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 "fetch-object.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 void *entry,
 937                            const void *entry_or_key,
 938                            const void *keydata)
 939 {
 940         const struct diff_options *diffopt = hashmap_cmp_fn_data;
 941         const struct moved_entry *a = entry;
 942         const struct moved_entry *b = entry_or_key;
 943         unsigned flags = diffopt->color_moved_ws_handling
 944                          & XDF_WHITESPACE_FLAGS;
 945
 946         if (diffopt->color_moved_ws_handling &
 947             COLOR_MOVED_WS_ALLOW_INDENTATION_CHANGE)
 948                 /*
 949                  * As there is not specific white space config given,
 950                  * we'd need to check for a new block, so ignore all
 951                  * white space. The setup of the white space
 952                  * configuration for the next block is done else where
 953                  */
 954                 flags |= XDF_IGNORE_WHITESPACE;
 955
 956         return !xdiff_compare_lines(a->es->line, a->es->len,
 957                                     b->es->line, b->es->len,
 958                                     flags);
 959 }
 960
 961 static struct moved_entry *prepare_entry(struct diff_options *o,
 962                                          int line_no)
 963 {
 964         struct moved_entry *ret = xmalloc(sizeof(*ret));
 965         struct emitted_diff_symbol *l = &o->emitted_symbols->buf[line_no];
 966         unsigned flags = o->color_moved_ws_handling & XDF_WHITESPACE_FLAGS;
 967
 968         ret->ent.hash = xdiff_hash_string(l->line, l->len, flags);
 969         ret->es = l;
 970         ret->next_line = NULL;
 971
 972         return ret;
 973 }
 974
 975 static void add_lines_to_move_detection(struct diff_options *o,
 976                                         struct hashmap *add_lines,
 977                                         struct hashmap *del_lines)
 978 {
 979         struct moved_entry *prev_line = NULL;
 980
 981         int n;
 982         for (n = 0; n < o->emitted_symbols->nr; n++) {
 983                 struct hashmap *hm;
 984                 struct moved_entry *key;
 985
 986                 switch (o->emitted_symbols->buf[n].s) {
 987                 case DIFF_SYMBOL_PLUS:
 988                         hm = add_lines;
 989                         break;
 990                 case DIFF_SYMBOL_MINUS:
 991                         hm = del_lines;
 992                         break;
 993                 default:
 994                         prev_line = NULL;
 995                         continue;
 996                 }
 997
 998                 if (o->color_moved_ws_handling &
 999                     COLOR_MOVED_WS_ALLOW_INDENTATION_CHANGE)
1000                         fill_es_indent_data(&o->emitted_symbols->buf[n]);
1001                 key = prepare_entry(o, n);
1002                 if (prev_line && prev_line->es->s == o->emitted_symbols->buf[n].s)
1003                         prev_line->next_line = key;
1004
1005                 hashmap_add(hm, key);
1006                 prev_line = key;
1007         }
1008 }
1009
1010 static void pmb_advance_or_null(struct diff_options *o,
1011                                 struct moved_entry *match,
1012                                 struct hashmap *hm,
1013                                 struct moved_block *pmb,
1014                                 int pmb_nr)
1015 {
1016         int i;
1017         for (i = 0; i < pmb_nr; i++) {
1018                 struct moved_entry *prev = pmb[i].match;
1019                 struct moved_entry *cur = (prev && prev->next_line) ?
1020                                 prev->next_line : NULL;
1021                 if (cur && !hm->cmpfn(o, cur, match, NULL)) {
1022                         pmb[i].match = cur;
1023                 } else {
1024                         pmb[i].match = NULL;
1025                 }
1026         }
1027 }
1028
1029 static void pmb_advance_or_null_multi_match(struct diff_options *o,
1030                                             struct moved_entry *match,
1031                                             struct hashmap *hm,
1032                                             struct moved_block *pmb,
1033                                             int pmb_nr, int n)
1034 {
1035         int i;
1036         char *got_match = xcalloc(1, pmb_nr);
1037
1038         for (; match; match = hashmap_get_next(hm, match)) {
1039                 for (i = 0; i < pmb_nr; i++) {
1040                         struct moved_entry *prev = pmb[i].match;
1041                         struct moved_entry *cur = (prev && prev->next_line) ?
