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 "parse-options.h"
  27 #include "help.h"
  28 #include "promisor-remote.h"
  29
  30 #ifdef NO_FAST_WORKING_DIRECTORY
  31 #define FAST_WORKING_DIRECTORY 0
  32 #else
  33 #define FAST_WORKING_DIRECTORY 1
  34 #endif
  35
  36 static int diff_detect_rename_default;
  37 static int diff_indent_heuristic = 1;
  38 static int diff_rename_limit_default = 400;
  39 static int diff_suppress_blank_empty;
  40 static int diff_use_color_default = -1;
  41 static int diff_color_moved_default;
  42 static int diff_color_moved_ws_default;
  43 static int diff_context_default = 3;
  44 static int diff_interhunk_context_default;
  45 static const char *diff_word_regex_cfg;
  46 static const char *external_diff_cmd_cfg;
  47 static const char *diff_order_file_cfg;
  48 int diff_auto_refresh_index = 1;
  49 static int diff_mnemonic_prefix;
  50 static int diff_no_prefix;
  51 static int diff_stat_graph_width;
  52 static int diff_dirstat_permille_default = 30;
  53 static struct diff_options default_diff_options;
  54 static long diff_algorithm;
  55 static unsigned ws_error_highlight_default = WSEH_NEW;
  56
  57 static char diff_colors[][COLOR_MAXLEN] = {
  58         GIT_COLOR_RESET,
  59         GIT_COLOR_NORMAL,       /* CONTEXT */
  60         GIT_COLOR_BOLD,         /* METAINFO */
  61         GIT_COLOR_CYAN,         /* FRAGINFO */
  62         GIT_COLOR_RED,          /* OLD */
  63         GIT_COLOR_GREEN,        /* NEW */
  64         GIT_COLOR_YELLOW,       /* COMMIT */
  65         GIT_COLOR_BG_RED,       /* WHITESPACE */
  66         GIT_COLOR_NORMAL,       /* FUNCINFO */
  67         GIT_COLOR_BOLD_MAGENTA, /* OLD_MOVED */
  68         GIT_COLOR_BOLD_BLUE,    /* OLD_MOVED ALTERNATIVE */
  69         GIT_COLOR_FAINT,        /* OLD_MOVED_DIM */
  70         GIT_COLOR_FAINT_ITALIC, /* OLD_MOVED_ALTERNATIVE_DIM */
  71         GIT_COLOR_BOLD_CYAN,    /* NEW_MOVED */
  72         GIT_COLOR_BOLD_YELLOW,  /* NEW_MOVED ALTERNATIVE */
  73         GIT_COLOR_FAINT,        /* NEW_MOVED_DIM */
  74         GIT_COLOR_FAINT_ITALIC, /* NEW_MOVED_ALTERNATIVE_DIM */
  75         GIT_COLOR_FAINT,        /* CONTEXT_DIM */
  76         GIT_COLOR_FAINT_RED,    /* OLD_DIM */
  77         GIT_COLOR_FAINT_GREEN,  /* NEW_DIM */
  78         GIT_COLOR_BOLD,         /* CONTEXT_BOLD */
  79         GIT_COLOR_BOLD_RED,     /* OLD_BOLD */
  80         GIT_COLOR_BOLD_GREEN,   /* NEW_BOLD */
  81 };
  82
  83 static const char *color_diff_slots[] = {
  84         [DIFF_CONTEXT]                = "context",
  85         [DIFF_METAINFO]               = "meta",
  86         [DIFF_FRAGINFO]               = "frag",
  87         [DIFF_FILE_OLD]               = "old",
  88         [DIFF_FILE_NEW]               = "new",
  89         [DIFF_COMMIT]                 = "commit",
  90         [DIFF_WHITESPACE]             = "whitespace",
  91         [DIFF_FUNCINFO]               = "func",
  92         [DIFF_FILE_OLD_MOVED]         = "oldMoved",
  93         [DIFF_FILE_OLD_MOVED_ALT]     = "oldMovedAlternative",
  94         [DIFF_FILE_OLD_MOVED_DIM]     = "oldMovedDimmed",
  95         [DIFF_FILE_OLD_MOVED_ALT_DIM] = "oldMovedAlternativeDimmed",
  96         [DIFF_FILE_NEW_MOVED]         = "newMoved",
  97         [DIFF_FILE_NEW_MOVED_ALT]     = "newMovedAlternative",
  98         [DIFF_FILE_NEW_MOVED_DIM]     = "newMovedDimmed",
  99         [DIFF_FILE_NEW_MOVED_ALT_DIM] = "newMovedAlternativeDimmed",
 100         [DIFF_CONTEXT_DIM]            = "contextDimmed",
 101         [DIFF_FILE_OLD_DIM]           = "oldDimmed",
 102         [DIFF_FILE_NEW_DIM]           = "newDimmed",
 103         [DIFF_CONTEXT_BOLD]           = "contextBold",
 104         [DIFF_FILE_OLD_BOLD]          = "oldBold",
 105         [DIFF_FILE_NEW_BOLD]          = "newBold",
 106 };
 107
 108 define_list_config_array_extra(color_diff_slots, {"plain"});
 109
 110 static int parse_diff_color_slot(const char *var)
 111 {
 112         if (!strcasecmp(var, "plain"))
 113                 return DIFF_CONTEXT;
 114         return LOOKUP_CONFIG(color_diff_slots, var);
 115 }
 116
 117 static int parse_dirstat_params(struct diff_options *options, const char *params_string,
 118                                 struct strbuf *errmsg)
 119 {
 120         char *params_copy = xstrdup(params_string);
 121         struct string_list params = STRING_LIST_INIT_NODUP;
 122         int ret = 0;
 123         int i;
 124
 125         if (*params_copy)
 126                 string_list_split_in_place(&params, params_copy, ',', -1);
 127         for (i = 0; i < params.nr; i++) {
 128                 const char *p = params.items[i].string;
 129                 if (!strcmp(p, "changes")) {
 130                         options->flags.dirstat_by_line = 0;
 131                         options->flags.dirstat_by_file = 0;
 132                 } else if (!strcmp(p, "lines")) {
 133                         options->flags.dirstat_by_line = 1;
 134                         options->flags.dirstat_by_file = 0;
 135                 } else if (!strcmp(p, "files")) {
 136                         options->flags.dirstat_by_line = 0;
 137                         options->flags.dirstat_by_file = 1;
 138                 } else if (!strcmp(p, "noncumulative")) {
 139                         options->flags.dirstat_cumulative = 0;
 140                 } else if (!strcmp(p, "cumulative")) {
 141                         options->flags.dirstat_cumulative = 1;
 142                 } else if (isdigit(*p)) {
 143                         char *end;
 144                         int permille = strtoul(p, &end, 10) * 10;
 145                         if (*end == '.' && isdigit(*++end)) {
 146                                 /* only use first digit */
 147                                 permille += *end - '0';
 148                                 /* .. and ignore any further digits */
 149                                 while (isdigit(*++end))
 150                                         ; /* nothing */
 151                         }
 152                         if (!*end)
 153                                 options->dirstat_permille = permille;
 154                         else {
 155                                 strbuf_addf(errmsg, _("  Failed to parse dirstat cut-off percentage '%s'\n"),
 156                                             p);
 157                                 ret++;
 158                         }
 159                 } else {
 160                         strbuf_addf(errmsg, _("  Unknown dirstat parameter '%s'\n"), p);
 161                         ret++;
 162                 }
 163
 164         }
 165         string_list_clear(&params, 0);
 166         free(params_copy);
 167         return ret;
 168 }
 169
 170 static int parse_submodule_params(struct diff_options *options, const char *value)
 171 {
 172         if (!strcmp(value, "log"))
 173                 options->submodule_format = DIFF_SUBMODULE_LOG;
 174         else if (!strcmp(value, "short"))
 175                 options->submodule_format = DIFF_SUBMODULE_SHORT;
 176         else if (!strcmp(value, "diff"))
 177                 options->submodule_format = DIFF_SUBMODULE_INLINE_DIFF;
 178         /*
 179          * Please update $__git_diff_submodule_formats in
 180          * git-completion.bash when you add new formats.
 181          */
 182         else
 183                 return -1;
 184         return 0;
 185 }
 186
 187 int git_config_rename(const char *var, const char *value)
 188 {
 189         if (!value)
 190                 return DIFF_DETECT_RENAME;
 191         if (!strcasecmp(value, "copies") || !strcasecmp(value, "copy"))
 192                 return  DIFF_DETECT_COPY;
 193         return git_config_bool(var,value) ? DIFF_DETECT_RENAME : 0;
 194 }
 195
 196 long parse_algorithm_value(const char *value)
 197 {
 198         if (!value)
 199                 return -1;
 200         else if (!strcasecmp(value, "myers") || !strcasecmp(value, "default"))
 201                 return 0;
 202         else if (!strcasecmp(value, "minimal"))
 203                 return XDF_NEED_MINIMAL;
 204         else if (!strcasecmp(value, "patience"))
 205                 return XDF_PATIENCE_DIFF;
 206         else if (!strcasecmp(value, "histogram"))
 207                 return XDF_HISTOGRAM_DIFF;
 208         /*
 209          * Please update $__git_diff_algorithms in git-completion.bash
 210          * when you add new algorithms.
 211          */
 212         return -1;
 213 }
 214
 215 static int parse_one_token(const char **arg, const char *token)
 216 {
 217         const char *rest;
 218         if (skip_prefix(*arg, token, &rest) && (!*rest || *rest == ',')) {
 219                 *arg = rest;
 220                 return 1;
 221         }
 222         return 0;
 223 }
 224
 225 static int parse_ws_error_highlight(const char *arg)
 226 {
 227         const char *orig_arg = arg;
 228         unsigned val = 0;
 229
 230         while (*arg) {
 231                 if (parse_one_token(&arg, "none"))
 232                         val = 0;
 233                 else if (parse_one_token(&arg, "default"))
 234                         val = WSEH_NEW;
 235                 else if (parse_one_token(&arg, "all"))
 236                         val = WSEH_NEW | WSEH_OLD | WSEH_CONTEXT;
 237                 else if (parse_one_token(&arg, "new"))
 238                         val |= WSEH_NEW;
 239                 else if (parse_one_token(&arg, "old"))
 240                         val |= WSEH_OLD;
 241                 else if (parse_one_token(&arg, "context"))
 242                         val |= WSEH_CONTEXT;
 243                 else {
 244                         return -1 - (int)(arg - orig_arg);
 245                 }
 246                 if (*arg)
 247                         arg++;
 248         }
 249         return val;
 250 }
 251
 252 /*
 253  * These are to give UI layer defaults.
 254  * The core-level commands such as git-diff-files should
 255  * never be affected by the setting of diff.renames
 256  * the user happens to have in the configuration file.
 257  */
 258 void init_diff_ui_defaults(void)
 259 {
 260         diff_detect_rename_default = DIFF_DETECT_RENAME;
 261 }
 262
 263 int git_diff_heuristic_config(const char *var, const char *value, void *cb)
 264 {
 265         if (!strcmp(var, "diff.indentheuristic"))
 266                 diff_indent_heuristic = git_config_bool(var, value);
 267         return 0;
 268 }
 269
 270 static int parse_color_moved(const char *arg)
 271 {
 272         switch (git_parse_maybe_bool(arg)) {
 273         case 0:
 274                 return COLOR_MOVED_NO;
 275         case 1:
 276                 return COLOR_MOVED_DEFAULT;
 277         default:
 278                 break;
 279         }
 280
 281         if (!strcmp(arg, "no"))
 282                 return COLOR_MOVED_NO;
 283         else if (!strcmp(arg, "plain"))
 284                 return COLOR_MOVED_PLAIN;
 285         else if (!strcmp(arg, "blocks"))
 286                 return COLOR_MOVED_BLOCKS;
 287         else if (!strcmp(arg, "zebra"))
 288                 return COLOR_MOVED_ZEBRA;
 289         else if (!strcmp(arg, "default"))
 290                 return COLOR_MOVED_DEFAULT;
 291         else if (!strcmp(arg, "dimmed-zebra"))
 292                 return COLOR_MOVED_ZEBRA_DIM;
 293         else if (!strcmp(arg, "dimmed_zebra"))
 294                 return COLOR_MOVED_ZEBRA_DIM;
 295         else
 296                 return error(_("color moved setting must be one of 'no', 'default', 'blocks', 'zebra', 'dimmed-zebra', 'plain'"));
 297 }
 298
 299 static unsigned parse_color_moved_ws(const char *arg)
 300 {
 301         int ret = 0;
 302         struct string_list l = STRING_LIST_INIT_DUP;
 303         struct string_list_item *i;
 304
 305         string_list_split(&l, arg, ',', -1);
 306
 307         for_each_string_list_item(i, &l) {
 308                 struct strbuf sb = STRBUF_INIT;
 309                 strbuf_addstr(&sb, i->string);
 310                 strbuf_trim(&sb);
 311
 312                 if (!strcmp(sb.buf, "no"))
 313                         ret = 0;
 314                 else if (!strcmp(sb.buf, "ignore-space-change"))
 315                         ret |= XDF_IGNORE_WHITESPACE_CHANGE;
 316                 else if (!strcmp(sb.buf, "ignore-space-at-eol"))
 317                         ret |= XDF_IGNORE_WHITESPACE_AT_EOL;
 318                 else if (!strcmp(sb.buf, "ignore-all-space"))
 319                         ret |= XDF_IGNORE_WHITESPACE;
 320                 else if (!strcmp(sb.buf, "allow-indentation-change"))
 321                         ret |= COLOR_MOVED_WS_ALLOW_INDENTATION_CHANGE;
 322                 else {
 323                         ret |= COLOR_MOVED_WS_ERROR;
 324                         error(_("unknown color-moved-ws mode '%s', possible values are 'ignore-space-change', 'ignore-space-at-eol', 'ignore-all-space', 'allow-indentation-change'"), sb.buf);
 325                 }
 326
 327                 strbuf_release(&sb);
 328         }
 329
 330         if ((ret & COLOR_MOVED_WS_ALLOW_INDENTATION_CHANGE) &&
 331             (ret & XDF_WHITESPACE_FLAGS)) {
 332                 error(_("color-moved-ws: allow-indentation-change cannot be combined with other whitespace modes"));
 333                 ret |= COLOR_MOVED_WS_ERROR;
 334         }
 335
 336         string_list_clear(&l, 0);
 337
 338         return ret;
 339 }
 340
 341 int git_diff_ui_config(const char *var, const char *value, void *cb)
 342 {
 343         if (!strcmp(var, "diff.color") || !strcmp(var, "color.diff")) {
 344                 diff_use_color_default = git_config_colorbool(var, value);
 345                 return 0;
 346         }
 347         if (!strcmp(var, "diff.colormoved")) {
 348                 int cm = parse_color_moved(value);
 349                 if (cm < 0)
 350                         return -1;
 351                 diff_color_moved_default = cm;
 352                 return 0;
 353         }
 354         if (!strcmp(var, "diff.colormovedws")) {
 355                 unsigned cm = parse_color_moved_ws(value);
 356                 if (cm & COLOR_MOVED_WS_ERROR)
 357                         return -1;
 358                 diff_color_moved_ws_default = cm;
 359                 return 0;
 360         }
 361         if (!strcmp(var, "diff.context")) {
 362                 diff_context_default = git_config_int(var, value);
 363                 if (diff_context_default < 0)
 364                         return -1;
 365                 return 0;
 366         }
 367         if (!strcmp(var, "diff.interhunkcontext")) {
 368                 diff_interhunk_context_default = git_config_int(var, value);
 369                 if (diff_interhunk_context_default < 0)
 370                         return -1;
 371                 return 0;
 372         }
 373         if (!strcmp(var, "diff.renames")) {
 374                 diff_detect_rename_default = git_config_rename(var, value);
 375                 return 0;
 376         }
 377         if (!strcmp(var, "diff.autorefreshindex")) {
 378                 diff_auto_refresh_index = git_config_bool(var, value);
 379                 return 0;
 380         }
 381         if (!strcmp(var, "diff.mnemonicprefix")) {
 382                 diff_mnemonic_prefix = git_config_bool(var, value);
 383                 return 0;
 384         }
 385         if (!strcmp(var, "diff.noprefix")) {
 386                 diff_no_prefix = git_config_bool(var, value);
 387                 return 0;
 388         }
 389         if (!strcmp(var, "diff.statgraphwidth")) {
 390                 diff_stat_graph_width = git_config_int(var, value);
 391                 return 0;
 392         }
 393         if (!strcmp(var, "diff.external"))
 394                 return git_config_string(&external_diff_cmd_cfg, var, value);
 395         if (!strcmp(var, "diff.wordregex"))
 396                 return git_config_string(&diff_word_regex_cfg, var, value);
 397         if (!strcmp(var, "diff.orderfile"))
 398                 return git_config_pathname(&diff_order_file_cfg, var, value);
 399
 400         if (!strcmp(var, "diff.ignoresubmodules"))
 401                 handle_ignore_submodules_arg(&default_diff_options, value);
 402
 403         if (!strcmp(var, "diff.submodule")) {
 404                 if (parse_submodule_params(&default_diff_options, value))
 405                         warning(_("Unknown value for 'diff.submodule' config variable: '%s'"),
 406                                 value);
 407                 return 0;
 408         }
 409
 410         if (!strcmp(var, "diff.algorithm")) {
 411                 diff_algorithm = parse_algorithm_value(value);
 412                 if (diff_algorithm < 0)
 413                         return -1;
 414                 return 0;
 415         }
 416
 417         if (!strcmp(var, "diff.wserrorhighlight")) {
 418                 int val = parse_ws_error_highlight(value);
 419                 if (val < 0)
 420                         return -1;
 421                 ws_error_highlight_default = val;
 422                 return 0;
 423         }
 424
 425         if (git_color_config(var, value, cb) < 0)
 426                 return -1;
 427
 428         return git_diff_basic_config(var, value, cb);
 429 }
 430
 431 int git_diff_basic_config(const char *var, const char *value, void *cb)
 432 {
 433         const char *name;
 434
 435         if (!strcmp(var, "diff.renamelimit")) {
 436                 diff_rename_limit_default = git_config_int(var, value);
 437                 return 0;
 438         }
 439
 440         if (userdiff_config(var, value) < 0)
 441                 return -1;
 442
 443         if (skip_prefix(var, "diff.color.", &name) ||
 444             skip_prefix(var, "color.diff.", &name)) {
 445                 int slot = parse_diff_color_slot(name);
 446                 if (slot < 0)
 447                         return 0;
 448                 if (!value)
 449                         return config_error_nonbool(var);
 450                 return color_parse(value, diff_colors[slot]);
 451         }
 452
 453         /* like GNU diff's --suppress-blank-empty option  */
 454         if (!strcmp(var, "diff.suppressblankempty") ||
 455                         /* for backwards compatibility */
 456                         !strcmp(var, "diff.suppress-blank-empty")) {
 457                 diff_suppress_blank_empty = git_config_bool(var, value);
 458                 return 0;
 459         }
 460
 461         if (!strcmp(var, "diff.dirstat")) {
 462                 struct strbuf errmsg = STRBUF_INIT;
 463                 default_diff_options.dirstat_permille = diff_dirstat_permille_default;
 464                 if (parse_dirstat_params(&default_diff_options, value, &errmsg))
 465                         warning(_("Found errors in 'diff.dirstat' config variable:\n%s"),
 466                                 errmsg.buf);
 467                 strbuf_release(&errmsg);
 468                 diff_dirstat_permille_default = default_diff_options.dirstat_permille;
 469                 return 0;
 470         }
 471
 472         if (git_diff_heuristic_config(var, value, cb) < 0)
 473                 return -1;
 474
 475         return git_default_config(var, value, cb);
 476 }
 477
 478 static char *quote_two(const char *one, const char *two)
 479 {
 480         int need_one = quote_c_style(one, NULL, NULL, 1);
 481         int need_two = quote_c_style(two, NULL, NULL, 1);
 482         struct strbuf res = STRBUF_INIT;
 483
 484         if (need_one + need_two) {
 485                 strbuf_addch(&res, '"');
 486                 quote_c_style(one, &res, NULL, 1);
 487                 quote_c_style(two, &res, NULL, 1);
 488                 strbuf_addch(&res, '"');
 489         } else {
 490                 strbuf_addstr(&res, one);
 491                 strbuf_addstr(&res, two);
 492         }
 493         return strbuf_detach(&res, NULL);
 494 }
 495
 496 static const char *external_diff(void)
 497 {
 498         static const char *external_diff_cmd = NULL;
 499         static int done_preparing = 0;
 500
 501         if (done_preparing)
 502                 return external_diff_cmd;
 503         external_diff_cmd = xstrdup_or_null(getenv("GIT_EXTERNAL_DIFF"));
 504         if (!external_diff_cmd)
 505                 external_diff_cmd = external_diff_cmd_cfg;
 506         done_preparing = 1;
 507         return external_diff_cmd;
 508 }
 509
 510 /*
 511  * Keep track of files used for diffing. Sometimes such an entry
 512  * refers to a temporary file, sometimes to an existing file, and
 513  * sometimes to "/dev/null".
 514  */
 515 static struct diff_tempfile {
 516         /*
 517          * filename external diff should read from, or NULL if this
 518          * entry is currently not in use:
 519          */
 520         const char *name;
 521
 522         char hex[GIT_MAX_HEXSZ + 1];
 523         char mode[10];
 524
 525         /*
 526          * If this diff_tempfile instance refers to a temporary file,
 527          * this tempfile object is used to manage its lifetime.
 528          */
 529         struct tempfile *tempfile;
 530 } diff_temp[2];
 531
 532 struct emit_callback {
 533         int color_diff;
 534         unsigned ws_rule;
 535         int blank_at_eof_in_preimage;
 536         int blank_at_eof_in_postimage;
 537         int lno_in_preimage;
 538         int lno_in_postimage;
 539         const char **label_path;
 540         struct diff_words_data *diff_words;
 541         struct diff_options *opt;
 542         struct strbuf *header;
 543 };
 544
 545 static int count_lines(const char *data, int size)
 546 {
 547         int count, ch, completely_empty = 1, nl_just_seen = 0;
 548         count = 0;
 549         while (0 < size--) {
 550                 ch = *data++;
 551                 if (ch == '\n') {
 552                         count++;
 553                         nl_just_seen = 1;
 554                         completely_empty = 0;
 555                 }
 556                 else {
 557                         nl_just_seen = 0;
 558                         completely_empty = 0;
 559                 }
 560         }
 561         if (completely_empty)
 562                 return 0;
 563         if (!nl_just_seen)
 564                 count++; /* no trailing newline */
 565         return count;
 566 }
 567
 568 static int fill_mmfile(struct repository *r, mmfile_t *mf,
 569                        struct diff_filespec *one)
 570 {
 571         if (!DIFF_FILE_VALID(one)) {
 572                 mf->ptr = (char *)""; /* does not matter */
 573                 mf->size = 0;
 574                 return 0;
 575         }
 576         else if (diff_populate_filespec(r, one, 0))
 577                 return -1;
 578
 579         mf->ptr = one->data;
 580         mf->size = one->size;
 581         return 0;
 582 }
 583
 584 /* like fill_mmfile, but only for size, so we can avoid retrieving blob */
 585 static unsigned long diff_filespec_size(struct repository *r,
 586                                         struct diff_filespec *one)
 587 {
 588         if (!DIFF_FILE_VALID(one))
 589                 return 0;
 590         diff_populate_filespec(r, one, CHECK_SIZE_ONLY);
 591         return one->size;
 592 }
 593
 594 static int count_trailing_blank(mmfile_t *mf, unsigned ws_rule)
 595 {
 596         char *ptr = mf->ptr;
 597         long size = mf->size;
 598         int cnt = 0;
 599
 600         if (!size)
 601                 return cnt;
 602         ptr += size - 1; /* pointing at the very end */
 603         if (*ptr != '\n')
 604                 ; /* incomplete line */
 605         else
 606                 ptr--; /* skip the last LF */
 607         while (mf->ptr < ptr) {
 608                 char *prev_eol;
 609                 for (prev_eol = ptr; mf->ptr <= prev_eol; prev_eol--)
 610                         if (*prev_eol == '\n')
 611                                 break;
 612                 if (!ws_blank_line(prev_eol + 1, ptr - prev_eol, ws_rule))
 613                         break;
 614                 cnt++;
 615                 ptr = prev_eol - 1;
 616         }
 617         return cnt;
 618 }
 619
 620 static void check_blank_at_eof(mmfile_t *mf1, mmfile_t *mf2,
 621                                struct emit_callback *ecbdata)
 622 {
 623         int l1, l2, at;
 624         unsigned ws_rule = ecbdata->ws_rule;
 625         l1 = count_trailing_blank(mf1, ws_rule);
 626         l2 = count_trailing_blank(mf2, ws_rule);
 627         if (l2 <= l1) {
 628                 ecbdata->blank_at_eof_in_preimage = 0;
 629                 ecbdata->blank_at_eof_in_postimage = 0;
 630                 return;
 631         }
 632         at = count_lines(mf1->ptr, mf1->size);
 633         ecbdata->blank_at_eof_in_preimage = (at - l1) + 1;
 634
 635         at = count_lines(mf2->ptr, mf2->size);
 636         ecbdata->blank_at_eof_in_postimage = (at - l2) + 1;
 637 }
 638
 639 static void emit_line_0(struct diff_options *o,
 640                         const char *set_sign, const char *set, unsigned reverse, const char *reset,
 641                         int first, const char *line, int len)
 642 {
 643         int has_trailing_newline, has_trailing_carriage_return;
 644         int needs_reset = 0; /* at the end of the line */
 645         FILE *file = o->file;
 646
 647         fputs(diff_line_prefix(o), file);
 648
 649         has_trailing_newline = (len > 0 && line[len-1] == '\n');
 650         if (has_trailing_newline)
 651                 len--;
 652
 653         has_trailing_carriage_return = (len > 0 && line[len-1] == '\r');
 654         if (has_trailing_carriage_return)
 655                 len--;
 656
 657         if (!len && !first)
 658                 goto end_of_line;
 659
 660         if (reverse && want_color(o->use_color)) {
 661                 fputs(GIT_COLOR_REVERSE, file);
 662                 needs_reset = 1;
 663         }
 664
 665         if (set_sign) {
 666                 fputs(set_sign, file);
 667                 needs_reset = 1;
 668         }
 669
 670         if (first)
 671                 fputc(first, file);
 672
 673         if (!len)
 674                 goto end_of_line;
 675
 676         if (set) {
 677                 if (set_sign && set != set_sign)
 678                         fputs(reset, file);
 679                 fputs(set, file);
 680                 needs_reset = 1;
 681         }
 682         fwrite(line, len, 1, file);
 683         needs_reset = 1; /* 'line' may contain color codes. */
 684
 685 end_of_line:
 686         if (needs_reset)
 687                 fputs(reset, file);
 688         if (has_trailing_carriage_return)
 689                 fputc('\r', file);
 690         if (has_trailing_newline)
 691                 fputc('\n', file);
 692 }
 693
 694 static void emit_line(struct diff_options *o, const char *set, const char *reset,
 695                       const char *line, int len)
 696 {
 697         emit_line_0(o, set, NULL, 0, reset, 0, line, len);
 698 }
 699
 700 enum diff_symbol {
 701         DIFF_SYMBOL_BINARY_DIFF_HEADER,
 702         DIFF_SYMBOL_BINARY_DIFF_HEADER_DELTA,
 703         DIFF_SYMBOL_BINARY_DIFF_HEADER_LITERAL,
 704         DIFF_SYMBOL_BINARY_DIFF_BODY,
 705         DIFF_SYMBOL_BINARY_DIFF_FOOTER,
 706         DIFF_SYMBOL_STATS_SUMMARY_NO_FILES,
 707         DIFF_SYMBOL_STATS_SUMMARY_ABBREV,
 708         DIFF_SYMBOL_STATS_SUMMARY_INSERTS_DELETES,
 709         DIFF_SYMBOL_STATS_LINE,
 710         DIFF_SYMBOL_WORD_DIFF,
 711         DIFF_SYMBOL_STAT_SEP,
 712         DIFF_SYMBOL_SUMMARY,
 713         DIFF_SYMBOL_SUBMODULE_ADD,
 714         DIFF_SYMBOL_SUBMODULE_DEL,
 715         DIFF_SYMBOL_SUBMODULE_UNTRACKED,
 716         DIFF_SYMBOL_SUBMODULE_MODIFIED,
 717         DIFF_SYMBOL_SUBMODULE_HEADER,
 718         DIFF_SYMBOL_SUBMODULE_ERROR,
 719         DIFF_SYMBOL_SUBMODULE_PIPETHROUGH,
 720         DIFF_SYMBOL_REWRITE_DIFF,
 721         DIFF_SYMBOL_BINARY_FILES,
 722         DIFF_SYMBOL_HEADER,
 723         DIFF_SYMBOL_FILEPAIR_PLUS,
 724         DIFF_SYMBOL_FILEPAIR_MINUS,
 725         DIFF_SYMBOL_WORDS_PORCELAIN,
 726         DIFF_SYMBOL_WORDS,
 727         DIFF_SYMBOL_CONTEXT,
 728         DIFF_SYMBOL_CONTEXT_INCOMPLETE,
 729         DIFF_SYMBOL_PLUS,
 730         DIFF_SYMBOL_MINUS,
 731         DIFF_SYMBOL_NO_LF_EOF,
 732         DIFF_SYMBOL_CONTEXT_FRAGINFO,
 733         DIFF_SYMBOL_CONTEXT_MARKER,
 734         DIFF_SYMBOL_SEPARATOR
 735 };
 736 /*
 737  * Flags for content lines:
 738  * 0..12 are whitespace rules
 739  * 13-15 are WSEH_NEW | WSEH_OLD | WSEH_CONTEXT
 740  * 16 is marking if the line is blank at EOF
 741  */
 742 #define DIFF_SYMBOL_CONTENT_BLANK_LINE_EOF      (1<<16)
 743 #define DIFF_SYMBOL_MOVED_LINE                  (1<<17)
 744 #define DIFF_SYMBOL_MOVED_LINE_ALT              (1<<18)
 745 #define DIFF_SYMBOL_MOVED_LINE_UNINTERESTING    (1<<19)
 746 #define DIFF_SYMBOL_CONTENT_WS_MASK (WSEH_NEW | WSEH_OLD | WSEH_CONTEXT | WS_RULE_MASK)
 747
 748 /*
 749  * This struct is used when we need to buffer the output of the diff output.
 750  *
 751  * NEEDSWORK: Instead of storing a copy of the line, add an offset pointer
 752  * into the pre/post image file. This pointer could be a union with the
 753  * line pointer. By storing an offset into the file instead of the literal line,
 754  * we can decrease the memory footprint for the buffered output. At first we
 755  * may want to only have indirection for the content lines, but we could also
 756  * enhance the state for emitting prefabricated lines, e.g. the similarity
 757  * score line or hunk/file headers would only need to store a number or path
 758  * and then the output can be constructed later on depending on state.
 759  */
 760 struct emitted_diff_symbol {
 761         const char *line;
 762         int len;
 763         int flags;
 764         int indent_off;   /* Offset to first non-whitespace character */
 765         int indent_width; /* The visual width of the indentation */
 766         enum diff_symbol s;
 767 };
 768 #define EMITTED_DIFF_SYMBOL_INIT {NULL}
 769
 770 struct emitted_diff_symbols {
 771         struct emitted_diff_symbol *buf;
 772         int nr, alloc;
 773 };
 774 #define EMITTED_DIFF_SYMBOLS_INIT {NULL, 0, 0}
 775
 776 static void append_emitted_diff_symbol(struct diff_options *o,
 777                                        struct emitted_diff_symbol *e)
 778 {
 779         struct emitted_diff_symbol *f;
 780
 781         ALLOC_GROW(o->emitted_symbols->buf,
 782                    o->emitted_symbols->nr + 1,
 783                    o->emitted_symbols->alloc);
 784         f = &o->emitted_symbols->buf[o->emitted_symbols->nr++];
 785
 786         memcpy(f, e, sizeof(struct emitted_diff_symbol));
 787         f->line = e->line ? xmemdupz(e->line, e->len) : NULL;
 788 }
 789
 790 struct moved_entry {
 791         struct hashmap_entry ent;
 792         const struct emitted_diff_symbol *es;
 793         struct moved_entry *next_line;
 794 };
 795
 796 struct moved_block {
 797         struct moved_entry *match;
 798         int wsd; /* The whitespace delta of this block */
 799 };
 800
 801 static void moved_block_clear(struct moved_block *b)
 802 {
 803         memset(b, 0, sizeof(*b));
 804 }
 805
 806 #define INDENT_BLANKLINE INT_MIN
 807
 808 static void fill_es_indent_data(struct emitted_diff_symbol *es)
 809 {
 810         unsigned int off = 0, i;
 811         int width = 0, tab_width = es->flags & WS_TAB_WIDTH_MASK;
 812         const char *s = es->line;
 813         const int len = es->len;
 814
 815         /* skip any \v \f \r at start of indentation */
 816         while (s[off] == '\f' || s[off] == '\v' ||
 817                (s[off] == '\r' && off < len - 1))
 818                 off++;
 819
 820         /* calculate the visual width of indentation */
 821         while(1) {
 822                 if (s[off] == ' ') {
 823                         width++;
 824                         off++;
 825                 } else if (s[off] == '\t') {
 826                         width += tab_width - (width % tab_width);
 827                         while (s[++off] == '\t')
 828                                 width += tab_width;
 829                 } else {
 830                         break;
 831                 }
 832         }
 833
 834         /* check if this line is blank */
 835         for (i = off; i < len; i++)
 836                 if (!isspace(s[i]))
 837                     break;
 838
 839         if (i == len) {
 840                 es->indent_width = INDENT_BLANKLINE;
 841                 es->indent_off = len;
 842         } else {
 843                 es->indent_off = off;
 844                 es->indent_width = width;
 845         }
 846 }
 847
 848 static int compute_ws_delta(const struct emitted_diff_symbol *a,
 849                             const struct emitted_diff_symbol *b,
 850                             int *out)
 851 {
 852         int a_len = a->len,
 853             b_len = b->len,
 854             a_off = a->indent_off,
 855             a_width = a->indent_width,
 856             b_off = b->indent_off,
 857             b_width = b->indent_width;
 858         int delta;
 859
 860         if (a_width == INDENT_BLANKLINE && b_width == INDENT_BLANKLINE) {
 861                 *out = INDENT_BLANKLINE;
 862                 return 1;
 863         }
 864
 865         if (a->s == DIFF_SYMBOL_PLUS)
 866                 delta = a_width - b_width;
 867         else
 868                 delta = b_width - a_width;
 869
 870         if (a_len - a_off != b_len - b_off ||
 871             memcmp(a->line + a_off, b->line + b_off, a_len - a_off))
 872                 return 0;
 873
 874         *out = delta;
 875
 876         return 1;
 877 }
 878
 879 static int cmp_in_block_with_wsd(const struct diff_options *o,
 880                                  const struct moved_entry *cur,
 881                                  const struct moved_entry *match,
 882                                  struct moved_block *pmb,
 883                                  int n)
 884 {
 885         struct emitted_diff_symbol *l = &o->emitted_symbols->buf[n];
 886         int al = cur->es->len, bl = match->es->len, cl = l->len;
 887         const char *a = cur->es->line,
 888                    *b = match->es->line,
 889                    *c = l->line;
 890         int a_off = cur->es->indent_off,
 891             a_width = cur->es->indent_width,
 892             c_off = l->indent_off,
 893             c_width = l->indent_width;
 894         int delta;
 895
 896         /*
 897          * We need to check if 'cur' is equal to 'match'.  As those
 898          * are from the same (+/-) side, we do not need to adjust for
 899          * indent changes. However these were found using fuzzy
 900          * matching so we do have to check if they are equal. Here we
 901          * just check the lengths. We delay calling memcmp() to check
 902          * the contents until later as if the length comparison for a
 903          * and c fails we can avoid the call all together.
