         /* deletion, addition, mode or type change
5380          * and rename are all interesting.
5381          */
5382         if (DIFF_FILE_VALID(one) != DIFF_FILE_VALID(two) ||
5383             DIFF_PAIR_MODE_CHANGED(p) ||
5384             strcmp(one->path, two->path))
5385                 return 0;
5386
5387         /* both are valid and point at the same path.  that is, we are
5388          * dealing with a change.
5389          */
5390         if (one->oid_valid && two->oid_valid &&
5391             oideq(&one->oid, &two->oid) &&
5392             !one->dirty_submodule && !two->dirty_submodule)
5393                 return 1; /* no change */
5394         if (!one->oid_valid && !two->oid_valid)
5395                 return 1; /* both look at the same file on the filesystem. */
5396         return 0;
5397 }
5398
5399 static void diff_flush_patch(struct diff_filepair *p, struct diff_options *o)
5400 {
5401         if (diff_unmodified_pair(p))
5402                 return;
5403
5404         if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
5405             (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
5406                 return; /* no tree diffs in patch format */
5407
5408         run_diff(p, o);
5409 }
5410
5411 static void diff_flush_stat(struct diff_filepair *p, struct diff_options *o,
5412                             struct diffstat_t *diffstat)
5413 {
5414         if (diff_unmodified_pair(p))
5415                 return;
5416
5417         if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
5418             (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
5419                 return; /* no useful stat for tree diffs */
5420
5421         run_diffstat(p, o, diffstat);
5422 }
5423
5424 static void diff_flush_checkdiff(struct diff_filepair *p,
5425                 struct diff_options *o)
5426 {
5427         if (diff_unmodified_pair(p))
5428                 return;
5429
5430         if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
5431             (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
5432                 return; /* nothing to check in tree diffs */
5433
5434         run_checkdiff(p, o);
5435 }
5436
5437 int diff_queue_is_empty(void)
5438 {
5439         struct diff_queue_struct *q = &diff_queued_diff;
5440         int i;
5441         for (i = 0; i < q->nr; i++)
5442                 if (!diff_unmodified_pair(q->queue[i]))
5443                         return 0;
5444         return 1;
5445 }
5446
5447 #if DIFF_DEBUG
5448 void diff_debug_filespec(struct diff_filespec *s, int x, const char *one)
5449 {
5450         fprintf(stderr, "queue[%d] %s (%s) %s %06o %s\n",
5451                 x, one ? one : "",
5452                 s->path,
5453                 DIFF_FILE_VALID(s) ? "valid" : "invalid",
5454                 s->mode,
5455                 s->oid_valid ? oid_to_hex(&s->oid) : "");
5456         fprintf(stderr, "queue[%d] %s size %lu\n",
5457                 x, one ? one : "",
5458                 s->size);
5459 }
5460
5461 void diff_debug_filepair(const struct diff_filepair *p, int i)
5462 {
5463         diff_debug_filespec(p->one, i, "one");
5464         diff_debug_filespec(p->two, i, "two");
5465         fprintf(stderr, "score %d, status %c rename_used %d broken %d\n",
5466                 p->score, p->status ? p->status : '?',
5467                 p->one->rename_used, p->broken_pair);
5468 }
5469
5470 void diff_debug_queue(const char *msg, struct diff_queue_struct *q)
5471 {
5472         int i;
5473         if (msg)
5474                 fprintf(stderr, "%s\n", msg);
5475         fprintf(stderr, "q->nr = %d\n", q->nr);
5476         for (i = 0; i < q->nr; i++) {
5477                 struct diff_filepair *p = q->queue[i];
5478                 diff_debug_filepair(p, i);
5479         }
5480 }
5481 #endif
5482
5483 static void diff_resolve_rename_copy(void)
5484 {
5485         int i;
5486         struct diff_filepair *p;
5487         struct diff_queue_struct *q = &diff_queued_diff;
5488
5489         diff_debug_queue("resolve-rename-copy", q);
5490
5491         for (i = 0; i < q->nr; i++) {
5492                 p = q->queue[i];
5493                 p->status = 0; /* undecided */
5494                 if (DIFF_PAIR_UNMERGED(p))
5495                         p->status = DIFF_STATUS_UNMERGED;
5496                 else if (!DIFF_FILE_VALID(p->one))
5497                         p->status = DIFF_STATUS_ADDED;
5498                 else if (!DIFF_FILE_VALID(p->two))
5499                         p->status = DIFF_STATUS_DELETED;
5500                 else if (DIFF_PAIR_TYPE_CHANGED(p))
5501                         p->status = DIFF_STATUS_TYPE_CHANGED;
5502
5503                 /* from this point on, we are dealing with a pair
5504                  * whose both sides are valid and of the same type, i.e.
5505                  * either in-place edit or rename/copy edit.
5506                  */
5507                 else if (DIFF_PAIR_RENAME(p)) {
5508                         /*
5509                          * A rename might have re-connected a broken
5510                          * pair up, causing the pathnames to be the
5511                          * same again. If so, that's not a rename at
5512                          * all, just a modification..
5513                          *
5514                          * Otherwise, see if this source was used for
5515                          * multiple renames, in which case we decrement
5516                          * the count, and call it a copy.