1042                                         prev->next_line : NULL;
1043                         if (!cur)
1044                                 continue;
1045                         if (!cmp_in_block_with_wsd(o, cur, match, &pmb[i], n))
1046                                 got_match[i] |= 1;
1047                 }
1048         }
1049
1050         for (i = 0; i < pmb_nr; i++) {
1051                 if (got_match[i]) {
1052                         /* Advance to the next line */
1053                         pmb[i].match = pmb[i].match->next_line;
1054                 } else {
1055                         moved_block_clear(&pmb[i]);
1056                 }
1057         }
1058
1059         free(got_match);
1060 }
1061
1062 static int shrink_potential_moved_blocks(struct moved_block *pmb,
1063                                          int pmb_nr)
1064 {
1065         int lp, rp;
1066
1067         /* Shrink the set of potential block to the remaining running */
1068         for (lp = 0, rp = pmb_nr - 1; lp <= rp;) {
1069                 while (lp < pmb_nr && pmb[lp].match)
1070                         lp++;
1071                 /* lp points at the first NULL now */
1072
1073                 while (rp > -1 && !pmb[rp].match)
1074                         rp--;
1075                 /* rp points at the last non-NULL */
1076
1077                 if (lp < pmb_nr && rp > -1 && lp < rp) {
1078                         pmb[lp] = pmb[rp];
1079                         memset(&pmb[rp], 0, sizeof(pmb[rp]));
1080                         rp--;
1081                         lp++;
1082                 }
1083         }
1084
1085         /* Remember the number of running sets */
1086         return rp + 1;
1087 }
1088
1089 /*
1090  * If o->color_moved is COLOR_MOVED_PLAIN, this function does nothing.
1091  *
1092  * Otherwise, if the last block has fewer alphanumeric characters than
1093  * COLOR_MOVED_MIN_ALNUM_COUNT, unset DIFF_SYMBOL_MOVED_LINE on all lines in
1094  * that block.
1095  *
1096  * The last block consists of the (n - block_length)'th line up to but not
1097  * including the nth line.
1098  *
1099  * Returns 0 if the last block is empty or is unset by this function, non zero
1100  * otherwise.
1101  *
1102  * NEEDSWORK: This uses the same heuristic as blame_entry_score() in blame.c.
1103  * Think of a way to unify them.
1104  */
1105 static int adjust_last_block(struct diff_options *o, int n, int block_length)
1106 {
1107         int i, alnum_count = 0;
1108         if (o->color_moved == COLOR_MOVED_PLAIN)
1109                 return block_length;
1110         for (i = 1; i < block_length + 1; i++) {
1111                 const char *c = o->emitted_symbols->buf[n - i].line;
1112                 for (; *c; c++) {
1113                         if (!isalnum(*c))
1114                                 continue;
1115                         alnum_count++;
1116                         if (alnum_count >= COLOR_MOVED_MIN_ALNUM_COUNT)
1117                                 return 1;
1118                 }
1119         }
1120         for (i = 1; i < block_length + 1; i++)
1121                 o->emitted_symbols->buf[n - i].flags &= ~DIFF_SYMBOL_MOVED_LINE;
1122         return 0;
1123 }
1124
1125 /* Find blocks of moved code, delegate actual coloring decision to helper */
1126 static void mark_color_as_moved(struct diff_options *o,
1127                                 struct hashmap *add_lines,
1128                                 struct hashmap *del_lines)
1129 {
1130         struct moved_block *pmb = NULL; /* potentially moved blocks */
1131         int pmb_nr = 0, pmb_alloc = 0;
1132         int n, flipped_block = 0, block_length = 0;
1133
1134
1135         for (n = 0; n < o->emitted_symbols->nr; n++) {
1136                 struct hashmap *hm = NULL;
1137                 struct moved_entry *key;
1138                 struct moved_entry *match = NULL;
1139                 struct emitted_diff_symbol *l = &o->emitted_symbols->buf[n];
1140                 enum diff_symbol last_symbol = 0;
1141
1142                 switch (l->s) {
1143                 case DIFF_SYMBOL_PLUS:
1144                         hm = del_lines;
1145                         key = prepare_entry(o, n);
1146                         match = hashmap_get(hm, key, NULL);
1147                         free(key);
1148                         break;
1149                 case DIFF_SYMBOL_MINUS:
1150                         hm = add_lines;
1151                         key = prepare_entry(o, n);
1152                         match = hashmap_get(hm, key, NULL);
1153                         free(key);
1154                         break;
1155                 default:
1156                         flipped_block = 0;
1157                 }
1158
1159                 if (!match) {
1160                         int i;
1161
1162                         adjust_last_block(o, n, block_length);
1163                         for(i = 0; i < pmb_nr; i++)
1164                                 moved_block_clear(&pmb[i]);
1165                         pmb_nr = 0;
1166                         block_length = 0;
1167                         flipped_block = 0;
1168                         last_symbol = l->s;
1169                         continue;
1170                 }
1171
1172                 if (o->color_moved == COLOR_MOVED_PLAIN) {
1173                         last_symbol = l->s;
1174                         l->flags |= DIFF_SYMBOL_MOVED_LINE;
1175                         continue;
1176                 }
1177
1178                 if (o->color_moved_ws_handling &
1179                     COLOR_MOVED_WS_ALLOW_INDENTATION_CHANGE)
1180                         pmb_advance_or_null_multi_match(o, match, hm, pmb, pmb_nr, n);
1181                 else
1182                         pmb_advance_or_null(o, match, hm, pmb, pmb_nr);
1183
1184                 pmb_nr = shrink_potential_moved_blocks(pmb, pmb_nr);
1185
1186                 if (pmb_nr == 0) {
1187                         /*
1188                          * The current line is the start of a new block.
