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