 904          */
 905         if (al != bl)
 906                 return 1;
 907
 908         /* If 'l' and 'cur' are both blank then they match. */
 909         if (a_width == INDENT_BLANKLINE && c_width == INDENT_BLANKLINE)
 910                 return 0;
 911
 912         /*
 913          * The indent changes of the block are known and stored in pmb->wsd;
 914          * however we need to check if the indent changes of the current line
 915          * match those of the current block and that the text of 'l' and 'cur'
 916          * after the indentation match.
 917          */
 918         if (cur->es->s == DIFF_SYMBOL_PLUS)
 919                 delta = a_width - c_width;
 920         else
 921                 delta = c_width - a_width;
 922
 923         /*
 924          * If the previous lines of this block were all blank then set its
 925          * whitespace delta.
 926          */
 927         if (pmb->wsd == INDENT_BLANKLINE)
 928                 pmb->wsd = delta;
 929
 930         return !(delta == pmb->wsd && al - a_off == cl - c_off &&
 931                  !memcmp(a, b, al) && !
 932                  memcmp(a + a_off, c + c_off, al - a_off));
 933 }
 934
 935 static int moved_entry_cmp(const void *hashmap_cmp_fn_data,
 936                            const struct hashmap_entry *eptr,
 937                            const struct hashmap_entry *entry_or_key,
 938                            const void *keydata)
 939 {
 940         const struct diff_options *diffopt = hashmap_cmp_fn_data;
 941         const struct moved_entry *a, *b;
 942         unsigned flags = diffopt->color_moved_ws_handling
 943                          & XDF_WHITESPACE_FLAGS;
 944
 945         a = container_of(eptr, const struct moved_entry, ent);
 946         b = container_of(entry_or_key, const struct moved_entry, ent);
 947
 948         if (diffopt->color_moved_ws_handling &
 949             COLOR_MOVED_WS_ALLOW_INDENTATION_CHANGE)
 950                 /*
 951                  * As there is not specific white space config given,
 952                  * we'd need to check for a new block, so ignore all
 953                  * white space. The setup of the white space
 954                  * configuration for the next block is done else where
 955                  */
 956                 flags |= XDF_IGNORE_WHITESPACE;
 957
 958         return !xdiff_compare_lines(a->es->line, a->es->len,
 959                                     b->es->line, b->es->len,
 960                                     flags);
 961 }
 962
 963 static struct moved_entry *prepare_entry(struct diff_options *o,
 964                                          int line_no)
 965 {
 966         struct moved_entry *ret = xmalloc(sizeof(*ret));
 967         struct emitted_diff_symbol *l = &o->emitted_symbols->buf[line_no];
 968         unsigned flags = o->color_moved_ws_handling & XDF_WHITESPACE_FLAGS;
 969         unsigned int hash = xdiff_hash_string(l->line, l->len, flags);
 970
 971         hashmap_entry_init(&ret->ent, hash);
 972         ret->es = l;
 973         ret->next_line = NULL;
 974
 975         return ret;
 976 }
 977
 978 static void add_lines_to_move_detection(struct diff_options *o,
 979                                         struct hashmap *add_lines,
 980                                         struct hashmap *del_lines)
 981 {
 982         struct moved_entry *prev_line = NULL;
 983
 984         int n;
 985         for (n = 0; n < o->emitted_symbols->nr; n++) {
 986                 struct hashmap *hm;
 987                 struct moved_entry *key;
 988
 989                 switch (o->emitted_symbols->buf[n].s) {
 990                 case DIFF_SYMBOL_PLUS:
 991                         hm = add_lines;
 992                         break;
 993                 case DIFF_SYMBOL_MINUS:
 994                         hm = del_lines;
 995                         break;
 996                 default:
 997                         prev_line = NULL;
 998                         continue;
 999                 }
1000
1001                 if (o->color_moved_ws_handling &
1002                     COLOR_MOVED_WS_ALLOW_INDENTATION_CHANGE)
1003                         fill_es_indent_data(&o->emitted_symbols->buf[n]);
1004                 key = prepare_entry(o, n);
1005                 if (prev_line && prev_line->es->s == o->emitted_symbols->buf[n].s)
1006                         prev_line->next_line = key;
1007
1008                 hashmap_add(hm, &key->ent);
1009                 prev_line = key;
1010         }
1011 }
1012
1013 static void pmb_advance_or_null(struct diff_options *o,
1014                                 struct moved_entry *match,
1015                                 struct hashmap *hm,
1016                                 struct moved_block *pmb,
1017                                 int pmb_nr)
1018 {
1019         int i;
1020         for (i = 0; i < pmb_nr; i++) {
1021                 struct moved_entry *prev = pmb[i].match;
1022                 struct moved_entry *cur = (prev && prev->next_line) ?
1023                                 prev->next_line : NULL;
1024                 if (cur && !hm->cmpfn(o, &cur->ent, &match->ent, NULL)) {
1025                         pmb[i].match = cur;
1026                 } else {
1027                         pmb[i].match = NULL;
1028                 }
1029         }
1030 }
1031
1032 static void pmb_advance_or_null_multi_match(struct diff_options *o,
1033                                             struct moved_entry *match,
1034                                             struct hashmap *hm,
1035                                             struct moved_block *pmb,
1036                                             int pmb_nr, int n)
1037 {
1038         int i;
1039         char *got_match = xcalloc(1, pmb_nr);
1040
1041         hashmap_for_each_entry_from(hm, match, ent) {
1042                 for (i = 0; i < pmb_nr; i++) {
1043                         struct moved_entry *prev = pmb[i].match;
1044                         struct moved_entry *cur = (prev && prev->next_line) ?
1045                                         prev->next_line : NULL;
1046                         if (!cur)
1047                                 continue;
1048                         if (!cmp_in_block_with_wsd(o, cur, match, &pmb[i], n))
1049                                 got_match[i] |= 1;
1050                 }
1051         }
1052
1053         for (i = 0; i < pmb_nr; i++) {
1054                 if (got_match[i]) {
1055                         /* Advance to the next line */
1056                         pmb[i].match = pmb[i].match->next_line;
1057                 } else {
1058                         moved_block_clear(&pmb[i]);
1059                 }
1060         }
1061
1062         free(got_match);
1063 }
1064
1065 static int shrink_potential_moved_blocks(struct moved_block *pmb,
1066                                          int pmb_nr)
1067 {
1068         int lp, rp;
1069
1070         /* Shrink the set of potential block to the remaining running */
1071         for (lp = 0, rp = pmb_nr - 1; lp <= rp;) {
1072                 while (lp < pmb_nr && pmb[lp].match)
1073                         lp++;
1074                 /* lp points at the first NULL now */
1075
1076                 while (rp > -1 && !pmb[rp].match)
1077                         rp--;
1078                 /* rp points at the last non-NULL */
1079
1080                 if (lp < pmb_nr && rp > -1 && lp < rp) {
1081                         pmb[lp] = pmb[rp];
1082                         memset(&pmb[rp], 0, sizeof(pmb[rp]));
1083                         rp--;
1084                         lp++;
1085                 }
1086         }
1087
1088         /* Remember the number of running sets */
1089         return rp + 1;
1090 }
1091
1092 /*
1093  * If o->color_moved is COLOR_MOVED_PLAIN, this function does nothing.
1094  *
1095  * Otherwise, if the last block has fewer alphanumeric characters than
1096  * COLOR_MOVED_MIN_ALNUM_COUNT, unset DIFF_SYMBOL_MOVED_LINE on all lines in
1097  * that block.
1098  *
1099  * The last block consists of the (n - block_length)'th line up to but not
1100  * including the nth line.
1101  *
1102  * Returns 0 if the last block is empty or is unset by this function, non zero
1103  * otherwise.
1104  *
1105  * NEEDSWORK: This uses the same heuristic as blame_entry_score() in blame.c.
1106  * Think of a way to unify them.
1107  */
1108 static int adjust_last_block(struct diff_options *o, int n, int block_length)
1109 {
1110         int i, alnum_count = 0;
1111         if (o->color_moved == COLOR_MOVED_PLAIN)
1112                 return block_length;
1113         for (i = 1; i < block_length + 1; i++) {
1114                 const char *c = o->emitted_symbols->buf[n - i].line;
1115                 for (; *c; c++) {
1116                         if (!isalnum(*c))
1117                                 continue;
1118                         alnum_count++;
1119                         if (alnum_count >= COLOR_MOVED_MIN_ALNUM_COUNT)
1120                                 return 1;
1121                 }
1122         }
1123         for (i = 1; i < block_length + 1; i++)
1124                 o->emitted_symbols->buf[n - i].flags &= ~DIFF_SYMBOL_MOVED_LINE;
1125         return 0;
1126 }
1127
1128 /* Find blocks of moved code, delegate actual coloring decision to helper */
1129 static void mark_color_as_moved(struct diff_options *o,
1130                                 struct hashmap *add_lines,
1131                                 struct hashmap *del_lines)
1132 {
1133         struct moved_block *pmb = NULL; /* potentially moved blocks */
1134         int pmb_nr = 0, pmb_alloc = 0;
1135         int n, flipped_block = 0, block_length = 0;
1136
1137
1138         for (n = 0; n < o->emitted_symbols->nr; n++) {
1139                 struct hashmap *hm = NULL;
1140                 struct moved_entry *key;
1141                 struct moved_entry *match = NULL;
1142                 struct emitted_diff_symbol *l = &o->emitted_symbols->buf[n];
1143                 enum diff_symbol last_symbol = 0;
1144
1145                 switch (l->s) {
1146                 case DIFF_SYMBOL_PLUS:
1147                         hm = del_lines;
1148                         key = prepare_entry(o, n);
1149                         match = hashmap_get_entry(hm, key, ent, NULL);
1150                         free(key);
1151                         break;
1152                 case DIFF_SYMBOL_MINUS:
1153                         hm = add_lines;
1154                         key = prepare_entry(o, n);
1155                         match = hashmap_get_entry(hm, key, ent, NULL);
1156                         free(key);
1157                         break;
1158                 default:
1159                         flipped_block = 0;
1160                 }
1161
1162                 if (!match) {
1163                         int i;
1164
1165                         adjust_last_block(o, n, block_length);
1166                         for(i = 0; i < pmb_nr; i++)
1167                                 moved_block_clear(&pmb[i]);
1168                         pmb_nr = 0;
1169                         block_length = 0;
1170                         flipped_block = 0;
1171                         last_symbol = l->s;
1172                         continue;
1173                 }
1174
1175                 if (o->color_moved == COLOR_MOVED_PLAIN) {
1176                         last_symbol = l->s;
1177                         l->flags |= DIFF_SYMBOL_MOVED_LINE;
1178                         continue;
1179                 }
1180
1181                 if (o->color_moved_ws_handling &
1182                     COLOR_MOVED_WS_ALLOW_INDENTATION_CHANGE)
1183                         pmb_advance_or_null_multi_match(o, match, hm, pmb, pmb_nr, n);
1184                 else
1185                         pmb_advance_or_null(o, match, hm, pmb, pmb_nr);
1186
1187                 pmb_nr = shrink_potential_moved_blocks(pmb, pmb_nr);
1188
1189                 if (pmb_nr == 0) {
1190                         /*
1191                          * The current line is the start of a new block.
1192                          * Setup the set of potential blocks.
1193                          */
1194                         hashmap_for_each_entry_from(hm, match, ent) {
1195                                 ALLOC_GROW(pmb, pmb_nr + 1, pmb_alloc);
1196                                 if (o->color_moved_ws_handling &
1197                                     COLOR_MOVED_WS_ALLOW_INDENTATION_CHANGE) {
1198                                         if (compute_ws_delta(l, match->es,
1199                                                              &pmb[pmb_nr].wsd))
1200                                                 pmb[pmb_nr++].match = match;
1201                                 } else {
1202                                         pmb[pmb_nr].wsd = 0;
1203                                         pmb[pmb_nr++].match = match;
1204                                 }
1205                         }
1206
1207                         if (adjust_last_block(o, n, block_length) &&
1208                             pmb_nr && last_symbol != l->s)
1209                                 flipped_block = (flipped_block + 1) % 2;
1210                         else
1211                                 flipped_block = 0;
1212
1213                         block_length = 0;
1214                 }
1215
1216                 if (pmb_nr) {
1217                         block_length++;
1218                         l->flags |= DIFF_SYMBOL_MOVED_LINE;
1219                         if (flipped_block && o->color_moved != COLOR_MOVED_BLOCKS)
1220                                 l->flags |= DIFF_SYMBOL_MOVED_LINE_ALT;
1221                 }
1222                 last_symbol = l->s;
1223         }
1224         adjust_last_block(o, n, block_length);
1225
1226         for(n = 0; n < pmb_nr; n++)
1227                 moved_block_clear(&pmb[n]);
1228         free(pmb);
1229 }
1230
1231 #define DIFF_SYMBOL_MOVED_LINE_ZEBRA_MASK \
1232   (DIFF_SYMBOL_MOVED_LINE | DIFF_SYMBOL_MOVED_LINE_ALT)
1233 static void dim_moved_lines(struct diff_options *o)
1234 {
1235         int n;
1236         for (n = 0; n < o->emitted_symbols->nr; n++) {
1237                 struct emitted_diff_symbol *prev = (n != 0) ?
1238                                 &o->emitted_symbols->buf[n - 1] : NULL;
1239                 struct emitted_diff_symbol *l = &o->emitted_symbols->buf[n];
1240                 struct emitted_diff_symbol *next =
1241                                 (n < o->emitted_symbols->nr - 1) ?
1242                                 &o->emitted_symbols->buf[n + 1] : NULL;
1243
1244                 /* Not a plus or minus line? */
1245                 if (l->s != DIFF_SYMBOL_PLUS && l->s != DIFF_SYMBOL_MINUS)
1246                         continue;
1247
1248                 /* Not a moved line? */
1249                 if (!(l->flags & DIFF_SYMBOL_MOVED_LINE))
1250                         continue;
1251
1252                 /*
1253                  * If prev or next are not a plus or minus line,
1254                  * pretend they don't exist
1255                  */
1256                 if (prev && prev->s != DIFF_SYMBOL_PLUS &&
1257                             prev->s != DIFF_SYMBOL_MINUS)
1258                         prev = NULL;
1259                 if (next && next->s != DIFF_SYMBOL_PLUS &&
1260                             next->s != DIFF_SYMBOL_MINUS)
1261                         next = NULL;
1262
1263                 /* Inside a block? */
1264                 if ((prev &&
1265                     (prev->flags & DIFF_SYMBOL_MOVED_LINE_ZEBRA_MASK) ==
1266                     (l->flags & DIFF_SYMBOL_MOVED_LINE_ZEBRA_MASK)) &&
1267                     (next &&
1268                     (next->flags & DIFF_SYMBOL_MOVED_LINE_ZEBRA_MASK) ==
1269                     (l->flags & DIFF_SYMBOL_MOVED_LINE_ZEBRA_MASK))) {
1270                         l->flags |= DIFF_SYMBOL_MOVED_LINE_UNINTERESTING;
1271                         continue;
1272                 }
1273
1274                 /* Check if we are at an interesting bound: */
1275                 if (prev && (prev->flags & DIFF_SYMBOL_MOVED_LINE) &&
1276                     (prev->flags & DIFF_SYMBOL_MOVED_LINE_ALT) !=
1277                        (l->flags & DIFF_SYMBOL_MOVED_LINE_ALT))
1278                         continue;
1279                 if (next && (next->flags & DIFF_SYMBOL_MOVED_LINE) &&
1280                     (next->flags & DIFF_SYMBOL_MOVED_LINE_ALT) !=
1281                        (l->flags & DIFF_SYMBOL_MOVED_LINE_ALT))
1282                         continue;
1283
1284                 /*
1285                  * The boundary to prev and next are not interesting,
1286                  * so this line is not interesting as a whole
1287                  */
1288                 l->flags |= DIFF_SYMBOL_MOVED_LINE_UNINTERESTING;
1289         }
1290 }
1291
1292 static void emit_line_ws_markup(struct diff_options *o,
1293                                 const char *set_sign, const char *set,
1294                                 const char *reset,
1295                                 int sign_index, const char *line, int len,
1296                                 unsigned ws_rule, int blank_at_eof)
1297 {
1298         const char *ws = NULL;
1299         int sign = o->output_indicators[sign_index];
1300
1301         if (o->ws_error_highlight & ws_rule) {
1302                 ws = diff_get_color_opt(o, DIFF_WHITESPACE);
1303                 if (!*ws)
1304                         ws = NULL;
1305         }
1306
1307         if (!ws && !set_sign)
1308                 emit_line_0(o, set, NULL, 0, reset, sign, line, len);
1309         else if (!ws) {
1310                 emit_line_0(o, set_sign, set, !!set_sign, reset, sign, line, len);
1311         } else if (blank_at_eof)
1312                 /* Blank line at EOF - paint '+' as well */
1313                 emit_line_0(o, ws, NULL, 0, reset, sign, line, len);
1314         else {
1315                 /* Emit just the prefix, then the rest. */
1316                 emit_line_0(o, set_sign ? set_sign : set, NULL, !!set_sign, reset,
1317                             sign, "", 0);
1318                 ws_check_emit(line, len, ws_rule,
1319                               o->file, set, reset, ws);
1320         }
1321 }
1322
1323 static void emit_diff_symbol_from_struct(struct diff_options *o,
1324                                          struct emitted_diff_symbol *eds)
1325 {
1326         static const char *nneof = " No newline at end of file\n";
1327         const char *context, *reset, *set, *set_sign, *meta, *fraginfo;
1328         struct strbuf sb = STRBUF_INIT;
1329
1330         enum diff_symbol s = eds->s;
1331         const char *line = eds->line;
1332         int len = eds->len;
1333         unsigned flags = eds->flags;
1334
1335         switch (s) {
1336         case DIFF_SYMBOL_NO_LF_EOF:
1337                 context = diff_get_color_opt(o, DIFF_CONTEXT);
1338                 reset = diff_get_color_opt(o, DIFF_RESET);
1339                 putc('\n', o->file);
1340                 emit_line_0(o, context, NULL, 0, reset, '\\',
1341                             nneof, strlen(nneof));
1342                 break;
1343         case DIFF_SYMBOL_SUBMODULE_HEADER:
1344         case DIFF_SYMBOL_SUBMODULE_ERROR:
1345         case DIFF_SYMBOL_SUBMODULE_PIPETHROUGH:
1346         case DIFF_SYMBOL_STATS_SUMMARY_INSERTS_DELETES:
1347         case DIFF_SYMBOL_SUMMARY:
1348         case DIFF_SYMBOL_STATS_LINE:
1349         case DIFF_SYMBOL_BINARY_DIFF_BODY:
1350         case DIFF_SYMBOL_CONTEXT_FRAGINFO:
1351                 emit_line(o, "", "", line, len);
1352                 break;
1353         case DIFF_SYMBOL_CONTEXT_INCOMPLETE:
1354         case DIFF_SYMBOL_CONTEXT_MARKER:
1355                 context = diff_get_color_opt(o, DIFF_CONTEXT);
1356                 reset = diff_get_color_opt(o, DIFF_RESET);
1357                 emit_line(o, context, reset, line, len);
1358                 break;
1359         case DIFF_SYMBOL_SEPARATOR:
1360                 fprintf(o->file, "%s%c",
1361                         diff_line_prefix(o),
1362                         o->line_termination);
1363                 break;
1364         case DIFF_SYMBOL_CONTEXT:
1365                 set = diff_get_color_opt(o, DIFF_CONTEXT);
1366                 reset = diff_get_color_opt(o, DIFF_RESET);
1367                 set_sign = NULL;
1368                 if (o->flags.dual_color_diffed_diffs) {
1369                         char c = !len ? 0 : line[0];
1370
1371                         if (c == '+')
1372                                 set = diff_get_color_opt(o, DIFF_FILE_NEW);
1373                         else if (c == '@')
1374                                 set = diff_get_color_opt(o, DIFF_FRAGINFO);
1375                         else if (c == '-')
1376                                 set = diff_get_color_opt(o, DIFF_FILE_OLD);
1377                 }
1378                 emit_line_ws_markup(o, set_sign, set, reset,
1379                                     OUTPUT_INDICATOR_CONTEXT, line, len,
1380                                     flags & (DIFF_SYMBOL_CONTENT_WS_MASK), 0);
1381                 break;
1382         case DIFF_SYMBOL_PLUS:
1383                 switch (flags & (DIFF_SYMBOL_MOVED_LINE |
1384                                  DIFF_SYMBOL_MOVED_LINE_ALT |
1385                                  DIFF_SYMBOL_MOVED_LINE_UNINTERESTING)) {
1386                 case DIFF_SYMBOL_MOVED_LINE |
1387                      DIFF_SYMBOL_MOVED_LINE_ALT |
1388                      DIFF_SYMBOL_MOVED_LINE_UNINTERESTING:
1389                         set = diff_get_color_opt(o, DIFF_FILE_NEW_MOVED_ALT_DIM);
1390                         break;
1391                 case DIFF_SYMBOL_MOVED_LINE |
1392                      DIFF_SYMBOL_MOVED_LINE_ALT:
1393                         set = diff_get_color_opt(o, DIFF_FILE_NEW_MOVED_ALT);
1394                         break;
1395                 case DIFF_SYMBOL_MOVED_LINE |
1396                      DIFF_SYMBOL_MOVED_LINE_UNINTERESTING:
1397                         set = diff_get_color_opt(o, DIFF_FILE_NEW_MOVED_DIM);
1398                         break;
1399                 case DIFF_SYMBOL_MOVED_LINE:
1400                         set = diff_get_color_opt(o, DIFF_FILE_NEW_MOVED);
1401                         break;
1402                 default:
1403                         set = diff_get_color_opt(o, DIFF_FILE_NEW);
1404                 }
1405                 reset = diff_get_color_opt(o, DIFF_RESET);
1406                 if (!o->flags.dual_color_diffed_diffs)
1407                         set_sign = NULL;
1408                 else {
1409                         char c = !len ? 0 : line[0];
1410
1411                         set_sign = set;
1412                         if (c == '-')
1413                                 set = diff_get_color_opt(o, DIFF_FILE_OLD_BOLD);
1414                         else if (c == '@')
1415                                 set = diff_get_color_opt(o, DIFF_FRAGINFO);
1416                         else if (c == '+')
1417                                 set = diff_get_color_opt(o, DIFF_FILE_NEW_BOLD);
1418                         else
1419                                 set = diff_get_color_opt(o, DIFF_CONTEXT_BOLD);
1420                         flags &= ~DIFF_SYMBOL_CONTENT_WS_MASK;
1421                 }
1422                 emit_line_ws_markup(o, set_sign, set, reset,
1423                                     OUTPUT_INDICATOR_NEW, line, len,
1424                                     flags & DIFF_SYMBOL_CONTENT_WS_MASK,
1425                                     flags & DIFF_SYMBOL_CONTENT_BLANK_LINE_EOF);
1426                 break;
1427         case DIFF_SYMBOL_MINUS:
1428                 switch (flags & (DIFF_SYMBOL_MOVED_LINE |
1429                                  DIFF_SYMBOL_MOVED_LINE_ALT |
1430                                  DIFF_SYMBOL_MOVED_LINE_UNINTERESTING)) {
1431                 case DIFF_SYMBOL_MOVED_LINE |
1432                      DIFF_SYMBOL_MOVED_LINE_ALT |
1433                      DIFF_SYMBOL_MOVED_LINE_UNINTERESTING:
1434                         set = diff_get_color_opt(o, DIFF_FILE_OLD_MOVED_ALT_DIM);
1435                         break;
1436                 case DIFF_SYMBOL_MOVED_LINE |
1437                      DIFF_SYMBOL_MOVED_LINE_ALT:
1438                         set = diff_get_color_opt(o, DIFF_FILE_OLD_MOVED_ALT);
1439                         break;
1440                 case DIFF_SYMBOL_MOVED_LINE |
1441                      DIFF_SYMBOL_MOVED_LINE_UNINTERESTING:
1442                         set = diff_get_color_opt(o, DIFF_FILE_OLD_MOVED_DIM);
1443                         break;
1444                 case DIFF_SYMBOL_MOVED_LINE:
1445                         set = diff_get_color_opt(o, DIFF_FILE_OLD_MOVED);
1446                         break;
1447                 default:
1448                         set = diff_get_color_opt(o, DIFF_FILE_OLD);
1449                 }
1450                 reset = diff_get_color_opt(o, DIFF_RESET);
1451                 if (!o->flags.dual_color_diffed_diffs)
1452                         set_sign = NULL;
1453                 else {
1454                         char c = !len ? 0 : line[0];
1455
1456                         set_sign = set;
1457                         if (c == '+')
1458                                 set = diff_get_color_opt(o, DIFF_FILE_NEW_DIM);
1459                         else if (c == '@')
1460                                 set = diff_get_color_opt(o, DIFF_FRAGINFO);
1461                         else if (c == '-')
1462                                 set = diff_get_color_opt(o, DIFF_FILE_OLD_DIM);
1463                         else
1464                                 set = diff_get_color_opt(o, DIFF_CONTEXT_DIM);
1465                 }
1466                 emit_line_ws_markup(o, set_sign, set, reset,
1467                                     OUTPUT_INDICATOR_OLD, line, len,
1468                                     flags & DIFF_SYMBOL_CONTENT_WS_MASK, 0);
1469                 break;
1470         case DIFF_SYMBOL_WORDS_PORCELAIN:
1471                 context = diff_get_color_opt(o, DIFF_CONTEXT);
1472                 reset = diff_get_color_opt(o, DIFF_RESET);
1473                 emit_line(o, context, reset, line, len);
1474                 fputs("~\n", o->file);
1475                 break;
1476         case DIFF_SYMBOL_WORDS:
1477                 context = diff_get_color_opt(o, DIFF_CONTEXT);
1478                 reset = diff_get_color_opt(o, DIFF_RESET);
1479                 /*
1480                  * Skip the prefix character, if any.  With
1481                  * diff_suppress_blank_empty, there may be
1482                  * none.
1483                  */
1484                 if (line[0] != '\n') {
1485                         line++;
1486                         len--;
1487                 }
1488                 emit_line(o, context, reset, line, len);
1489                 break;
1490         case DIFF_SYMBOL_FILEPAIR_PLUS:
1491                 meta = diff_get_color_opt(o, DIFF_METAINFO);
1492                 reset = diff_get_color_opt(o, DIFF_RESET);
1493                 fprintf(o->file, "%s%s+++ %s%s%s\n", diff_line_prefix(o), meta,
1494                         line, reset,
1495                         strchr(line, ' ') ? "\t" : "");
1496                 break;
1497         case DIFF_SYMBOL_FILEPAIR_MINUS:
1498                 meta = diff_get_color_opt(o, DIFF_METAINFO);
1499                 reset = diff_get_color_opt(o, DIFF_RESET);
1500                 fprintf(o->file, "%s%s--- %s%s%s\n", diff_line_prefix(o), meta,
1501                         line, reset,
1502                         strchr(line, ' ') ? "\t" : "");
1503                 break;
1504         case DIFF_SYMBOL_BINARY_FILES:
1505         case DIFF_SYMBOL_HEADER:
1506                 fprintf(o->file, "%s", line);
1507                 break;
1508         case DIFF_SYMBOL_BINARY_DIFF_HEADER:
1509                 fprintf(o->file, "%sGIT binary patch\n", diff_line_prefix(o));
1510                 break;
1511         case DIFF_SYMBOL_BINARY_DIFF_HEADER_DELTA:
1512                 fprintf(o->file, "%sdelta %s\n", diff_line_prefix(o), line);
1513                 break;
1514         case DIFF_SYMBOL_BINARY_DIFF_HEADER_LITERAL:
1515                 fprintf(o->file, "%sliteral %s\n", diff_line_prefix(o), line);
1516                 break;
1517         case DIFF_SYMBOL_BINARY_DIFF_FOOTER:
1518                 fputs(diff_line_prefix(o), o->file);
1519                 fputc('\n', o->file);
1520                 break;
1521         case DIFF_SYMBOL_REWRITE_DIFF:
1522                 fraginfo = diff_get_color(o->use_color, DIFF_FRAGINFO);
1523                 reset = diff_get_color_opt(o, DIFF_RESET);
1524                 emit_line(o, fraginfo, reset, line, len);
1525                 break;
1526         case DIFF_SYMBOL_SUBMODULE_ADD:
1527                 set = diff_get_color_opt(o, DIFF_FILE_NEW);
1528                 reset = diff_get_color_opt(o, DIFF_RESET);
1529                 emit_line(o, set, reset, line, len);
1530                 break;
1531         case DIFF_SYMBOL_SUBMODULE_DEL:
1532                 set = diff_get_color_opt(o, DIFF_FILE_OLD);
1533                 reset = diff_get_color_opt(o, DIFF_RESET);
1534                 emit_line(o, set, reset, line, len);
1535                 break;
1536         case DIFF_SYMBOL_SUBMODULE_UNTRACKED:
1537                 fprintf(o->file, "%sSubmodule %s contains untracked content\n",
1538                         diff_line_prefix(o), line);
1539                 break;
1540         case DIFF_SYMBOL_SUBMODULE_MODIFIED:
1541                 fprintf(o->file, "%sSubmodule %s contains modified content\n",
1542                         diff_line_prefix(o), line);
1543                 break;
1544         case DIFF_SYMBOL_STATS_SUMMARY_NO_FILES:
1545                 emit_line(o, "", "", " 0 files changed\n",
1546                           strlen(" 0 files changed\n"));
1547                 break;
1548         case DIFF_SYMBOL_STATS_SUMMARY_ABBREV:
1549                 emit_line(o, "", "", " ...\n", strlen(" ...\n"));
1550                 break;
1551         case DIFF_SYMBOL_WORD_DIFF:
1552                 fprintf(o->file, "%.*s", len, line);
1553                 break;
1554         case DIFF_SYMBOL_STAT_SEP:
1555                 fputs(o->stat_sep, o->file);
1556                 break;
1557         default:
1558                 BUG("unknown diff symbol");
1559         }
1560         strbuf_release(&sb);
1561 }
1562
1563 static void emit_diff_symbol(struct diff_options *o, enum diff_symbol s,
1564                              const char *line, int len, unsigned flags)
1565 {
1566         struct emitted_diff_symbol e = {line, len, flags, 0, 0, s};
1567
1568         if (o->emitted_symbols)
1569                 append_emitted_diff_symbol(o, &e);
1570         else
1571                 emit_diff_symbol_from_struct(o, &e);
1572 }
1573
1574 void diff_emit_submodule_del(struct diff_options *o, const char *line)
1575 {
1576         emit_diff_symbol(o, DIFF_SYMBOL_SUBMODULE_DEL, line, strlen(line), 0);
1577 }
1578
1579 void diff_emit_submodule_add(struct diff_options *o, const char *line)
1580 {
1581         emit_diff_symbol(o, DIFF_SYMBOL_SUBMODULE_ADD, line, strlen(line), 0);
1582 }
1583
1584 void diff_emit_submodule_untracked(struct diff_options *o, const char *path)
1585 {
1586         emit_diff_symbol(o, DIFF_SYMBOL_SUBMODULE_UNTRACKED,
1587                          path, strlen(path), 0);
1588 }
1589
1590 void diff_emit_submodule_modified(struct diff_options *o, const char *path)
1591 {
1592         emit_diff_symbol(o, DIFF_SYMBOL_SUBMODULE_MODIFIED,
1593                          path, strlen(path), 0);
1594 }
1595
1596 void diff_emit_submodule_header(struct diff_options *o, const char *header)
1597 {
1598         emit_diff_symbol(o, DIFF_SYMBOL_SUBMODULE_HEADER,
1599                          header, strlen(header), 0);
1600 }
1601
1602 void diff_emit_submodule_error(struct diff_options *o, const char *err)
1603 {
1604         emit_diff_symbol(o, DIFF_SYMBOL_SUBMODULE_ERROR, err, strlen(err), 0);
1605 }
1606
1607 void diff_emit_submodule_pipethrough(struct diff_options *o,
1608                                      const char *line, int len)
1609 {
1610         emit_diff_symbol(o, DIFF_SYMBOL_SUBMODULE_PIPETHROUGH, line, len, 0);
1611 }
1612
1613 static int new_blank_line_at_eof(struct emit_callback *ecbdata, const char *line, int len)
1614 {
1615         if (!((ecbdata->ws_rule & WS_BLANK_AT_EOF) &&
1616               ecbdata->blank_at_eof_in_preimage &&
1617               ecbdata->blank_at_eof_in_postimage &&
1618               ecbdata->blank_at_eof_in_preimage <= ecbdata->lno_in_preimage &&
1619               ecbdata->blank_at_eof_in_postimage <= ecbdata->lno_in_postimage))
1620                 return 0;
1621         return ws_blank_line(line, len, ecbdata->ws_rule);
1622 }
1623
1624 static void emit_add_line(struct emit_callback *ecbdata,
1625                           const char *line, int len)
1626 {
1627         unsigned flags = WSEH_NEW | ecbdata->ws_rule;
1628         if (new_blank_line_at_eof(ecbdata, line, len))
1629                 flags |= DIFF_SYMBOL_CONTENT_BLANK_LINE_EOF;
1630
1631         emit_diff_symbol(ecbdata->opt, DIFF_SYMBOL_PLUS, line, len, flags);
1632 }
1633
1634 static void emit_del_line(struct emit_callback *ecbdata,
1635                           const char *line, int len)
1636 {
1637         unsigned flags = WSEH_OLD | ecbdata->ws_rule;
1638         emit_diff_symbol(ecbdata->opt, DIFF_SYMBOL_MINUS, line, len, flags);
1639 }
1640
1641 static void emit_context_line(struct emit_callback *ecbdata,
1642                               const char *line, int len)
1643 {
1644         unsigned flags = WSEH_CONTEXT | ecbdata->ws_rule;
1645         emit_diff_symbol(ecbdata->opt, DIFF_SYMBOL_CONTEXT, line, len, flags);
1646 }
1647
1648 static void emit_hunk_header(struct emit_callback *ecbdata,
1649                              const char *line, int len)
1650 {
1651         const char *context = diff_get_color(ecbdata->color_diff, DIFF_CONTEXT);
1652         const char *frag = diff_get_color(ecbdata->color_diff, DIFF_FRAGINFO);
1653         const char *func = diff_get_color(ecbdata->color_diff, DIFF_FUNCINFO);
1654         const char *reset = diff_get_color(ecbdata->color_diff, DIFF_RESET);
1655         const char *reverse = ecbdata->color_diff ? GIT_COLOR_REVERSE : "";
1656         static const char atat[2] = { '@', '@' };
1657         const char *cp, *ep;
1658         struct strbuf msgbuf = STRBUF_INIT;
1659         int org_len = len;
1660         int i = 1;
1661
1662         /*
1663          * As a hunk header must begin with "@@ -<old>, +<new> @@",
1664          * it always is at least 10 bytes long.