1189                          * Setup the set of potential blocks.
1190                          */
1191                         for (; match; match = hashmap_get_next(hm, match)) {
1192                                 ALLOC_GROW(pmb, pmb_nr + 1, pmb_alloc);
1193                                 if (o->color_moved_ws_handling &
1194                                     COLOR_MOVED_WS_ALLOW_INDENTATION_CHANGE) {
1195                                         if (compute_ws_delta(l, match->es,
1196                                                              &pmb[pmb_nr].wsd))
1197                                                 pmb[pmb_nr++].match = match;
1198                                 } else {
1199                                         pmb[pmb_nr].wsd = 0;
1200                                         pmb[pmb_nr++].match = match;
1201                                 }
1202                         }
1203
1204                         if (adjust_last_block(o, n, block_length) &&
1205                             pmb_nr && last_symbol != l->s)
1206                                 flipped_block = (flipped_block + 1) % 2;
1207                         else
1208                                 flipped_block = 0;
1209
1210                         block_length = 0;
1211                 }
1212
1213                 if (pmb_nr) {
1214                         block_length++;
1215                         l->flags |= DIFF_SYMBOL_MOVED_LINE;
1216                         if (flipped_block && o->color_moved != COLOR_MOVED_BLOCKS)
1217                                 l->flags |= DIFF_SYMBOL_MOVED_LINE_ALT;
1218                 }
1219                 last_symbol = l->s;
1220         }
1221         adjust_last_block(o, n, block_length);
1222
1223         for(n = 0; n < pmb_nr; n++)
1224                 moved_block_clear(&pmb[n]);
1225         free(pmb);
1226 }
1227
1228 #define DIFF_SYMBOL_MOVED_LINE_ZEBRA_MASK \
1229   (DIFF_SYMBOL_MOVED_LINE | DIFF_SYMBOL_MOVED_LINE_ALT)
1230 static void dim_moved_lines(struct diff_options *o)
1231 {
1232         int n;
1233         for (n = 0; n < o->emitted_symbols->nr; n++) {
1234                 struct emitted_diff_symbol *prev = (n != 0) ?
1235                                 &o->emitted_symbols->buf[n - 1] : NULL;
1236                 struct emitted_diff_symbol *l = &o->emitted_symbols->buf[n];
1237                 struct emitted_diff_symbol *next =
1238                                 (n < o->emitted_symbols->nr - 1) ?
1239                                 &o->emitted_symbols->buf[n + 1] : NULL;
1240
1241                 /* Not a plus or minus line? */
1242                 if (l->s != DIFF_SYMBOL_PLUS && l->s != DIFF_SYMBOL_MINUS)
1243                         continue;
1244
1245                 /* Not a moved line? */
1246                 if (!(l->flags & DIFF_SYMBOL_MOVED_LINE))
1247                         continue;
1248
1249                 /*
1250                  * If prev or next are not a plus or minus line,
1251                  * pretend they don't exist
1252                  */
1253                 if (prev && prev->s != DIFF_SYMBOL_PLUS &&
1254                             prev->s != DIFF_SYMBOL_MINUS)
1255                         prev = NULL;
1256                 if (next && next->s != DIFF_SYMBOL_PLUS &&
1257                             next->s != DIFF_SYMBOL_MINUS)
1258                         next = NULL;
1259
1260                 /* Inside a block? */
1261                 if ((prev &&
1262                     (prev->flags & DIFF_SYMBOL_MOVED_LINE_ZEBRA_MASK) ==
1263                     (l->flags & DIFF_SYMBOL_MOVED_LINE_ZEBRA_MASK)) &&
1264                     (next &&
1265                     (next->flags & DIFF_SYMBOL_MOVED_LINE_ZEBRA_MASK) ==
1266                     (l->flags & DIFF_SYMBOL_MOVED_LINE_ZEBRA_MASK))) {
1267                         l->flags |= DIFF_SYMBOL_MOVED_LINE_UNINTERESTING;
1268                         continue;
1269                 }
1270
1271                 /* Check if we are at an interesting bound: */
1272                 if (prev && (prev->flags & DIFF_SYMBOL_MOVED_LINE) &&
1273                     (prev->flags & DIFF_SYMBOL_MOVED_LINE_ALT) !=
1274                        (l->flags & DIFF_SYMBOL_MOVED_LINE_ALT))
1275                         continue;
1276                 if (next && (next->flags & DIFF_SYMBOL_MOVED_LINE) &&
1277                     (next->flags & DIFF_SYMBOL_MOVED_LINE_ALT) !=
1278                        (l->flags & DIFF_SYMBOL_MOVED_LINE_ALT))
1279                         continue;
1280
1281                 /*