5517                          */
5518                         if (!strcmp(p->one->path, p->two->path))
5519                                 p->status = DIFF_STATUS_MODIFIED;
5520                         else if (--p->one->rename_used > 0)
5521                                 p->status = DIFF_STATUS_COPIED;
5522                         else
5523                                 p->status = DIFF_STATUS_RENAMED;
5524                 }
5525                 else if (!oideq(&p->one->oid, &p->two->oid) ||
5526                          p->one->mode != p->two->mode ||
5527                          p->one->dirty_submodule ||
5528                          p->two->dirty_submodule ||
5529                          is_null_oid(&p->one->oid))
5530                         p->status = DIFF_STATUS_MODIFIED;
5531                 else {
5532                         /* This is a "no-change" entry and should not
5533                          * happen anymore, but prepare for broken callers.
5534                          */
5535                         error("feeding unmodified %s to diffcore",
5536                               p->one->path);
5537                         p->status = DIFF_STATUS_UNKNOWN;
5538                 }
5539         }
5540         diff_debug_queue("resolve-rename-copy done", q);
5541 }
5542
5543 static int check_pair_status(struct diff_filepair *p)
5544 {
5545         switch (p->status) {
5546         case DIFF_STATUS_UNKNOWN:
5547                 return 0;
5548         case 0:
5549                 die("internal error in diff-resolve-rename-copy");
5550         default:
5551                 return 1;
5552         }
5553 }
5554
5555 static void flush_one_pair(struct diff_filepair *p, struct diff_options *opt)
5556 {
5557         int fmt = opt->output_format;
5558
5559         if (fmt & DIFF_FORMAT_CHECKDIFF)
5560                 diff_flush_checkdiff(p, opt);
5561         else if (fmt & (DIFF_FORMAT_RAW | DIFF_FORMAT_NAME_STATUS))
5562                 diff_flush_raw(p, opt);
5563         else if (fmt & DIFF_FORMAT_NAME) {
5564                 const char *name_a, *name_b;
5565                 name_a = p->two->path;
5566                 name_b = NULL;
5567                 strip_prefix(opt->prefix_length, &name_a, &name_b);
5568                 fprintf(opt->file, "%s", diff_line_prefix(opt));
5569                 write_name_quoted(name_a, opt->file, opt->line_termination);
5570         }
5571 }
5572
5573 static void show_file_mode_name(struct diff_options *opt, const char *newdelete, struct diff_filespec *fs)
5574 {
5575         struct strbuf sb = STRBUF_INIT;
5576         if (fs->mode)
5577                 strbuf_addf(&sb, " %s mode %06o ", newdelete, fs->mode);
5578         else
5579                 strbuf_addf(&sb, " %s ", newdelete);
5580
5581         quote_c_style(fs->path, &sb, NULL, 0);
5582         strbuf_addch(&sb, '\n');
5583         emit_diff_symbol(opt, DIFF_SYMBOL_SUMMARY,
5584                          sb.buf, sb.len, 0);
5585         strbuf_release(&sb);
5586 }
5587
5588 static void show_mode_change(struct diff_options *opt, struct diff_filepair *p,
5589                 int show_name)
5590 {
5591         if (p->one->mode && p->two->mode && p->one->mode != p->two->mode) {
5592                 struct strbuf sb = STRBUF_INIT;
5593                 strbuf_addf(&sb, " mode change %06o => %06o",
5594                             p->one->mode, p->two->mode);
5595                 if (show_name) {
5596                         strbuf_addch(&sb, ' ');
5597                         quote_c_style(p->two->path, &sb, NULL, 0);
5598                 }
5599                 strbuf_addch(&sb, '\n');
5600                 emit_diff_symbol(opt, DIFF_SYMBOL_SUMMARY,
5601                                  sb.buf, sb.len, 0);
5602                 strbuf_release(&sb);
5603         }
5604 }
5605
5606 static void show_rename_copy(struct diff_options *opt, const char *renamecopy,
5607                 struct diff_filepair *p)
5608 {
5609         struct strbuf sb = STRBUF_INIT;
5610         struct strbuf names = STRBUF_INIT;
5611
5612         pprint_rename(&names, p->one->path, p->two->path);
5613         strbuf_addf(&sb, " %s %s (%d%%)\n",
5614                     renamecopy, names.buf, similarity_index(p));
5615         strbuf_release(&names);
5616         emit_diff_symbol(opt, DIFF_SYMBOL_SUMMARY,
5617                                  sb.buf, sb.len, 0);
5618         show_mode_change(opt, p, 0);
5619         strbuf_release(&sb);
5620 }
5621
5622 static void diff_summary(struct diff_options *opt, struct diff_filepair *p)
5623 {
5624         switch(p->status) {
5625         case DIFF_STATUS_DELETED:
5626                 show_file_mode_name(opt, "delete", p->one);
5627                 break;
5628         case DIFF_STATUS_ADDED:
5629                 show_file_mode_name(opt, "create", p->two);
5630                 break;
5631         case DIFF_STATUS_COPIED:
5632                 show_rename_copy(opt, "copy", p);
5633                 break;
5634         case DIFF_STATUS_RENAMED:
5635                 show_rename_copy(opt, "rename", p);
5636                 break;
5637         default:
5638                 if (p->score) {
5639                         struct strbuf sb = STRBUF_INIT;
5640                         strbuf_addstr(&sb, " rewrite ");
5641                         quote_c_style(p->two->path, &sb, NULL, 0);
5642                         strbuf_addf(&sb, " (%d%%)\n", similarity_index(p));
5643                         emit_diff_symbol(opt, DIFF_SYMBOL_SUMMARY,
5644                                          sb.buf, sb.len, 0);