1665          */
1666         if (len < 10 ||
1667             memcmp(line, atat, 2) ||
1668             !(ep = memmem(line + 2, len - 2, atat, 2))) {
1669                 emit_diff_symbol(ecbdata->opt,
1670                                  DIFF_SYMBOL_CONTEXT_MARKER, line, len, 0);
1671                 return;
1672         }
1673         ep += 2; /* skip over @@ */
1674
1675         /* The hunk header in fraginfo color */
1676         if (ecbdata->opt->flags.dual_color_diffed_diffs)
1677                 strbuf_addstr(&msgbuf, reverse);
1678         strbuf_addstr(&msgbuf, frag);
1679         if (ecbdata->opt->flags.suppress_hunk_header_line_count)
1680                 strbuf_add(&msgbuf, atat, sizeof(atat));
1681         else
1682                 strbuf_add(&msgbuf, line, ep - line);
1683         strbuf_addstr(&msgbuf, reset);
1684
1685         /*
1686          * trailing "\r\n"
1687          */
1688         for ( ; i < 3; i++)
1689                 if (line[len - i] == '\r' || line[len - i] == '\n')
1690                         len--;
1691
1692         /* blank before the func header */
1693         for (cp = ep; ep - line < len; ep++)
1694                 if (*ep != ' ' && *ep != '\t')
1695                         break;
1696         if (ep != cp) {
1697                 strbuf_addstr(&msgbuf, context);
1698                 strbuf_add(&msgbuf, cp, ep - cp);
1699                 strbuf_addstr(&msgbuf, reset);
1700         }
1701
1702         if (ep < line + len) {
1703                 strbuf_addstr(&msgbuf, func);
1704                 strbuf_add(&msgbuf, ep, line + len - ep);
1705                 strbuf_addstr(&msgbuf, reset);
1706         }
1707
1708         strbuf_add(&msgbuf, line + len, org_len - len);
1709         strbuf_complete_line(&msgbuf);
1710         emit_diff_symbol(ecbdata->opt,
1711                          DIFF_SYMBOL_CONTEXT_FRAGINFO, msgbuf.buf, msgbuf.len, 0);
1712         strbuf_release(&msgbuf);
1713 }
1714
1715 static struct diff_tempfile *claim_diff_tempfile(void)
1716 {
1717         int i;
1718         for (i = 0; i < ARRAY_SIZE(diff_temp); i++)
1719                 if (!diff_temp[i].name)
1720                         return diff_temp + i;
1721         BUG("diff is failing to clean up its tempfiles");
1722 }
1723
1724 static void remove_tempfile(void)
1725 {
1726         int i;
1727         for (i = 0; i < ARRAY_SIZE(diff_temp); i++) {
1728                 if (is_tempfile_active(diff_temp[i].tempfile))
1729                         delete_tempfile(&diff_temp[i].tempfile);
1730                 diff_temp[i].name = NULL;
1731         }
1732 }
1733
1734 static void add_line_count(struct strbuf *out, int count)
1735 {
1736         switch (count) {
1737         case 0:
1738                 strbuf_addstr(out, "0,0");
1739                 break;
1740         case 1:
1741                 strbuf_addstr(out, "1");
1742                 break;
1743         default:
1744                 strbuf_addf(out, "1,%d", count);
1745                 break;
1746         }
1747 }
1748
1749 static void emit_rewrite_lines(struct emit_callback *ecb,
1750                                int prefix, const char *data, int size)
1751 {
1752         const char *endp = NULL;
1753
1754         while (0 < size) {
1755                 int len;
1756
1757                 endp = memchr(data, '\n', size);
1758                 len = endp ? (endp - data + 1) : size;
1759                 if (prefix != '+') {
1760                         ecb->lno_in_preimage++;
1761                         emit_del_line(ecb, data, len);
1762                 } else {
1763                         ecb->lno_in_postimage++;
1764                         emit_add_line(ecb, data, len);
1765                 }
1766                 size -= len;
1767                 data += len;
1768         }
1769         if (!endp)
1770                 emit_diff_symbol(ecb->opt, DIFF_SYMBOL_NO_LF_EOF, NULL, 0, 0);
1771 }
1772
1773 static void emit_rewrite_diff(const char *name_a,
1774                               const char *name_b,
1775                               struct diff_filespec *one,
1776                               struct diff_filespec *two,
1777                               struct userdiff_driver *textconv_one,
1778                               struct userdiff_driver *textconv_two,
1779                               struct diff_options *o)
1780 {
1781         int lc_a, lc_b;
1782         static struct strbuf a_name = STRBUF_INIT, b_name = STRBUF_INIT;
1783         const char *a_prefix, *b_prefix;
1784         char *data_one, *data_two;
1785         size_t size_one, size_two;
1786         struct emit_callback ecbdata;
1787         struct strbuf out = STRBUF_INIT;
1788
1789         if (diff_mnemonic_prefix && o->flags.reverse_diff) {
1790                 a_prefix = o->b_prefix;
1791                 b_prefix = o->a_prefix;
1792         } else {
1793                 a_prefix = o->a_prefix;
1794                 b_prefix = o->b_prefix;
1795         }
1796
1797         name_a += (*name_a == '/');
1798         name_b += (*name_b == '/');
1799
1800         strbuf_reset(&a_name);
1801         strbuf_reset(&b_name);
1802         quote_two_c_style(&a_name, a_prefix, name_a, 0);
1803         quote_two_c_style(&b_name, b_prefix, name_b, 0);
1804
1805         size_one = fill_textconv(o->repo, textconv_one, one, &data_one);
1806         size_two = fill_textconv(o->repo, textconv_two, two, &data_two);
1807
1808         memset(&ecbdata, 0, sizeof(ecbdata));
1809         ecbdata.color_diff = want_color(o->use_color);
1810         ecbdata.ws_rule = whitespace_rule(o->repo->index, name_b);
1811         ecbdata.opt = o;
1812         if (ecbdata.ws_rule & WS_BLANK_AT_EOF) {
1813                 mmfile_t mf1, mf2;
1814                 mf1.ptr = (char *)data_one;
1815                 mf2.ptr = (char *)data_two;
1816                 mf1.size = size_one;
1817                 mf2.size = size_two;
1818                 check_blank_at_eof(&mf1, &mf2, &ecbdata);
1819         }
1820         ecbdata.lno_in_preimage = 1;
1821         ecbdata.lno_in_postimage = 1;
1822
1823         lc_a = count_lines(data_one, size_one);
1824         lc_b = count_lines(data_two, size_two);
1825
1826         emit_diff_symbol(o, DIFF_SYMBOL_FILEPAIR_MINUS,
1827                          a_name.buf, a_name.len, 0);
1828         emit_diff_symbol(o, DIFF_SYMBOL_FILEPAIR_PLUS,
1829                          b_name.buf, b_name.len, 0);
1830
1831         strbuf_addstr(&out, "@@ -");
1832         if (!o->irreversible_delete)
1833                 add_line_count(&out, lc_a);
1834         else
1835                 strbuf_addstr(&out, "?,?");
1836         strbuf_addstr(&out, " +");
1837         add_line_count(&out, lc_b);
1838         strbuf_addstr(&out, " @@\n");
1839         emit_diff_symbol(o, DIFF_SYMBOL_REWRITE_DIFF, out.buf, out.len, 0);
1840         strbuf_release(&out);
1841
1842         if (lc_a && !o->irreversible_delete)
1843                 emit_rewrite_lines(&ecbdata, '-', data_one, size_one);
1844         if (lc_b)
1845                 emit_rewrite_lines(&ecbdata, '+', data_two, size_two);
1846         if (textconv_one)
1847                 free((char *)data_one);
1848         if (textconv_two)
1849                 free((char *)data_two);
1850 }
1851
1852 struct diff_words_buffer {
1853         mmfile_t text;
1854         unsigned long alloc;
1855         struct diff_words_orig {
1856                 const char *begin, *end;
1857         } *orig;
1858         int orig_nr, orig_alloc;
1859 };
1860
1861 static void diff_words_append(char *line, unsigned long len,
1862                 struct diff_words_buffer *buffer)
1863 {
1864         ALLOC_GROW(buffer->text.ptr, buffer->text.size + len, buffer->alloc);
1865         line++;
1866         len--;
1867         memcpy(buffer->text.ptr + buffer->text.size, line, len);
1868         buffer->text.size += len;
1869         buffer->text.ptr[buffer->text.size] = '\0';
1870 }
1871
1872 struct diff_words_style_elem {
1873         const char *prefix;
1874         const char *suffix;
1875         const char *color; /* NULL; filled in by the setup code if
1876                             * color is enabled */
1877 };
1878
1879 struct diff_words_style {
1880         enum diff_words_type type;
1881         struct diff_words_style_elem new_word, old_word, ctx;
1882         const char *newline;
1883 };
1884
1885 static struct diff_words_style diff_words_styles[] = {
1886         { DIFF_WORDS_PORCELAIN, {"+", "\n"}, {"-", "\n"}, {" ", "\n"}, "~\n" },
1887         { DIFF_WORDS_PLAIN, {"{+", "+}"}, {"[-", "-]"}, {"", ""}, "\n" },
1888         { DIFF_WORDS_COLOR, {"", ""}, {"", ""}, {"", ""}, "\n" }
1889 };
1890
1891 struct diff_words_data {
1892         struct diff_words_buffer minus, plus;
1893         const char *current_plus;
1894         int last_minus;
1895         struct diff_options *opt;
1896         regex_t *word_regex;
1897         enum diff_words_type type;
1898         struct diff_words_style *style;
1899 };
1900
1901 static int fn_out_diff_words_write_helper(struct diff_options *o,
1902                                           struct diff_words_style_elem *st_el,
1903                                           const char *newline,
1904                                           size_t count, const char *buf)
1905 {
1906         int print = 0;
1907         struct strbuf sb = STRBUF_INIT;
1908
1909         while (count) {
1910                 char *p = memchr(buf, '\n', count);
1911                 if (print)
1912                         strbuf_addstr(&sb, diff_line_prefix(o));
1913
1914                 if (p != buf) {
1915                         const char *reset = st_el->color && *st_el->color ?
1916                                             GIT_COLOR_RESET : NULL;
1917                         if (st_el->color && *st_el->color)
1918                                 strbuf_addstr(&sb, st_el->color);
1919                         strbuf_addstr(&sb, st_el->prefix);
1920                         strbuf_add(&sb, buf, p ? p - buf : count);
1921                         strbuf_addstr(&sb, st_el->suffix);
1922                         if (reset)
1923                                 strbuf_addstr(&sb, reset);
1924                 }
1925                 if (!p)
1926                         goto out;
1927
1928                 strbuf_addstr(&sb, newline);
1929                 count -= p + 1 - buf;
1930                 buf = p + 1;
1931                 print = 1;
1932                 if (count) {
1933                         emit_diff_symbol(o, DIFF_SYMBOL_WORD_DIFF,
1934                                          sb.buf, sb.len, 0);
1935                         strbuf_reset(&sb);
1936                 }
1937         }
1938
1939 out:
1940         if (sb.len)
1941                 emit_diff_symbol(o, DIFF_SYMBOL_WORD_DIFF,
1942                                  sb.buf, sb.len, 0);
1943         strbuf_release(&sb);
1944         return 0;
1945 }
1946
1947 /*
1948  * '--color-words' algorithm can be described as:
1949  *
1950  *   1. collect the minus/plus lines of a diff hunk, divided into
1951  *      minus-lines and plus-lines;
1952  *
1953  *   2. break both minus-lines and plus-lines into words and
1954  *      place them into two mmfile_t with one word for each line;
1955  *
1956  *   3. use xdiff to run diff on the two mmfile_t to get the words level diff;
1957  *
1958  * And for the common parts of the both file, we output the plus side text.
1959  * diff_words->current_plus is used to trace the current position of the plus file
1960  * which printed. diff_words->last_minus is used to trace the last minus word
1961  * printed.
1962  *
1963  * For '--graph' to work with '--color-words', we need to output the graph prefix
1964  * on each line of color words output. Generally, there are two conditions on
1965  * which we should output the prefix.
1966  *
1967  *   1. diff_words->last_minus == 0 &&
1968  *      diff_words->current_plus == diff_words->plus.text.ptr
1969  *
1970  *      that is: the plus text must start as a new line, and if there is no minus
1971  *      word printed, a graph prefix must be printed.
1972  *
1973  *   2. diff_words->current_plus > diff_words->plus.text.ptr &&
1974  *      *(diff_words->current_plus - 1) == '\n'
1975  *
1976  *      that is: a graph prefix must be printed following a '\n'
1977  */
1978 static int color_words_output_graph_prefix(struct diff_words_data *diff_words)
1979 {
1980         if ((diff_words->last_minus == 0 &&
1981                 diff_words->current_plus == diff_words->plus.text.ptr) ||
1982                 (diff_words->current_plus > diff_words->plus.text.ptr &&
1983                 *(diff_words->current_plus - 1) == '\n')) {
1984                 return 1;
1985         } else {
1986                 return 0;
1987         }
1988 }
1989
1990 static void fn_out_diff_words_aux(void *priv,
1991                                   long minus_first, long minus_len,
1992                                   long plus_first, long plus_len,
1993                                   const char *func, long funclen)
1994 {
1995         struct diff_words_data *diff_words = priv;
1996         struct diff_words_style *style = diff_words->style;
1997         const char *minus_begin, *minus_end, *plus_begin, *plus_end;
1998         struct diff_options *opt = diff_words->opt;
1999         const char *line_prefix;
2000
2001         assert(opt);
2002         line_prefix = diff_line_prefix(opt);
2003
2004         /* POSIX requires that first be decremented by one if len == 0... */
2005         if (minus_len) {
2006                 minus_begin = diff_words->minus.orig[minus_first].begin;
2007                 minus_end =
2008                         diff_words->minus.orig[minus_first + minus_len - 1].end;
2009         } else
2010                 minus_begin = minus_end =
2011                         diff_words->minus.orig[minus_first].end;
2012
2013         if (plus_len) {
2014                 plus_begin = diff_words->plus.orig[plus_first].begin;
2015                 plus_end = diff_words->plus.orig[plus_first + plus_len - 1].end;
2016         } else
2017                 plus_begin = plus_end = diff_words->plus.orig[plus_first].end;
2018
2019         if (color_words_output_graph_prefix(diff_words)) {
2020                 fputs(line_prefix, diff_words->opt->file);
2021         }
2022         if (diff_words->current_plus != plus_begin) {
2023                 fn_out_diff_words_write_helper(diff_words->opt,
2024                                 &style->ctx, style->newline,
2025                                 plus_begin - diff_words->current_plus,
2026                                 diff_words->current_plus);
2027         }
2028         if (minus_begin != minus_end) {
2029                 fn_out_diff_words_write_helper(diff_words->opt,
2030                                 &style->old_word, style->newline,
2031                                 minus_end - minus_begin, minus_begin);
2032         }
2033         if (plus_begin != plus_end) {
2034                 fn_out_diff_words_write_helper(diff_words->opt,
2035                                 &style->new_word, style->newline,
2036                                 plus_end - plus_begin, plus_begin);
2037         }
2038
2039         diff_words->current_plus = plus_end;
2040         diff_words->last_minus = minus_first;
2041 }
2042
2043 /* This function starts looking at *begin, and returns 0 iff a word was found. */
2044 static int find_word_boundaries(mmfile_t *buffer, regex_t *word_regex,
2045                 int *begin, int *end)
2046 {
2047         if (word_regex && *begin < buffer->size) {
2048                 regmatch_t match[1];
2049                 if (!regexec_buf(word_regex, buffer->ptr + *begin,
2050                                  buffer->size - *begin, 1, match, 0)) {
2051                         char *p = memchr(buffer->ptr + *begin + match[0].rm_so,
2052                                         '\n', match[0].rm_eo - match[0].rm_so);
2053                         *end = p ? p - buffer->ptr : match[0].rm_eo + *begin;
2054                         *begin += match[0].rm_so;
2055                         return *begin >= *end;
2056                 }
2057                 return -1;
2058         }
2059
2060         /* find the next word */
2061         while (*begin < buffer->size && isspace(buffer->ptr[*begin]))
2062                 (*begin)++;
2063         if (*begin >= buffer->size)
2064                 return -1;
2065
2066         /* find the end of the word */
2067         *end = *begin + 1;
2068         while (*end < buffer->size && !isspace(buffer->ptr[*end]))
2069                 (*end)++;
2070
2071         return 0;
2072 }
2073
2074 /*
2075  * This function splits the words in buffer->text, stores the list with
2076  * newline separator into out, and saves the offsets of the original words
2077  * in buffer->orig.
2078  */
2079 static void diff_words_fill(struct diff_words_buffer *buffer, mmfile_t *out,
2080                 regex_t *word_regex)
2081 {
2082         int i, j;
2083         long alloc = 0;
2084
2085         out->size = 0;
2086         out->ptr = NULL;
2087
2088         /* fake an empty "0th" word */
2089         ALLOC_GROW(buffer->orig, 1, buffer->orig_alloc);
2090         buffer->orig[0].begin = buffer->orig[0].end = buffer->text.ptr;
2091         buffer->orig_nr = 1;
2092
2093         for (i = 0; i < buffer->text.size; i++) {
2094                 if (find_word_boundaries(&buffer->text, word_regex, &i, &j))
2095                         return;
2096
2097                 /* store original boundaries */
2098                 ALLOC_GROW(buffer->orig, buffer->orig_nr + 1,
2099                                 buffer->orig_alloc);
2100                 buffer->orig[buffer->orig_nr].begin = buffer->text.ptr + i;
2101                 buffer->orig[buffer->orig_nr].end = buffer->text.ptr + j;
2102                 buffer->orig_nr++;
2103
2104                 /* store one word */
2105                 ALLOC_GROW(out->ptr, out->size + j - i + 1, alloc);
2106                 memcpy(out->ptr + out->size, buffer->text.ptr + i, j - i);
2107                 out->ptr[out->size + j - i] = '\n';
2108                 out->size += j - i + 1;
2109
2110                 i = j - 1;
2111         }
2112 }
2113
2114 /* this executes the word diff on the accumulated buffers */
2115 static void diff_words_show(struct diff_words_data *diff_words)
2116 {
2117         xpparam_t xpp;
2118         xdemitconf_t xecfg;
2119         mmfile_t minus, plus;
2120         struct diff_words_style *style = diff_words->style;
2121
2122         struct diff_options *opt = diff_words->opt;
2123         const char *line_prefix;
2124
2125         assert(opt);
2126         line_prefix = diff_line_prefix(opt);
2127
2128         /* special case: only removal */
2129         if (!diff_words->plus.text.size) {
2130                 emit_diff_symbol(diff_words->opt, DIFF_SYMBOL_WORD_DIFF,
2131                                  line_prefix, strlen(line_prefix), 0);
2132                 fn_out_diff_words_write_helper(diff_words->opt,
2133                         &style->old_word, style->newline,
2134                         diff_words->minus.text.size,
2135                         diff_words->minus.text.ptr);
2136                 diff_words->minus.text.size = 0;
2137                 return;
2138         }
2139
2140         diff_words->current_plus = diff_words->plus.text.ptr;
2141         diff_words->last_minus = 0;
2142
2143         memset(&xpp, 0, sizeof(xpp));
2144         memset(&xecfg, 0, sizeof(xecfg));
2145         diff_words_fill(&diff_words->minus, &minus, diff_words->word_regex);
2146         diff_words_fill(&diff_words->plus, &plus, diff_words->word_regex);
2147         xpp.flags = 0;
2148         /* as only the hunk header will be parsed, we need a 0-context */
2149         xecfg.ctxlen = 0;
2150         if (xdi_diff_outf(&minus, &plus, fn_out_diff_words_aux, NULL,
2151                           diff_words, &xpp, &xecfg))
2152                 die("unable to generate word diff");
2153         free(minus.ptr);
2154         free(plus.ptr);
2155         if (diff_words->current_plus != diff_words->plus.text.ptr +
2156                         diff_words->plus.text.size) {
2157                 if (color_words_output_graph_prefix(diff_words))
2158                         emit_diff_symbol(diff_words->opt, DIFF_SYMBOL_WORD_DIFF,
2159                                          line_prefix, strlen(line_prefix), 0);
2160                 fn_out_diff_words_write_helper(diff_words->opt,
2161                         &style->ctx, style->newline,
2162                         diff_words->plus.text.ptr + diff_words->plus.text.size
2163                         - diff_words->current_plus, diff_words->current_plus);
2164         }
2165         diff_words->minus.text.size = diff_words->plus.text.size = 0;
2166 }
2167
2168 /* In "color-words" mode, show word-diff of words accumulated in the buffer */
2169 static void diff_words_flush(struct emit_callback *ecbdata)
2170 {
2171         struct diff_options *wo = ecbdata->diff_words->opt;
2172
2173         if (ecbdata->diff_words->minus.text.size ||
2174             ecbdata->diff_words->plus.text.size)
2175                 diff_words_show(ecbdata->diff_words);
2176
2177         if (wo->emitted_symbols) {
2178                 struct diff_options *o = ecbdata->opt;
2179                 struct emitted_diff_symbols *wol = wo->emitted_symbols;
2180                 int i;
2181
2182                 /*
2183                  * NEEDSWORK:
2184                  * Instead of appending each, concat all words to a line?
2185                  */
2186                 for (i = 0; i < wol->nr; i++)
2187                         append_emitted_diff_symbol(o, &wol->buf[i]);
2188
2189                 for (i = 0; i < wol->nr; i++)
2190                         free((void *)wol->buf[i].line);
2191
2192                 wol->nr = 0;
2193         }
2194 }
2195
2196 static void diff_filespec_load_driver(struct diff_filespec *one,
2197                                       struct index_state *istate)
2198 {
2199         /* Use already-loaded driver */
2200         if (one->driver)
2201                 return;
2202
2203         if (S_ISREG(one->mode))
2204                 one->driver = userdiff_find_by_path(istate, one->path);
2205
2206         /* Fallback to default settings */
2207         if (!one->driver)
2208                 one->driver = userdiff_find_by_name("default");
2209 }
2210
2211 static const char *userdiff_word_regex(struct diff_filespec *one,
2212                                        struct index_state *istate)
2213 {
2214         diff_filespec_load_driver(one, istate);
2215         return one->driver->word_regex;
2216 }
2217
2218 static void init_diff_words_data(struct emit_callback *ecbdata,
2219                                  struct diff_options *orig_opts,
2220                                  struct diff_filespec *one,
2221                                  struct diff_filespec *two)
2222 {
2223         int i;
2224         struct diff_options *o = xmalloc(sizeof(struct diff_options));
2225         memcpy(o, orig_opts, sizeof(struct diff_options));
2226
2227         ecbdata->diff_words =
2228                 xcalloc(1, sizeof(struct diff_words_data));
2229         ecbdata->diff_words->type = o->word_diff;
2230         ecbdata->diff_words->opt = o;
2231
2232         if (orig_opts->emitted_symbols)
2233                 o->emitted_symbols =
2234                         xcalloc(1, sizeof(struct emitted_diff_symbols));
2235
2236         if (!o->word_regex)
2237                 o->word_regex = userdiff_word_regex(one, o->repo->index);
2238         if (!o->word_regex)
2239                 o->word_regex = userdiff_word_regex(two, o->repo->index);
2240         if (!o->word_regex)
2241                 o->word_regex = diff_word_regex_cfg;
2242         if (o->word_regex) {
2243                 ecbdata->diff_words->word_regex = (regex_t *)
2244                         xmalloc(sizeof(regex_t));
2245                 if (regcomp(ecbdata->diff_words->word_regex,
2246                             o->word_regex,
2247                             REG_EXTENDED | REG_NEWLINE))
2248                         die("invalid regular expression: %s",
2249                             o->word_regex);
2250         }
2251         for (i = 0; i < ARRAY_SIZE(diff_words_styles); i++) {
2252                 if (o->word_diff == diff_words_styles[i].type) {
2253                         ecbdata->diff_words->style =
2254                                 &diff_words_styles[i];
2255                         break;
2256                 }
2257         }
2258         if (want_color(o->use_color)) {
2259                 struct diff_words_style *st = ecbdata->diff_words->style;
2260                 st->old_word.color = diff_get_color_opt(o, DIFF_FILE_OLD);
2261                 st->new_word.color = diff_get_color_opt(o, DIFF_FILE_NEW);
2262                 st->ctx.color = diff_get_color_opt(o, DIFF_CONTEXT);
2263         }
2264 }
2265
2266 static void free_diff_words_data(struct emit_callback *ecbdata)
2267 {
2268         if (ecbdata->diff_words) {
2269                 diff_words_flush(ecbdata);
2270                 free (ecbdata->diff_words->opt->emitted_symbols);
2271                 free (ecbdata->diff_words->opt);
2272                 free (ecbdata->diff_words->minus.text.ptr);
2273                 free (ecbdata->diff_words->minus.orig);
2274                 free (ecbdata->diff_words->plus.text.ptr);
2275                 free (ecbdata->diff_words->plus.orig);
2276                 if (ecbdata->diff_words->word_regex) {
2277                         regfree(ecbdata->diff_words->word_regex);
2278                         free(ecbdata->diff_words->word_regex);
2279                 }
2280                 FREE_AND_NULL(ecbdata->diff_words);
2281         }
2282 }
2283
2284 const char *diff_get_color(int diff_use_color, enum color_diff ix)
2285 {
2286         if (want_color(diff_use_color))
2287                 return diff_colors[ix];
2288         return "";
2289 }
2290
2291 const char *diff_line_prefix(struct diff_options *opt)
2292 {
2293         struct strbuf *msgbuf;
2294         if (!opt->output_prefix)
2295                 return "";
2296
2297         msgbuf = opt->output_prefix(opt, opt->output_prefix_data);
2298         return msgbuf->buf;
2299 }
2300
2301 static unsigned long sane_truncate_line(char *line, unsigned long len)
2302 {
2303         const char *cp;
2304         unsigned long allot;
2305         size_t l = len;
2306
2307         cp = line;
2308         allot = l;
2309         while (0 < l) {
2310                 (void) utf8_width(&cp, &l);
2311                 if (!cp)
2312                         break; /* truncated in the middle? */
2313         }
2314         return allot - l;
2315 }
2316
2317 static void find_lno(const char *line, struct emit_callback *ecbdata)
2318 {
2319         const char *p;
2320         ecbdata->lno_in_preimage = 0;
2321         ecbdata->lno_in_postimage = 0;
2322         p = strchr(line, '-');
2323         if (!p)
2324                 return; /* cannot happen */
2325         ecbdata->lno_in_preimage = strtol(p + 1, NULL, 10);
2326         p = strchr(p, '+');
2327         if (!p)
2328                 return; /* cannot happen */
2329         ecbdata->lno_in_postimage = strtol(p + 1, NULL, 10);
2330 }
2331
2332 static void fn_out_consume(void *priv, char *line, unsigned long len)
2333 {
2334         struct emit_callback *ecbdata = priv;
2335         struct diff_options *o = ecbdata->opt;
2336
2337         o->found_changes = 1;
2338
2339         if (ecbdata->header) {
2340                 emit_diff_symbol(o, DIFF_SYMBOL_HEADER,
2341                                  ecbdata->header->buf, ecbdata->header->len, 0);
2342                 strbuf_reset(ecbdata->header);
2343                 ecbdata->header = NULL;
2344         }
2345
2346         if (ecbdata->label_path[0]) {
2347                 emit_diff_symbol(o, DIFF_SYMBOL_FILEPAIR_MINUS,
2348                                  ecbdata->label_path[0],
2349                                  strlen(ecbdata->label_path[0]), 0);
2350                 emit_diff_symbol(o, DIFF_SYMBOL_FILEPAIR_PLUS,
2351                                  ecbdata->label_path[1],
2352                                  strlen(ecbdata->label_path[1]), 0);
2353                 ecbdata->label_path[0] = ecbdata->label_path[1] = NULL;
2354         }
2355
2356         if (diff_suppress_blank_empty
2357             && len == 2 && line[0] == ' ' && line[1] == '\n') {
2358                 line[0] = '\n';
2359                 len = 1;
2360         }
2361
2362         if (line[0] == '@') {
2363                 if (ecbdata->diff_words)
2364                         diff_words_flush(ecbdata);
2365                 len = sane_truncate_line(line, len);
2366                 find_lno(line, ecbdata);
2367                 emit_hunk_header(ecbdata, line, len);
2368                 return;
2369         }
2370
2371         if (ecbdata->diff_words) {
2372                 enum diff_symbol s =
2373                         ecbdata->diff_words->type == DIFF_WORDS_PORCELAIN ?
2374                         DIFF_SYMBOL_WORDS_PORCELAIN : DIFF_SYMBOL_WORDS;
2375                 if (line[0] == '-') {
2376                         diff_words_append(line, len,
2377                                           &ecbdata->diff_words->minus);
2378                         return;
2379                 } else if (line[0] == '+') {
2380                         diff_words_append(line, len,
2381                                           &ecbdata->diff_words->plus);
2382                         return;
2383                 } else if (starts_with(line, "\\ ")) {
2384                         /*
2385                          * Eat the "no newline at eof" marker as if we
2386                          * saw a "+" or "-" line with nothing on it,
2387                          * and return without diff_words_flush() to
2388                          * defer processing. If this is the end of
2389                          * preimage, more "+" lines may come after it.
2390                          */
2391                         return;
2392                 }
2393                 diff_words_flush(ecbdata);
2394                 emit_diff_symbol(o, s, line, len, 0);
2395                 return;
2396         }
2397
2398         switch (line[0]) {
2399         case '+':
2400                 ecbdata->lno_in_postimage++;
2401                 emit_add_line(ecbdata, line + 1, len - 1);
2402                 break;
2403         case '-':
2404                 ecbdata->lno_in_preimage++;
2405                 emit_del_line(ecbdata, line + 1, len - 1);
2406                 break;
2407         case ' ':
2408                 ecbdata->lno_in_postimage++;
2409                 ecbdata->lno_in_preimage++;
2410                 emit_context_line(ecbdata, line + 1, len - 1);
2411                 break;
2412         default:
2413                 /* incomplete line at the end */
2414                 ecbdata->lno_in_preimage++;
2415                 emit_diff_symbol(o, DIFF_SYMBOL_CONTEXT_INCOMPLETE,
2416                                  line, len, 0);
2417                 break;
2418         }
2419 }
2420
2421 static void pprint_rename(struct strbuf *name, const char *a, const char *b)
2422 {
2423         const char *old_name = a;
2424         const char *new_name = b;
2425         int pfx_length, sfx_length;
2426         int pfx_adjust_for_slash;
2427         int len_a = strlen(a);
2428         int len_b = strlen(b);
2429         int a_midlen, b_midlen;
2430         int qlen_a = quote_c_style(a, NULL, NULL, 0);
2431         int qlen_b = quote_c_style(b, NULL, NULL, 0);
2432
2433         if (qlen_a || qlen_b) {
2434                 quote_c_style(a, name, NULL, 0);
2435                 strbuf_addstr(name, " => ");
2436                 quote_c_style(b, name, NULL, 0);
2437                 return;
2438         }
2439
2440         /* Find common prefix */
2441         pfx_length = 0;
2442         while (*old_name && *new_name && *old_name == *new_name) {
2443                 if (*old_name == '/')
2444                         pfx_length = old_name - a + 1;
2445                 old_name++;
2446                 new_name++;
2447         }
2448
2449         /* Find common suffix */
2450         old_name = a + len_a;
2451         new_name = b + len_b;
2452         sfx_length = 0;
2453         /*
2454          * If there is a common prefix, it must end in a slash.  In
2455          * that case we let this loop run 1 into the prefix to see the
2456          * same slash.
2457          *
2458          * If there is no common prefix, we cannot do this as it would
2459          * underrun the input strings.