1282                  * The boundary to prev and next are not interesting,
1283                  * so this line is not interesting as a whole
1284                  */
1285                 l->flags |= DIFF_SYMBOL_MOVED_LINE_UNINTERESTING;
1286         }
1287 }
1288
1289 static void emit_line_ws_markup(struct diff_options *o,
1290                                 const char *set_sign, const char *set,
1291                                 const char *reset,
1292                                 int sign_index, const char *line, int len,
1293                                 unsigned ws_rule, int blank_at_eof)
1294 {
1295         const char *ws = NULL;
1296         int sign = o->output_indicators[sign_index];
1297
1298         if (o->ws_error_highlight & ws_rule) {
1299                 ws = diff_get_color_opt(o, DIFF_WHITESPACE);
1300                 if (!*ws)
1301                         ws = NULL;
1302         }
1303
1304         if (!ws && !set_sign)
1305                 emit_line_0(o, set, NULL, 0, reset, sign, line, len);
1306         else if (!ws) {
1307                 emit_line_0(o, set_sign, set, !!set_sign, reset, sign, line, len);
1308         } else if (blank_at_eof)
1309                 /* Blank line at EOF - paint '+' as well */
1310                 emit_line_0(o, ws, NULL, 0, reset, sign, line, len);
1311         else {
1312                 /* Emit just the prefix, then the rest. */
1313                 emit_line_0(o, set_sign ? set_sign : set, NULL, !!set_sign, reset,
1314                             sign, "", 0);
1315                 ws_check_emit(line, len, ws_rule,
1316                               o->file, set, reset, ws);
1317         }
1318 }
1319
1320 static void emit_diff_symbol_from_struct(struct diff_options *o,
1321                                          struct emitted_diff_symbol *eds)
1322 {
1323         static const char *nneof = " No newline at end of file\n";
1324         const char *context, *reset, *set, *set_sign, *meta, *fraginfo;
1325         struct strbuf sb = STRBUF_INIT;
1326
1327         enum diff_symbol s = eds->s;
1328         const char *line = eds->line;
1329         int len = eds->len;
1330         unsigned flags = eds->flags;
1331
1332         switch (s) {
1333         case DIFF_SYMBOL_NO_LF_EOF:
1334                 context = diff_get_color_opt(o, DIFF_CONTEXT);
1335                 reset = diff_get_color_opt(o, DIFF_RESET);
1336                 putc('\n', o->file);
1337                 emit_line_0(o, context, NULL, 0, reset, '\\',
1338                             nneof, strlen(nneof));
1339                 break;
1340         case DIFF_SYMBOL_SUBMODULE_HEADER:
1341         case DIFF_SYMBOL_SUBMODULE_ERROR:
1342         case DIFF_SYMBOL_SUBMODULE_PIPETHROUGH:
1343         case DIFF_SYMBOL_STATS_SUMMARY_INSERTS_DELETES:
1344         case DIFF_SYMBOL_SUMMARY:
1345         case DIFF_SYMBOL_STATS_LINE:
1346         case DIFF_SYMBOL_BINARY_DIFF_BODY:
1347         case DIFF_SYMBOL_CONTEXT_FRAGINFO:
1348                 emit_line(o, "", "", line, len);
1349                 break;
1350         case DIFF_SYMBOL_CONTEXT_INCOMPLETE:
1351         case DIFF_SYMBOL_CONTEXT_MARKER:
1352                 context = diff_get_color_opt(o, DIFF_CONTEXT);
1353                 reset = diff_get_color_opt(o, DIFF_RESET);
1354                 emit_line(o, context, reset, line, len);
1355                 break;
1356         case DIFF_SYMBOL_SEPARATOR:
1357                 fprintf(o->file, "%s%c",
1358                         diff_line_prefix(o),
1359                         o->line_termination);
1360                 break;
1361         case DIFF_SYMBOL_CONTEXT:
1362                 set = diff_get_color_opt(o, DIFF_CONTEXT);
1363                 reset = diff_get_color_opt(o, DIFF_RESET);
1364                 set_sign = NULL;
1365                 if (o->flags.dual_color_diffed_diffs) {
1366                         char c = !len ? 0 : line[0];
1367
1368                         if (c == '+')
1369                                 set = diff_get_color_opt(o, DIFF_FILE_NEW);
1370                         else if (c == '@')
1371                                 set = diff_get_color_opt(o, DIFF_FRAGINFO);
1372                         else if (c == '-')
1373                                 set = diff_get_color_opt(o, DIFF_FILE_OLD);
1374                 }
1375                 emit_line_ws_markup(o, set_sign, set, reset,
1376                                     OUTPUT_INDICATOR_CONTEXT, line, len,
1377                                     flags & (DIFF_SYMBOL_CONTENT_WS_MASK), 0);