5645                         strbuf_release(&sb);
5646                 }
5647                 show_mode_change(opt, p, !p->score);
5648                 break;
5649         }
5650 }
5651
5652 struct patch_id_t {
5653         git_SHA_CTX *ctx;
5654         int patchlen;
5655 };
5656
5657 static int remove_space(char *line, int len)
5658 {
5659         int i;
5660         char *dst = line;
5661         unsigned char c;
5662
5663         for (i = 0; i < len; i++)
5664                 if (!isspace((c = line[i])))
5665                         *dst++ = c;
5666
5667         return dst - line;
5668 }
5669
5670 static void patch_id_consume(void *priv, char *line, unsigned long len)
5671 {
5672         struct patch_id_t *data = priv;
5673         int new_len;
5674
5675         new_len = remove_space(line, len);
5676
5677         git_SHA1_Update(data->ctx, line, new_len);
5678         data->patchlen += new_len;
5679 }
5680
5681 static void patch_id_add_string(git_SHA_CTX *ctx, const char *str)
5682 {
5683         git_SHA1_Update(ctx, str, strlen(str));
5684 }
5685
5686 static void patch_id_add_mode(git_SHA_CTX *ctx, unsigned mode)
5687 {
5688         /* large enough for 2^32 in octal */
5689         char buf[12];
5690         int len = xsnprintf(buf, sizeof(buf), "%06o", mode);
5691         git_SHA1_Update(ctx, buf, len);
5692 }
5693
5694 /* returns 0 upon success, and writes result into sha1 */
5695 static int diff_get_patch_id(struct diff_options *options, struct object_id *oid, int diff_header_only)
5696 {
5697         struct diff_queue_struct *q = &diff_queued_diff;
5698         int i;
5699         git_SHA_CTX ctx;
5700         struct patch_id_t data;
5701
5702         git_SHA1_Init(&ctx);
5703         memset(&data, 0, sizeof(struct patch_id_t));
5704         data.ctx = &ctx;
5705
5706         for (i = 0; i < q->nr; i++) {
5707                 xpparam_t xpp;
5708                 xdemitconf_t xecfg;
5709                 mmfile_t mf1, mf2;
5710                 struct diff_filepair *p = q->queue[i];
5711                 int len1, len2;
5712
5713                 memset(&xpp, 0, sizeof(xpp));
5714                 memset(&xecfg, 0, sizeof(xecfg));
5715                 if (p->status == 0)
5716                         return error("internal diff status error");
5717                 if (p->status == DIFF_STATUS_UNKNOWN)
5718                         continue;
5719                 if (diff_unmodified_pair(p))
5720                         continue;
5721                 if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
5722                     (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
5723                         continue;
5724                 if (DIFF_PAIR_UNMERGED(p))
5725                         continue;
5726
5727                 diff_fill_oid_info(p->one, options->repo->index);
5728                 diff_fill_oid_info(p->two, options->repo->index);
5729
5730                 len1 = remove_space(p->one->path, strlen(p->one->path));
5731                 len2 = remove_space(p->two->path, strlen(p->two->path));
5732                 patch_id_add_string(&ctx, "diff--git");
5733                 patch_id_add_string(&ctx, "a/");
5734                 git_SHA1_Update(&ctx, p->one->path, len1);
5735                 patch_id_add_string(&ctx, "b/");
5736                 git_SHA1_Update(&ctx, p->two->path, len2);
5737
5738                 if (p->one->mode == 0) {
5739                         patch_id_add_string(&ctx, "newfilemode");
5740                         patch_id_add_mode(&ctx, p->two->mode);
5741                         patch_id_add_string(&ctx, "---/dev/null");
5742                         patch_id_add_string(&ctx, "+++b/");
5743                         git_SHA1_Update(&ctx, p->two->path, len2);
5744                 } else if (p->two->mode == 0) {
5745                         patch_id_add_string(&ctx, "deletedfilemode");
5746                         patch_id_add_mode(&ctx, p->one->mode);
5747                         patch_id_add_string(&ctx, "---a/");
5748                         git_SHA1_Update(&ctx, p->one->path, len1);
5749                         patch_id_add_string(&ctx, "+++/dev/null");
5750                 } else {
5751                         patch_id_add_string(&ctx, "---a/");
5752                         git_SHA1_Update(&ctx, p->one->path, len1);
5753                         patch_id_add_string(&ctx, "+++b/");
5754                         git_SHA1_Update(&ctx, p->two->path, len2);
5755                 }
5756
5757                 if (diff_header_only)
5758                         continue;
5759
5760                 if (fill_mmfile(options->repo, &mf1, p->one) < 0 ||
5761                     fill_mmfile(options->repo, &mf2, p->two) < 0)
5762                         return error("unable to read files to diff");
5763
5764                 if (diff_filespec_is_binary(options->repo, p->one) ||
5765                     diff_filespec_is_binary(options->repo, p->two)) {
5766                         git_SHA1_Update(&ctx, oid_to_hex(&p->one->oid),
5767                                         GIT_SHA1_HEXSZ);
5768                         git_SHA1_Update(&ctx, oid_to_hex(&p->two->oid),
5769                                         GIT_SHA1_HEXSZ);
5770                         continue;
5771                 }
5772
5773                 xpp.flags = 0;
5774                 xecfg.ctxlen = 3;
5775                 xecfg.flags = 0;
5776                 if (xdi_diff_outf(&mf1, &mf2, discard_hunk_line,
5777                                   patch_id_consume, &data, &xpp, &xecfg))