2460          */
2461         pfx_adjust_for_slash = (pfx_length ? 1 : 0);
2462         while (a + pfx_length - pfx_adjust_for_slash <= old_name &&
2463                b + pfx_length - pfx_adjust_for_slash <= new_name &&
2464                *old_name == *new_name) {
2465                 if (*old_name == '/')
2466                         sfx_length = len_a - (old_name - a);
2467                 old_name--;
2468                 new_name--;
2469         }
2470
2471         /*
2472          * pfx{mid-a => mid-b}sfx
2473          * {pfx-a => pfx-b}sfx
2474          * pfx{sfx-a => sfx-b}
2475          * name-a => name-b
2476          */
2477         a_midlen = len_a - pfx_length - sfx_length;
2478         b_midlen = len_b - pfx_length - sfx_length;
2479         if (a_midlen < 0)
2480                 a_midlen = 0;
2481         if (b_midlen < 0)
2482                 b_midlen = 0;
2483
2484         strbuf_grow(name, pfx_length + a_midlen + b_midlen + sfx_length + 7);
2485         if (pfx_length + sfx_length) {
2486                 strbuf_add(name, a, pfx_length);
2487                 strbuf_addch(name, '{');
2488         }
2489         strbuf_add(name, a + pfx_length, a_midlen);
2490         strbuf_addstr(name, " => ");
2491         strbuf_add(name, b + pfx_length, b_midlen);
2492         if (pfx_length + sfx_length) {
2493                 strbuf_addch(name, '}');
2494                 strbuf_add(name, a + len_a - sfx_length, sfx_length);
2495         }
2496 }
2497
2498 struct diffstat_t {
2499         int nr;
2500         int alloc;
2501         struct diffstat_file {
2502                 char *from_name;
2503                 char *name;
2504                 char *print_name;
2505                 const char *comments;
2506                 unsigned is_unmerged:1;
2507                 unsigned is_binary:1;
2508                 unsigned is_renamed:1;
2509                 unsigned is_interesting:1;
2510                 uintmax_t added, deleted;
2511         } **files;
2512 };
2513
2514 static struct diffstat_file *diffstat_add(struct diffstat_t *diffstat,
2515                                           const char *name_a,
2516                                           const char *name_b)
2517 {
2518         struct diffstat_file *x;
2519         x = xcalloc(1, sizeof(*x));
2520         ALLOC_GROW(diffstat->files, diffstat->nr + 1, diffstat->alloc);
2521         diffstat->files[diffstat->nr++] = x;
2522         if (name_b) {
2523                 x->from_name = xstrdup(name_a);
2524                 x->name = xstrdup(name_b);
2525                 x->is_renamed = 1;
2526         }
2527         else {
2528                 x->from_name = NULL;
2529                 x->name = xstrdup(name_a);
2530         }
2531         return x;
2532 }
2533
2534 static void diffstat_consume(void *priv, char *line, unsigned long len)
2535 {
2536         struct diffstat_t *diffstat = priv;
2537         struct diffstat_file *x = diffstat->files[diffstat->nr - 1];
2538
2539         if (line[0] == '+')
2540                 x->added++;
2541         else if (line[0] == '-')
2542                 x->deleted++;
2543 }
2544
2545 const char mime_boundary_leader[] = "------------";
2546
2547 static int scale_linear(int it, int width, int max_change)
2548 {
2549         if (!it)
2550                 return 0;
2551         /*
2552          * make sure that at least one '-' or '+' is printed if
2553          * there is any change to this path. The easiest way is to
2554          * scale linearly as if the alloted width is one column shorter
2555          * than it is, and then add 1 to the result.
2556          */
2557         return 1 + (it * (width - 1) / max_change);
2558 }
2559
2560 static void show_graph(struct strbuf *out, char ch, int cnt,
2561                        const char *set, const char *reset)
2562 {
2563         if (cnt <= 0)
2564                 return;
2565         strbuf_addstr(out, set);
2566         strbuf_addchars(out, ch, cnt);
2567         strbuf_addstr(out, reset);
2568 }
2569
2570 static void fill_print_name(struct diffstat_file *file)
2571 {
2572         struct strbuf pname = STRBUF_INIT;
2573
2574         if (file->print_name)
2575                 return;
2576
2577         if (file->is_renamed)
2578                 pprint_rename(&pname, file->from_name, file->name);
2579         else
2580                 quote_c_style(file->name, &pname, NULL, 0);
2581
2582         if (file->comments)
2583                 strbuf_addf(&pname, " (%s)", file->comments);
2584
2585         file->print_name = strbuf_detach(&pname, NULL);
2586 }
2587
2588 static void print_stat_summary_inserts_deletes(struct diff_options *options,
2589                 int files, int insertions, int deletions)
2590 {
2591         struct strbuf sb = STRBUF_INIT;
2592
2593         if (!files) {
2594                 assert(insertions == 0 && deletions == 0);
2595                 emit_diff_symbol(options, DIFF_SYMBOL_STATS_SUMMARY_NO_FILES,
2596                                  NULL, 0, 0);
2597                 return;
2598         }
2599
2600         strbuf_addf(&sb,
2601                     (files == 1) ? " %d file changed" : " %d files changed",
2602                     files);
2603
2604         /*
2605          * For binary diff, the caller may want to print "x files
2606          * changed" with insertions == 0 && deletions == 0.
2607          *
2608          * Not omitting "0 insertions(+), 0 deletions(-)" in this case
2609          * is probably less confusing (i.e skip over "2 files changed
2610          * but nothing about added/removed lines? Is this a bug in Git?").
2611          */
2612         if (insertions || deletions == 0) {
2613                 strbuf_addf(&sb,
2614                             (insertions == 1) ? ", %d insertion(+)" : ", %d insertions(+)",
2615                             insertions);
2616         }
2617
2618         if (deletions || insertions == 0) {
2619                 strbuf_addf(&sb,
2620                             (deletions == 1) ? ", %d deletion(-)" : ", %d deletions(-)",
2621                             deletions);
2622         }
2623         strbuf_addch(&sb, '\n');
2624         emit_diff_symbol(options, DIFF_SYMBOL_STATS_SUMMARY_INSERTS_DELETES,
2625                          sb.buf, sb.len, 0);
2626         strbuf_release(&sb);
2627 }
2628
2629 void print_stat_summary(FILE *fp, int files,
2630                         int insertions, int deletions)
2631 {
2632         struct diff_options o;
2633         memset(&o, 0, sizeof(o));
2634         o.file = fp;
2635
2636         print_stat_summary_inserts_deletes(&o, files, insertions, deletions);
2637 }
2638
2639 static void show_stats(struct diffstat_t *data, struct diff_options *options)
2640 {
2641         int i, len, add, del, adds = 0, dels = 0;
2642         uintmax_t max_change = 0, max_len = 0;
2643         int total_files = data->nr, count;
2644         int width, name_width, graph_width, number_width = 0, bin_width = 0;
2645         const char *reset, *add_c, *del_c;
2646         int extra_shown = 0;
2647         const char *line_prefix = diff_line_prefix(options);
2648         struct strbuf out = STRBUF_INIT;
2649
2650         if (data->nr == 0)
2651                 return;
2652
2653         count = options->stat_count ? options->stat_count : data->nr;
2654
2655         reset = diff_get_color_opt(options, DIFF_RESET);
2656         add_c = diff_get_color_opt(options, DIFF_FILE_NEW);
2657         del_c = diff_get_color_opt(options, DIFF_FILE_OLD);
2658
2659         /*
2660          * Find the longest filename and max number of changes
2661          */
2662         for (i = 0; (i < count) && (i < data->nr); i++) {
2663                 struct diffstat_file *file = data->files[i];
2664                 uintmax_t change = file->added + file->deleted;
2665
2666                 if (!file->is_interesting && (change == 0)) {
2667                         count++; /* not shown == room for one more */
2668                         continue;
2669                 }
2670                 fill_print_name(file);
2671                 len = strlen(file->print_name);
2672                 if (max_len < len)
2673                         max_len = len;
2674
2675                 if (file->is_unmerged) {
2676                         /* "Unmerged" is 8 characters */
2677                         bin_width = bin_width < 8 ? 8 : bin_width;
2678                         continue;
2679                 }
2680                 if (file->is_binary) {
2681                         /* "Bin XXX -> YYY bytes" */
2682                         int w = 14 + decimal_width(file->added)
2683                                 + decimal_width(file->deleted);
2684                         bin_width = bin_width < w ? w : bin_width;
2685                         /* Display change counts aligned with "Bin" */
2686                         number_width = 3;
2687                         continue;
2688                 }
2689
2690                 if (max_change < change)
2691                         max_change = change;
2692         }
2693         count = i; /* where we can stop scanning in data->files[] */
2694
2695         /*
2696          * We have width = stat_width or term_columns() columns total.
2697          * We want a maximum of min(max_len, stat_name_width) for the name part.
2698          * We want a maximum of min(max_change, stat_graph_width) for the +- part.
2699          * We also need 1 for " " and 4 + decimal_width(max_change)
2700          * for " | NNNN " and one the empty column at the end, altogether
2701          * 6 + decimal_width(max_change).
2702          *
2703          * If there's not enough space, we will use the smaller of
2704          * stat_name_width (if set) and 5/8*width for the filename,
2705          * and the rest for constant elements + graph part, but no more
2706          * than stat_graph_width for the graph part.
2707          * (5/8 gives 50 for filename and 30 for the constant parts + graph
2708          * for the standard terminal size).
2709          *
2710          * In other words: stat_width limits the maximum width, and
2711          * stat_name_width fixes the maximum width of the filename,
2712          * and is also used to divide available columns if there
2713          * aren't enough.
2714          *
2715          * Binary files are displayed with "Bin XXX -> YYY bytes"
2716          * instead of the change count and graph. This part is treated
2717          * similarly to the graph part, except that it is not
2718          * "scaled". If total width is too small to accommodate the
2719          * guaranteed minimum width of the filename part and the
2720          * separators and this message, this message will "overflow"
2721          * making the line longer than the maximum width.
2722          */
2723
2724         if (options->stat_width == -1)
2725                 width = term_columns() - strlen(line_prefix);
2726         else
2727                 width = options->stat_width ? options->stat_width : 80;
2728         number_width = decimal_width(max_change) > number_width ?
2729                 decimal_width(max_change) : number_width;
2730
2731         if (options->stat_graph_width == -1)
2732                 options->stat_graph_width = diff_stat_graph_width;
2733
2734         /*
2735          * Guarantee 3/8*16==6 for the graph part
2736          * and 5/8*16==10 for the filename part
2737          */
2738         if (width < 16 + 6 + number_width)
2739                 width = 16 + 6 + number_width;
2740
2741         /*
2742          * First assign sizes that are wanted, ignoring available width.
2743          * strlen("Bin XXX -> YYY bytes") == bin_width, and the part
2744          * starting from "XXX" should fit in graph_width.
2745          */
2746         graph_width = max_change + 4 > bin_width ? max_change : bin_width - 4;
2747         if (options->stat_graph_width &&
2748             options->stat_graph_width < graph_width)
2749                 graph_width = options->stat_graph_width;
2750
2751         name_width = (options->stat_name_width > 0 &&
2752                       options->stat_name_width < max_len) ?
2753                 options->stat_name_width : max_len;
2754
2755         /*
2756          * Adjust adjustable widths not to exceed maximum width
2757          */
2758         if (name_width + number_width + 6 + graph_width > width) {
2759                 if (graph_width > width * 3/8 - number_width - 6) {
2760                         graph_width = width * 3/8 - number_width - 6;
2761                         if (graph_width < 6)
2762                                 graph_width = 6;
2763                 }
2764
2765                 if (options->stat_graph_width &&
2766                     graph_width > options->stat_graph_width)
2767                         graph_width = options->stat_graph_width;
2768                 if (name_width > width - number_width - 6 - graph_width)
2769                         name_width = width - number_width - 6 - graph_width;
2770                 else
2771                         graph_width = width - number_width - 6 - name_width;
2772         }
2773
2774         /*
2775          * From here name_width is the width of the name area,
2776          * and graph_width is the width of the graph area.
2777          * max_change is used to scale graph properly.
2778          */
2779         for (i = 0; i < count; i++) {
2780                 const char *prefix = "";
2781                 struct diffstat_file *file = data->files[i];
2782                 char *name = file->print_name;
2783                 uintmax_t added = file->added;
2784                 uintmax_t deleted = file->deleted;
2785                 int name_len;
2786
2787                 if (!file->is_interesting && (added + deleted == 0))
2788                         continue;
2789
2790                 /*
2791                  * "scale" the filename
2792                  */
2793                 len = name_width;
2794                 name_len = strlen(name);
2795                 if (name_width < name_len) {
2796                         char *slash;
2797                         prefix = "...";
2798                         len -= 3;
2799                         name += name_len - len;
2800                         slash = strchr(name, '/');
2801                         if (slash)
2802                                 name = slash;
2803                 }
2804
2805                 if (file->is_binary) {
2806                         strbuf_addf(&out, " %s%-*s |", prefix, len, name);
2807                         strbuf_addf(&out, " %*s", number_width, "Bin");
2808                         if (!added && !deleted) {
2809                                 strbuf_addch(&out, '\n');
2810                                 emit_diff_symbol(options, DIFF_SYMBOL_STATS_LINE,
2811                                                  out.buf, out.len, 0);
2812                                 strbuf_reset(&out);
2813                                 continue;
2814                         }
2815                         strbuf_addf(&out, " %s%"PRIuMAX"%s",
2816                                 del_c, deleted, reset);
2817                         strbuf_addstr(&out, " -> ");
2818                         strbuf_addf(&out, "%s%"PRIuMAX"%s",
2819                                 add_c, added, reset);
2820                         strbuf_addstr(&out, " bytes\n");
2821                         emit_diff_symbol(options, DIFF_SYMBOL_STATS_LINE,
2822                                          out.buf, out.len, 0);
2823                         strbuf_reset(&out);
2824                         continue;
2825                 }
2826                 else if (file->is_unmerged) {
2827                         strbuf_addf(&out, " %s%-*s |", prefix, len, name);
2828                         strbuf_addstr(&out, " Unmerged\n");
2829                         emit_diff_symbol(options, DIFF_SYMBOL_STATS_LINE,
2830                                          out.buf, out.len, 0);
2831                         strbuf_reset(&out);
2832                         continue;
2833                 }
2834
2835                 /*
2836                  * scale the add/delete
2837                  */
2838                 add = added;
2839                 del = deleted;
2840
2841                 if (graph_width <= max_change) {
2842                         int total = scale_linear(add + del, graph_width, max_change);
2843                         if (total < 2 && add && del)
2844                                 /* width >= 2 due to the sanity check */
2845                                 total = 2;
2846                         if (add < del) {
2847                                 add = scale_linear(add, graph_width, max_change);
2848                                 del = total - add;
2849                         } else {
2850                                 del = scale_linear(del, graph_width, max_change);
2851                                 add = total - del;
2852                         }
2853                 }
2854                 strbuf_addf(&out, " %s%-*s |", prefix, len, name);
2855                 strbuf_addf(&out, " %*"PRIuMAX"%s",
2856                         number_width, added + deleted,
2857                         added + deleted ? " " : "");
2858                 show_graph(&out, '+', add, add_c, reset);
2859                 show_graph(&out, '-', del, del_c, reset);
2860                 strbuf_addch(&out, '\n');
2861                 emit_diff_symbol(options, DIFF_SYMBOL_STATS_LINE,
2862                                  out.buf, out.len, 0);
2863                 strbuf_reset(&out);
2864         }
2865
2866         for (i = 0; i < data->nr; i++) {
2867                 struct diffstat_file *file = data->files[i];
2868                 uintmax_t added = file->added;
2869                 uintmax_t deleted = file->deleted;
2870
2871                 if (file->is_unmerged ||
2872                     (!file->is_interesting && (added + deleted == 0))) {
2873                         total_files--;
2874                         continue;
2875                 }
2876
2877                 if (!file->is_binary) {
2878                         adds += added;
2879                         dels += deleted;
2880                 }
2881                 if (i < count)
2882                         continue;
2883                 if (!extra_shown)
2884                         emit_diff_symbol(options,
2885                                          DIFF_SYMBOL_STATS_SUMMARY_ABBREV,
2886                                          NULL, 0, 0);
2887                 extra_shown = 1;
2888         }
2889
2890         print_stat_summary_inserts_deletes(options, total_files, adds, dels);
2891         strbuf_release(&out);
2892 }
2893
2894 static void show_shortstats(struct diffstat_t *data, struct diff_options *options)
2895 {
2896         int i, adds = 0, dels = 0, total_files = data->nr;
2897
2898         if (data->nr == 0)
2899                 return;
2900
2901         for (i = 0; i < data->nr; i++) {
2902                 int added = data->files[i]->added;
2903                 int deleted = data->files[i]->deleted;
2904
2905                 if (data->files[i]->is_unmerged ||
2906                     (!data->files[i]->is_interesting && (added + deleted == 0))) {
2907                         total_files--;
2908                 } else if (!data->files[i]->is_binary) { /* don't count bytes */
2909                         adds += added;
2910                         dels += deleted;
2911                 }
2912         }
2913         print_stat_summary_inserts_deletes(options, total_files, adds, dels);
2914 }
2915
2916 static void show_numstat(struct diffstat_t *data, struct diff_options *options)
2917 {
2918         int i;
2919
2920         if (data->nr == 0)
2921                 return;
2922
2923         for (i = 0; i < data->nr; i++) {
2924                 struct diffstat_file *file = data->files[i];
2925
2926                 fprintf(options->file, "%s", diff_line_prefix(options));
2927
2928                 if (file->is_binary)
2929                         fprintf(options->file, "-\t-\t");
2930                 else
2931                         fprintf(options->file,
2932                                 "%"PRIuMAX"\t%"PRIuMAX"\t",
2933                                 file->added, file->deleted);
2934                 if (options->line_termination) {
2935                         fill_print_name(file);
2936                         if (!file->is_renamed)
2937                                 write_name_quoted(file->name, options->file,
2938                                                   options->line_termination);
2939                         else {
2940                                 fputs(file->print_name, options->file);
2941                                 putc(options->line_termination, options->file);
2942                         }
2943                 } else {
2944                         if (file->is_renamed) {
2945                                 putc('\0', options->file);
2946                                 write_name_quoted(file->from_name, options->file, '\0');
2947                         }
2948                         write_name_quoted(file->name, options->file, '\0');
2949                 }
2950         }
2951 }
2952
2953 struct dirstat_file {
2954         const char *name;
2955         unsigned long changed;
2956 };
2957
2958 struct dirstat_dir {
2959         struct dirstat_file *files;
2960         int alloc, nr, permille, cumulative;
2961 };
2962
2963 static long gather_dirstat(struct diff_options *opt, struct dirstat_dir *dir,
2964                 unsigned long changed, const char *base, int baselen)
2965 {
2966         unsigned long sum_changes = 0;
2967         unsigned int sources = 0;
2968         const char *line_prefix = diff_line_prefix(opt);
2969
2970         while (dir->nr) {
2971                 struct dirstat_file *f = dir->files;
2972                 int namelen = strlen(f->name);
2973                 unsigned long changes;
2974                 char *slash;
2975
2976                 if (namelen < baselen)
2977                         break;
2978                 if (memcmp(f->name, base, baselen))
2979                         break;
2980                 slash = strchr(f->name + baselen, '/');
2981                 if (slash) {
2982                         int newbaselen = slash + 1 - f->name;
2983                         changes = gather_dirstat(opt, dir, changed, f->name, newbaselen);
2984                         sources++;
2985                 } else {
2986                         changes = f->changed;
2987                         dir->files++;
2988                         dir->nr--;
2989                         sources += 2;
2990                 }
2991                 sum_changes += changes;
2992         }
2993
2994         /*
2995          * We don't report dirstat's for
2996          *  - the top level
2997          *  - or cases where everything came from a single directory
2998          *    under this directory (sources == 1).
2999          */
3000         if (baselen && sources != 1) {
3001                 if (sum_changes) {
3002                         int permille = sum_changes * 1000 / changed;
3003                         if (permille >= dir->permille) {
3004                                 fprintf(opt->file, "%s%4d.%01d%% %.*s\n", line_prefix,
3005                                         permille / 10, permille % 10, baselen, base);
3006                                 if (!dir->cumulative)
3007                                         return 0;
3008                         }
3009                 }
3010         }
3011         return sum_changes;
3012 }
3013
3014 static int dirstat_compare(const void *_a, const void *_b)
3015 {
3016         const struct dirstat_file *a = _a;
3017         const struct dirstat_file *b = _b;
3018         return strcmp(a->name, b->name);
3019 }
3020
3021 static void show_dirstat(struct diff_options *options)
3022 {
3023         int i;
3024         unsigned long changed;
3025         struct dirstat_dir dir;
3026         struct diff_queue_struct *q = &diff_queued_diff;
3027
3028         dir.files = NULL;
3029         dir.alloc = 0;
3030         dir.nr = 0;
3031         dir.permille = options->dirstat_permille;
3032         dir.cumulative = options->flags.dirstat_cumulative;
3033
3034         changed = 0;
3035         for (i = 0; i < q->nr; i++) {
3036                 struct diff_filepair *p = q->queue[i];
3037                 const char *name;
3038                 unsigned long copied, added, damage;
3039
3040                 name = p->two->path ? p->two->path : p->one->path;
3041
3042                 if (p->one->oid_valid && p->two->oid_valid &&
3043                     oideq(&p->one->oid, &p->two->oid)) {
3044                         /*
3045                          * The SHA1 has not changed, so pre-/post-content is
3046                          * identical. We can therefore skip looking at the
3047                          * file contents altogether.
3048                          */
3049                         damage = 0;
3050                         goto found_damage;
3051                 }
3052
3053                 if (options->flags.dirstat_by_file) {
3054                         /*
3055                          * In --dirstat-by-file mode, we don't really need to
3056                          * look at the actual file contents at all.
3057                          * The fact that the SHA1 changed is enough for us to
3058                          * add this file to the list of results
3059                          * (with each file contributing equal damage).
3060                          */
3061                         damage = 1;
3062                         goto found_damage;
3063                 }
3064
3065                 if (DIFF_FILE_VALID(p->one) && DIFF_FILE_VALID(p->two)) {
3066                         diff_populate_filespec(options->repo, p->one, 0);
3067                         diff_populate_filespec(options->repo, p->two, 0);
3068                         diffcore_count_changes(options->repo,
3069                                                p->one, p->two, NULL, NULL,
3070                                                &copied, &added);
3071                         diff_free_filespec_data(p->one);
3072                         diff_free_filespec_data(p->two);
3073                 } else if (DIFF_FILE_VALID(p->one)) {
3074                         diff_populate_filespec(options->repo, p->one, CHECK_SIZE_ONLY);
3075                         copied = added = 0;
3076                         diff_free_filespec_data(p->one);
3077                 } else if (DIFF_FILE_VALID(p->two)) {
3078                         diff_populate_filespec(options->repo, p->two, CHECK_SIZE_ONLY);
3079                         copied = 0;
3080                         added = p->two->size;
3081                         diff_free_filespec_data(p->two);
3082                 } else
3083                         continue;
3084
3085                 /*
3086                  * Original minus copied is the removed material,
3087                  * added is the new material.  They are both damages
3088                  * made to the preimage.
3089                  * If the resulting damage is zero, we know that
3090                  * diffcore_count_changes() considers the two entries to
3091                  * be identical, but since the oid changed, we
3092                  * know that there must have been _some_ kind of change,
3093                  * so we force all entries to have damage > 0.
3094                  */
3095                 damage = (p->one->size - copied) + added;
3096                 if (!damage)
3097                         damage = 1;
3098
3099 found_damage:
3100                 ALLOC_GROW(dir.files, dir.nr + 1, dir.alloc);
3101                 dir.files[dir.nr].name = name;
3102                 dir.files[dir.nr].changed = damage;
3103                 changed += damage;
3104                 dir.nr++;
3105         }
3106
3107         /* This can happen even with many files, if everything was renames */
3108         if (!changed)
3109                 return;
3110
3111         /* Show all directories with more than x% of the changes */
3112         QSORT(dir.files, dir.nr, dirstat_compare);
3113         gather_dirstat(options, &dir, changed, "", 0);
3114 }
3115
3116 static void show_dirstat_by_line(struct diffstat_t *data, struct diff_options *options)
3117 {
3118         int i;
3119         unsigned long changed;
3120         struct dirstat_dir dir;
3121
3122         if (data->nr == 0)
3123                 return;
3124
3125         dir.files = NULL;
3126         dir.alloc = 0;
3127         dir.nr = 0;
3128         dir.permille = options->dirstat_permille;
3129         dir.cumulative = options->flags.dirstat_cumulative;
3130
3131         changed = 0;
3132         for (i = 0; i < data->nr; i++) {
3133                 struct diffstat_file *file = data->files[i];
3134                 unsigned long damage = file->added + file->deleted;
3135                 if (file->is_binary)
3136                         /*
3137                          * binary files counts bytes, not lines. Must find some
3138                          * way to normalize binary bytes vs. textual lines.
3139                          * The following heuristic assumes that there are 64
3140                          * bytes per "line".
3141                          * This is stupid and ugly, but very cheap...
3142                          */
3143                         damage = DIV_ROUND_UP(damage, 64);
3144                 ALLOC_GROW(dir.files, dir.nr + 1, dir.alloc);
3145                 dir.files[dir.nr].name = file->name;
3146                 dir.files[dir.nr].changed = damage;
3147                 changed += damage;
3148                 dir.nr++;
3149         }
3150
3151         /* This can happen even with many files, if everything was renames */
3152         if (!changed)
3153                 return;
3154
3155         /* Show all directories with more than x% of the changes */
3156         QSORT(dir.files, dir.nr, dirstat_compare);
3157         gather_dirstat(options, &dir, changed, "", 0);
3158 }
3159
3160 static void free_diffstat_info(struct diffstat_t *diffstat)
3161 {
3162         int i;
3163         for (i = 0; i < diffstat->nr; i++) {
3164                 struct diffstat_file *f = diffstat->files[i];
3165                 free(f->print_name);
3166                 free(f->name);
3167                 free(f->from_name);
3168                 free(f);
3169         }
3170         free(diffstat->files);
3171 }
3172
3173 struct checkdiff_t {
3174         const char *filename;
3175         int lineno;
3176         int conflict_marker_size;
3177         struct diff_options *o;
3178         unsigned ws_rule;
3179         unsigned status;
3180 };
3181
3182 static int is_conflict_marker(const char *line, int marker_size, unsigned long len)
3183 {
3184         char firstchar;
3185         int cnt;
3186
3187         if (len < marker_size + 1)
3188                 return 0;
3189         firstchar = line[0];
3190         switch (firstchar) {
3191         case '=': case '>': case '<': case '|':
3192                 break;
3193         default:
3194                 return 0;
3195         }
3196         for (cnt = 1; cnt < marker_size; cnt++)
3197                 if (line[cnt] != firstchar)
3198                         return 0;
3199         /* line[1] thru line[marker_size-1] are same as firstchar */
3200         if (len < marker_size + 1 || !isspace(line[marker_size]))
3201                 return 0;
3202         return 1;
3203 }
3204
3205 static void checkdiff_consume_hunk(void *priv,
3206                                    long ob, long on, long nb, long nn,
3207                                    const char *func, long funclen)
3208
3209 {
3210         struct checkdiff_t *data = priv;
3211         data->lineno = nb - 1;
3212 }
3213
3214 static void checkdiff_consume(void *priv, char *line, unsigned long len)
3215 {
3216         struct checkdiff_t *data = priv;
3217         int marker_size = data->conflict_marker_size;
3218         const char *ws = diff_get_color(data->o->use_color, DIFF_WHITESPACE);
3219         const char *reset = diff_get_color(data->o->use_color, DIFF_RESET);
3220         const char *set = diff_get_color(data->o->use_color, DIFF_FILE_NEW);
3221         char *err;
3222         const char *line_prefix;
3223
3224         assert(data->o);
3225         line_prefix = diff_line_prefix(data->o);
3226
3227         if (line[0] == '+') {
3228                 unsigned bad;
3229                 data->lineno++;
3230                 if (is_conflict_marker(line + 1, marker_size, len - 1)) {
3231                         data->status |= 1;
3232                         fprintf(data->o->file,
3233                                 "%s%s:%d: leftover conflict marker\n",
3234                                 line_prefix, data->filename, data->lineno);
3235                 }
3236                 bad = ws_check(line + 1, len - 1, data->ws_rule);
3237                 if (!bad)
3238                         return;
3239                 data->status |= bad;
3240                 err = whitespace_error_string(bad);
3241                 fprintf(data->o->file, "%s%s:%d: %s.\n",
3242                         line_prefix, data->filename, data->lineno, err);
3243                 free(err);
3244                 emit_line(data->o, set, reset, line, 1);
3245                 ws_check_emit(line + 1, len - 1, data->ws_rule,
3246                               data->o->file, set, reset, ws);
3247         } else if (line[0] == ' ') {
3248                 data->lineno++;
3249         }
3250 }
3251
3252 static unsigned char *deflate_it(char *data,
3253                                  unsigned long size,
3254                                  unsigned long *result_size)
3255 {
3256         int bound;
3257         unsigned char *deflated;
3258         git_zstream stream;
3259
3260         git_deflate_init(&stream, zlib_compression_level);
3261         bound = git_deflate_bound(&stream, size);
3262         deflated = xmalloc(bound);
3263         stream.next_out = deflated;
3264         stream.avail_out = bound;
3265
3266         stream.next_in = (unsigned char *)data;
3267         stream.avail_in = size;
3268         while (git_deflate(&stream, Z_FINISH) == Z_OK)
3269                 ; /* nothing */
3270         git_deflate_end(&stream);
3271         *result_size = stream.total_out;
3272         return deflated;
3273 }
3274
3275 static void emit_binary_diff_body(struct diff_options *o,
3276                                   mmfile_t *one, mmfile_t *two)
3277 {
3278         void *cp;
3279         void *delta;
3280         void *deflated;
3281         void *data;
3282         unsigned long orig_size;
3283         unsigned long delta_size;
3284         unsigned long deflate_size;
3285         unsigned long data_size;
3286
3287         /* We could do deflated delta, or we could do just deflated two,
3288          * whichever is smaller.
3289          */
3290         delta = NULL;
3291         deflated = deflate_it(two->ptr, two->size, &deflate_size);
3292         if (one->size && two->size) {
3293                 delta = diff_delta(one->ptr, one->size,
3294                                    two->ptr, two->size,
3295                                    &delta_size, deflate_size);
3296                 if (delta) {
3297                         void *to_free = delta;
3298                         orig_size = delta_size;
3299                         delta = deflate_it(delta, delta_size, &delta_size);
3300                         free(to_free);
3301                 }
3302         }
3303
3304         if (delta && delta_size < deflate_size) {
3305                 char *s = xstrfmt("%"PRIuMAX , (uintmax_t)orig_size);
3306                 emit_diff_symbol(o, DIFF_SYMBOL_BINARY_DIFF_HEADER_DELTA,
3307                                  s, strlen(s), 0);
3308                 free(s);
3309                 free(deflated);
3310                 data = delta;
3311                 data_size = delta_size;
3312         } else {
3313                 char *s = xstrfmt("%lu", two->size);
3314                 emit_diff_symbol(o, DIFF_SYMBOL_BINARY_DIFF_HEADER_LITERAL,
3315                                  s, strlen(s), 0);
3316                 free(s);
3317                 free(delta);
3318                 data = deflated;
3319                 data_size = deflate_size;
3320         }
3321
3322         /* emit data encoded in base85 */
3323         cp = data;
3324         while (data_size) {
3325                 int len;
3326                 int bytes = (52 < data_size) ? 52 : data_size;
3327                 char line[71];
3328                 data_size -= bytes;
3329                 if (bytes <= 26)
3330                         line[0] = bytes + 'A' - 1;
3331                 else
3332                         line[0] = bytes - 26 + 'a' - 1;
3333                 encode_85(line + 1, cp, bytes);
3334                 cp = (char *) cp + bytes;
3335
3336                 len = strlen(line);
3337                 line[len++] = '\n';
3338                 line[len] = '\0';
3339
3340                 emit_diff_symbol(o, DIFF_SYMBOL_BINARY_DIFF_BODY,
3341                                  line, len, 0);
3342         }
3343         emit_diff_symbol(o, DIFF_SYMBOL_BINARY_DIFF_FOOTER, NULL, 0, 0);
3344         free(data);
3345 }
3346
3347 static void emit_binary_diff(struct diff_options *o,
3348                              mmfile_t *one, mmfile_t *two)
3349 {
3350         emit_diff_symbol(o, DIFF_SYMBOL_BINARY_DIFF_HEADER, NULL, 0, 0);
3351         emit_binary_diff_body(o, one, two);
3352         emit_binary_diff_body(o, two, one);
3353 }
3354
3355 int diff_filespec_is_binary(struct repository *r,
3356                             struct diff_filespec *one)
3357 {
3358         if (one->is_binary == -1) {
3359                 diff_filespec_load_driver(one, r->index);
3360                 if (one->driver->binary != -1)
3361                         one->is_binary = one->driver->binary;
3362                 else {
3363                         if (!one->data && DIFF_FILE_VALID(one))
3364                                 diff_populate_filespec(r, one, CHECK_BINARY);
3365                         if (one->is_binary == -1 && one->data)
3366                                 one->is_binary = buffer_is_binary(one->data,
3367                                                 one->size);
3368                         if (one->is_binary == -1)
3369                                 one->is_binary = 0;
3370                 }
3371         }
3372         return one->is_binary;
3373 }
3374
3375 static const struct userdiff_funcname *
3376 diff_funcname_pattern(struct diff_options *o, struct diff_filespec *one)
3377 {
3378         diff_filespec_load_driver(one, o->repo->index);
3379         return one->driver->funcname.pattern ? &one->driver->funcname : NULL;
3380 }
3381
3382 void diff_set_mnemonic_prefix(struct diff_options *options, const char *a, const char *b)
3383 {
3384         if (!options->a_prefix)
3385                 options->a_prefix = a;
3386         if (!options->b_prefix)
3387                 options->b_prefix = b;
3388 }
3389
3390 struct userdiff_driver *get_textconv(struct repository *r,
3391                                      struct diff_filespec *one)
3392 {
3393         if (!DIFF_FILE_VALID(one))
3394                 return NULL;
3395
3396         diff_filespec_load_driver(one, r->index);
3397         return userdiff_get_textconv(r, one->driver);
3398 }
3399
3400 static void builtin_diff(const char *name_a,
3401                          const char *name_b,
3402                          struct diff_filespec *one,
3403                          struct diff_filespec *two,
3404                          const char *xfrm_msg,
3405                          int must_show_header,
3406                          struct diff_options *o,
3407                          int complete_rewrite)
3408 {
3409         mmfile_t mf1, mf2;
3410         const char *lbl[2];
3411         char *a_one, *b_two;
3412         const char *meta = diff_get_color_opt(o, DIFF_METAINFO);
3413         const char *reset = diff_get_color_opt(o, DIFF_RESET);
3414         const char *a_prefix, *b_prefix;
3415         struct userdiff_driver *textconv_one = NULL;
3416         struct userdiff_driver *textconv_two = NULL;
3417         struct strbuf header = STRBUF_INIT;
3418         const char *line_prefix = diff_line_prefix(o);
3419
3420         diff_set_mnemonic_prefix(o, "a/", "b/");
3421         if (o->flags.reverse_diff) {
3422                 a_prefix = o->b_prefix;
3423                 b_prefix = o->a_prefix;
3424         } else {
3425                 a_prefix = o->a_prefix;
3426                 b_prefix = o->b_prefix;
3427         }
3428
3429         if (o->submodule_format == DIFF_SUBMODULE_LOG &&
3430             (!one->mode || S_ISGITLINK(one->mode)) &&
3431             (!two->mode || S_ISGITLINK(two->mode))) {
3432                 show_submodule_summary(o, one->path ? one->path : two->path,
3433                                 &one->oid, &two->oid,
3434                                 two->dirty_submodule);
3435                 return;
3436         } else if (o->submodule_format == DIFF_SUBMODULE_INLINE_DIFF &&
3437                    (!one->mode || S_ISGITLINK(one->mode)) &&
3438                    (!two->mode || S_ISGITLINK(two->mode))) {
3439                 show_submodule_inline_diff(o, one->path ? one->path : two->path,
3440                                 &one->oid, &two->oid,
3441                                 two->dirty_submodule);
3442                 return;
3443         }
3444
3445         if (o->flags.allow_textconv) {
3446                 textconv_one = get_textconv(o->repo, one);
3447                 textconv_two = get_textconv(o->repo, two);
3448         }
3449
3450         /* Never use a non-valid filename anywhere if at all possible */
3451         name_a = DIFF_FILE_VALID(one) ? name_a : name_b;
3452         name_b = DIFF_FILE_VALID(two) ? name_b : name_a;
3453
3454         a_one = quote_two(a_prefix, name_a + (*name_a == '/'));
3455         b_two = quote_two(b_prefix, name_b + (*name_b == '/'));
3456         lbl[0] = DIFF_FILE_VALID(one) ? a_one : "/dev/null";
3457         lbl[1] = DIFF_FILE_VALID(two) ? b_two : "/dev/null";
3458         strbuf_addf(&header, "%s%sdiff --git %s %s%s\n", line_prefix, meta, a_one, b_two, reset);
3459         if (lbl[0][0] == '/') {
3460                 /* /dev/null */
3461                 strbuf_addf(&header, "%s%snew file mode %06o%s\n", line_prefix, meta, two->mode, reset);
3462                 if (xfrm_msg)
3463                         strbuf_addstr(&header, xfrm_msg);
3464                 must_show_header = 1;
3465         }
3466         else if (lbl[1][0] == '/') {
3467                 strbuf_addf(&header, "%s%sdeleted file mode %06o%s\n", line_prefix, meta, one->mode, reset);
3468                 if (xfrm_msg)
3469                         strbuf_addstr(&header, xfrm_msg);
3470                 must_show_header = 1;
3471         }
3472         else {
3473                 if (one->mode != two->mode) {
3474                         strbuf_addf(&header, "%s%sold mode %06o%s\n", line_prefix, meta, one->mode, reset);
3475                         strbuf_addf(&header, "%s%snew mode %06o%s\n", line_prefix, meta, two->mode, reset);
3476                         must_show_header = 1;
3477                 }
3478                 if (xfrm_msg)
3479                         strbuf_addstr(&header, xfrm_msg);
3480
3481                 /*
3482                  * we do not run diff between different kind
3483                  * of objects.