1378                 break;
1379         case DIFF_SYMBOL_PLUS:
1380                 switch (flags & (DIFF_SYMBOL_MOVED_LINE |
1381                                  DIFF_SYMBOL_MOVED_LINE_ALT |
1382                                  DIFF_SYMBOL_MOVED_LINE_UNINTERESTING)) {
1383                 case DIFF_SYMBOL_MOVED_LINE |
1384                      DIFF_SYMBOL_MOVED_LINE_ALT |
1385                      DIFF_SYMBOL_MOVED_LINE_UNINTERESTING:
1386                         set = diff_get_color_opt(o, DIFF_FILE_NEW_MOVED_ALT_DIM);
1387                         break;
1388                 case DIFF_SYMBOL_MOVED_LINE |
1389                      DIFF_SYMBOL_MOVED_LINE_ALT:
1390                         set = diff_get_color_opt(o, DIFF_FILE_NEW_MOVED_ALT);
1391                         break;
1392                 case DIFF_SYMBOL_MOVED_LINE |
1393                      DIFF_SYMBOL_MOVED_LINE_UNINTERESTING:
1394                         set = diff_get_color_opt(o, DIFF_FILE_NEW_MOVED_DIM);
1395                         break;
1396                 case DIFF_SYMBOL_MOVED_LINE:
1397                         set = diff_get_color_opt(o, DIFF_FILE_NEW_MOVED);
1398                         break;
1399                 default:
1400                         set = diff_get_color_opt(o, DIFF_FILE_NEW);
1401                 }
1402                 reset = diff_get_color_opt(o, DIFF_RESET);
1403                 if (!o->flags.dual_color_diffed_diffs)
1404                         set_sign = NULL;
1405                 else {
1406                         char c = !len ? 0 : line[0];
1407
1408                         set_sign = set;
1409                         if (c == '-')
1410                                 set = diff_get_color_opt(o, DIFF_FILE_OLD_BOLD);
1411                         else if (c == '@')
1412                                 set = diff_get_color_opt(o, DIFF_FRAGINFO);
1413                         else if (c == '+')
1414                                 set = diff_get_color_opt(o, DIFF_FILE_NEW_BOLD);
1415                         else
1416                                 set = diff_get_color_opt(o, DIFF_CONTEXT_BOLD);
1417                         flags &= ~DIFF_SYMBOL_CONTENT_WS_MASK;
1418                 }
1419                 emit_line_ws_markup(o, set_sign, set, reset,
1420                                     OUTPUT_INDICATOR_NEW, line, len,
1421                                     flags & DIFF_SYMBOL_CONTENT_WS_MASK,
1422                                     flags & DIFF_SYMBOL_CONTENT_BLANK_LINE_EOF);
1423                 break;
1424         case DIFF_SYMBOL_MINUS:
1425                 switch (flags & (DIFF_SYMBOL_MOVED_LINE |
1426                                  DIFF_SYMBOL_MOVED_LINE_ALT |
1427                                  DIFF_SYMBOL_MOVED_LINE_UNINTERESTING)) {
1428                 case DIFF_SYMBOL_MOVED_LINE |
1429                      DIFF_SYMBOL_MOVED_LINE_ALT |
1430                      DIFF_SYMBOL_MOVED_LINE_UNINTERESTING:
1431                         set = diff_get_color_opt(o, DIFF_FILE_OLD_MOVED_ALT_DIM);
1432                         break;
1433                 case DIFF_SYMBOL_MOVED_LINE |
1434                      DIFF_SYMBOL_MOVED_LINE_ALT:
1435                         set = diff_get_color_opt(o, DIFF_FILE_OLD_MOVED_ALT);
1436                         break;
1437                 case DIFF_SYMBOL_MOVED_LINE |
1438                      DIFF_SYMBOL_MOVED_LINE_UNINTERESTING:
1439                         set = diff_get_color_opt(o, DIFF_FILE_OLD_MOVED_DIM);
1440                         break;
1441                 case DIFF_SYMBOL_MOVED_LINE:
1442                         set = diff_get_color_opt(o, DIFF_FILE_OLD_MOVED);
1443                         break;
1444                 default:
1445                         set = diff_get_color_opt(o, DIFF_FILE_OLD);
1446                 }
1447                 reset = diff_get_color_opt(o, DIFF_RESET);
1448                 if (!o->flags.dual_color_diffed_diffs)
1449                         set_sign = NULL;
1450                 else {
1451                         char c = !len ? 0 : line[0];
1452
1453                         set_sign = set;
1454                         if (c == '+')
1455                                 set = diff_get_color_opt(o, DIFF_FILE_NEW_DIM);
1456                         else if (c == '@')
1457                                 set = diff_get_color_opt(o, DIFF_FRAGINFO);
1458                         else if (c == '-')
1459                                 set = diff_get_color_opt(o, DIFF_FILE_OLD_DIM);
1460                         else