5778                         return error("unable to generate patch-id diff for %s",
5779                                      p->one->path);
5780         }
5781
5782         git_SHA1_Final(oid->hash, &ctx);
5783         return 0;
5784 }
5785
5786 int diff_flush_patch_id(struct diff_options *options, struct object_id *oid, int diff_header_only)
5787 {
5788         struct diff_queue_struct *q = &diff_queued_diff;
5789         int i;
5790         int result = diff_get_patch_id(options, oid, diff_header_only);
5791
5792         for (i = 0; i < q->nr; i++)
5793                 diff_free_filepair(q->queue[i]);
5794
5795         free(q->queue);
5796         DIFF_QUEUE_CLEAR(q);
5797
5798         return result;
5799 }
5800
5801 static int is_summary_empty(const struct diff_queue_struct *q)
5802 {
5803         int i;
5804
5805         for (i = 0; i < q->nr; i++) {
5806                 const struct diff_filepair *p = q->queue[i];
5807
5808                 switch (p->status) {
5809                 case DIFF_STATUS_DELETED:
5810                 case DIFF_STATUS_ADDED:
5811                 case DIFF_STATUS_COPIED:
5812                 case DIFF_STATUS_RENAMED:
5813                         return 0;
5814                 default:
5815                         if (p->score)
5816                                 return 0;
5817                         if (p->one->mode && p->two->mode &&
5818                             p->one->mode != p->two->mode)
5819                                 return 0;
5820                         break;
5821                 }
5822         }
5823         return 1;
5824 }
5825
5826 static const char rename_limit_warning[] =
5827 N_("inexact rename detection was skipped due to too many files.");
5828
5829 static const char degrade_cc_to_c_warning[] =
5830 N_("only found copies from modified paths due to too many files.");
5831
5832 static const char rename_limit_advice[] =
5833 N_("you may want to set your %s variable to at least "
5834    "%d and retry the command.");
5835
5836 void diff_warn_rename_limit(const char *varname, int needed, int degraded_cc)
5837 {
5838         fflush(stdout);
5839         if (degraded_cc)
5840                 warning(_(degrade_cc_to_c_warning));
5841         else if (needed)
5842                 warning(_(rename_limit_warning));
5843         else
5844                 return;
5845         if (0 < needed)
5846                 warning(_(rename_limit_advice), varname, needed);
5847 }
5848
5849 static void diff_flush_patch_all_file_pairs(struct diff_options *o)
5850 {
5851         int i;
5852         static struct emitted_diff_symbols esm = EMITTED_DIFF_SYMBOLS_INIT;
5853         struct diff_queue_struct *q = &diff_queued_diff;
5854
5855         if (WSEH_NEW & WS_RULE_MASK)
5856                 BUG("WS rules bit mask overlaps with diff symbol flags");
5857
5858         if (o->color_moved)
5859                 o->emitted_symbols = &esm;
5860
5861         for (i = 0; i < q->nr; i++) {
5862                 struct diff_filepair *p = q->queue[i];
5863                 if (check_pair_status(p))
5864                         diff_flush_patch(p, o);
5865         }
5866
5867         if (o->emitted_symbols) {
5868                 if (o->color_moved) {
5869                         struct hashmap add_lines, del_lines;
5870
5871                         if (o->color_moved_ws_handling &
5872                             COLOR_MOVED_WS_ALLOW_INDENTATION_CHANGE)
5873                                 o->color_moved_ws_handling |= XDF_IGNORE_WHITESPACE;
5874
5875                         hashmap_init(&del_lines, moved_entry_cmp, o, 0);
5876                         hashmap_init(&add_lines, moved_entry_cmp, o, 0);
5877
5878                         add_lines_to_move_detection(o, &add_lines, &del_lines);
5879                         mark_color_as_moved(o, &add_lines, &del_lines);
5880                         if (o->color_moved == COLOR_MOVED_ZEBRA_DIM)
5881                                 dim_moved_lines(o);
5882
5883                         hashmap_free(&add_lines, 1);
5884                         hashmap_free(&del_lines, 1);
5885                 }
5886
5887                 for (i = 0; i < esm.nr; i++)
5888                         emit_diff_symbol_from_struct(o, &esm.buf[i]);
5889
5890                 for (i = 0; i < esm.nr; i++)
5891                         free((void *)esm.buf[i].line);
5892         }
5893         esm.nr = 0;
5894 }
5895
5896 void diff_flush(struct diff_options *options)
5897 {
5898         struct diff_queue_struct *q = &diff_queued_diff;
5899         int i, output_format = options->output_format;
5900         int separator = 0;
5901         int dirstat_by_line = 0;
5902
5903         /*
5904          * Order: raw, stat, summary, patch
5905          * or:    name/name-status/checkdiff (other bits clear)
5906          */
5907         if (!q->nr)
5908                 goto free_queue;
5909
5910         if (output_format & (DIFF_FORMAT_RAW |
5911                              DIFF_FORMAT_NAME |
5912                              DIFF_FORMAT_NAME_STATUS |
5913                              DIFF_FORMAT_CHECKDIFF)) {
5914                 for (i = 0; i < q->nr; i++) {
5915                         struct diff_filepair *p = q->queue[i];
5916                         if (check_pair_status(p))
5917                                 flush_one_pair(p, options);
5918                 }
5919                 separator++;
5920         }
5921
5922         if (output_format & DIFF_FORMAT_DIRSTAT && options->flags.dirstat_by_line)
5923                 dirstat_by_line = 1;
5924