3484                  */
3485                 if ((one->mode ^ two->mode) & S_IFMT)
3486                         goto free_ab_and_return;
3487                 if (complete_rewrite &&
3488                     (textconv_one || !diff_filespec_is_binary(o->repo, one)) &&
3489                     (textconv_two || !diff_filespec_is_binary(o->repo, two))) {
3490                         emit_diff_symbol(o, DIFF_SYMBOL_HEADER,
3491                                          header.buf, header.len, 0);
3492                         strbuf_reset(&header);
3493                         emit_rewrite_diff(name_a, name_b, one, two,
3494                                           textconv_one, textconv_two, o);
3495                         o->found_changes = 1;
3496                         goto free_ab_and_return;
3497                 }
3498         }
3499
3500         if (o->irreversible_delete && lbl[1][0] == '/') {
3501                 emit_diff_symbol(o, DIFF_SYMBOL_HEADER, header.buf,
3502                                  header.len, 0);
3503                 strbuf_reset(&header);
3504                 goto free_ab_and_return;
3505         } else if (!o->flags.text &&
3506                    ( (!textconv_one && diff_filespec_is_binary(o->repo, one)) ||
3507                      (!textconv_two && diff_filespec_is_binary(o->repo, two)) )) {
3508                 struct strbuf sb = STRBUF_INIT;
3509                 if (!one->data && !two->data &&
3510                     S_ISREG(one->mode) && S_ISREG(two->mode) &&
3511                     !o->flags.binary) {
3512                         if (oideq(&one->oid, &two->oid)) {
3513                                 if (must_show_header)
3514                                         emit_diff_symbol(o, DIFF_SYMBOL_HEADER,
3515                                                          header.buf, header.len,
3516                                                          0);
3517                                 goto free_ab_and_return;
3518                         }
3519                         emit_diff_symbol(o, DIFF_SYMBOL_HEADER,
3520                                          header.buf, header.len, 0);
3521                         strbuf_addf(&sb, "%sBinary files %s and %s differ\n",
3522                                     diff_line_prefix(o), lbl[0], lbl[1]);
3523                         emit_diff_symbol(o, DIFF_SYMBOL_BINARY_FILES,
3524                                          sb.buf, sb.len, 0);
3525                         strbuf_release(&sb);
3526                         goto free_ab_and_return;
3527                 }
3528                 if (fill_mmfile(o->repo, &mf1, one) < 0 ||
3529                     fill_mmfile(o->repo, &mf2, two) < 0)
3530                         die("unable to read files to diff");
3531                 /* Quite common confusing case */
3532                 if (mf1.size == mf2.size &&
3533                     !memcmp(mf1.ptr, mf2.ptr, mf1.size)) {
3534                         if (must_show_header)
3535                                 emit_diff_symbol(o, DIFF_SYMBOL_HEADER,
3536                                                  header.buf, header.len, 0);
3537                         goto free_ab_and_return;
3538                 }
3539                 emit_diff_symbol(o, DIFF_SYMBOL_HEADER, header.buf, header.len, 0);
3540                 strbuf_reset(&header);
3541                 if (o->flags.binary)
3542                         emit_binary_diff(o, &mf1, &mf2);
3543                 else {
3544                         strbuf_addf(&sb, "%sBinary files %s and %s differ\n",
3545                                     diff_line_prefix(o), lbl[0], lbl[1]);
3546                         emit_diff_symbol(o, DIFF_SYMBOL_BINARY_FILES,
3547                                          sb.buf, sb.len, 0);
3548                         strbuf_release(&sb);
3549                 }
3550                 o->found_changes = 1;
3551         } else {
3552                 /* Crazy xdl interfaces.. */
3553                 const char *diffopts;
3554                 const char *v;
3555                 xpparam_t xpp;
3556                 xdemitconf_t xecfg;
3557                 struct emit_callback ecbdata;
3558                 const struct userdiff_funcname *pe;
3559
3560                 if (must_show_header) {
3561                         emit_diff_symbol(o, DIFF_SYMBOL_HEADER,
3562                                          header.buf, header.len, 0);
3563                         strbuf_reset(&header);
3564                 }
3565
3566                 mf1.size = fill_textconv(o->repo, textconv_one, one, &mf1.ptr);
3567                 mf2.size = fill_textconv(o->repo, textconv_two, two, &mf2.ptr);
3568
3569                 pe = diff_funcname_pattern(o, one);
3570                 if (!pe)
3571                         pe = diff_funcname_pattern(o, two);
3572
3573                 memset(&xpp, 0, sizeof(xpp));
3574                 memset(&xecfg, 0, sizeof(xecfg));
3575                 memset(&ecbdata, 0, sizeof(ecbdata));
3576                 if (o->flags.suppress_diff_headers)
3577                         lbl[0] = NULL;
3578                 ecbdata.label_path = lbl;
3579                 ecbdata.color_diff = want_color(o->use_color);
3580                 ecbdata.ws_rule = whitespace_rule(o->repo->index, name_b);
3581                 if (ecbdata.ws_rule & WS_BLANK_AT_EOF)
3582                         check_blank_at_eof(&mf1, &mf2, &ecbdata);
3583                 ecbdata.opt = o;
3584                 if (header.len && !o->flags.suppress_diff_headers)
3585                         ecbdata.header = &header;
3586                 xpp.flags = o->xdl_opts;
3587                 xpp.anchors = o->anchors;
3588                 xpp.anchors_nr = o->anchors_nr;
3589                 xecfg.ctxlen = o->context;
3590                 xecfg.interhunkctxlen = o->interhunkcontext;
3591                 xecfg.flags = XDL_EMIT_FUNCNAMES;
3592                 if (o->flags.funccontext)
3593                         xecfg.flags |= XDL_EMIT_FUNCCONTEXT;
3594                 if (pe)
3595                         xdiff_set_find_func(&xecfg, pe->pattern, pe->cflags);
3596
3597                 diffopts = getenv("GIT_DIFF_OPTS");
3598                 if (!diffopts)
3599                         ;
3600                 else if (skip_prefix(diffopts, "--unified=", &v))
3601                         xecfg.ctxlen = strtoul(v, NULL, 10);
3602                 else if (skip_prefix(diffopts, "-u", &v))
3603                         xecfg.ctxlen = strtoul(v, NULL, 10);
3604
3605                 if (o->word_diff)
3606                         init_diff_words_data(&ecbdata, o, one, two);
3607                 if (xdi_diff_outf(&mf1, &mf2, NULL, fn_out_consume,
3608                                   &ecbdata, &xpp, &xecfg))
3609                         die("unable to generate diff for %s", one->path);
3610                 if (o->word_diff)
3611                         free_diff_words_data(&ecbdata);
3612                 if (textconv_one)
3613                         free(mf1.ptr);
3614                 if (textconv_two)
3615                         free(mf2.ptr);
3616                 xdiff_clear_find_func(&xecfg);
3617         }
3618
3619  free_ab_and_return:
3620         strbuf_release(&header);
3621         diff_free_filespec_data(one);
3622         diff_free_filespec_data(two);
3623         free(a_one);
3624         free(b_two);
3625         return;
3626 }
3627
3628 static char *get_compact_summary(const struct diff_filepair *p, int is_renamed)
3629 {
3630         if (!is_renamed) {
3631                 if (p->status == DIFF_STATUS_ADDED) {
3632                         if (S_ISLNK(p->two->mode))
3633                                 return "new +l";
3634                         else if ((p->two->mode & 0777) == 0755)
3635                                 return "new +x";
3636                         else
3637                                 return "new";
3638                 } else if (p->status == DIFF_STATUS_DELETED)
3639                         return "gone";
3640         }
3641         if (S_ISLNK(p->one->mode) && !S_ISLNK(p->two->mode))
3642                 return "mode -l";
3643         else if (!S_ISLNK(p->one->mode) && S_ISLNK(p->two->mode))
3644                 return "mode +l";
3645         else if ((p->one->mode & 0777) == 0644 &&
3646                  (p->two->mode & 0777) == 0755)
3647                 return "mode +x";
3648         else if ((p->one->mode & 0777) == 0755 &&
3649                  (p->two->mode & 0777) == 0644)
3650                 return "mode -x";
3651         return NULL;
3652 }
3653
3654 static void builtin_diffstat(const char *name_a, const char *name_b,
3655                              struct diff_filespec *one,
3656                              struct diff_filespec *two,
3657                              struct diffstat_t *diffstat,
3658                              struct diff_options *o,
3659                              struct diff_filepair *p)
3660 {
3661         mmfile_t mf1, mf2;
3662         struct diffstat_file *data;
3663         int same_contents;
3664         int complete_rewrite = 0;
3665
3666         if (!DIFF_PAIR_UNMERGED(p)) {
3667                 if (p->status == DIFF_STATUS_MODIFIED && p->score)
3668                         complete_rewrite = 1;
3669         }
3670
3671         data = diffstat_add(diffstat, name_a, name_b);
3672         data->is_interesting = p->status != DIFF_STATUS_UNKNOWN;
3673         if (o->flags.stat_with_summary)
3674                 data->comments = get_compact_summary(p, data->is_renamed);
3675
3676         if (!one || !two) {
3677                 data->is_unmerged = 1;
3678                 return;
3679         }
3680
3681         same_contents = oideq(&one->oid, &two->oid);
3682
3683         if (diff_filespec_is_binary(o->repo, one) ||
3684             diff_filespec_is_binary(o->repo, two)) {
3685                 data->is_binary = 1;
3686                 if (same_contents) {
3687                         data->added = 0;
3688                         data->deleted = 0;
3689                 } else {
3690                         data->added = diff_filespec_size(o->repo, two);
3691                         data->deleted = diff_filespec_size(o->repo, one);
3692                 }
3693         }
3694
3695         else if (complete_rewrite) {
3696                 diff_populate_filespec(o->repo, one, 0);
3697                 diff_populate_filespec(o->repo, two, 0);
3698                 data->deleted = count_lines(one->data, one->size);
3699                 data->added = count_lines(two->data, two->size);
3700         }
3701
3702         else if (!same_contents) {
3703                 /* Crazy xdl interfaces.. */
3704                 xpparam_t xpp;
3705                 xdemitconf_t xecfg;
3706
3707                 if (fill_mmfile(o->repo, &mf1, one) < 0 ||
3708                     fill_mmfile(o->repo, &mf2, two) < 0)
3709                         die("unable to read files to diff");
3710
3711                 memset(&xpp, 0, sizeof(xpp));
3712                 memset(&xecfg, 0, sizeof(xecfg));
3713                 xpp.flags = o->xdl_opts;
3714                 xpp.anchors = o->anchors;
3715                 xpp.anchors_nr = o->anchors_nr;
3716                 xecfg.ctxlen = o->context;
3717                 xecfg.interhunkctxlen = o->interhunkcontext;
3718                 if (xdi_diff_outf(&mf1, &mf2, discard_hunk_line,
3719                                   diffstat_consume, diffstat, &xpp, &xecfg))
3720                         die("unable to generate diffstat for %s", one->path);
3721         }
3722
3723         diff_free_filespec_data(one);
3724         diff_free_filespec_data(two);
3725 }
3726
3727 static void builtin_checkdiff(const char *name_a, const char *name_b,
3728                               const char *attr_path,
3729                               struct diff_filespec *one,
3730                               struct diff_filespec *two,
3731                               struct diff_options *o)
3732 {
3733         mmfile_t mf1, mf2;
3734         struct checkdiff_t data;
3735
3736         if (!two)
3737                 return;
3738
3739         memset(&data, 0, sizeof(data));
3740         data.filename = name_b ? name_b : name_a;
3741         data.lineno = 0;
3742         data.o = o;
3743         data.ws_rule = whitespace_rule(o->repo->index, attr_path);
3744         data.conflict_marker_size = ll_merge_marker_size(o->repo->index, attr_path);
3745
3746         if (fill_mmfile(o->repo, &mf1, one) < 0 ||
3747             fill_mmfile(o->repo, &mf2, two) < 0)
3748                 die("unable to read files to diff");
3749
3750         /*
3751          * All the other codepaths check both sides, but not checking
3752          * the "old" side here is deliberate.  We are checking the newly
3753          * introduced changes, and as long as the "new" side is text, we
3754          * can and should check what it introduces.
3755          */
3756         if (diff_filespec_is_binary(o->repo, two))
3757                 goto free_and_return;
3758         else {
3759                 /* Crazy xdl interfaces.. */
3760                 xpparam_t xpp;
3761                 xdemitconf_t xecfg;
3762
3763                 memset(&xpp, 0, sizeof(xpp));
3764                 memset(&xecfg, 0, sizeof(xecfg));
3765                 xecfg.ctxlen = 1; /* at least one context line */
3766                 xpp.flags = 0;
3767                 if (xdi_diff_outf(&mf1, &mf2, checkdiff_consume_hunk,
3768                                   checkdiff_consume, &data,
3769                                   &xpp, &xecfg))
3770                         die("unable to generate checkdiff for %s", one->path);
3771
3772                 if (data.ws_rule & WS_BLANK_AT_EOF) {
3773                         struct emit_callback ecbdata;
3774                         int blank_at_eof;
3775
3776                         ecbdata.ws_rule = data.ws_rule;
3777                         check_blank_at_eof(&mf1, &mf2, &ecbdata);
3778                         blank_at_eof = ecbdata.blank_at_eof_in_postimage;
3779
3780                         if (blank_at_eof) {
3781                                 static char *err;
3782                                 if (!err)
3783                                         err = whitespace_error_string(WS_BLANK_AT_EOF);
3784                                 fprintf(o->file, "%s:%d: %s.\n",
3785                                         data.filename, blank_at_eof, err);
3786                                 data.status = 1; /* report errors */
3787                         }
3788                 }
3789         }
3790  free_and_return:
3791         diff_free_filespec_data(one);
3792         diff_free_filespec_data(two);
3793         if (data.status)
3794                 o->flags.check_failed = 1;
3795 }
3796
3797 struct diff_filespec *alloc_filespec(const char *path)
3798 {
3799         struct diff_filespec *spec;
3800
3801         FLEXPTR_ALLOC_STR(spec, path, path);
3802         spec->count = 1;
3803         spec->is_binary = -1;
3804         return spec;
3805 }
3806
3807 void free_filespec(struct diff_filespec *spec)
3808 {
3809         if (!--spec->count) {
3810                 diff_free_filespec_data(spec);
3811                 free(spec);
3812         }
3813 }
3814
3815 void fill_filespec(struct diff_filespec *spec, const struct object_id *oid,
3816                    int oid_valid, unsigned short mode)
3817 {
3818         if (mode) {
3819                 spec->mode = canon_mode(mode);
3820                 oidcpy(&spec->oid, oid);
3821                 spec->oid_valid = oid_valid;
3822         }
3823 }
3824
3825 /*
3826  * Given a name and sha1 pair, if the index tells us the file in
3827  * the work tree has that object contents, return true, so that
3828  * prepare_temp_file() does not have to inflate and extract.
3829  */
3830 static int reuse_worktree_file(struct index_state *istate,
3831                                const char *name,
3832                                const struct object_id *oid,
3833                                int want_file)
3834 {
3835         const struct cache_entry *ce;
3836         struct stat st;
3837         int pos, len;
3838
3839         /*
3840          * We do not read the cache ourselves here, because the
3841          * benchmark with my previous version that always reads cache
3842          * shows that it makes things worse for diff-tree comparing
3843          * two linux-2.6 kernel trees in an already checked out work
3844          * tree.  This is because most diff-tree comparisons deal with
3845          * only a small number of files, while reading the cache is
3846          * expensive for a large project, and its cost outweighs the
3847          * savings we get by not inflating the object to a temporary
3848          * file.  Practically, this code only helps when we are used
3849          * by diff-cache --cached, which does read the cache before
3850          * calling us.
3851          */
3852         if (!istate->cache)
3853                 return 0;
3854
3855         /* We want to avoid the working directory if our caller
3856          * doesn't need the data in a normal file, this system
3857          * is rather slow with its stat/open/mmap/close syscalls,
3858          * and the object is contained in a pack file.  The pack
3859          * is probably already open and will be faster to obtain
3860          * the data through than the working directory.  Loose
3861          * objects however would tend to be slower as they need
3862          * to be individually opened and inflated.
3863          */
3864         if (!FAST_WORKING_DIRECTORY && !want_file && has_object_pack(oid))
3865                 return 0;
3866
3867         /*
3868          * Similarly, if we'd have to convert the file contents anyway, that
3869          * makes the optimization not worthwhile.
3870          */
3871         if (!want_file && would_convert_to_git(istate, name))
3872                 return 0;
3873
3874         len = strlen(name);
3875         pos = index_name_pos(istate, name, len);
3876         if (pos < 0)
3877                 return 0;
3878         ce = istate->cache[pos];
3879
3880         /*
3881          * This is not the sha1 we are looking for, or
3882          * unreusable because it is not a regular file.
3883          */
3884         if (!oideq(oid, &ce->oid) || !S_ISREG(ce->ce_mode))
3885                 return 0;
3886
3887         /*
3888          * If ce is marked as "assume unchanged", there is no
3889          * guarantee that work tree matches what we are looking for.
3890          */
3891         if ((ce->ce_flags & CE_VALID) || ce_skip_worktree(ce))
3892                 return 0;
3893
3894         /*
3895          * If ce matches the file in the work tree, we can reuse it.
3896          */
3897         if (ce_uptodate(ce) ||
3898             (!lstat(name, &st) && !ie_match_stat(istate, ce, &st, 0)))
3899                 return 1;
3900
3901         return 0;
3902 }
3903
3904 static int diff_populate_gitlink(struct diff_filespec *s, int size_only)
3905 {
3906         struct strbuf buf = STRBUF_INIT;
3907         char *dirty = "";
3908
3909         /* Are we looking at the work tree? */
3910         if (s->dirty_submodule)
3911                 dirty = "-dirty";
3912
3913         strbuf_addf(&buf, "Subproject commit %s%s\n",
3914                     oid_to_hex(&s->oid), dirty);
3915         s->size = buf.len;
3916         if (size_only) {
3917                 s->data = NULL;
3918                 strbuf_release(&buf);
3919         } else {
3920                 s->data = strbuf_detach(&buf, NULL);
3921                 s->should_free = 1;
3922         }
3923         return 0;
3924 }
3925
3926 /*
3927  * While doing rename detection and pickaxe operation, we may need to
3928  * grab the data for the blob (or file) for our own in-core comparison.
3929  * diff_filespec has data and size fields for this purpose.
3930  */
3931 int diff_populate_filespec(struct repository *r,
3932                            struct diff_filespec *s,
3933                            unsigned int flags)
3934 {
3935         int size_only = flags & CHECK_SIZE_ONLY;
3936         int err = 0;
3937         int conv_flags = global_conv_flags_eol;
3938         /*
3939          * demote FAIL to WARN to allow inspecting the situation
3940          * instead of refusing.
3941          */
3942         if (conv_flags & CONV_EOL_RNDTRP_DIE)
3943                 conv_flags = CONV_EOL_RNDTRP_WARN;
3944
3945         if (!DIFF_FILE_VALID(s))
3946                 die("internal error: asking to populate invalid file.");
3947         if (S_ISDIR(s->mode))
3948                 return -1;
3949
3950         if (s->data)
3951                 return 0;
3952
3953         if (size_only && 0 < s->size)
3954                 return 0;
3955
3956         if (S_ISGITLINK(s->mode))
3957                 return diff_populate_gitlink(s, size_only);
3958
3959         if (!s->oid_valid ||
3960             reuse_worktree_file(r->index, s->path, &s->oid, 0)) {
3961                 struct strbuf buf = STRBUF_INIT;
3962                 struct stat st;
3963                 int fd;
3964
3965                 if (lstat(s->path, &st) < 0) {
3966                 err_empty:
3967                         err = -1;
3968                 empty:
3969                         s->data = (char *)"";
3970                         s->size = 0;
3971                         return err;
3972                 }
3973                 s->size = xsize_t(st.st_size);
3974                 if (!s->size)
3975                         goto empty;
3976                 if (S_ISLNK(st.st_mode)) {
3977                         struct strbuf sb = STRBUF_INIT;
3978
3979                         if (strbuf_readlink(&sb, s->path, s->size))
3980                                 goto err_empty;
3981                         s->size = sb.len;
3982                         s->data = strbuf_detach(&sb, NULL);
3983                         s->should_free = 1;
3984                         return 0;
3985                 }
3986
3987                 /*
3988                  * Even if the caller would be happy with getting
3989                  * only the size, we cannot return early at this
3990                  * point if the path requires us to run the content
3991                  * conversion.
3992                  */
3993                 if (size_only && !would_convert_to_git(r->index, s->path))
3994                         return 0;
3995
3996                 /*
3997                  * Note: this check uses xsize_t(st.st_size) that may
3998                  * not be the true size of the blob after it goes
3999                  * through convert_to_git().  This may not strictly be
4000                  * correct, but the whole point of big_file_threshold
4001                  * and is_binary check being that we want to avoid
4002                  * opening the file and inspecting the contents, this
4003                  * is probably fine.
4004                  */
4005                 if ((flags & CHECK_BINARY) &&
4006                     s->size > big_file_threshold && s->is_binary == -1) {
4007                         s->is_binary = 1;
4008                         return 0;
4009                 }
4010                 fd = open(s->path, O_RDONLY);
4011                 if (fd < 0)
4012                         goto err_empty;
4013                 s->data = xmmap(NULL, s->size, PROT_READ, MAP_PRIVATE, fd, 0);
4014                 close(fd);
4015                 s->should_munmap = 1;
4016
4017                 /*
4018                  * Convert from working tree format to canonical git format
4019                  */
4020                 if (convert_to_git(r->index, s->path, s->data, s->size, &buf, conv_flags)) {
4021                         size_t size = 0;
4022                         munmap(s->data, s->size);
4023                         s->should_munmap = 0;
4024                         s->data = strbuf_detach(&buf, &size);
4025                         s->size = size;
4026                         s->should_free = 1;
4027                 }
4028         }
4029         else {
4030                 enum object_type type;
4031                 if (size_only || (flags & CHECK_BINARY)) {
4032                         type = oid_object_info(r, &s->oid, &s->size);
4033                         if (type < 0)
4034                                 die("unable to read %s",
4035                                     oid_to_hex(&s->oid));
4036                         if (size_only)
4037                                 return 0;
4038                         if (s->size > big_file_threshold && s->is_binary == -1) {
4039                                 s->is_binary = 1;
4040                                 return 0;
4041                         }
4042                 }
4043                 s->data = read_object_file(&s->oid, &type, &s->size);
4044                 if (!s->data)
4045                         die("unable to read %s", oid_to_hex(&s->oid));
4046                 s->should_free = 1;
4047         }
4048         return 0;
4049 }
4050
4051 void diff_free_filespec_blob(struct diff_filespec *s)
4052 {
4053         if (s->should_free)
4054                 free(s->data);
4055         else if (s->should_munmap)
4056                 munmap(s->data, s->size);
4057
4058         if (s->should_free || s->should_munmap) {
4059                 s->should_free = s->should_munmap = 0;
4060                 s->data = NULL;
4061         }
4062 }
4063
4064 void diff_free_filespec_data(struct diff_filespec *s)
4065 {
4066         diff_free_filespec_blob(s);
4067         FREE_AND_NULL(s->cnt_data);
4068 }
4069
4070 static void prep_temp_blob(struct index_state *istate,
4071                            const char *path, struct diff_tempfile *temp,
4072                            void *blob,
4073                            unsigned long size,
4074                            const struct object_id *oid,
4075                            int mode)
4076 {
4077         struct strbuf buf = STRBUF_INIT;
4078         struct strbuf tempfile = STRBUF_INIT;
4079         char *path_dup = xstrdup(path);
4080         const char *base = basename(path_dup);
4081
4082         /* Generate "XXXXXX_basename.ext" */
4083         strbuf_addstr(&tempfile, "XXXXXX_");
4084         strbuf_addstr(&tempfile, base);
4085
4086         temp->tempfile = mks_tempfile_ts(tempfile.buf, strlen(base) + 1);
4087         if (!temp->tempfile)
4088                 die_errno("unable to create temp-file");
4089         if (convert_to_working_tree(istate, path,
4090                         (const char *)blob, (size_t)size, &buf)) {
4091                 blob = buf.buf;
4092                 size = buf.len;
4093         }
4094         if (write_in_full(temp->tempfile->fd, blob, size) < 0 ||
4095             close_tempfile_gently(temp->tempfile))
4096                 die_errno("unable to write temp-file");
4097         temp->name = get_tempfile_path(temp->tempfile);
4098         oid_to_hex_r(temp->hex, oid);
4099         xsnprintf(temp->mode, sizeof(temp->mode), "%06o", mode);
4100         strbuf_release(&buf);
4101         strbuf_release(&tempfile);
4102         free(path_dup);
4103 }
4104
4105 static struct diff_tempfile *prepare_temp_file(struct repository *r,
4106                                                const char *name,
4107                                                struct diff_filespec *one)
4108 {
4109         struct diff_tempfile *temp = claim_diff_tempfile();
4110
4111         if (!DIFF_FILE_VALID(one)) {
4112         not_a_valid_file:
4113                 /* A '-' entry produces this for file-2, and
4114                  * a '+' entry produces this for file-1.
4115                  */
4116                 temp->name = "/dev/null";
4117                 xsnprintf(temp->hex, sizeof(temp->hex), ".");
4118                 xsnprintf(temp->mode, sizeof(temp->mode), ".");
4119                 return temp;
4120         }
4121
4122         if (!S_ISGITLINK(one->mode) &&
4123             (!one->oid_valid ||
4124              reuse_worktree_file(r->index, name, &one->oid, 1))) {
4125                 struct stat st;
4126                 if (lstat(name, &st) < 0) {
4127                         if (errno == ENOENT)
4128                                 goto not_a_valid_file;
4129                         die_errno("stat(%s)", name);
4130                 }
4131                 if (S_ISLNK(st.st_mode)) {
4132                         struct strbuf sb = STRBUF_INIT;
4133                         if (strbuf_readlink(&sb, name, st.st_size) < 0)
4134                                 die_errno("readlink(%s)", name);
4135                         prep_temp_blob(r->index, name, temp, sb.buf, sb.len,
4136                                        (one->oid_valid ?
4137                                         &one->oid : &null_oid),
4138                                        (one->oid_valid ?
4139                                         one->mode : S_IFLNK));
4140                         strbuf_release(&sb);
4141                 }
4142                 else {
4143                         /* we can borrow from the file in the work tree */
4144                         temp->name = name;
4145                         if (!one->oid_valid)
4146                                 oid_to_hex_r(temp->hex, &null_oid);
4147                         else
4148                                 oid_to_hex_r(temp->hex, &one->oid);
4149                         /* Even though we may sometimes borrow the
4150                          * contents from the work tree, we always want
4151                          * one->mode.  mode is trustworthy even when
4152                          * !(one->oid_valid), as long as
4153                          * DIFF_FILE_VALID(one).
4154                          */
4155                         xsnprintf(temp->mode, sizeof(temp->mode), "%06o", one->mode);
4156                 }
4157                 return temp;
4158         }
4159         else {
4160                 if (diff_populate_filespec(r, one, 0))
4161                         die("cannot read data blob for %s", one->path);
4162                 prep_temp_blob(r->index, name, temp,
4163                                one->data, one->size,
4164                                &one->oid, one->mode);
4165         }
4166         return temp;
4167 }
4168
4169 static void add_external_diff_name(struct repository *r,
4170                                    struct argv_array *argv,
4171                                    const char *name,
4172                                    struct diff_filespec *df)
4173 {
4174         struct diff_tempfile *temp = prepare_temp_file(r, name, df);
4175         argv_array_push(argv, temp->name);
4176         argv_array_push(argv, temp->hex);
4177         argv_array_push(argv, temp->mode);
4178 }
4179
4180 /* An external diff command takes:
4181  *
4182  * diff-cmd name infile1 infile1-sha1 infile1-mode \
4183  *               infile2 infile2-sha1 infile2-mode [ rename-to ]
4184  *
4185  */
4186 static void run_external_diff(const char *pgm,
4187                               const char *name,
4188                               const char *other,
4189                               struct diff_filespec *one,
4190                               struct diff_filespec *two,
4191                               const char *xfrm_msg,
4192                               struct diff_options *o)
4193 {
4194         struct argv_array argv = ARGV_ARRAY_INIT;
4195         struct argv_array env = ARGV_ARRAY_INIT;
4196         struct diff_queue_struct *q = &diff_queued_diff;
4197
4198         argv_array_push(&argv, pgm);
4199         argv_array_push(&argv, name);
4200
4201         if (one && two) {
4202                 add_external_diff_name(o->repo, &argv, name, one);
4203                 if (!other)
4204                         add_external_diff_name(o->repo, &argv, name, two);
4205                 else {
4206                         add_external_diff_name(o->repo, &argv, other, two);
4207                         argv_array_push(&argv, other);
4208                         argv_array_push(&argv, xfrm_msg);
4209                 }
4210         }
4211
4212         argv_array_pushf(&env, "GIT_DIFF_PATH_COUNTER=%d", ++o->diff_path_counter);
4213         argv_array_pushf(&env, "GIT_DIFF_PATH_TOTAL=%d", q->nr);
4214
4215         diff_free_filespec_data(one);
4216         diff_free_filespec_data(two);
4217         if (run_command_v_opt_cd_env(argv.argv, RUN_USING_SHELL, NULL, env.argv))
4218                 die(_("external diff died, stopping at %s"), name);
4219
4220         remove_tempfile();
4221         argv_array_clear(&argv);
4222         argv_array_clear(&env);
4223 }
4224
4225 static int similarity_index(struct diff_filepair *p)
4226 {
4227         return p->score * 100 / MAX_SCORE;
4228 }
4229
4230 static const char *diff_abbrev_oid(const struct object_id *oid, int abbrev)
4231 {
4232         if (startup_info->have_repository)
4233                 return find_unique_abbrev(oid, abbrev);
4234         else {
4235                 char *hex = oid_to_hex(oid);
4236                 if (abbrev < 0)
4237                         abbrev = FALLBACK_DEFAULT_ABBREV;
4238                 if (abbrev > the_hash_algo->hexsz)
4239                         BUG("oid abbreviation out of range: %d", abbrev);
4240                 if (abbrev)
4241                         hex[abbrev] = '\0';
4242                 return hex;
4243         }
4244 }
4245
4246 static void fill_metainfo(struct strbuf *msg,
4247                           const char *name,
4248                           const char *other,
4249                           struct diff_filespec *one,
4250                           struct diff_filespec *two,
4251                           struct diff_options *o,
4252                           struct diff_filepair *p,
4253                           int *must_show_header,
4254                           int use_color)
4255 {
4256         const char *set = diff_get_color(use_color, DIFF_METAINFO);
4257         const char *reset = diff_get_color(use_color, DIFF_RESET);
4258         const char *line_prefix = diff_line_prefix(o);
4259
4260         *must_show_header = 1;
4261         strbuf_init(msg, PATH_MAX * 2 + 300);
4262         switch (p->status) {
4263         case DIFF_STATUS_COPIED:
4264                 strbuf_addf(msg, "%s%ssimilarity index %d%%",
4265                             line_prefix, set, similarity_index(p));
4266                 strbuf_addf(msg, "%s\n%s%scopy from ",
4267                             reset,  line_prefix, set);
4268                 quote_c_style(name, msg, NULL, 0);
4269                 strbuf_addf(msg, "%s\n%s%scopy to ", reset, line_prefix, set);
4270                 quote_c_style(other, msg, NULL, 0);
4271                 strbuf_addf(msg, "%s\n", reset);
4272                 break;
4273         case DIFF_STATUS_RENAMED:
4274                 strbuf_addf(msg, "%s%ssimilarity index %d%%",
4275                             line_prefix, set, similarity_index(p));
4276                 strbuf_addf(msg, "%s\n%s%srename from ",
4277                             reset, line_prefix, set);
4278                 quote_c_style(name, msg, NULL, 0);
4279                 strbuf_addf(msg, "%s\n%s%srename to ",
4280                             reset, line_prefix, set);
4281                 quote_c_style(other, msg, NULL, 0);
4282                 strbuf_addf(msg, "%s\n", reset);
4283                 break;
4284         case DIFF_STATUS_MODIFIED:
4285                 if (p->score) {
4286                         strbuf_addf(msg, "%s%sdissimilarity index %d%%%s\n",
4287                                     line_prefix,
4288                                     set, similarity_index(p), reset);
4289                         break;
4290                 }
4291                 /* fallthru */
4292         default:
4293                 *must_show_header = 0;
4294         }
4295         if (one && two && !oideq(&one->oid, &two->oid)) {
4296                 const unsigned hexsz = the_hash_algo->hexsz;
4297                 int abbrev = o->flags.full_index ? hexsz : DEFAULT_ABBREV;
4298
4299                 if (o->flags.binary) {
4300                         mmfile_t mf;
4301                         if ((!fill_mmfile(o->repo, &mf, one) &&
4302                              diff_filespec_is_binary(o->repo, one)) ||
4303                             (!fill_mmfile(o->repo, &mf, two) &&
4304                              diff_filespec_is_binary(o->repo, two)))
4305                                 abbrev = hexsz;
4306                 }
4307                 strbuf_addf(msg, "%s%sindex %s..%s", line_prefix, set,
4308                             diff_abbrev_oid(&one->oid, abbrev),
4309                             diff_abbrev_oid(&two->oid, abbrev));
4310                 if (one->mode == two->mode)
4311                         strbuf_addf(msg, " %06o", one->mode);
4312                 strbuf_addf(msg, "%s\n", reset);
4313         }
4314 }
4315
4316 static void run_diff_cmd(const char *pgm,
4317                          const char *name,
4318                          const char *other,
4319                          const char *attr_path,
4320                          struct diff_filespec *one,
4321                          struct diff_filespec *two,
4322                          struct strbuf *msg,
4323                          struct diff_options *o,
4324                          struct diff_filepair *p)
4325 {
4326         const char *xfrm_msg = NULL;
4327         int complete_rewrite = (p->status == DIFF_STATUS_MODIFIED) && p->score;
4328         int must_show_header = 0;
4329
4330
4331         if (o->flags.allow_external) {
4332                 struct userdiff_driver *drv;
4333
4334                 drv = userdiff_find_by_path(o->repo->index, attr_path);
4335                 if (drv && drv->external)
4336                         pgm = drv->external;
4337         }
4338
4339         if (msg) {
4340                 /*
4341                  * don't use colors when the header is intended for an
4342                  * external diff driver
4343                  */
4344                 fill_metainfo(msg, name, other, one, two, o, p,
4345                               &must_show_header,
4346                               want_color(o->use_color) && !pgm);
4347                 xfrm_msg = msg->len ? msg->buf : NULL;
4348         }
4349
4350         if (pgm) {
4351                 run_external_diff(pgm, name, other, one, two, xfrm_msg, o);
4352                 return;
4353         }
4354         if (one && two)
4355                 builtin_diff(name, other ? other : name,
4356                              one, two, xfrm_msg, must_show_header,
4357                              o, complete_rewrite);
4358         else
4359                 fprintf(o->file, "* Unmerged path %s\n", name);
4360 }
4361
4362 static void diff_fill_oid_info(struct diff_filespec *one, struct index_state *istate)
4363 {
4364         if (DIFF_FILE_VALID(one)) {
4365                 if (!one->oid_valid) {
4366                         struct stat st;
4367                         if (one->is_stdin) {
4368                                 oidclr(&one->oid);
4369                                 return;
4370                         }
4371                         if (lstat(one->path, &st) < 0)
4372                                 die_errno("stat '%s'", one->path);
4373                         if (index_path(istate, &one->oid, one->path, &st, 0))
4374                                 die("cannot hash %s", one->path);
4375                 }
4376         }
4377         else
4378                 oidclr(&one->oid);
4379 }
4380
4381 static void strip_prefix(int prefix_length, const char **namep, const char **otherp)
4382 {
4383         /* Strip the prefix but do not molest /dev/null and absolute paths */
4384         if (*namep && !is_absolute_path(*namep)) {
4385                 *namep += prefix_length;
4386                 if (**namep == '/')
4387                         ++*namep;
4388         }
4389         if (*otherp && !is_absolute_path(*otherp)) {
4390                 *otherp += prefix_length;
4391                 if (**otherp == '/')
4392                         ++*otherp;
4393         }
4394 }
4395
4396 static void run_diff(struct diff_filepair *p, struct diff_options *o)
4397 {
4398         const char *pgm = external_diff();
4399         struct strbuf msg;
4400         struct diff_filespec *one = p->one;
4401         struct diff_filespec *two = p->two;
4402         const char *name;
4403         const char *other;
4404         const char *attr_path;
4405
4406         name  = one->path;
4407         other = (strcmp(name, two->path) ? two->path : NULL);
4408         attr_path = name;
4409         if (o->prefix_length)
4410                 strip_prefix(o->prefix_length, &name, &other);
4411
4412         if (!o->flags.allow_external)
4413                 pgm = NULL;
4414
4415         if (DIFF_PAIR_UNMERGED(p)) {
4416                 run_diff_cmd(pgm, name, NULL, attr_path,
4417                              NULL, NULL, NULL, o, p);
4418                 return;
4419         }
4420
4421         diff_fill_oid_info(one, o->repo->index);
4422         diff_fill_oid_info(two, o->repo->index);
4423
4424         if (!pgm &&
4425             DIFF_FILE_VALID(one) && DIFF_FILE_VALID(two) &&
4426             (S_IFMT & one->mode) != (S_IFMT & two->mode)) {
4427                 /*
4428                  * a filepair that changes between file and symlink
4429                  * needs to be split into deletion and creation.