1461                                 set = diff_get_color_opt(o, DIFF_CONTEXT_DIM);
1462                 }
1463                 emit_line_ws_markup(o, set_sign, set, reset,
1464                                     OUTPUT_INDICATOR_OLD, line, len,
1465                                     flags & DIFF_SYMBOL_CONTENT_WS_MASK, 0);
1466                 break;
1467         case DIFF_SYMBOL_WORDS_PORCELAIN:
1468                 context = diff_get_color_opt(o, DIFF_CONTEXT);
1469                 reset = diff_get_color_opt(o, DIFF_RESET);
1470                 emit_line(o, context, reset, line, len);
1471                 fputs("~\n", o->file);
1472                 break;
1473         case DIFF_SYMBOL_WORDS:
1474                 context = diff_get_color_opt(o, DIFF_CONTEXT);
1475                 reset = diff_get_color_opt(o, DIFF_RESET);
1476                 /*
1477                  * Skip the prefix character, if any.  With
1478                  * diff_suppress_blank_empty, there may be
1479                  * none.
1480                  */
1481                 if (line[0] != '\n') {
1482                         line++;
1483                         len--;
1484                 }
1485                 emit_line(o, context, reset, line, len);
1486                 break;
1487         case DIFF_SYMBOL_FILEPAIR_PLUS:
1488                 meta = diff_get_color_opt(o, DIFF_METAINFO);
1489                 reset = diff_get_color_opt(o, DIFF_RESET);
1490                 fprintf(o->file, "%s%s+++ %s%s%s\n", diff_line_prefix(o), meta,
1491                         line, reset,
1492                         strchr(line, ' ') ? "\t" : "");
1493                 break;
1494         case DIFF_SYMBOL_FILEPAIR_MINUS:
1495                 meta = diff_get_color_opt(o, DIFF_METAINFO);
1496                 reset = diff_get_color_opt(o, DIFF_RESET);
1497                 fprintf(o->file, "%s%s--- %s%s%s\n", diff_line_prefix(o), meta,
1498                         line, reset,
1499                         strchr(line, ' ') ? "\t" : "");
1500                 break;
1501         case DIFF_SYMBOL_BINARY_FILES:
1502         case DIFF_SYMBOL_HEADER:
1503                 fprintf(o->file, "%s", line);
1504                 break;
1505         case DIFF_SYMBOL_BINARY_DIFF_HEADER:
1506                 fprintf(o->file, "%sGIT binary patch\n", diff_line_prefix(o));
1507                 break;
1508         case DIFF_SYMBOL_BINARY_DIFF_HEADER_DELTA:
1509                 fprintf(o->file, "%sdelta %s\n", diff_line_prefix(o), line);
1510                 break;
1511         case DIFF_SYMBOL_BINARY_DIFF_HEADER_LITERAL:
1512                 fprintf(o->file, "%sliteral %s\n", diff_line_prefix(o), line);
1513                 break;
1514         case DIFF_SYMBOL_BINARY_DIFF_FOOTER:
1515                 fputs(diff_line_prefix(o), o->file);
1516                 fputc('\n', o->file);
1517                 break;
1518         case DIFF_SYMBOL_REWRITE_DIFF:
1519                 fraginfo = diff_get_color(o->use_color, DIFF_FRAGINFO);
1520                 reset = diff_get_color_opt(o, DIFF_RESET);
1521                 emit_line(o, fraginfo, reset, line, len);
1522                 break;
1523         case DIFF_SYMBOL_SUBMODULE_ADD:
1524                 set = diff_get_color_opt(o, DIFF_FILE_NEW);
1525                 reset = diff_get_color_opt(o, DIFF_RESET);
1526                 emit_line(o, set, reset, line, len);
1527                 break;
1528         case DIFF_SYMBOL_SUBMODULE_DEL:
1529                 set = diff_get_color_opt(o, DIFF_FILE_OLD);
1530                 reset = diff_get_color_opt(o, DIFF_RESET);
1531                 emit_line(o, set, reset, line, len);
1532                 break;
1533         case DIFF_SYMBOL_SUBMODULE_UNTRACKED:
1534                 fprintf(o->file, "%sSubmodule %s contains untracked content\n",
1535                         diff_line_prefix(o), line);
1536                 break;
1537         case DIFF_SYMBOL_SUBMODULE_MODIFIED:
1538                 fprintf(o->file, "%sSubmodule %s contains modified content\n",
1539                         diff_line_prefix(o), line);
1540                 break;
1541         case DIFF_SYMBOL_STATS_SUMMARY_NO_FILES:
1542                 emit_line(o, "", "", " 0 files changed\n",
1543                           strlen(" 0 files changed\n"));
1544                 break;
1545         case DIFF_SYMBOL_STATS_SUMMARY_ABBREV:
1546                 emit_line(o, "", "", " ...\n", strlen(" ...\n"));
1547                 break;
1548         case DIFF_SYMBOL_WORD_DIFF:
1549                 fprintf(o->file, "%.*s", len, line);