5925         if (output_format & (DIFF_FORMAT_DIFFSTAT|DIFF_FORMAT_SHORTSTAT|DIFF_FORMAT_NUMSTAT) ||
5926             dirstat_by_line) {
5927                 struct diffstat_t diffstat;
5928
5929                 memset(&diffstat, 0, sizeof(struct diffstat_t));
5930                 for (i = 0; i < q->nr; i++) {
5931                         struct diff_filepair *p = q->queue[i];
5932                         if (check_pair_status(p))
5933                                 diff_flush_stat(p, options, &diffstat);
5934                 }
5935                 if (output_format & DIFF_FORMAT_NUMSTAT)
5936                         show_numstat(&diffstat, options);
5937                 if (output_format & DIFF_FORMAT_DIFFSTAT)
5938                         show_stats(&diffstat, options);
5939                 if (output_format & DIFF_FORMAT_SHORTSTAT)
5940                         show_shortstats(&diffstat, options);
5941                 if (output_format & DIFF_FORMAT_DIRSTAT && dirstat_by_line)
5942                         show_dirstat_by_line(&diffstat, options);
5943                 free_diffstat_info(&diffstat);
5944                 separator++;
5945         }
5946         if ((output_format & DIFF_FORMAT_DIRSTAT) && !dirstat_by_line)
5947                 show_dirstat(options);
5948
5949         if (output_format & DIFF_FORMAT_SUMMARY && !is_summary_empty(q)) {
5950                 for (i = 0; i < q->nr; i++) {
5951                         diff_summary(options, q->queue[i]);
5952                 }
5953                 separator++;
5954         }
5955
5956         if (output_format & DIFF_FORMAT_NO_OUTPUT &&
5957             options->flags.exit_with_status &&
5958             options->flags.diff_from_contents) {
5959                 /*
5960                  * run diff_flush_patch for the exit status. setting
5961                  * options->file to /dev/null should be safe, because we
5962                  * aren't supposed to produce any output anyway.
5963                  */
5964                 if (options->close_file)
5965                         fclose(options->file);
5966                 options->file = xfopen("/dev/null", "w");
5967                 options->close_file = 1;
5968                 options->color_moved = 0;
5969                 for (i = 0; i < q->nr; i++) {
5970                         struct diff_filepair *p = q->queue[i];
5971                         if (check_pair_status(p))
5972                                 diff_flush_patch(p, options);
5973                         if (options->found_changes)
5974                                 break;
5975                 }
5976         }
5977
5978         if (output_format & DIFF_FORMAT_PATCH) {
5979                 if (separator) {
5980                         emit_diff_symbol(options, DIFF_SYMBOL_SEPARATOR, NULL, 0, 0);
5981                         if (options->stat_sep)
5982                                 /* attach patch instead of inline */
5983                                 emit_diff_symbol(options, DIFF_SYMBOL_STAT_SEP,
5984                                                  NULL, 0, 0);
5985                 }
5986
5987                 diff_flush_patch_all_file_pairs(options);
5988         }
5989
5990         if (output_format & DIFF_FORMAT_CALLBACK)
5991                 options->format_callback(q, options, options->format_callback_data);
5992
5993         for (i = 0; i < q->nr; i++)
5994                 diff_free_filepair(q->queue[i]);
5995 free_queue:
5996         free(q->queue);
5997         DIFF_QUEUE_CLEAR(q);
5998         if (options->close_file)
5999                 fclose(options->file);
6000
6001         /*
6002          * Report the content-level differences with HAS_CHANGES;
6003          * diff_addremove/diff_change does not set the bit when
6004          * DIFF_FROM_CONTENTS is in effect (e.g. with -w).
6005          */
6006         if (options->flags.diff_from_contents) {
6007                 if (options->found_changes)
6008                         options->flags.has_changes = 1;
6009                 else
6010                         options->flags.has_changes = 0;
6011         }
6012 }
6013
6014 static int match_filter(const struct diff_options *options, const struct diff_filepair *p)
6015 {
6016         return (((p->status == DIFF_STATUS_MODIFIED) &&
6017                  ((p->score &&
6018                    filter_bit_tst(DIFF_STATUS_FILTER_BROKEN, options)) ||
6019                   (!p->score &&
6020                    filter_bit_tst(DIFF_STATUS_MODIFIED, options)))) ||
6021                 ((p->status != DIFF_STATUS_MODIFIED) &&
6022                  filter_bit_tst(p->status, options)));
6023 }
6024
6025 static void diffcore_apply_filter(struct diff_options *options)
6026 {
6027         int i;
6028         struct diff_queue_struct *q = &diff_queued_diff;
6029         struct diff_queue_struct outq;
6030
6031         DIFF_QUEUE_CLEAR(&outq);
6032
6033         if (!options->filter)
6034                 return;
6035
6036         if (filter_bit_tst(DIFF_STATUS_FILTER_AON, options)) {
6037                 int found;
6038                 for (i = found = 0; !found && i < q->nr; i++) {
6039                         if (match_filter(options, q->queue[i]))
6040                                 found++;
6041                 }
6042                 if (found)
6043                         return;
6044
6045                 /* otherwise we will clear the whole queue
6046                  * by copying the empty outq at the end of this
6047                  * function, but first clear the current entries
6048                  * in the queue.