4430                  */
4431                 struct diff_filespec *null = alloc_filespec(two->path);
4432                 run_diff_cmd(NULL, name, other, attr_path,
4433                              one, null, &msg,
4434                              o, p);
4435                 free(null);
4436                 strbuf_release(&msg);
4437
4438                 null = alloc_filespec(one->path);
4439                 run_diff_cmd(NULL, name, other, attr_path,
4440                              null, two, &msg, o, p);
4441                 free(null);
4442         }
4443         else
4444                 run_diff_cmd(pgm, name, other, attr_path,
4445                              one, two, &msg, o, p);
4446
4447         strbuf_release(&msg);
4448 }
4449
4450 static void run_diffstat(struct diff_filepair *p, struct diff_options *o,
4451                          struct diffstat_t *diffstat)
4452 {
4453         const char *name;
4454         const char *other;
4455
4456         if (DIFF_PAIR_UNMERGED(p)) {
4457                 /* unmerged */
4458                 builtin_diffstat(p->one->path, NULL, NULL, NULL,
4459                                  diffstat, o, p);
4460                 return;
4461         }
4462
4463         name = p->one->path;
4464         other = (strcmp(name, p->two->path) ? p->two->path : NULL);
4465
4466         if (o->prefix_length)
4467                 strip_prefix(o->prefix_length, &name, &other);
4468
4469         diff_fill_oid_info(p->one, o->repo->index);
4470         diff_fill_oid_info(p->two, o->repo->index);
4471
4472         builtin_diffstat(name, other, p->one, p->two,
4473                          diffstat, o, p);
4474 }
4475
4476 static void run_checkdiff(struct diff_filepair *p, struct diff_options *o)
4477 {
4478         const char *name;
4479         const char *other;
4480         const char *attr_path;
4481
4482         if (DIFF_PAIR_UNMERGED(p)) {
4483                 /* unmerged */
4484                 return;
4485         }
4486
4487         name = p->one->path;
4488         other = (strcmp(name, p->two->path) ? p->two->path : NULL);
4489         attr_path = other ? other : name;
4490
4491         if (o->prefix_length)
4492                 strip_prefix(o->prefix_length, &name, &other);
4493
4494         diff_fill_oid_info(p->one, o->repo->index);
4495         diff_fill_oid_info(p->two, o->repo->index);
4496
4497         builtin_checkdiff(name, other, attr_path, p->one, p->two, o);
4498 }
4499
4500 static void prep_parse_options(struct diff_options *options);
4501
4502 void repo_diff_setup(struct repository *r, struct diff_options *options)
4503 {
4504         memcpy(options, &default_diff_options, sizeof(*options));
4505
4506         options->file = stdout;
4507         options->repo = r;
4508
4509         options->output_indicators[OUTPUT_INDICATOR_NEW] = '+';
4510         options->output_indicators[OUTPUT_INDICATOR_OLD] = '-';
4511         options->output_indicators[OUTPUT_INDICATOR_CONTEXT] = ' ';
4512         options->abbrev = DEFAULT_ABBREV;
4513         options->line_termination = '\n';
4514         options->break_opt = -1;
4515         options->rename_limit = -1;
4516         options->dirstat_permille = diff_dirstat_permille_default;
4517         options->context = diff_context_default;
4518         options->interhunkcontext = diff_interhunk_context_default;
4519         options->ws_error_highlight = ws_error_highlight_default;
4520         options->flags.rename_empty = 1;
4521         options->objfind = NULL;
4522
4523         /* pathchange left =NULL by default */
4524         options->change = diff_change;
4525         options->add_remove = diff_addremove;
4526         options->use_color = diff_use_color_default;
4527         options->detect_rename = diff_detect_rename_default;
4528         options->xdl_opts |= diff_algorithm;
4529         if (diff_indent_heuristic)
4530                 DIFF_XDL_SET(options, INDENT_HEURISTIC);
4531
4532         options->orderfile = diff_order_file_cfg;
4533
4534         if (diff_no_prefix) {
4535                 options->a_prefix = options->b_prefix = "";
4536         } else if (!diff_mnemonic_prefix) {
4537                 options->a_prefix = "a/";
4538                 options->b_prefix = "b/";
4539         }
4540
4541         options->color_moved = diff_color_moved_default;
4542         options->color_moved_ws_handling = diff_color_moved_ws_default;
4543
4544         prep_parse_options(options);
4545 }
4546
4547 void diff_setup_done(struct diff_options *options)
4548 {
4549         unsigned check_mask = DIFF_FORMAT_NAME |
4550                               DIFF_FORMAT_NAME_STATUS |
4551                               DIFF_FORMAT_CHECKDIFF |
4552                               DIFF_FORMAT_NO_OUTPUT;
4553         /*
4554          * This must be signed because we're comparing against a potentially
4555          * negative value.
4556          */
4557         const int hexsz = the_hash_algo->hexsz;
4558
4559         if (options->set_default)
4560                 options->set_default(options);
4561
4562         if (HAS_MULTI_BITS(options->output_format & check_mask))
4563                 die(_("--name-only, --name-status, --check and -s are mutually exclusive"));
4564
4565         if (HAS_MULTI_BITS(options->pickaxe_opts & DIFF_PICKAXE_KINDS_MASK))
4566                 die(_("-G, -S and --find-object are mutually exclusive"));
4567
4568         /*
4569          * Most of the time we can say "there are changes"
4570          * only by checking if there are changed paths, but
4571          * --ignore-whitespace* options force us to look
4572          * inside contents.
4573          */
4574
4575         if ((options->xdl_opts & XDF_WHITESPACE_FLAGS))
4576                 options->flags.diff_from_contents = 1;
4577         else
4578                 options->flags.diff_from_contents = 0;
4579
4580         if (options->flags.find_copies_harder)
4581                 options->detect_rename = DIFF_DETECT_COPY;
4582
4583         if (!options->flags.relative_name)
4584                 options->prefix = NULL;
4585         if (options->prefix)
4586                 options->prefix_length = strlen(options->prefix);
4587         else
4588                 options->prefix_length = 0;
4589
4590         if (options->output_format & (DIFF_FORMAT_NAME |
4591                                       DIFF_FORMAT_NAME_STATUS |
4592                                       DIFF_FORMAT_CHECKDIFF |
4593                                       DIFF_FORMAT_NO_OUTPUT))
4594                 options->output_format &= ~(DIFF_FORMAT_RAW |
4595                                             DIFF_FORMAT_NUMSTAT |
4596                                             DIFF_FORMAT_DIFFSTAT |
4597                                             DIFF_FORMAT_SHORTSTAT |
4598                                             DIFF_FORMAT_DIRSTAT |
4599                                             DIFF_FORMAT_SUMMARY |
4600                                             DIFF_FORMAT_PATCH);
4601
4602         /*
4603          * These cases always need recursive; we do not drop caller-supplied
4604          * recursive bits for other formats here.
4605          */
4606         if (options->output_format & (DIFF_FORMAT_PATCH |
4607                                       DIFF_FORMAT_NUMSTAT |
4608                                       DIFF_FORMAT_DIFFSTAT |
4609                                       DIFF_FORMAT_SHORTSTAT |
4610                                       DIFF_FORMAT_DIRSTAT |
4611                                       DIFF_FORMAT_SUMMARY |
4612                                       DIFF_FORMAT_CHECKDIFF))
4613                 options->flags.recursive = 1;
4614         /*
4615          * Also pickaxe would not work very well if you do not say recursive
4616          */
4617         if (options->pickaxe_opts & DIFF_PICKAXE_KINDS_MASK)
4618                 options->flags.recursive = 1;
4619         /*
4620          * When patches are generated, submodules diffed against the work tree
4621          * must be checked for dirtiness too so it can be shown in the output
4622          */
4623         if (options->output_format & DIFF_FORMAT_PATCH)
4624                 options->flags.dirty_submodules = 1;
4625
4626         if (options->detect_rename && options->rename_limit < 0)
4627                 options->rename_limit = diff_rename_limit_default;
4628         if (hexsz < options->abbrev)
4629                 options->abbrev = hexsz; /* full */
4630
4631         /*
4632          * It does not make sense to show the first hit we happened
4633          * to have found.  It does not make sense not to return with
4634          * exit code in such a case either.
4635          */
4636         if (options->flags.quick) {
4637                 options->output_format = DIFF_FORMAT_NO_OUTPUT;
4638                 options->flags.exit_with_status = 1;
4639         }
4640
4641         options->diff_path_counter = 0;
4642
4643         if (options->flags.follow_renames && options->pathspec.nr != 1)
4644                 die(_("--follow requires exactly one pathspec"));
4645
4646         if (!options->use_color || external_diff())
4647                 options->color_moved = 0;
4648
4649         FREE_AND_NULL(options->parseopts);
4650 }
4651
4652 int parse_long_opt(const char *opt, const char **argv,
4653                    const char **optarg)
4654 {
4655         const char *arg = argv[0];
4656         if (!skip_prefix(arg, "--", &arg))
4657                 return 0;
4658         if (!skip_prefix(arg, opt, &arg))
4659                 return 0;
4660         if (*arg == '=') { /* stuck form: --option=value */
4661                 *optarg = arg + 1;
4662                 return 1;
4663         }
4664         if (*arg != '\0')
4665                 return 0;
4666         /* separate form: --option value */
4667         if (!argv[1])
4668                 die("Option '--%s' requires a value", opt);
4669         *optarg = argv[1];
4670         return 2;
4671 }
4672
4673 static int diff_opt_stat(const struct option *opt, const char *value, int unset)
4674 {
4675         struct diff_options *options = opt->value;
4676         int width = options->stat_width;
4677         int name_width = options->stat_name_width;
4678         int graph_width = options->stat_graph_width;
4679         int count = options->stat_count;
4680         char *end;
4681
4682         BUG_ON_OPT_NEG(unset);
4683
4684         if (!strcmp(opt->long_name, "stat")) {
4685                 if (value) {
4686                         width = strtoul(value, &end, 10);
4687                         if (*end == ',')
4688                                 name_width = strtoul(end+1, &end, 10);
4689                         if (*end == ',')
4690                                 count = strtoul(end+1, &end, 10);
4691                         if (*end)
4692                                 return error(_("invalid --stat value: %s"), value);
4693                 }
4694         } else if (!strcmp(opt->long_name, "stat-width")) {
4695                 width = strtoul(value, &end, 10);
4696                 if (*end)
4697                         return error(_("%s expects a numerical value"),
4698                                      opt->long_name);
4699         } else if (!strcmp(opt->long_name, "stat-name-width")) {
4700                 name_width = strtoul(value, &end, 10);
4701                 if (*end)
4702                         return error(_("%s expects a numerical value"),
4703                                      opt->long_name);
4704         } else if (!strcmp(opt->long_name, "stat-graph-width")) {
4705                 graph_width = strtoul(value, &end, 10);
4706                 if (*end)
4707                         return error(_("%s expects a numerical value"),
4708                                      opt->long_name);
4709         } else if (!strcmp(opt->long_name, "stat-count")) {
4710                 count = strtoul(value, &end, 10);
4711                 if (*end)
4712                         return error(_("%s expects a numerical value"),
4713                                      opt->long_name);
4714         } else
4715                 BUG("%s should not get here", opt->long_name);
4716
4717         options->output_format |= DIFF_FORMAT_DIFFSTAT;
4718         options->stat_name_width = name_width;
4719         options->stat_graph_width = graph_width;
4720         options->stat_width = width;
4721         options->stat_count = count;
4722         return 0;
4723 }
4724
4725 static int parse_dirstat_opt(struct diff_options *options, const char *params)
4726 {
4727         struct strbuf errmsg = STRBUF_INIT;
4728         if (parse_dirstat_params(options, params, &errmsg))
4729                 die(_("Failed to parse --dirstat/-X option parameter:\n%s"),
4730                     errmsg.buf);
4731         strbuf_release(&errmsg);
4732         /*
4733          * The caller knows a dirstat-related option is given from the command
4734          * line; allow it to say "return this_function();"
4735          */
4736         options->output_format |= DIFF_FORMAT_DIRSTAT;
4737         return 1;
4738 }
4739
4740 static const char diff_status_letters[] = {
4741         DIFF_STATUS_ADDED,
4742         DIFF_STATUS_COPIED,
4743         DIFF_STATUS_DELETED,
4744         DIFF_STATUS_MODIFIED,
4745         DIFF_STATUS_RENAMED,
4746         DIFF_STATUS_TYPE_CHANGED,
4747         DIFF_STATUS_UNKNOWN,
4748         DIFF_STATUS_UNMERGED,
4749         DIFF_STATUS_FILTER_AON,
4750         DIFF_STATUS_FILTER_BROKEN,
4751         '\0',
4752 };
4753
4754 static unsigned int filter_bit['Z' + 1];
4755
4756 static void prepare_filter_bits(void)
4757 {
4758         int i;
4759
4760         if (!filter_bit[DIFF_STATUS_ADDED]) {
4761                 for (i = 0; diff_status_letters[i]; i++)
4762                         filter_bit[(int) diff_status_letters[i]] = (1 << i);
4763         }
4764 }
4765
4766 static unsigned filter_bit_tst(char status, const struct diff_options *opt)
4767 {
4768         return opt->filter & filter_bit[(int) status];
4769 }
4770
4771 unsigned diff_filter_bit(char status)
4772 {
4773         prepare_filter_bits();
4774         return filter_bit[(int) status];
4775 }
4776
4777 static int diff_opt_diff_filter(const struct option *option,
4778                                 const char *optarg, int unset)
4779 {
4780         struct diff_options *opt = option->value;
4781         int i, optch;
4782
4783         BUG_ON_OPT_NEG(unset);
4784         prepare_filter_bits();
4785
4786         /*
4787          * If there is a negation e.g. 'd' in the input, and we haven't
4788          * initialized the filter field with another --diff-filter, start
4789          * from full set of bits, except for AON.
4790          */
4791         if (!opt->filter) {
4792                 for (i = 0; (optch = optarg[i]) != '\0'; i++) {
4793                         if (optch < 'a' || 'z' < optch)
4794                                 continue;
4795                         opt->filter = (1 << (ARRAY_SIZE(diff_status_letters) - 1)) - 1;
4796                         opt->filter &= ~filter_bit[DIFF_STATUS_FILTER_AON];
4797                         break;
4798                 }
4799         }
4800
4801         for (i = 0; (optch = optarg[i]) != '\0'; i++) {
4802                 unsigned int bit;
4803                 int negate;
4804
4805                 if ('a' <= optch && optch <= 'z') {
4806                         negate = 1;
4807                         optch = toupper(optch);
4808                 } else {
4809                         negate = 0;
4810                 }
4811
4812                 bit = (0 <= optch && optch <= 'Z') ? filter_bit[optch] : 0;
4813                 if (!bit)
4814                         return error(_("unknown change class '%c' in --diff-filter=%s"),
4815                                      optarg[i], optarg);
4816                 if (negate)
4817                         opt->filter &= ~bit;
4818                 else
4819                         opt->filter |= bit;
4820         }
4821         return 0;
4822 }
4823
4824 static void enable_patch_output(int *fmt)
4825 {
4826         *fmt &= ~DIFF_FORMAT_NO_OUTPUT;
4827         *fmt |= DIFF_FORMAT_PATCH;
4828 }
4829
4830 static int diff_opt_ws_error_highlight(const struct option *option,
4831                                        const char *arg, int unset)
4832 {
4833         struct diff_options *opt = option->value;
4834         int val = parse_ws_error_highlight(arg);
4835
4836         BUG_ON_OPT_NEG(unset);
4837         if (val < 0)
4838                 return error(_("unknown value after ws-error-highlight=%.*s"),
4839                              -1 - val, arg);
4840         opt->ws_error_highlight = val;
4841         return 0;
4842 }
4843
4844 static int diff_opt_find_object(const struct option *option,
4845                                 const char *arg, int unset)
4846 {
4847         struct diff_options *opt = option->value;
4848         struct object_id oid;
4849
4850         BUG_ON_OPT_NEG(unset);
4851         if (get_oid(arg, &oid))
4852                 return error(_("unable to resolve '%s'"), arg);
4853
4854         if (!opt->objfind)
4855                 opt->objfind = xcalloc(1, sizeof(*opt->objfind));
4856
4857         opt->pickaxe_opts |= DIFF_PICKAXE_KIND_OBJFIND;
4858         opt->flags.recursive = 1;
4859         opt->flags.tree_in_recursive = 1;
4860         oidset_insert(opt->objfind, &oid);
4861         return 0;
4862 }
4863
4864 static int diff_opt_anchored(const struct option *opt,
4865                              const char *arg, int unset)
4866 {
4867         struct diff_options *options = opt->value;
4868
4869         BUG_ON_OPT_NEG(unset);
4870         options->xdl_opts = DIFF_WITH_ALG(options, PATIENCE_DIFF);
4871         ALLOC_GROW(options->anchors, options->anchors_nr + 1,
4872                    options->anchors_alloc);
4873         options->anchors[options->anchors_nr++] = xstrdup(arg);
4874         return 0;
4875 }
4876
4877 static int diff_opt_binary(const struct option *opt,
4878                            const char *arg, int unset)
4879 {
4880         struct diff_options *options = opt->value;
4881
4882         BUG_ON_OPT_NEG(unset);
4883         BUG_ON_OPT_ARG(arg);
4884         enable_patch_output(&options->output_format);
4885         options->flags.binary = 1;
4886         return 0;
4887 }
4888
4889 static int diff_opt_break_rewrites(const struct option *opt,
4890                                    const char *arg, int unset)
4891 {
4892         int *break_opt = opt->value;
4893         int opt1, opt2;
4894
4895         BUG_ON_OPT_NEG(unset);
4896         if (!arg)
4897                 arg = "";
4898         opt1 = parse_rename_score(&arg);
4899         if (*arg == 0)
4900                 opt2 = 0;
4901         else if (*arg != '/')
4902                 return error(_("%s expects <n>/<m> form"), opt->long_name);
4903         else {
4904                 arg++;
4905                 opt2 = parse_rename_score(&arg);
4906         }
4907         if (*arg != 0)
4908                 return error(_("%s expects <n>/<m> form"), opt->long_name);
4909         *break_opt = opt1 | (opt2 << 16);
4910         return 0;
4911 }
4912
4913 static int diff_opt_char(const struct option *opt,
4914                          const char *arg, int unset)
4915 {
4916         char *value = opt->value;
4917
4918         BUG_ON_OPT_NEG(unset);
4919         if (arg[1])
4920                 return error(_("%s expects a character, got '%s'"),
4921                              opt->long_name, arg);
4922         *value = arg[0];
4923         return 0;
4924 }
4925
4926 static int diff_opt_color_moved(const struct option *opt,
4927                                 const char *arg, int unset)
4928 {
4929         struct diff_options *options = opt->value;
4930
4931         if (unset) {
4932                 options->color_moved = COLOR_MOVED_NO;
4933         } else if (!arg) {
4934                 if (diff_color_moved_default)
4935                         options->color_moved = diff_color_moved_default;
4936                 if (options->color_moved == COLOR_MOVED_NO)
4937                         options->color_moved = COLOR_MOVED_DEFAULT;
4938         } else {
4939                 int cm = parse_color_moved(arg);
4940                 if (cm < 0)
4941                         return error(_("bad --color-moved argument: %s"), arg);
4942                 options->color_moved = cm;
4943         }
4944         return 0;
4945 }
4946
4947 static int diff_opt_color_moved_ws(const struct option *opt,
4948                                    const char *arg, int unset)
4949 {
4950         struct diff_options *options = opt->value;
4951         unsigned cm;
4952
4953         if (unset) {
4954                 options->color_moved_ws_handling = 0;
4955                 return 0;
4956         }
4957
4958         cm = parse_color_moved_ws(arg);
4959         if (cm & COLOR_MOVED_WS_ERROR)
4960                 return error(_("invalid mode '%s' in --color-moved-ws"), arg);
4961         options->color_moved_ws_handling = cm;
4962         return 0;
4963 }
4964
4965 static int diff_opt_color_words(const struct option *opt,
4966                                 const char *arg, int unset)
4967 {
4968         struct diff_options *options = opt->value;
4969
4970         BUG_ON_OPT_NEG(unset);
4971         options->use_color = 1;
4972         options->word_diff = DIFF_WORDS_COLOR;
4973         options->word_regex = arg;
4974         return 0;
4975 }
4976
4977 static int diff_opt_compact_summary(const struct option *opt,
4978                                     const char *arg, int unset)
4979 {
4980         struct diff_options *options = opt->value;
4981
4982         BUG_ON_OPT_ARG(arg);
4983         if (unset) {
4984                 options->flags.stat_with_summary = 0;
4985         } else {
4986                 options->flags.stat_with_summary = 1;
4987                 options->output_format |= DIFF_FORMAT_DIFFSTAT;
4988         }
4989         return 0;
4990 }
4991
4992 static int diff_opt_diff_algorithm(const struct option *opt,
4993                                    const char *arg, int unset)
4994 {
4995         struct diff_options *options = opt->value;
4996         long value = parse_algorithm_value(arg);
4997
4998         BUG_ON_OPT_NEG(unset);
4999         if (value < 0)
5000                 return error(_("option diff-algorithm accepts \"myers\", "
5001                                "\"minimal\", \"patience\" and \"histogram\""));
5002
5003         /* clear out previous settings */
5004         DIFF_XDL_CLR(options, NEED_MINIMAL);
5005         options->xdl_opts &= ~XDF_DIFF_ALGORITHM_MASK;
5006         options->xdl_opts |= value;
5007         return 0;
5008 }
5009
5010 static int diff_opt_dirstat(const struct option *opt,
5011                             const char *arg, int unset)
5012 {
5013         struct diff_options *options = opt->value;
5014
5015         BUG_ON_OPT_NEG(unset);
5016         if (!strcmp(opt->long_name, "cumulative")) {
5017                 if (arg)
5018                         BUG("how come --cumulative take a value?");
5019                 arg = "cumulative";
5020         } else if (!strcmp(opt->long_name, "dirstat-by-file"))
5021                 parse_dirstat_opt(options, "files");
5022         parse_dirstat_opt(options, arg ? arg : "");
5023         return 0;
5024 }
5025
5026 static int diff_opt_find_copies(const struct option *opt,
5027                                 const char *arg, int unset)
5028 {
5029         struct diff_options *options = opt->value;
5030
5031         BUG_ON_OPT_NEG(unset);
5032         if (!arg)
5033                 arg = "";
5034         options->rename_score = parse_rename_score(&arg);
5035         if (*arg != 0)
5036                 return error(_("invalid argument to %s"), opt->long_name);
5037
5038         if (options->detect_rename == DIFF_DETECT_COPY)
5039                 options->flags.find_copies_harder = 1;
5040         else
5041                 options->detect_rename = DIFF_DETECT_COPY;
5042
5043         return 0;
5044 }
5045
5046 static int diff_opt_find_renames(const struct option *opt,
5047                                  const char *arg, int unset)
5048 {
5049         struct diff_options *options = opt->value;
5050
5051         BUG_ON_OPT_NEG(unset);
5052         if (!arg)
5053                 arg = "";
5054         options->rename_score = parse_rename_score(&arg);
5055         if (*arg != 0)
5056                 return error(_("invalid argument to %s"), opt->long_name);
5057
5058         options->detect_rename = DIFF_DETECT_RENAME;
5059         return 0;
5060 }
5061
5062 static int diff_opt_follow(const struct option *opt,
5063                            const char *arg, int unset)
5064 {
5065         struct diff_options *options = opt->value;
5066
5067         BUG_ON_OPT_ARG(arg);
5068         if (unset) {
5069                 options->flags.follow_renames = 0;
5070                 options->flags.default_follow_renames = 0;
5071         } else {
5072                 options->flags.follow_renames = 1;
5073         }
5074         return 0;
5075 }
5076
5077 static int diff_opt_ignore_submodules(const struct option *opt,
5078                                       const char *arg, int unset)
5079 {
5080         struct diff_options *options = opt->value;
5081
5082         BUG_ON_OPT_NEG(unset);
5083         if (!arg)
5084                 arg = "all";
5085         options->flags.override_submodule_config = 1;
5086         handle_ignore_submodules_arg(options, arg);
5087         return 0;
5088 }
5089
5090 static int diff_opt_line_prefix(const struct option *opt,
5091                                 const char *optarg, int unset)
5092 {
5093         struct diff_options *options = opt->value;
5094
5095         BUG_ON_OPT_NEG(unset);
5096         options->line_prefix = optarg;
5097         options->line_prefix_length = strlen(options->line_prefix);
5098         graph_setup_line_prefix(options);
5099         return 0;
5100 }
5101
5102 static int diff_opt_no_prefix(const struct option *opt,
5103                               const char *optarg, int unset)
5104 {
5105         struct diff_options *options = opt->value;
5106
5107         BUG_ON_OPT_NEG(unset);
5108         BUG_ON_OPT_ARG(optarg);
5109         options->a_prefix = "";
5110         options->b_prefix = "";
5111         return 0;
5112 }
5113
5114 static enum parse_opt_result diff_opt_output(struct parse_opt_ctx_t *ctx,
5115                                              const struct option *opt,
5116                                              const char *arg, int unset)
5117 {
5118         struct diff_options *options = opt->value;
5119         char *path;
5120
5121         BUG_ON_OPT_NEG(unset);
5122         path = prefix_filename(ctx->prefix, arg);
5123         options->file = xfopen(path, "w");
5124         options->close_file = 1;
5125         if (options->use_color != GIT_COLOR_ALWAYS)
5126                 options->use_color = GIT_COLOR_NEVER;
5127         free(path);
5128         return 0;
5129 }
5130
5131 static int diff_opt_patience(const struct option *opt,
5132                              const char *arg, int unset)
5133 {
5134         struct diff_options *options = opt->value;
5135         int i;
5136
5137         BUG_ON_OPT_NEG(unset);
5138         BUG_ON_OPT_ARG(arg);
5139         options->xdl_opts = DIFF_WITH_ALG(options, PATIENCE_DIFF);
5140         /*
5141          * Both --patience and --anchored use PATIENCE_DIFF
5142          * internally, so remove any anchors previously
5143          * specified.