1550                 break;
1551         case DIFF_SYMBOL_STAT_SEP:
1552                 fputs(o->stat_sep, o->file);
1553                 break;
1554         default:
1555                 BUG("unknown diff symbol");
1556         }
1557         strbuf_release(&sb);
1558 }
1559
1560 static void emit_diff_symbol(struct diff_options *o, enum diff_symbol s,
1561                              const char *line, int len, unsigned flags)
1562 {
1563         struct emitted_diff_symbol e = {line, len, flags, 0, 0, s};
1564
1565         if (o->emitted_symbols)
1566                 append_emitted_diff_symbol(o, &e);
1567         else
1568                 emit_diff_symbol_from_struct(o, &e);
1569 }
1570
1571 void diff_emit_submodule_del(struct diff_options *o, const char *line)
1572 {
1573         emit_diff_symbol(o, DIFF_SYMBOL_SUBMODULE_DEL, line, strlen(line), 0);
1574 }
1575
1576 void diff_emit_submodule_add(struct diff_options *o, const char *line)
1577 {
1578         emit_diff_symbol(o, DIFF_SYMBOL_SUBMODULE_ADD, line, strlen(line), 0);
1579 }
1580
1581 void diff_emit_submodule_untracked(struct diff_options *o, const char *path)
1582 {
1583         emit_diff_symbol(o, DIFF_SYMBOL_SUBMODULE_UNTRACKED,
1584                          path, strlen(path), 0);
1585 }
1586
1587 void diff_emit_submodule_modified(struct diff_options *o, const char *path)
1588 {
1589         emit_diff_symbol(o, DIFF_SYMBOL_SUBMODULE_MODIFIED,
1590                          path, strlen(path), 0);
1591 }
1592
1593 void diff_emit_submodule_header(struct diff_options *o, const char *header)
1594 {
1595         emit_diff_symbol(o, DIFF_SYMBOL_SUBMODULE_HEADER,
1596                          header, strlen(header), 0);
1597 }
1598
1599 void diff_emit_submodule_error(struct diff_options *o, const char *err)
1600 {
1601         emit_diff_symbol(o, DIFF_SYMBOL_SUBMODULE_ERROR, err, strlen(err), 0);
1602 }
1603
1604 void diff_emit_submodule_pipethrough(struct diff_options *o,
1605                                      const char *line, int len)
1606 {
1607         emit_diff_symbol(o, DIFF_SYMBOL_SUBMODULE_PIPETHROUGH, line, len, 0);
1608 }
1609
1610 static int new_blank_line_at_eof(struct emit_callback *ecbdata, const char *line, int len)
1611 {
1612         if (!((ecbdata->ws_rule & WS_BLANK_AT_EOF) &&
1613               ecbdata->blank_at_eof_in_preimage &&
1614               ecbdata->blank_at_eof_in_postimage &&
1615               ecbdata->blank_at_eof_in_preimage <= ecbdata->lno_in_preimage &&
1616               ecbdata->blank_at_eof_in_postimage <= ecbdata->lno_in_postimage))
1617                 return 0;
1618         return ws_blank_line(line, len, ecbdata->ws_rule);
1619 }
1620
1621 static void emit_add_line(struct emit_callback *ecbdata,
1622                           const char *line, int len)
1623 {
1624         unsigned flags = WSEH_NEW | ecbdata->ws_rule;
1625         if (new_blank_line_at_eof(ecbdata, line, len))
1626                 flags |= DIFF_SYMBOL_CONTENT_BLANK_LINE_EOF;
1627
1628         emit_diff_symbol(ecbdata->opt, DIFF_SYMBOL_PLUS, line, len, flags);
1629 }
1630
1631 static void emit_del_line(struct emit_callback *ecbdata,
1632                           const char *line, int len)
1633 {
1634         unsigned flags = WSEH_OLD | ecbdata->ws_rule;
1635         emit_diff_symbol(ecbdata->opt, DIFF_SYMBOL_MINUS, line, len, flags);
1636 }
1637
1638 static void emit_context_line(struct emit_callback *ecbdata,
1639                               const char *line, int len)
1640 {
1641         unsigned flags = WSEH_CONTEXT | ecbdata->ws_rule;
1642         emit_diff_symbol(ecbdata->opt, DIFF_SYMBOL_CONTEXT, line, len, flags);
1643 }
1644
1645 static void emit_hunk_header(struct emit_callback *ecbdata,
1646                              const char *line, int len)
1647 {
1648         const char *context = diff_get_color(ecbdata->color_diff, DIFF_CONTEXT);
1649         const char *frag = diff_get_color(ecbdata->color_diff, DIFF_FRAGINFO);
1650         const char *func = diff_get_color(ecbdata->color_diff, DIFF_FUNCINFO);
1651         const char *reset = diff_get_color(ecbdata->color_diff, DIFF_RESET);
1652         const char *reverse = ecbdata->color_diff ? GIT_COLOR_REVERSE : "";
1653         static const char atat[2] = { '@', '@' };
1654         const char *cp, *ep;
1655         struct strbuf msgbuf = STRBUF_INIT;
1656         int org_len = len;
1657         int i = 1;
1658
1659         /*
1660          * As a hunk header must begin with "@@ -<old>, +<new> @@",
1661          * it always is at least 10 bytes long.