6049                  */
6050                 for (i = 0; i < q->nr; i++)
6051                         diff_free_filepair(q->queue[i]);
6052         }
6053         else {
6054                 /* Only the matching ones */
6055                 for (i = 0; i < q->nr; i++) {
6056                         struct diff_filepair *p = q->queue[i];
6057                         if (match_filter(options, p))
6058                                 diff_q(&outq, p);
6059                         else
6060                                 diff_free_filepair(p);
6061                 }
6062         }
6063         free(q->queue);
6064         *q = outq;
6065 }
6066
6067 /* Check whether two filespecs with the same mode and size are identical */
6068 static int diff_filespec_is_identical(struct repository *r,
6069                                       struct diff_filespec *one,
6070                                       struct diff_filespec *two)
6071 {
6072         if (S_ISGITLINK(one->mode))
6073                 return 0;
6074         if (diff_populate_filespec(r, one, 0))
6075                 return 0;
6076         if (diff_populate_filespec(r, two, 0))
6077                 return 0;
6078         return !memcmp(one->data, two->data, one->size);
6079 }
6080
6081 static int diff_filespec_check_stat_unmatch(struct repository *r,
6082                                             struct diff_filepair *p)
6083 {
6084         if (p->done_skip_stat_unmatch)
6085                 return p->skip_stat_unmatch_result;
6086
6087         p->done_skip_stat_unmatch = 1;
6088         p->skip_stat_unmatch_result = 0;
6089         /*
6090          * 1. Entries that come from stat info dirtiness
6091          *    always have both sides (iow, not create/delete),
6092          *    one side of the object name is unknown, with
6093          *    the same mode and size.  Keep the ones that
6094          *    do not match these criteria.  They have real
6095          *    differences.
6096          *
6097          * 2. At this point, the file is known to be modified,
6098          *    with the same mode and size, and the object
6099          *    name of one side is unknown.  Need to inspect
6100          *    the identical contents.
6101          */
6102         if (!DIFF_FILE_VALID(p->one) || /* (1) */
6103             !DIFF_FILE_VALID(p->two) ||
6104             (p->one->oid_valid && p->two->oid_valid) ||
6105             (p->one->mode != p->two->mode) ||
6106             diff_populate_filespec(r, p->one, CHECK_SIZE_ONLY) ||
6107             diff_populate_filespec(r, p->two, CHECK_SIZE_ONLY) ||
6108             (p->one->size != p->two->size) ||
6109             !diff_filespec_is_identical(r, p->one, p->two)) /* (2) */
6110                 p->skip_stat_unmatch_result = 1;
6111         return p->skip_stat_unmatch_result;
6112 }
6113
6114 static void diffcore_skip_stat_unmatch(struct diff_options *diffopt)
6115 {
6116         int i;
6117         struct diff_queue_struct *q = &diff_queued_diff;
6118         struct diff_queue_struct outq;
6119         DIFF_QUEUE_CLEAR(&outq);
6120
6121         for (i = 0; i < q->nr; i++) {
6122                 struct diff_filepair *p = q->queue[i];
6123
6124                 if (diff_filespec_check_stat_unmatch(diffopt->repo, p))
6125                         diff_q(&outq, p);
6126                 else {
6127                         /*
6128                          * The caller can subtract 1 from skip_stat_unmatch
6129                          * to determine how many paths were dirty only
6130                          * due to stat info mismatch.
6131                          */
6132                         if (!diffopt->flags.no_index)
6133                                 diffopt->skip_stat_unmatch++;
6134                         diff_free_filepair(p);
6135                 }
6136         }
6137         free(q->queue);
6138         *q = outq;
6139 }
6140
6141 static int diffnamecmp(const void *a_, const void *b_)
6142 {
6143         const struct diff_filepair *a = *((const struct diff_filepair **)a_);
6144         const struct diff_filepair *b = *((const struct diff_filepair **)b_);
6145         const char *name_a, *name_b;
6146
6147         name_a = a->one ? a->one->path : a->two->path;
6148         name_b = b->one ? b->one->path : b->two->path;
6149         return strcmp(name_a, name_b);
6150 }
6151
6152 void diffcore_fix_diff_index(struct diff_options *options)
6153 {
6154         struct diff_queue_struct *q = &diff_queued_diff;
6155         QSORT(q->queue, q->nr, diffnamecmp);
6156 }
6157
6158 void diffcore_std(struct diff_options *options)
6159 {
6160         /* NOTE please keep the following in sync with diff_tree_combined() */
6161         if (options->skip_stat_unmatch)
6162                 diffcore_skip_stat_unmatch(options);
6163         if (!options->found_follow) {
6164                 /* See try_to_follow_renames() in tree-diff.c */
6165                 if (options->break_opt != -1)
6166                         diffcore_break(options->repo,
6167                                        options->break_opt);
6168                 if (options->detect_rename)
6169                         diffcore_rename(options);
6170                 if (options->break_opt != -1)
6171                         diffcore_merge_broken();
6172         }
6173         if (options->pickaxe_opts & DIFF_PICKAXE_KINDS_MASK)
6174                 diffcore_pickaxe(options);
6175         if (options->orderfile)
6176                 diffcore_order(options->orderfile);
6177         if (!options->found_follow)
6178                 /* See try_to_follow_renames() in tree-diff.c */