5144          */
5145         for (i = 0; i < options->anchors_nr; i++)
5146                 free(options->anchors[i]);
5147         options->anchors_nr = 0;
5148         return 0;
5149 }
5150
5151 static int diff_opt_pickaxe_regex(const struct option *opt,
5152                                   const char *arg, int unset)
5153 {
5154         struct diff_options *options = opt->value;
5155
5156         BUG_ON_OPT_NEG(unset);
5157         options->pickaxe = arg;
5158         options->pickaxe_opts |= DIFF_PICKAXE_KIND_G;
5159         return 0;
5160 }
5161
5162 static int diff_opt_pickaxe_string(const struct option *opt,
5163                                    const char *arg, int unset)
5164 {
5165         struct diff_options *options = opt->value;
5166
5167         BUG_ON_OPT_NEG(unset);
5168         options->pickaxe = arg;
5169         options->pickaxe_opts |= DIFF_PICKAXE_KIND_S;
5170         return 0;
5171 }
5172
5173 static int diff_opt_relative(const struct option *opt,
5174                              const char *arg, int unset)
5175 {
5176         struct diff_options *options = opt->value;
5177
5178         BUG_ON_OPT_NEG(unset);
5179         options->flags.relative_name = 1;
5180         if (arg)
5181                 options->prefix = arg;
5182         return 0;
5183 }
5184
5185 static int diff_opt_submodule(const struct option *opt,
5186                               const char *arg, int unset)
5187 {
5188         struct diff_options *options = opt->value;
5189
5190         BUG_ON_OPT_NEG(unset);
5191         if (!arg)
5192                 arg = "log";
5193         if (parse_submodule_params(options, arg))
5194                 return error(_("failed to parse --submodule option parameter: '%s'"),
5195                              arg);
5196         return 0;
5197 }
5198
5199 static int diff_opt_textconv(const struct option *opt,
5200                              const char *arg, int unset)
5201 {
5202         struct diff_options *options = opt->value;
5203
5204         BUG_ON_OPT_ARG(arg);
5205         if (unset) {
5206                 options->flags.allow_textconv = 0;
5207         } else {
5208                 options->flags.allow_textconv = 1;
5209                 options->flags.textconv_set_via_cmdline = 1;
5210         }
5211         return 0;
5212 }
5213
5214 static int diff_opt_unified(const struct option *opt,
5215                             const char *arg, int unset)
5216 {
5217         struct diff_options *options = opt->value;
5218         char *s;
5219
5220         BUG_ON_OPT_NEG(unset);
5221
5222         if (arg) {
5223                 options->context = strtol(arg, &s, 10);
5224                 if (*s)
5225                         return error(_("%s expects a numerical value"), "--unified");
5226         }
5227         enable_patch_output(&options->output_format);
5228
5229         return 0;
5230 }
5231
5232 static int diff_opt_word_diff(const struct option *opt,
5233                               const char *arg, int unset)
5234 {
5235         struct diff_options *options = opt->value;
5236
5237         BUG_ON_OPT_NEG(unset);
5238         if (arg) {
5239                 if (!strcmp(arg, "plain"))
5240                         options->word_diff = DIFF_WORDS_PLAIN;
5241                 else if (!strcmp(arg, "color")) {
5242                         options->use_color = 1;
5243                         options->word_diff = DIFF_WORDS_COLOR;
5244                 }
5245                 else if (!strcmp(arg, "porcelain"))
5246                         options->word_diff = DIFF_WORDS_PORCELAIN;
5247                 else if (!strcmp(arg, "none"))
5248                         options->word_diff = DIFF_WORDS_NONE;
5249                 else
5250                         return error(_("bad --word-diff argument: %s"), arg);
5251         } else {
5252                 if (options->word_diff == DIFF_WORDS_NONE)
5253                         options->word_diff = DIFF_WORDS_PLAIN;
5254         }
5255         return 0;
5256 }
5257
5258 static int diff_opt_word_diff_regex(const struct option *opt,
5259                                     const char *arg, int unset)
5260 {
5261         struct diff_options *options = opt->value;
5262
5263         BUG_ON_OPT_NEG(unset);
5264         if (options->word_diff == DIFF_WORDS_NONE)
5265                 options->word_diff = DIFF_WORDS_PLAIN;
5266         options->word_regex = arg;
5267         return 0;
5268 }
5269
5270 static void prep_parse_options(struct diff_options *options)
5271 {
5272         struct option parseopts[] = {
5273                 OPT_GROUP(N_("Diff output format options")),
5274                 OPT_BITOP('p', "patch", &options->output_format,
5275                           N_("generate patch"),
5276                           DIFF_FORMAT_PATCH, DIFF_FORMAT_NO_OUTPUT),
5277                 OPT_BIT_F('s', "no-patch", &options->output_format,
5278                           N_("suppress diff output"),
5279                           DIFF_FORMAT_NO_OUTPUT, PARSE_OPT_NONEG),
5280                 OPT_BITOP('u', NULL, &options->output_format,
5281                           N_("generate patch"),
5282                           DIFF_FORMAT_PATCH, DIFF_FORMAT_NO_OUTPUT),
5283                 OPT_CALLBACK_F('U', "unified", options, N_("<n>"),
5284                                N_("generate diffs with <n> lines context"),
5285                                PARSE_OPT_NONEG | PARSE_OPT_OPTARG, diff_opt_unified),
5286                 OPT_BOOL('W', "function-context", &options->flags.funccontext,
5287                          N_("generate diffs with <n> lines context")),
5288                 OPT_BIT_F(0, "raw", &options->output_format,
5289                           N_("generate the diff in raw format"),
5290                           DIFF_FORMAT_RAW, PARSE_OPT_NONEG),
5291                 OPT_BITOP(0, "patch-with-raw", &options->output_format,
5292                           N_("synonym for '-p --raw'"),
5293                           DIFF_FORMAT_PATCH | DIFF_FORMAT_RAW,
5294                           DIFF_FORMAT_NO_OUTPUT),
5295                 OPT_BITOP(0, "patch-with-stat", &options->output_format,
5296                           N_("synonym for '-p --stat'"),
5297                           DIFF_FORMAT_PATCH | DIFF_FORMAT_DIFFSTAT,
5298                           DIFF_FORMAT_NO_OUTPUT),
5299                 OPT_BIT_F(0, "numstat", &options->output_format,
5300                           N_("machine friendly --stat"),
5301                           DIFF_FORMAT_NUMSTAT, PARSE_OPT_NONEG),
5302                 OPT_BIT_F(0, "shortstat", &options->output_format,
5303                           N_("output only the last line of --stat"),
5304                           DIFF_FORMAT_SHORTSTAT, PARSE_OPT_NONEG),
5305                 OPT_CALLBACK_F('X', "dirstat", options, N_("<param1,param2>..."),
5306                                N_("output the distribution of relative amount of changes for each sub-directory"),
5307                                PARSE_OPT_NONEG | PARSE_OPT_OPTARG,
5308                                diff_opt_dirstat),
5309                 OPT_CALLBACK_F(0, "cumulative", options, NULL,
5310                                N_("synonym for --dirstat=cumulative"),
5311                                PARSE_OPT_NONEG | PARSE_OPT_NOARG,
5312                                diff_opt_dirstat),
5313                 OPT_CALLBACK_F(0, "dirstat-by-file", options, N_("<param1,param2>..."),
5314                                N_("synonym for --dirstat=files,param1,param2..."),
5315                                PARSE_OPT_NONEG | PARSE_OPT_OPTARG,
5316                                diff_opt_dirstat),
5317                 OPT_BIT_F(0, "check", &options->output_format,
5318                           N_("warn if changes introduce conflict markers or whitespace errors"),
5319                           DIFF_FORMAT_CHECKDIFF, PARSE_OPT_NONEG),
5320                 OPT_BIT_F(0, "summary", &options->output_format,
5321                           N_("condensed summary such as creations, renames and mode changes"),
5322                           DIFF_FORMAT_SUMMARY, PARSE_OPT_NONEG),
5323                 OPT_BIT_F(0, "name-only", &options->output_format,
5324                           N_("show only names of changed files"),
5325                           DIFF_FORMAT_NAME, PARSE_OPT_NONEG),
5326                 OPT_BIT_F(0, "name-status", &options->output_format,
5327                           N_("show only names and status of changed files"),
5328                           DIFF_FORMAT_NAME_STATUS, PARSE_OPT_NONEG),
5329                 OPT_CALLBACK_F(0, "stat", options, N_("<width>[,<name-width>[,<count>]]"),
5330                                N_("generate diffstat"),
5331                                PARSE_OPT_NONEG | PARSE_OPT_OPTARG, diff_opt_stat),
5332                 OPT_CALLBACK_F(0, "stat-width", options, N_("<width>"),
5333                                N_("generate diffstat with a given width"),
5334                                PARSE_OPT_NONEG, diff_opt_stat),
5335                 OPT_CALLBACK_F(0, "stat-name-width", options, N_("<width>"),
5336                                N_("generate diffstat with a given name width"),
5337                                PARSE_OPT_NONEG, diff_opt_stat),
5338                 OPT_CALLBACK_F(0, "stat-graph-width", options, N_("<width>"),
5339                                N_("generate diffstat with a given graph width"),
5340                                PARSE_OPT_NONEG, diff_opt_stat),
5341                 OPT_CALLBACK_F(0, "stat-count", options, N_("<count>"),
5342                                N_("generate diffstat with limited lines"),
5343                                PARSE_OPT_NONEG, diff_opt_stat),
5344                 OPT_CALLBACK_F(0, "compact-summary", options, NULL,
5345                                N_("generate compact summary in diffstat"),
5346                                PARSE_OPT_NOARG, diff_opt_compact_summary),
5347                 OPT_CALLBACK_F(0, "binary", options, NULL,
5348                                N_("output a binary diff that can be applied"),
5349                                PARSE_OPT_NONEG | PARSE_OPT_NOARG, diff_opt_binary),
5350                 OPT_BOOL(0, "full-index", &options->flags.full_index,
5351                          N_("show full pre- and post-image object names on the \"index\" lines")),
5352                 OPT_COLOR_FLAG(0, "color", &options->use_color,
5353                                N_("show colored diff")),
5354                 OPT_CALLBACK_F(0, "ws-error-highlight", options, N_("<kind>"),
5355                                N_("highlight whitespace errors in the 'context', 'old' or 'new' lines in the diff"),
5356                                PARSE_OPT_NONEG, diff_opt_ws_error_highlight),
5357                 OPT_SET_INT('z', NULL, &options->line_termination,
5358                             N_("do not munge pathnames and use NULs as output field terminators in --raw or --numstat"),
5359                             0),
5360                 OPT__ABBREV(&options->abbrev),
5361                 OPT_STRING_F(0, "src-prefix", &options->a_prefix, N_("<prefix>"),
5362                              N_("show the given source prefix instead of \"a/\""),
5363                              PARSE_OPT_NONEG),
5364                 OPT_STRING_F(0, "dst-prefix", &options->b_prefix, N_("<prefix>"),
5365                              N_("show the given destination prefix instead of \"b/\""),
5366                              PARSE_OPT_NONEG),
5367                 OPT_CALLBACK_F(0, "line-prefix", options, N_("<prefix>"),
5368                                N_("prepend an additional prefix to every line of output"),
5369                                PARSE_OPT_NONEG, diff_opt_line_prefix),
5370                 OPT_CALLBACK_F(0, "no-prefix", options, NULL,
5371                                N_("do not show any source or destination prefix"),
5372                                PARSE_OPT_NONEG | PARSE_OPT_NOARG, diff_opt_no_prefix),
5373                 OPT_INTEGER_F(0, "inter-hunk-context", &options->interhunkcontext,
5374                               N_("show context between diff hunks up to the specified number of lines"),
5375                               PARSE_OPT_NONEG),
5376                 OPT_CALLBACK_F(0, "output-indicator-new",
5377                                &options->output_indicators[OUTPUT_INDICATOR_NEW],
5378                                N_("<char>"),
5379                                N_("specify the character to indicate a new line instead of '+'"),
5380                                PARSE_OPT_NONEG, diff_opt_char),
5381                 OPT_CALLBACK_F(0, "output-indicator-old",
5382                                &options->output_indicators[OUTPUT_INDICATOR_OLD],
5383                                N_("<char>"),
5384                                N_("specify the character to indicate an old line instead of '-'"),
5385                                PARSE_OPT_NONEG, diff_opt_char),
5386                 OPT_CALLBACK_F(0, "output-indicator-context",
5387                                &options->output_indicators[OUTPUT_INDICATOR_CONTEXT],
5388                                N_("<char>"),
5389                                N_("specify the character to indicate a context instead of ' '"),
5390                                PARSE_OPT_NONEG, diff_opt_char),
5391
5392                 OPT_GROUP(N_("Diff rename options")),
5393                 OPT_CALLBACK_F('B', "break-rewrites", &options->break_opt, N_("<n>[/<m>]"),
5394                                N_("break complete rewrite changes into pairs of delete and create"),
5395                                PARSE_OPT_NONEG | PARSE_OPT_OPTARG,
5396                                diff_opt_break_rewrites),
5397                 OPT_CALLBACK_F('M', "find-renames", options, N_("<n>"),
5398                                N_("detect renames"),
5399                                PARSE_OPT_NONEG | PARSE_OPT_OPTARG,
5400                                diff_opt_find_renames),
5401                 OPT_SET_INT_F('D', "irreversible-delete", &options->irreversible_delete,
5402                               N_("omit the preimage for deletes"),
5403                               1, PARSE_OPT_NONEG),
5404                 OPT_CALLBACK_F('C', "find-copies", options, N_("<n>"),
5405                                N_("detect copies"),
5406                                PARSE_OPT_NONEG | PARSE_OPT_OPTARG,
5407                                diff_opt_find_copies),
5408                 OPT_BOOL(0, "find-copies-harder", &options->flags.find_copies_harder,
5409                          N_("use unmodified files as source to find copies")),
5410                 OPT_SET_INT_F(0, "no-renames", &options->detect_rename,
5411                               N_("disable rename detection"),
5412                               0, PARSE_OPT_NONEG),
5413                 OPT_BOOL(0, "rename-empty", &options->flags.rename_empty,
5414                          N_("use empty blobs as rename source")),
5415                 OPT_CALLBACK_F(0, "follow", options, NULL,
5416                                N_("continue listing the history of a file beyond renames"),
5417                                PARSE_OPT_NOARG, diff_opt_follow),
5418                 OPT_INTEGER('l', NULL, &options->rename_limit,
5419                             N_("prevent rename/copy detection if the number of rename/copy targets exceeds given limit")),
5420
5421                 OPT_GROUP(N_("Diff algorithm options")),
5422                 OPT_BIT(0, "minimal", &options->xdl_opts,
5423                         N_("produce the smallest possible diff"),
5424                         XDF_NEED_MINIMAL),
5425                 OPT_BIT_F('w', "ignore-all-space", &options->xdl_opts,
5426                           N_("ignore whitespace when comparing lines"),
5427                           XDF_IGNORE_WHITESPACE, PARSE_OPT_NONEG),
5428                 OPT_BIT_F('b', "ignore-space-change", &options->xdl_opts,
5429                           N_("ignore changes in amount of whitespace"),
5430                           XDF_IGNORE_WHITESPACE_CHANGE, PARSE_OPT_NONEG),
5431                 OPT_BIT_F(0, "ignore-space-at-eol", &options->xdl_opts,
5432                           N_("ignore changes in whitespace at EOL"),
5433                           XDF_IGNORE_WHITESPACE_AT_EOL, PARSE_OPT_NONEG),
5434                 OPT_BIT_F(0, "ignore-cr-at-eol", &options->xdl_opts,
5435                           N_("ignore carrier-return at the end of line"),
5436                           XDF_IGNORE_CR_AT_EOL, PARSE_OPT_NONEG),
5437                 OPT_BIT_F(0, "ignore-blank-lines", &options->xdl_opts,
5438                           N_("ignore changes whose lines are all blank"),
5439                           XDF_IGNORE_BLANK_LINES, PARSE_OPT_NONEG),
5440                 OPT_BIT(0, "indent-heuristic", &options->xdl_opts,
5441                         N_("heuristic to shift diff hunk boundaries for easy reading"),
5442                         XDF_INDENT_HEURISTIC),
5443                 OPT_CALLBACK_F(0, "patience", options, NULL,
5444                                N_("generate diff using the \"patience diff\" algorithm"),
5445                                PARSE_OPT_NONEG | PARSE_OPT_NOARG,
5446                                diff_opt_patience),
5447                 OPT_BITOP(0, "histogram", &options->xdl_opts,
5448                           N_("generate diff using the \"histogram diff\" algorithm"),
5449                           XDF_HISTOGRAM_DIFF, XDF_DIFF_ALGORITHM_MASK),
5450                 OPT_CALLBACK_F(0, "diff-algorithm", options, N_("<algorithm>"),
5451                                N_("choose a diff algorithm"),
5452                                PARSE_OPT_NONEG, diff_opt_diff_algorithm),
5453                 OPT_CALLBACK_F(0, "anchored", options, N_("<text>"),
5454                                N_("generate diff using the \"anchored diff\" algorithm"),
5455                                PARSE_OPT_NONEG, diff_opt_anchored),
5456                 OPT_CALLBACK_F(0, "word-diff", options, N_("<mode>"),
5457                                N_("show word diff, using <mode> to delimit changed words"),
5458                                PARSE_OPT_NONEG | PARSE_OPT_OPTARG, diff_opt_word_diff),
5459                 OPT_CALLBACK_F(0, "word-diff-regex", options, N_("<regex>"),
5460                                N_("use <regex> to decide what a word is"),
5461                                PARSE_OPT_NONEG, diff_opt_word_diff_regex),
5462                 OPT_CALLBACK_F(0, "color-words", options, N_("<regex>"),
5463                                N_("equivalent to --word-diff=color --word-diff-regex=<regex>"),
5464                                PARSE_OPT_NONEG | PARSE_OPT_OPTARG, diff_opt_color_words),
5465                 OPT_CALLBACK_F(0, "color-moved", options, N_("<mode>"),
5466                                N_("moved lines of code are colored differently"),
5467                                PARSE_OPT_OPTARG, diff_opt_color_moved),
5468                 OPT_CALLBACK_F(0, "color-moved-ws", options, N_("<mode>"),
5469                                N_("how white spaces are ignored in --color-moved"),
5470                                0, diff_opt_color_moved_ws),
5471
5472                 OPT_GROUP(N_("Other diff options")),
5473                 OPT_CALLBACK_F(0, "relative", options, N_("<prefix>"),
5474                                N_("when run from subdir, exclude changes outside and show relative paths"),
5475                                PARSE_OPT_NONEG | PARSE_OPT_OPTARG,
5476                                diff_opt_relative),
5477                 OPT_BOOL('a', "text", &options->flags.text,
5478                          N_("treat all files as text")),
5479                 OPT_BOOL('R', NULL, &options->flags.reverse_diff,
5480                          N_("swap two inputs, reverse the diff")),
5481                 OPT_BOOL(0, "exit-code", &options->flags.exit_with_status,
5482                          N_("exit with 1 if there were differences, 0 otherwise")),
5483                 OPT_BOOL(0, "quiet", &options->flags.quick,
5484                          N_("disable all output of the program")),
5485                 OPT_BOOL(0, "ext-diff", &options->flags.allow_external,
5486                          N_("allow an external diff helper to be executed")),
5487                 OPT_CALLBACK_F(0, "textconv", options, NULL,
5488                                N_("run external text conversion filters when comparing binary files"),
5489                                PARSE_OPT_NOARG, diff_opt_textconv),
5490                 OPT_CALLBACK_F(0, "ignore-submodules", options, N_("<when>"),
5491                                N_("ignore changes to submodules in the diff generation"),
5492                                PARSE_OPT_NONEG | PARSE_OPT_OPTARG,
5493                                diff_opt_ignore_submodules),
5494                 OPT_CALLBACK_F(0, "submodule", options, N_("<format>"),
5495                                N_("specify how differences in submodules are shown"),
5496                                PARSE_OPT_NONEG | PARSE_OPT_OPTARG,
5497                                diff_opt_submodule),
5498                 OPT_SET_INT_F(0, "ita-invisible-in-index", &options->ita_invisible_in_index,
5499                               N_("hide 'git add -N' entries from the index"),
5500                               1, PARSE_OPT_NONEG),
5501                 OPT_SET_INT_F(0, "ita-visible-in-index", &options->ita_invisible_in_index,
5502                               N_("treat 'git add -N' entries as real in the index"),
5503                               0, PARSE_OPT_NONEG),
5504                 OPT_CALLBACK_F('S', NULL, options, N_("<string>"),
5505                                N_("look for differences that change the number of occurrences of the specified string"),
5506                                0, diff_opt_pickaxe_string),
5507                 OPT_CALLBACK_F('G', NULL, options, N_("<regex>"),
5508                                N_("look for differences that change the number of occurrences of the specified regex"),
5509                                0, diff_opt_pickaxe_regex),
5510                 OPT_BIT_F(0, "pickaxe-all", &options->pickaxe_opts,
5511                           N_("show all changes in the changeset with -S or -G"),
5512                           DIFF_PICKAXE_ALL, PARSE_OPT_NONEG),
5513                 OPT_BIT_F(0, "pickaxe-regex", &options->pickaxe_opts,
5514                           N_("treat <string> in -S as extended POSIX regular expression"),
5515                           DIFF_PICKAXE_REGEX, PARSE_OPT_NONEG),
5516                 OPT_FILENAME('O', NULL, &options->orderfile,
5517                              N_("control the order in which files appear in the output")),
5518                 OPT_CALLBACK_F(0, "find-object", options, N_("<object-id>"),
5519                                N_("look for differences that change the number of occurrences of the specified object"),
5520                                PARSE_OPT_NONEG, diff_opt_find_object),
5521                 OPT_CALLBACK_F(0, "diff-filter", options, N_("[(A|C|D|M|R|T|U|X|B)...[*]]"),
5522                                N_("select files by diff type"),
5523                                PARSE_OPT_NONEG, diff_opt_diff_filter),
5524                 { OPTION_CALLBACK, 0, "output", options, N_("<file>"),
5525                   N_("Output to a specific file"),
5526                   PARSE_OPT_NONEG, NULL, 0, diff_opt_output },
5527
5528                 OPT_END()
5529         };
5530
5531         ALLOC_ARRAY(options->parseopts, ARRAY_SIZE(parseopts));
5532         memcpy(options->parseopts, parseopts, sizeof(parseopts));
5533 }
5534
5535 int diff_opt_parse(struct diff_options *options,
5536                    const char **av, int ac, const char *prefix)
5537 {
5538         if (!prefix)
5539                 prefix = "";
5540
5541         ac = parse_options(ac, av, prefix, options->parseopts, NULL,
5542                            PARSE_OPT_KEEP_DASHDASH |
5543                            PARSE_OPT_KEEP_UNKNOWN |
5544                            PARSE_OPT_NO_INTERNAL_HELP |
5545                            PARSE_OPT_ONE_SHOT |
5546                            PARSE_OPT_STOP_AT_NON_OPTION);
5547
5548         return ac;
5549 }
5550
5551 int parse_rename_score(const char **cp_p)
5552 {
5553         unsigned long num, scale;
5554         int ch, dot;
5555         const char *cp = *cp_p;
5556
5557         num = 0;
5558         scale = 1;
5559         dot = 0;
5560         for (;;) {
5561                 ch = *cp;
5562                 if ( !dot && ch == '.' ) {
5563                         scale = 1;
5564                         dot = 1;
5565                 } else if ( ch == '%' ) {
5566                         scale = dot ? scale*100 : 100;
5567                         cp++;   /* % is always at the end */
5568                         break;
5569                 } else if ( ch >= '0' && ch <= '9' ) {
5570                         if ( scale < 100000 ) {
5571                                 scale *= 10;
5572                                 num = (num*10) + (ch-'0');
5573                         }
5574                 } else {
5575                         break;
5576                 }
5577                 cp++;
5578         }
5579         *cp_p = cp;
5580
5581         /* user says num divided by scale and we say internally that
5582          * is MAX_SCORE * num / scale.
5583          */
5584         return (int)((num >= scale) ? MAX_SCORE : (MAX_SCORE * num / scale));
5585 }
5586
5587 struct diff_queue_struct diff_queued_diff;
5588
5589 void diff_q(struct diff_queue_struct *queue, struct diff_filepair *dp)
5590 {
5591         ALLOC_GROW(queue->queue, queue->nr + 1, queue->alloc);
5592         queue->queue[queue->nr++] = dp;
5593 }
5594
5595 struct diff_filepair *diff_queue(struct diff_queue_struct *queue,
5596                                  struct diff_filespec *one,
5597                                  struct diff_filespec *two)
5598 {
5599         struct diff_filepair *dp = xcalloc(1, sizeof(*dp));
5600         dp->one = one;
5601         dp->two = two;
5602         if (queue)
5603                 diff_q(queue, dp);
5604         return dp;
5605 }
5606
5607 void diff_free_filepair(struct diff_filepair *p)
5608 {
5609         free_filespec(p->one);
5610         free_filespec(p->two);
5611         free(p);
5612 }
5613
5614 const char *diff_aligned_abbrev(const struct object_id *oid, int len)
5615 {
5616         int abblen;
5617         const char *abbrev;
5618
5619         /* Do we want all 40 hex characters? */
5620         if (len == the_hash_algo->hexsz)
5621                 return oid_to_hex(oid);
5622
5623         /* An abbreviated value is fine, possibly followed by an ellipsis. */
5624         abbrev = diff_abbrev_oid(oid, len);
5625
5626         if (!print_sha1_ellipsis())
5627                 return abbrev;
5628
5629         abblen = strlen(abbrev);
5630
5631         /*
5632          * In well-behaved cases, where the abbreviated result is the
5633          * same as the requested length, append three dots after the
5634          * abbreviation (hence the whole logic is limited to the case
5635          * where abblen < 37); when the actual abbreviated result is a
5636          * bit longer than the requested length, we reduce the number
5637          * of dots so that they match the well-behaved ones.  However,
5638          * if the actual abbreviation is longer than the requested
5639          * length by more than three, we give up on aligning, and add
5640          * three dots anyway, to indicate that the output is not the
5641          * full object name.  Yes, this may be suboptimal, but this
5642          * appears only in "diff --raw --abbrev" output and it is not
5643          * worth the effort to change it now.  Note that this would
5644          * likely to work fine when the automatic sizing of default
5645          * abbreviation length is used--we would be fed -1 in "len" in
5646          * that case, and will end up always appending three-dots, but
5647          * the automatic sizing is supposed to give abblen that ensures
5648          * uniqueness across all objects (statistically speaking).
5649          */
5650         if (abblen < the_hash_algo->hexsz - 3) {
5651                 static char hex[GIT_MAX_HEXSZ + 1];
5652                 if (len < abblen && abblen <= len + 2)
5653                         xsnprintf(hex, sizeof(hex), "%s%.*s", abbrev, len+3-abblen, "..");
5654                 else
5655                         xsnprintf(hex, sizeof(hex), "%s...", abbrev);
5656                 return hex;
5657         }
5658
5659         return oid_to_hex(oid);
5660 }
5661
5662 static void diff_flush_raw(struct diff_filepair *p, struct diff_options *opt)
5663 {
5664         int line_termination = opt->line_termination;
5665         int inter_name_termination = line_termination ? '\t' : '\0';
5666
5667         fprintf(opt->file, "%s", diff_line_prefix(opt));
5668         if (!(opt->output_format & DIFF_FORMAT_NAME_STATUS)) {
5669                 fprintf(opt->file, ":%06o %06o %s ", p->one->mode, p->two->mode,
5670                         diff_aligned_abbrev(&p->one->oid, opt->abbrev));
5671                 fprintf(opt->file, "%s ",
5672                         diff_aligned_abbrev(&p->two->oid, opt->abbrev));
5673         }
5674         if (p->score) {
5675                 fprintf(opt->file, "%c%03d%c", p->status, similarity_index(p),
5676                         inter_name_termination);
5677         } else {
5678                 fprintf(opt->file, "%c%c", p->status, inter_name_termination);
5679         }
5680
5681         if (p->status == DIFF_STATUS_COPIED ||
5682             p->status == DIFF_STATUS_RENAMED) {
5683                 const char *name_a, *name_b;
5684                 name_a = p->one->path;
5685                 name_b = p->two->path;
5686                 strip_prefix(opt->prefix_length, &name_a, &name_b);
5687                 write_name_quoted(name_a, opt->file, inter_name_termination);
5688                 write_name_quoted(name_b, opt->file, line_termination);
5689         } else {
5690                 const char *name_a, *name_b;
5691                 name_a = p->one->mode ? p->one->path : p->two->path;
5692                 name_b = NULL;
5693                 strip_prefix(opt->prefix_length, &name_a, &name_b);
5694                 write_name_quoted(name_a, opt->file, line_termination);
5695         }
5696 }
5697
5698 int diff_unmodified_pair(struct diff_filepair *p)
5699 {
5700         /* This function is written stricter than necessary to support
5701          * the currently implemented transformers, but the idea is to
5702          * let transformers to produce diff_filepairs any way they want,
5703          * and filter and clean them up here before producing the output.
5704          */
5705         struct diff_filespec *one = p->one, *two = p->two;
5706
5707         if (DIFF_PAIR_UNMERGED(p))
5708                 return 0; /* unmerged is interesting */
5709
5710         /* deletion, addition, mode or type change
5711          * and rename are all interesting.
5712          */
5713         if (DIFF_FILE_VALID(one) != DIFF_FILE_VALID(two) ||
5714             DIFF_PAIR_MODE_CHANGED(p) ||
5715             strcmp(one->path, two->path))
5716                 return 0;
5717
5718         /* both are valid and point at the same path.  that is, we are
5719          * dealing with a change.
5720          */
5721         if (one->oid_valid && two->oid_valid &&
5722             oideq(&one->oid, &two->oid) &&
5723             !one->dirty_submodule && !two->dirty_submodule)
5724                 return 1; /* no change */
5725         if (!one->oid_valid && !two->oid_valid)
5726                 return 1; /* both look at the same file on the filesystem. */
5727         return 0;
5728 }
5729
5730 static void diff_flush_patch(struct diff_filepair *p, struct diff_options *o)
5731 {
5732         if (diff_unmodified_pair(p))
5733                 return;
5734
5735         if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
5736             (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
5737                 return; /* no tree diffs in patch format */
5738
5739         run_diff(p, o);
5740 }
5741
5742 static void diff_flush_stat(struct diff_filepair *p, struct diff_options *o,
5743                             struct diffstat_t *diffstat)
5744 {
5745         if (diff_unmodified_pair(p))
5746                 return;
5747
5748         if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
5749             (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
5750                 return; /* no useful stat for tree diffs */
5751
5752         run_diffstat(p, o, diffstat);
5753 }
5754
5755 static void diff_flush_checkdiff(struct diff_filepair *p,
5756                 struct diff_options *o)
5757 {
5758         if (diff_unmodified_pair(p))
5759                 return;
5760
5761         if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
5762             (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
5763                 return; /* nothing to check in tree diffs */
5764
5765         run_checkdiff(p, o);
5766 }
5767
5768 int diff_queue_is_empty(void)
5769 {
5770         struct diff_queue_struct *q = &diff_queued_diff;
5771         int i;
5772         for (i = 0; i < q->nr; i++)
5773                 if (!diff_unmodified_pair(q->queue[i]))
5774                         return 0;
5775         return 1;
5776 }
5777
5778 #if DIFF_DEBUG
5779 void diff_debug_filespec(struct diff_filespec *s, int x, const char *one)
5780 {
5781         fprintf(stderr, "queue[%d] %s (%s) %s %06o %s\n",
5782                 x, one ? one : "",
5783                 s->path,
5784                 DIFF_FILE_VALID(s) ? "valid" : "invalid",
5785                 s->mode,
5786                 s->oid_valid ? oid_to_hex(&s->oid) : "");
5787         fprintf(stderr, "queue[%d] %s size %lu\n",
5788                 x, one ? one : "",
5789                 s->size);
5790 }
5791
5792 void diff_debug_filepair(const struct diff_filepair *p, int i)
5793 {
5794         diff_debug_filespec(p->one, i, "one");
5795         diff_debug_filespec(p->two, i, "two");
5796         fprintf(stderr, "score %d, status %c rename_used %d broken %d\n",
5797                 p->score, p->status ? p->status : '?',
5798                 p->one->rename_used, p->broken_pair);
5799 }
5800
5801 void diff_debug_queue(const char *msg, struct diff_queue_struct *q)
5802 {
5803         int i;
5804         if (msg)
5805                 fprintf(stderr, "%s\n", msg);
5806         fprintf(stderr, "q->nr = %d\n", q->nr);
5807         for (i = 0; i < q->nr; i++) {
5808                 struct diff_filepair *p = q->queue[i];
5809                 diff_debug_filepair(p, i);
5810         }
5811 }
5812 #endif
5813
5814 static void diff_resolve_rename_copy(void)
5815 {
5816         int i;
5817         struct diff_filepair *p;
5818         struct diff_queue_struct *q = &diff_queued_diff;
5819
5820         diff_debug_queue("resolve-rename-copy", q);
5821
5822         for (i = 0; i < q->nr; i++) {
5823                 p = q->queue[i];
5824                 p->status = 0; /* undecided */
5825                 if (DIFF_PAIR_UNMERGED(p))
5826                         p->status = DIFF_STATUS_UNMERGED;
5827                 else if (!DIFF_FILE_VALID(p->one))
5828                         p->status = DIFF_STATUS_ADDED;
5829                 else if (!DIFF_FILE_VALID(p->two))
5830                         p->status = DIFF_STATUS_DELETED;
5831                 else if (DIFF_PAIR_TYPE_CHANGED(p))
5832                         p->status = DIFF_STATUS_TYPE_CHANGED;
5833
5834                 /* from this point on, we are dealing with a pair
5835                  * whose both sides are valid and of the same type, i.e.
5836                  * either in-place edit or rename/copy edit.
5837                  */
5838                 else if (DIFF_PAIR_RENAME(p)) {
5839                         /*
5840                          * A rename might have re-connected a broken
5841                          * pair up, causing the pathnames to be the
5842                          * same again. If so, that's not a rename at
5843                          * all, just a modification..
5844                          *
5845                          * Otherwise, see if this source was used for
5846                          * multiple renames, in which case we decrement
5847                          * the count, and call it a copy.
5848                          */
5849                         if (!strcmp(p->one->path, p->two->path))
5850                                 p->status = DIFF_STATUS_MODIFIED;
5851                         else if (--p->one->rename_used > 0)
5852                                 p->status = DIFF_STATUS_COPIED;
5853                         else
5854                                 p->status = DIFF_STATUS_RENAMED;
5855                 }
5856                 else if (!oideq(&p->one->oid, &p->two->oid) ||
5857                          p->one->mode != p->two->mode ||
5858                          p->one->dirty_submodule ||
5859                          p->two->dirty_submodule ||
5860                          is_null_oid(&p->one->oid))
5861                         p->status = DIFF_STATUS_MODIFIED;
5862                 else {
5863                         /* This is a "no-change" entry and should not
5864                          * happen anymore, but prepare for broken callers.