6179                 diff_resolve_rename_copy();
6180         diffcore_apply_filter(options);
6181
6182         if (diff_queued_diff.nr && !options->flags.diff_from_contents)
6183                 options->flags.has_changes = 1;
6184         else
6185                 options->flags.has_changes = 0;
6186
6187         options->found_follow = 0;
6188 }
6189
6190 int diff_result_code(struct diff_options *opt, int status)
6191 {
6192         int result = 0;
6193
6194         diff_warn_rename_limit("diff.renameLimit",
6195                                opt->needed_rename_limit,
6196                                opt->degraded_cc_to_c);
6197         if (!opt->flags.exit_with_status &&
6198             !(opt->output_format & DIFF_FORMAT_CHECKDIFF))
6199                 return status;
6200         if (opt->flags.exit_with_status &&
6201             opt->flags.has_changes)
6202                 result |= 01;
6203         if ((opt->output_format & DIFF_FORMAT_CHECKDIFF) &&
6204             opt->flags.check_failed)
6205                 result |= 02;
6206         return result;
6207 }
6208
6209 int diff_can_quit_early(struct diff_options *opt)
6210 {
6211         return (opt->flags.quick &&
6212                 !opt->filter &&
6213                 opt->flags.has_changes);
6214 }
6215
6216 /*
6217  * Shall changes to this submodule be ignored?
6218  *
6219  * Submodule changes can be configured to be ignored separately for each path,
6220  * but that configuration can be overridden from the command line.
6221  */
6222 static int is_submodule_ignored(const char *path, struct diff_options *options)
6223 {
6224         int ignored = 0;
6225         struct diff_flags orig_flags = options->flags;
6226         if (!options->flags.override_submodule_config)
6227                 set_diffopt_flags_from_submodule_config(options, path);
6228         if (options->flags.ignore_submodules)
6229                 ignored = 1;
6230         options->flags = orig_flags;
6231         return ignored;
6232 }
6233
6234 void diff_addremove(struct diff_options *options,
6235                     int addremove, unsigned mode,
6236                     const struct object_id *oid,
6237                     int oid_valid,
6238                     const char *concatpath, unsigned dirty_submodule)
6239 {
6240         struct diff_filespec *one, *two;
6241
6242         if (S_ISGITLINK(mode) && is_submodule_ignored(concatpath, options))
6243                 return;
6244
6245         /* This may look odd, but it is a preparation for
6246          * feeding "there are unchanged files which should
6247          * not produce diffs, but when you are doing copy
6248          * detection you would need them, so here they are"
6249          * entries to the diff-core.  They will be prefixed
6250          * with something like '=' or '*' (I haven't decided
6251          * which but should not make any difference).
6252          * Feeding the same new and old to diff_change()
6253          * also has the same effect.
6254          * Before the final output happens, they are pruned after
6255          * merged into rename/copy pairs as appropriate.
6256          */
6257         if (options->flags.reverse_diff)
6258                 addremove = (addremove == '+' ? '-' :
6259                              addremove == '-' ? '+' : addremove);
6260
6261         if (options->prefix &&
6262             strncmp(concatpath, options->prefix, options->prefix_length))
6263                 return;
6264
6265         one = alloc_filespec(concatpath);
6266         two = alloc_filespec(concatpath);
6267
6268         if (addremove != '+')
6269                 fill_filespec(one, oid, oid_valid, mode);
6270         if (addremove != '-') {
6271                 fill_filespec(two, oid, oid_valid, mode);
6272                 two->dirty_submodule = dirty_submodule;
6273         }
6274
6275         diff_queue(&diff_queued_diff, one, two);
6276         if (!options->flags.diff_from_contents)
6277                 options->flags.has_changes = 1;
6278 }
6279
6280 void diff_change(struct diff_options *options,
6281                  unsigned old_mode, unsigned new_mode,
6282                  const struct object_id *old_oid,
6283                  const struct object_id *new_oid,
6284                  int old_oid_valid, int new_oid_valid,
6285                  const char *concatpath,
6286                  unsigned old_dirty_submodule, unsigned new_dirty_submodule)
6287 {
6288         struct diff_filespec *one, *two;
6289         struct diff_filepair *p;
6290
6291         if (S_ISGITLINK(old_mode) && S_ISGITLINK(new_mode) &&
6292             is_submodule_ignored(concatpath, options))
6293                 return;
6294
6295         if (options->flags.reverse_diff) {
6296                 SWAP(old_mode, new_mode);
6297                 SWAP(old_oid, new_oid);
6298                 SWAP(old_oid_valid, new_oid_valid);
6299                 SWAP(old_dirty_submodule, new_dirty_submodule);
6300         }
6301
6302         if (options->prefix &&
6303             strncmp(concatpath, options->prefix, options->prefix_length))
6304                 return;
6305
6306         one = alloc_filespec(concatpath);
6307         two = alloc_filespec(concatpath);
6308         fill_filespec(one, old_oid, old_oid_valid, old_mode);
6309         fill_filespec(two, new_oid, new_oid_valid, new_mode);
6310         one->dirty_submodule = old_dirty_submodule;
6311         two->dirty_submodule = new_dirty_submodule;
6312         p = diff_queue(&diff_queued_diff, one, two);
6313
6314         if (options->flags.diff_from_contents)
6315                 return;
6316
6317         if (options->flags.quick && options->skip_stat_unmatch &&