5865                          */
5866                         error("feeding unmodified %s to diffcore",
5867                               p->one->path);
5868                         p->status = DIFF_STATUS_UNKNOWN;
5869                 }
5870         }
5871         diff_debug_queue("resolve-rename-copy done", q);
5872 }
5873
5874 static int check_pair_status(struct diff_filepair *p)
5875 {
5876         switch (p->status) {
5877         case DIFF_STATUS_UNKNOWN:
5878                 return 0;
5879         case 0:
5880                 die("internal error in diff-resolve-rename-copy");
5881         default:
5882                 return 1;
5883         }
5884 }
5885
5886 static void flush_one_pair(struct diff_filepair *p, struct diff_options *opt)
5887 {
5888         int fmt = opt->output_format;
5889
5890         if (fmt & DIFF_FORMAT_CHECKDIFF)
5891                 diff_flush_checkdiff(p, opt);
5892         else if (fmt & (DIFF_FORMAT_RAW | DIFF_FORMAT_NAME_STATUS))
5893                 diff_flush_raw(p, opt);
5894         else if (fmt & DIFF_FORMAT_NAME) {
5895                 const char *name_a, *name_b;
5896                 name_a = p->two->path;
5897                 name_b = NULL;
5898                 strip_prefix(opt->prefix_length, &name_a, &name_b);
5899                 fprintf(opt->file, "%s", diff_line_prefix(opt));
5900                 write_name_quoted(name_a, opt->file, opt->line_termination);
5901         }
5902 }
5903
5904 static void show_file_mode_name(struct diff_options *opt, const char *newdelete, struct diff_filespec *fs)
5905 {
5906         struct strbuf sb = STRBUF_INIT;
5907         if (fs->mode)
5908                 strbuf_addf(&sb, " %s mode %06o ", newdelete, fs->mode);
5909         else
5910                 strbuf_addf(&sb, " %s ", newdelete);
5911
5912         quote_c_style(fs->path, &sb, NULL, 0);
5913         strbuf_addch(&sb, '\n');
5914         emit_diff_symbol(opt, DIFF_SYMBOL_SUMMARY,
5915                          sb.buf, sb.len, 0);
5916         strbuf_release(&sb);
5917 }
5918
5919 static void show_mode_change(struct diff_options *opt, struct diff_filepair *p,
5920                 int show_name)
5921 {
5922         if (p->one->mode && p->two->mode && p->one->mode != p->two->mode) {
5923                 struct strbuf sb = STRBUF_INIT;
5924                 strbuf_addf(&sb, " mode change %06o => %06o",
5925                             p->one->mode, p->two->mode);
5926                 if (show_name) {
5927                         strbuf_addch(&sb, ' ');
5928                         quote_c_style(p->two->path, &sb, NULL, 0);
5929                 }
5930                 strbuf_addch(&sb, '\n');
5931                 emit_diff_symbol(opt, DIFF_SYMBOL_SUMMARY,
5932                                  sb.buf, sb.len, 0);
5933                 strbuf_release(&sb);
5934         }
5935 }
5936
5937 static void show_rename_copy(struct diff_options *opt, const char *renamecopy,
5938                 struct diff_filepair *p)
5939 {
5940         struct strbuf sb = STRBUF_INIT;
5941         struct strbuf names = STRBUF_INIT;
5942
5943         pprint_rename(&names, p->one->path, p->two->path);
5944         strbuf_addf(&sb, " %s %s (%d%%)\n",
5945                     renamecopy, names.buf, similarity_index(p));
5946         strbuf_release(&names);
5947         emit_diff_symbol(opt, DIFF_SYMBOL_SUMMARY,
5948                                  sb.buf, sb.len, 0);
5949         show_mode_change(opt, p, 0);
5950         strbuf_release(&sb);
5951 }
5952
5953 static void diff_summary(struct diff_options *opt, struct diff_filepair *p)
5954 {
5955         switch(p->status) {
5956         case DIFF_STATUS_DELETED:
5957                 show_file_mode_name(opt, "delete", p->one);
5958                 break;
5959         case DIFF_STATUS_ADDED:
5960                 show_file_mode_name(opt, "create", p->two);
5961                 break;
5962         case DIFF_STATUS_COPIED:
5963                 show_rename_copy(opt, "copy", p);
5964                 break;
5965         case DIFF_STATUS_RENAMED:
5966                 show_rename_copy(opt, "rename", p);
5967                 break;
5968         default:
5969                 if (p->score) {
5970                         struct strbuf sb = STRBUF_INIT;
5971                         strbuf_addstr(&sb, " rewrite ");
5972                         quote_c_style(p->two->path, &sb, NULL, 0);
5973                         strbuf_addf(&sb, " (%d%%)\n", similarity_index(p));
5974                         emit_diff_symbol(opt, DIFF_SYMBOL_SUMMARY,
5975                                          sb.buf, sb.len, 0);
5976                         strbuf_release(&sb);
5977                 }
5978                 show_mode_change(opt, p, !p->score);
5979                 break;
5980         }
5981 }
5982
5983 struct patch_id_t {
5984         git_hash_ctx *ctx;
5985         int patchlen;
5986 };
5987
5988 static int remove_space(char *line, int len)
5989 {
5990         int i;
5991         char *dst = line;
5992         unsigned char c;
5993
5994         for (i = 0; i < len; i++)
5995                 if (!isspace((c = line[i])))
5996                         *dst++ = c;
5997
5998         return dst - line;
5999 }
6000
6001 void flush_one_hunk(struct object_id *result, git_hash_ctx *ctx)
6002 {
6003         unsigned char hash[GIT_MAX_RAWSZ];
6004         unsigned short carry = 0;
6005         int i;
6006
6007         the_hash_algo->final_fn(hash, ctx);
6008         the_hash_algo->init_fn(ctx);
6009         /* 20-byte sum, with carry */
6010         for (i = 0; i < the_hash_algo->rawsz; ++i) {
6011                 carry += result->hash[i] + hash[i];
6012                 result->hash[i] = carry;
6013                 carry >>= 8;
6014         }
6015 }
6016
6017 static void patch_id_consume(void *priv, char *line, unsigned long len)
6018 {
6019         struct patch_id_t *data = priv;
6020         int new_len;
6021
6022         new_len = remove_space(line, len);
6023
6024         the_hash_algo->update_fn(data->ctx, line, new_len);
6025         data->patchlen += new_len;
6026 }
6027
6028 static void patch_id_add_string(git_hash_ctx *ctx, const char *str)
6029 {
6030         the_hash_algo->update_fn(ctx, str, strlen(str));
6031 }
6032
6033 static void patch_id_add_mode(git_hash_ctx *ctx, unsigned mode)
6034 {
6035         /* large enough for 2^32 in octal */
6036         char buf[12];
6037         int len = xsnprintf(buf, sizeof(buf), "%06o", mode);
6038         the_hash_algo->update_fn(ctx, buf, len);
6039 }
6040
6041 /* returns 0 upon success, and writes result into oid */
6042 static int diff_get_patch_id(struct diff_options *options, struct object_id *oid, int diff_header_only, int stable)
6043 {
6044         struct diff_queue_struct *q = &diff_queued_diff;
6045         int i;
6046         git_hash_ctx ctx;
6047         struct patch_id_t data;
6048
6049         the_hash_algo->init_fn(&ctx);
6050         memset(&data, 0, sizeof(struct patch_id_t));
6051         data.ctx = &ctx;
6052         oidclr(oid);
6053
6054         for (i = 0; i < q->nr; i++) {
6055                 xpparam_t xpp;
6056                 xdemitconf_t xecfg;
6057                 mmfile_t mf1, mf2;
6058                 struct diff_filepair *p = q->queue[i];
6059                 int len1, len2;
6060
6061                 memset(&xpp, 0, sizeof(xpp));
6062                 memset(&xecfg, 0, sizeof(xecfg));
6063                 if (p->status == 0)
6064                         return error("internal diff status error");
6065                 if (p->status == DIFF_STATUS_UNKNOWN)
6066                         continue;
6067                 if (diff_unmodified_pair(p))
6068                         continue;
6069                 if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
6070                     (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
6071                         continue;
6072                 if (DIFF_PAIR_UNMERGED(p))
6073                         continue;
6074
6075                 diff_fill_oid_info(p->one, options->repo->index);
6076                 diff_fill_oid_info(p->two, options->repo->index);
6077
6078                 len1 = remove_space(p->one->path, strlen(p->one->path));
6079                 len2 = remove_space(p->two->path, strlen(p->two->path));
6080                 patch_id_add_string(&ctx, "diff--git");
6081                 patch_id_add_string(&ctx, "a/");
6082                 the_hash_algo->update_fn(&ctx, p->one->path, len1);
6083                 patch_id_add_string(&ctx, "b/");
6084                 the_hash_algo->update_fn(&ctx, p->two->path, len2);
6085
6086                 if (p->one->mode == 0) {
6087                         patch_id_add_string(&ctx, "newfilemode");
6088                         patch_id_add_mode(&ctx, p->two->mode);
6089                         patch_id_add_string(&ctx, "---/dev/null");
6090                         patch_id_add_string(&ctx, "+++b/");
6091                         the_hash_algo->update_fn(&ctx, p->two->path, len2);
6092                 } else if (p->two->mode == 0) {
6093                         patch_id_add_string(&ctx, "deletedfilemode");
6094                         patch_id_add_mode(&ctx, p->one->mode);
6095                         patch_id_add_string(&ctx, "---a/");
6096                         the_hash_algo->update_fn(&ctx, p->one->path, len1);
6097                         patch_id_add_string(&ctx, "+++/dev/null");
6098                 } else {
6099                         patch_id_add_string(&ctx, "---a/");
6100                         the_hash_algo->update_fn(&ctx, p->one->path, len1);
6101                         patch_id_add_string(&ctx, "+++b/");
6102                         the_hash_algo->update_fn(&ctx, p->two->path, len2);
6103                 }
6104
6105                 if (diff_header_only)
6106                         continue;
6107
6108                 if (fill_mmfile(options->repo, &mf1, p->one) < 0 ||
6109                     fill_mmfile(options->repo, &mf2, p->two) < 0)
6110                         return error("unable to read files to diff");
6111
6112                 if (diff_filespec_is_binary(options->repo, p->one) ||
6113                     diff_filespec_is_binary(options->repo, p->two)) {
6114                         the_hash_algo->update_fn(&ctx, oid_to_hex(&p->one->oid),
6115                                         the_hash_algo->hexsz);
6116                         the_hash_algo->update_fn(&ctx, oid_to_hex(&p->two->oid),
6117                                         the_hash_algo->hexsz);
6118                         continue;
6119                 }
6120
6121                 xpp.flags = 0;
6122                 xecfg.ctxlen = 3;
6123                 xecfg.flags = 0;
6124                 if (xdi_diff_outf(&mf1, &mf2, discard_hunk_line,
6125                                   patch_id_consume, &data, &xpp, &xecfg))
6126                         return error("unable to generate patch-id diff for %s",
6127                                      p->one->path);
6128
6129                 if (stable)
6130                         flush_one_hunk(oid, &ctx);
6131         }
6132
6133         if (!stable)
6134                 the_hash_algo->final_fn(oid->hash, &ctx);
6135
6136         return 0;
6137 }
6138
6139 int diff_flush_patch_id(struct diff_options *options, struct object_id *oid, int diff_header_only, int stable)
6140 {
6141         struct diff_queue_struct *q = &diff_queued_diff;
6142         int i;
6143         int result = diff_get_patch_id(options, oid, diff_header_only, stable);
6144
6145         for (i = 0; i < q->nr; i++)
6146                 diff_free_filepair(q->queue[i]);
6147
6148         free(q->queue);
6149         DIFF_QUEUE_CLEAR(q);
6150
6151         return result;
6152 }
6153
6154 static int is_summary_empty(const struct diff_queue_struct *q)
6155 {
6156         int i;
6157
6158         for (i = 0; i < q->nr; i++) {
6159                 const struct diff_filepair *p = q->queue[i];
6160
6161                 switch (p->status) {
6162                 case DIFF_STATUS_DELETED:
6163                 case DIFF_STATUS_ADDED:
6164                 case DIFF_STATUS_COPIED:
6165                 case DIFF_STATUS_RENAMED:
6166                         return 0;
6167                 default:
6168                         if (p->score)
6169                                 return 0;
6170                         if (p->one->mode && p->two->mode &&
6171                             p->one->mode != p->two->mode)
6172                                 return 0;
6173                         break;
6174                 }
6175         }
6176         return 1;
6177 }
6178
6179 static const char rename_limit_warning[] =
6180 N_("inexact rename detection was skipped due to too many files.");
6181
6182 static const char degrade_cc_to_c_warning[] =
6183 N_("only found copies from modified paths due to too many files.");
6184
6185 static const char rename_limit_advice[] =
6186 N_("you may want to set your %s variable to at least "
6187    "%d and retry the command.");
6188
6189 void diff_warn_rename_limit(const char *varname, int needed, int degraded_cc)
6190 {
6191         fflush(stdout);
6192         if (degraded_cc)
6193                 warning(_(degrade_cc_to_c_warning));
6194         else if (needed)
6195                 warning(_(rename_limit_warning));
6196         else
6197                 return;
6198         if (0 < needed)
6199                 warning(_(rename_limit_advice), varname, needed);
6200 }
6201
6202 static void diff_flush_patch_all_file_pairs(struct diff_options *o)
6203 {
6204         int i;
6205         static struct emitted_diff_symbols esm = EMITTED_DIFF_SYMBOLS_INIT;
6206         struct diff_queue_struct *q = &diff_queued_diff;
6207
6208         if (WSEH_NEW & WS_RULE_MASK)
6209                 BUG("WS rules bit mask overlaps with diff symbol flags");
6210
6211         if (o->color_moved)
6212                 o->emitted_symbols = &esm;
6213
6214         for (i = 0; i < q->nr; i++) {
6215                 struct diff_filepair *p = q->queue[i];
6216                 if (check_pair_status(p))
6217                         diff_flush_patch(p, o);
6218         }
6219
6220         if (o->emitted_symbols) {
6221                 if (o->color_moved) {
6222                         struct hashmap add_lines, del_lines;
6223
6224                         if (o->color_moved_ws_handling &
6225                             COLOR_MOVED_WS_ALLOW_INDENTATION_CHANGE)
6226                                 o->color_moved_ws_handling |= XDF_IGNORE_WHITESPACE;
6227
6228                         hashmap_init(&del_lines, moved_entry_cmp, o, 0);
6229                         hashmap_init(&add_lines, moved_entry_cmp, o, 0);
6230
6231                         add_lines_to_move_detection(o, &add_lines, &del_lines);
6232                         mark_color_as_moved(o, &add_lines, &del_lines);
6233                         if (o->color_moved == COLOR_MOVED_ZEBRA_DIM)
6234                                 dim_moved_lines(o);
6235
6236                         hashmap_free_entries(&add_lines, struct moved_entry,
6237                                                 ent);
6238                         hashmap_free_entries(&del_lines, struct moved_entry,
6239                                                 ent);
6240                 }
6241
6242                 for (i = 0; i < esm.nr; i++)
6243                         emit_diff_symbol_from_struct(o, &esm.buf[i]);
6244
6245                 for (i = 0; i < esm.nr; i++)
6246                         free((void *)esm.buf[i].line);
6247                 esm.nr = 0;
6248
6249                 o->emitted_symbols = NULL;
6250         }
6251 }
6252
6253 void diff_flush(struct diff_options *options)
6254 {
6255         struct diff_queue_struct *q = &diff_queued_diff;
6256         int i, output_format = options->output_format;
6257         int separator = 0;
6258         int dirstat_by_line = 0;
6259
6260         /*
6261          * Order: raw, stat, summary, patch
6262          * or:    name/name-status/checkdiff (other bits clear)
6263          */
6264         if (!q->nr)
6265                 goto free_queue;
6266
6267         if (output_format & (DIFF_FORMAT_RAW |
6268                              DIFF_FORMAT_NAME |
6269                              DIFF_FORMAT_NAME_STATUS |
6270                              DIFF_FORMAT_CHECKDIFF)) {
6271                 for (i = 0; i < q->nr; i++) {
6272                         struct diff_filepair *p = q->queue[i];
6273                         if (check_pair_status(p))
6274                                 flush_one_pair(p, options);
6275                 }
6276                 separator++;
6277         }
6278
6279         if (output_format & DIFF_FORMAT_DIRSTAT && options->flags.dirstat_by_line)
6280                 dirstat_by_line = 1;
6281
6282         if (output_format & (DIFF_FORMAT_DIFFSTAT|DIFF_FORMAT_SHORTSTAT|DIFF_FORMAT_NUMSTAT) ||
6283             dirstat_by_line) {
6284                 struct diffstat_t diffstat;
6285
6286                 memset(&diffstat, 0, sizeof(struct diffstat_t));
6287                 for (i = 0; i < q->nr; i++) {
6288                         struct diff_filepair *p = q->queue[i];
6289                         if (check_pair_status(p))
6290                                 diff_flush_stat(p, options, &diffstat);
6291                 }
6292                 if (output_format & DIFF_FORMAT_NUMSTAT)
6293                         show_numstat(&diffstat, options);
6294                 if (output_format & DIFF_FORMAT_DIFFSTAT)
6295                         show_stats(&diffstat, options);
6296                 if (output_format & DIFF_FORMAT_SHORTSTAT)
6297                         show_shortstats(&diffstat, options);
6298                 if (output_format & DIFF_FORMAT_DIRSTAT && dirstat_by_line)
6299                         show_dirstat_by_line(&diffstat, options);
6300                 free_diffstat_info(&diffstat);
6301                 separator++;
6302         }
6303         if ((output_format & DIFF_FORMAT_DIRSTAT) && !dirstat_by_line)
6304                 show_dirstat(options);
6305
6306         if (output_format & DIFF_FORMAT_SUMMARY && !is_summary_empty(q)) {
6307                 for (i = 0; i < q->nr; i++) {
6308                         diff_summary(options, q->queue[i]);
6309                 }
6310                 separator++;
6311         }
6312
6313         if (output_format & DIFF_FORMAT_NO_OUTPUT &&
6314             options->flags.exit_with_status &&
6315             options->flags.diff_from_contents) {
6316                 /*
6317                  * run diff_flush_patch for the exit status. setting
6318                  * options->file to /dev/null should be safe, because we
6319                  * aren't supposed to produce any output anyway.
6320                  */
6321                 if (options->close_file)
6322                         fclose(options->file);
6323                 options->file = xfopen("/dev/null", "w");
6324                 options->close_file = 1;
6325                 options->color_moved = 0;
6326                 for (i = 0; i < q->nr; i++) {
6327                         struct diff_filepair *p = q->queue[i];
6328                         if (check_pair_status(p))
6329                                 diff_flush_patch(p, options);
6330                         if (options->found_changes)
6331                                 break;
6332                 }
6333         }
6334
6335         if (output_format & DIFF_FORMAT_PATCH) {
6336                 if (separator) {
6337                         emit_diff_symbol(options, DIFF_SYMBOL_SEPARATOR, NULL, 0, 0);
6338                         if (options->stat_sep)
6339                                 /* attach patch instead of inline */
6340                                 emit_diff_symbol(options, DIFF_SYMBOL_STAT_SEP,
6341                                                  NULL, 0, 0);
6342                 }
6343
6344                 diff_flush_patch_all_file_pairs(options);
6345         }
6346
6347         if (output_format & DIFF_FORMAT_CALLBACK)
6348                 options->format_callback(q, options, options->format_callback_data);
6349
6350         for (i = 0; i < q->nr; i++)
6351                 diff_free_filepair(q->queue[i]);
6352 free_queue:
6353         free(q->queue);
6354         DIFF_QUEUE_CLEAR(q);
6355         if (options->close_file)
6356                 fclose(options->file);
6357
6358         /*
6359          * Report the content-level differences with HAS_CHANGES;
6360          * diff_addremove/diff_change does not set the bit when
6361          * DIFF_FROM_CONTENTS is in effect (e.g. with -w).
6362          */
6363         if (options->flags.diff_from_contents) {
6364                 if (options->found_changes)
6365                         options->flags.has_changes = 1;
6366                 else
6367                         options->flags.has_changes = 0;
6368         }
6369 }
6370
6371 static int match_filter(const struct diff_options *options, const struct diff_filepair *p)
6372 {
6373         return (((p->status == DIFF_STATUS_MODIFIED) &&
6374                  ((p->score &&
6375                    filter_bit_tst(DIFF_STATUS_FILTER_BROKEN, options)) ||
6376                   (!p->score &&
6377                    filter_bit_tst(DIFF_STATUS_MODIFIED, options)))) ||
6378                 ((p->status != DIFF_STATUS_MODIFIED) &&
6379                  filter_bit_tst(p->status, options)));
6380 }
6381
6382 static void diffcore_apply_filter(struct diff_options *options)
6383 {
6384         int i;
6385         struct diff_queue_struct *q = &diff_queued_diff;
6386         struct diff_queue_struct outq;
6387
6388         DIFF_QUEUE_CLEAR(&outq);
6389
6390         if (!options->filter)
6391                 return;
6392
6393         if (filter_bit_tst(DIFF_STATUS_FILTER_AON, options)) {
6394                 int found;
6395                 for (i = found = 0; !found && i < q->nr; i++) {
6396                         if (match_filter(options, q->queue[i]))
6397                                 found++;
6398                 }
6399                 if (found)
6400                         return;
6401
6402                 /* otherwise we will clear the whole queue
6403                  * by copying the empty outq at the end of this
6404                  * function, but first clear the current entries
6405                  * in the queue.
6406                  */
6407                 for (i = 0; i < q->nr; i++)
6408                         diff_free_filepair(q->queue[i]);
6409         }
6410         else {
6411                 /* Only the matching ones */
6412                 for (i = 0; i < q->nr; i++) {
6413                         struct diff_filepair *p = q->queue[i];
6414                         if (match_filter(options, p))
6415                                 diff_q(&outq, p);
6416                         else
6417                                 diff_free_filepair(p);
6418                 }
6419         }
6420         free(q->queue);
6421         *q = outq;
6422 }
6423
6424 /* Check whether two filespecs with the same mode and size are identical */
6425 static int diff_filespec_is_identical(struct repository *r,
6426                                       struct diff_filespec *one,
6427                                       struct diff_filespec *two)
6428 {
6429         if (S_ISGITLINK(one->mode))
6430                 return 0;
6431         if (diff_populate_filespec(r, one, 0))
6432                 return 0;
6433         if (diff_populate_filespec(r, two, 0))
6434                 return 0;
6435         return !memcmp(one->data, two->data, one->size);
6436 }
6437
6438 static int diff_filespec_check_stat_unmatch(struct repository *r,
6439                                             struct diff_filepair *p)
6440 {
6441         if (p->done_skip_stat_unmatch)
6442                 return p->skip_stat_unmatch_result;
6443
6444         p->done_skip_stat_unmatch = 1;
6445         p->skip_stat_unmatch_result = 0;
6446         /*
6447          * 1. Entries that come from stat info dirtiness
6448          *    always have both sides (iow, not create/delete),
6449          *    one side of the object name is unknown, with
6450          *    the same mode and size.  Keep the ones that
6451          *    do not match these criteria.  They have real
6452          *    differences.
6453          *
6454          * 2. At this point, the file is known to be modified,
6455          *    with the same mode and size, and the object
6456          *    name of one side is unknown.  Need to inspect
6457          *    the identical contents.
6458          */
6459         if (!DIFF_FILE_VALID(p->one) || /* (1) */
6460             !DIFF_FILE_VALID(p->two) ||
6461             (p->one->oid_valid && p->two->oid_valid) ||
6462             (p->one->mode != p->two->mode) ||
6463             diff_populate_filespec(r, p->one, CHECK_SIZE_ONLY) ||
6464             diff_populate_filespec(r, p->two, CHECK_SIZE_ONLY) ||
6465             (p->one->size != p->two->size) ||
6466             !diff_filespec_is_identical(r, p->one, p->two)) /* (2) */
6467                 p->skip_stat_unmatch_result = 1;
6468         return p->skip_stat_unmatch_result;
6469 }
6470
6471 static void diffcore_skip_stat_unmatch(struct diff_options *diffopt)
6472 {
6473         int i;
6474         struct diff_queue_struct *q = &diff_queued_diff;
6475         struct diff_queue_struct outq;
6476         DIFF_QUEUE_CLEAR(&outq);
6477
6478         for (i = 0; i < q->nr; i++) {
6479                 struct diff_filepair *p = q->queue[i];
6480
6481                 if (diff_filespec_check_stat_unmatch(diffopt->repo, p))
6482                         diff_q(&outq, p);
6483                 else {
6484                         /*
6485                          * The caller can subtract 1 from skip_stat_unmatch
6486                          * to determine how many paths were dirty only
6487                          * due to stat info mismatch.
6488                          */
6489                         if (!diffopt->flags.no_index)
6490                                 diffopt->skip_stat_unmatch++;
6491                         diff_free_filepair(p);
6492                 }
6493         }
6494         free(q->queue);
6495         *q = outq;
6496 }
6497
6498 static int diffnamecmp(const void *a_, const void *b_)
6499 {
6500         const struct diff_filepair *a = *((const struct diff_filepair **)a_);
6501         const struct diff_filepair *b = *((const struct diff_filepair **)b_);
6502         const char *name_a, *name_b;
6503
6504         name_a = a->one ? a->one->path : a->two->path;
6505         name_b = b->one ? b->one->path : b->two->path;
6506         return strcmp(name_a, name_b);
6507 }
6508
6509 void diffcore_fix_diff_index(void)
6510 {
6511         struct diff_queue_struct *q = &diff_queued_diff;
6512         QSORT(q->queue, q->nr, diffnamecmp);
6513 }
6514
6515 static void add_if_missing(struct repository *r,
6516                            struct oid_array *to_fetch,
6517                            const struct diff_filespec *filespec)
6518 {
6519         if (filespec && filespec->oid_valid &&
6520             !S_ISGITLINK(filespec->mode) &&
6521             oid_object_info_extended(r, &filespec->oid, NULL,
6522                                      OBJECT_INFO_FOR_PREFETCH))
6523                 oid_array_append(to_fetch, &filespec->oid);
6524 }
6525
6526 void diffcore_std(struct diff_options *options)
6527 {
6528         if (options->repo == the_repository && has_promisor_remote()) {
6529                 /*
6530                  * Prefetch the diff pairs that are about to be flushed.
6531                  */
6532                 int i;
6533                 struct diff_queue_struct *q = &diff_queued_diff;
6534                 struct oid_array to_fetch = OID_ARRAY_INIT;
6535
6536                 for (i = 0; i < q->nr; i++) {
6537                         struct diff_filepair *p = q->queue[i];
6538                         add_if_missing(options->repo, &to_fetch, p->one);
6539                         add_if_missing(options->repo, &to_fetch, p->two);
6540                 }
6541                 if (to_fetch.nr)
6542                         /*
6543                          * NEEDSWORK: Consider deduplicating the OIDs sent.
6544                          */
6545                         promisor_remote_get_direct(options->repo,
6546                                                    to_fetch.oid, to_fetch.nr);
6547                 oid_array_clear(&to_fetch);
6548         }
6549
6550         /* NOTE please keep the following in sync with diff_tree_combined() */
6551         if (options->skip_stat_unmatch)
6552                 diffcore_skip_stat_unmatch(options);
6553         if (!options->found_follow) {
6554                 /* See try_to_follow_renames() in tree-diff.c */
6555                 if (options->break_opt != -1)
6556                         diffcore_break(options->repo,
6557                                        options->break_opt);
6558                 if (options->detect_rename)
6559                         diffcore_rename(options);
6560                 if (options->break_opt != -1)
6561                         diffcore_merge_broken();
6562         }
6563         if (options->pickaxe_opts & DIFF_PICKAXE_KINDS_MASK)
6564                 diffcore_pickaxe(options);
6565         if (options->orderfile)
6566                 diffcore_order(options->orderfile);
6567         if (!options->found_follow)
6568                 /* See try_to_follow_renames() in tree-diff.c */
6569                 diff_resolve_rename_copy();
6570         diffcore_apply_filter(options);
6571
6572         if (diff_queued_diff.nr && !options->flags.diff_from_contents)
6573                 options->flags.has_changes = 1;
6574         else
6575                 options->flags.has_changes = 0;
6576
6577         options->found_follow = 0;
6578 }
6579
6580 int diff_result_code(struct diff_options *opt, int status)
6581 {
6582         int result = 0;
6583
6584         diff_warn_rename_limit("diff.renameLimit",
6585                                opt->needed_rename_limit,
6586                                opt->degraded_cc_to_c);
6587         if (!opt->flags.exit_with_status &&
6588             !(opt->output_format & DIFF_FORMAT_CHECKDIFF))
6589                 return status;
6590         if (opt->flags.exit_with_status &&
6591             opt->flags.has_changes)
6592                 result |= 01;
6593         if ((opt->output_format & DIFF_FORMAT_CHECKDIFF) &&
6594             opt->flags.check_failed)
6595                 result |= 02;
6596         return result;
6597 }
6598
6599 int diff_can_quit_early(struct diff_options *opt)
6600 {
6601         return (opt->flags.quick &&
6602                 !opt->filter &&
6603                 opt->flags.has_changes);
6604 }
6605
6606 /*
6607  * Shall changes to this submodule be ignored?
6608  *
6609  * Submodule changes can be configured to be ignored separately for each path,
6610  * but that configuration can be overridden from the command line.
6611  */
6612 static int is_submodule_ignored(const char *path, struct diff_options *options)
6613 {
6614         int ignored = 0;
6615         struct diff_flags orig_flags = options->flags;
6616         if (!options->flags.override_submodule_config)
6617                 set_diffopt_flags_from_submodule_config(options, path);
6618         if (options->flags.ignore_submodules)
6619                 ignored = 1;
6620         options->flags = orig_flags;
6621         return ignored;
6622 }
6623
6624 void diff_addremove(struct diff_options *options,
6625                     int addremove, unsigned mode,
6626                     const struct object_id *oid,
6627                     int oid_valid,
6628                     const char *concatpath, unsigned dirty_submodule)
6629 {
6630         struct diff_filespec *one, *two;
6631
6632         if (S_ISGITLINK(mode) && is_submodule_ignored(concatpath, options))
6633                 return;
6634
6635         /* This may look odd, but it is a preparation for
6636          * feeding "there are unchanged files which should
6637          * not produce diffs, but when you are doing copy
6638          * detection you would need them, so here they are"
6639          * entries to the diff-core.  They will be prefixed
6640          * with something like '=' or '*' (I haven't decided
6641          * which but should not make any difference).
6642          * Feeding the same new and old to diff_change()
6643          * also has the same effect.
6644          * Before the final output happens, they are pruned after
6645          * merged into rename/copy pairs as appropriate.
6646          */
6647         if (options->flags.reverse_diff)
6648                 addremove = (addremove == '+' ? '-' :
6649                              addremove == '-' ? '+' : addremove);
6650
6651         if (options->prefix &&
6652             strncmp(concatpath, options->prefix, options->prefix_length))
6653                 return;
6654
6655         one = alloc_filespec(concatpath);
6656         two = alloc_filespec(concatpath);
6657
6658         if (addremove != '+')
6659                 fill_filespec(one, oid, oid_valid, mode);
6660         if (addremove != '-') {
6661                 fill_filespec(two, oid, oid_valid, mode);
6662                 two->dirty_submodule = dirty_submodule;
6663         }
6664
6665         diff_queue(&diff_queued_diff, one, two);
6666         if (!options->flags.diff_from_contents)
6667                 options->flags.has_changes = 1;
6668 }
6669
6670 void diff_change(struct diff_options *options,
6671                  unsigned old_mode, unsigned new_mode,
6672                  const struct object_id *old_oid,
6673                  const struct object_id *new_oid,
6674                  int old_oid_valid, int new_oid_valid,
6675                  const char *concatpath,
6676                  unsigned old_dirty_submodule, unsigned new_dirty_submodule)
6677 {
6678         struct diff_filespec *one, *two;
6679         struct diff_filepair *p;
6680
6681         if (S_ISGITLINK(old_mode) && S_ISGITLINK(new_mode) &&
6682             is_submodule_ignored(concatpath, options))
6683                 return;
6684
6685         if (options->flags.reverse_diff) {
6686                 SWAP(old_mode, new_mode);
6687                 SWAP(old_oid, new_oid);
6688                 SWAP(old_oid_valid, new_oid_valid);
6689                 SWAP(old_dirty_submodule, new_dirty_submodule);
6690         }
6691
6692         if (options->prefix &&
6693             strncmp(concatpath, options->prefix, options->prefix_length))
6694                 return;
6695
6696         one = alloc_filespec(concatpath);
6697         two = alloc_filespec(concatpath);
6698         fill_filespec(one, old_oid, old_oid_valid, old_mode);
6699         fill_filespec(two, new_oid, new_oid_valid, new_mode);
6700         one->dirty_submodule = old_dirty_submodule;
6701         two->dirty_submodule = new_dirty_submodule;
6702         p = diff_queue(&diff_queued_diff, one, two);
6703
6704         if (options->flags.diff_from_contents)
6705                 return;
6706
6707         if (options->flags.quick && options->skip_stat_unmatch &&
6708             !diff_filespec_check_stat_unmatch(options->repo, p))
6709                 return;
6710
6711         options->flags.has_changes = 1;
6712 }
6713
6714 struct diff_filepair *diff_unmerge(struct diff_options *options, const char *path)
6715 {
6716         struct diff_filepair *pair;
6717         struct diff_filespec *one, *two;
6718
6719         if (options->prefix &&
6720             strncmp(path, options->prefix, options->prefix_length))
6721                 return NULL;
6722
6723         one = alloc_filespec(path);
6724         two = alloc_filespec(path);
6725         pair = diff_queue(&diff_queued_diff, one, two);
6726         pair->is_unmerged = 1;
6727         return pair;
6728 }
6729
6730 static char *run_textconv(struct repository *r,
6731                           const char *pgm,
6732                           struct diff_filespec *spec,
6733                           size_t *outsize)
6734 {
6735         struct diff_tempfile *temp;
6736         const char *argv[3];
6737         const char **arg = argv;
6738         struct child_process child = CHILD_PROCESS_INIT;
6739         struct strbuf buf = STRBUF_INIT;
6740         int err = 0;
6741
6742         temp = prepare_temp_file(r, spec->path, spec);
6743         *arg++ = pgm;
6744         *arg++ = temp->name;
6745         *arg = NULL;
6746
6747         child.use_shell = 1;
6748         child.argv = argv;
6749         child.out = -1;
6750         if (start_command(&child)) {
6751                 remove_tempfile();
6752                 return NULL;
6753         }
6754
6755         if (strbuf_read(&buf, child.out, 0) < 0)
6756                 err = error("error reading from textconv command '%s'", pgm);
6757         close(child.out);
6758
6759         if (finish_command(&child) || err) {
6760                 strbuf_release(&buf);
6761                 remove_tempfile();
6762                 return NULL;
6763         }
6764         remove_tempfile();
6765
6766         return strbuf_detach(&buf, outsize);
6767 }
6768
6769 size_t fill_textconv(struct repository *r,
6770                      struct userdiff_driver *driver,
6771                      struct diff_filespec *df,
6772                      char **outbuf)
6773 {
6774         size_t size;
6775
6776         if (!driver) {
6777                 if (!DIFF_FILE_VALID(df)) {
6778                         *outbuf = "";
6779                         return 0;
6780                 }
6781                 if (diff_populate_filespec(r, df, 0))
6782                         die("unable to read files to diff");
6783                 *outbuf = df->data;
6784                 return df->size;
6785         }
6786
6787         if (!driver->textconv)
6788                 BUG("fill_textconv called with non-textconv driver");
6789
6790         if (driver->textconv_cache && df->oid_valid) {
6791                 *outbuf = notes_cache_get(driver->textconv_cache,
6792                                           &df->oid,
6793                                           &size);
6794                 if (*outbuf)
6795                         return size;
6796         }
6797
6798         *outbuf = run_textconv(r, driver->textconv, df, &size);
6799         if (!*outbuf)
6800                 die("unable to read files to diff");
6801
6802         if (driver->textconv_cache && df->oid_valid) {
6803                 /* ignore errors, as we might be in a readonly repository */
6804                 notes_cache_put(driver->textconv_cache, &df->oid, *outbuf,
6805                                 size);
6806                 /*
6807                  * we could save up changes and flush them all at the end,
6808                  * but we would need an extra call after all diffing is done.
6809                  * Since generating a cache entry is the slow path anyway,
6810                  * this extra overhead probably isn't a big deal.
6811                  */
6812                 notes_cache_write(driver->textconv_cache);
6813         }
6814
6815         return size;
6816 }
6817
6818 int textconv_object(struct repository *r,
6819                     const char *path,
6820                     unsigned mode,
6821                     const struct object_id *oid,
6822                     int oid_valid,
6823                     char **buf,
6824                     unsigned long *buf_size)
6825 {
6826         struct diff_filespec *df;
6827         struct userdiff_driver *textconv;
6828
6829         df = alloc_filespec(path);
6830         fill_filespec(df, oid, oid_valid, mode);
6831         textconv = get_textconv(r, df);
6832         if (!textconv) {
6833                 free_filespec(df);
6834                 return 0;
6835         }
6836
6837         *buf_size = fill_textconv(r, textconv, df, buf);
6838         free_filespec(df);
6839         return 1;
6840 }
6841
6842 void setup_diff_pager(struct diff_options *opt)
6843 {
6844         /*
6845          * If the user asked for our exit code, then either they want --quiet
6846          * or --exit-code. We should definitely not bother with a pager in the
6847          * former case, as we will generate no output. Since we still properly
6848          * report our exit code even when a pager is run, we _could_ run a
6849          * pager with --exit-code. But since we have not done so historically,
6850          * and because it is easy to find people oneline advising "git diff
6851          * --exit-code" in hooks and other scripts, we do not do so.
6852          */
6853         if (!opt->flags.exit_with_status &&
6854             check_pager_config("diff") != 0)
6855                 setup_pager();
6856 }