6318             !diff_filespec_check_stat_unmatch(options->repo, p))
6319                 return;
6320
6321         options->flags.has_changes = 1;
6322 }
6323
6324 struct diff_filepair *diff_unmerge(struct diff_options *options, const char *path)
6325 {
6326         struct diff_filepair *pair;
6327         struct diff_filespec *one, *two;
6328
6329         if (options->prefix &&
6330             strncmp(path, options->prefix, options->prefix_length))
6331                 return NULL;
6332
6333         one = alloc_filespec(path);
6334         two = alloc_filespec(path);
6335         pair = diff_queue(&diff_queued_diff, one, two);
6336         pair->is_unmerged = 1;
6337         return pair;
6338 }
6339
6340 static char *run_textconv(struct repository *r,
6341                           const char *pgm,
6342                           struct diff_filespec *spec,
6343                           size_t *outsize)
6344 {
6345         struct diff_tempfile *temp;
6346         const char *argv[3];
6347         const char **arg = argv;
6348         struct child_process child = CHILD_PROCESS_INIT;
6349         struct strbuf buf = STRBUF_INIT;
6350         int err = 0;
6351
6352         temp = prepare_temp_file(r, spec->path, spec);
6353         *arg++ = pgm;
6354         *arg++ = temp->name;
6355         *arg = NULL;
6356
6357         child.use_shell = 1;
6358         child.argv = argv;
6359         child.out = -1;
6360         if (start_command(&child)) {
6361                 remove_tempfile();
6362                 return NULL;
6363         }
6364
6365         if (strbuf_read(&buf, child.out, 0) < 0)
6366                 err = error("error reading from textconv command '%s'", pgm);
6367         close(child.out);
6368
6369         if (finish_command(&child) || err) {
6370                 strbuf_release(&buf);
6371                 remove_tempfile();
6372                 return NULL;
6373         }
6374         remove_tempfile();
6375
6376         return strbuf_detach(&buf, outsize);
6377 }
6378
6379 size_t fill_textconv(struct repository *r,
6380                      struct userdiff_driver *driver,
6381                      struct diff_filespec *df,
6382                      char **outbuf)
6383 {
6384         size_t size;
6385
6386         if (!driver) {
6387                 if (!DIFF_FILE_VALID(df)) {
6388                         *outbuf = "";
6389                         return 0;
6390                 }
6391                 if (diff_populate_filespec(r, df, 0))
6392                         die("unable to read files to diff");
6393                 *outbuf = df->data;
6394                 return df->size;
6395         }
6396
6397         if (!driver->textconv)
6398                 BUG("fill_textconv called with non-textconv driver");
6399
6400         if (driver->textconv_cache && df->oid_valid) {
6401                 *outbuf = notes_cache_get(driver->textconv_cache,
6402                                           &df->oid,
6403                                           &size);
6404                 if (*outbuf)
6405                         return size;
6406         }
6407
6408         *outbuf = run_textconv(r, driver->textconv, df, &size);
6409         if (!*outbuf)
6410                 die("unable to read files to diff");
6411
6412         if (driver->textconv_cache && df->oid_valid) {
6413                 /* ignore errors, as we might be in a readonly repository */
6414                 notes_cache_put(driver->textconv_cache, &df->oid, *outbuf,
6415                                 size);
6416                 /*
6417                  * we could save up changes and flush them all at the end,
6418                  * but we would need an extra call after all diffing is done.
6419                  * Since generating a cache entry is the slow path anyway,
6420                  * this extra overhead probably isn't a big deal.
6421                  */
6422                 notes_cache_write(driver->textconv_cache);
6423         }
6424
6425         return size;
6426 }
6427
6428 int textconv_object(struct repository *r,
6429                     const char *path,
6430                     unsigned mode,
6431                     const struct object_id *oid,
6432                     int oid_valid,
6433                     char **buf,
6434                     unsigned long *buf_size)
6435 {
6436         struct diff_filespec *df;
6437         struct userdiff_driver *textconv;
6438
6439         df = alloc_filespec(path);
6440         fill_filespec(df, oid, oid_valid, mode);
6441         textconv = get_textconv(r->index, df);
6442         if (!textconv) {
6443                 free_filespec(df);
6444                 return 0;
6445         }
6446
6447         *buf_size = fill_textconv(r, textconv, df, buf);
6448         free_filespec(df);
6449         return 1;
6450 }
6451
6452 void setup_diff_pager(struct diff_options *opt)
6453 {
6454         /*
6455          * If the user asked for our exit code, then either they want --quiet
6456          * or --exit-code. We should definitely not bother with a pager in the
6457          * former case, as we will generate no output. Since we still properly
6458          * report our exit code even when a pager is run, we _could_ run a
6459          * pager with --exit-code. But since we have not done so historically,
6460          * and because it is easy to find people oneline advising "git diff
6461          * --exit-code" in hooks and other scripts, we do not do so.
6462          */
6463         if (!opt->flags.exit_with_status &&
6464             check_pager_config("diff") != 0)
6465                 setup_pager();
6466 }