1662          */
1663         if (len < 10 ||
1664             memcmp(line, atat, 2) ||
1665             !(ep = memmem(line + 2, len - 2, atat, 2))) {
1666                 emit_diff_symbol(ecbdata->opt,
1667                                  DIFF_SYMBOL_CONTEXT_MARKER, line, len, 0);
1668                 return;
1669         }
1670         ep += 2; /* skip over @@ */
1671
1672         /* The hunk header in fraginfo color */
1673         if (ecbdata->opt->flags.dual_color_diffed_diffs)
1674                 strbuf_addstr(&msgbuf, reverse);
1675         strbuf_addstr(&msgbuf, frag);
1676         strbuf_add(&msgbuf, line, ep - line);
1677         strbuf_addstr(&msgbuf, reset);
1678
1679         /*
1680          * trailing "\r\n"
1681          */
1682         for ( ; i < 3; i++)
1683                 if (line[len - i] == '\r' || line[len - i] == '\n')
1684                         len--;
1685
1686         /* blank before the func header */
1687         for (cp = ep; ep - line < len; ep++)
1688                 if (*ep != ' ' && *ep != '\t')
1689                         break;
1690         if (ep != cp) {
1691                 strbuf_addstr(&msgbuf, context);
1692                 strbuf_add(&msgbuf, cp, ep - cp);
1693                 strbuf_addstr(&msgbuf, reset);
1694         }
1695
1696         if (ep < line + len) {
1697                 strbuf_addstr(&msgbuf, func);
1698                 strbuf_add(&msgbuf, ep, line + len - ep);
1699                 strbuf_addstr(&msgbuf, reset);
1700         }
1701
1702         strbuf_add(&msgbuf, line + len, org_len - len);
1703         strbuf_complete_line(&msgbuf);
1704         emit_diff_symbol(ecbdata->opt,
1705                          DIFF_SYMBOL_CONTEXT_FRAGINFO, msgbuf.buf, msgbuf.len, 0);
1706         strbuf_release(&msgbuf);
1707 }
1708
1709 static struct diff_tempfile *claim_diff_tempfile(void)
1710 {
1711         int i;
1712         for (i = 0; i < ARRAY_SIZE(diff_temp); i++)
1713                 if (!diff_temp[i].name)
1714                         return diff_temp + i;
1715         BUG("diff is failing to clean up its tempfiles");
1716 }
1717
1718 static void remove_tempfile(void)
1719 {
1720         int i;
1721         for (i = 0; i < ARRAY_SIZE(diff_temp); i++) {
1722                 if (is_tempfile_active(diff_temp[i].tempfile))
1723                         delete_tempfile(&diff_temp[i].tempfile);
1724                 diff_temp[i].name = NULL;
1725         }
1726 }
1727
1728 static void add_line_count(struct strbuf *out, int count)
1729 {
1730         switch (count) {
1731         case 0:
1732                 strbuf_addstr(out, "0,0");
1733                 break;
1734         case 1:
1735                 strbuf_addstr(out, "1");
1736                 break;
1737         default:
1738                 strbuf_addf(out, "1,%d", count);
1739                 break;
1740         }
1741 }
1742
1743 static void emit_rewrite_lines(struct emit_callback *ecb,
1744                                int prefix, const char *data, int size)
1745 {
1746         const char *endp = NULL;
1747
1748         while (0 < size) {
1749                 int len;
1750
1751                 endp = memchr(data, '\n', size);
1752                 len = endp ? (endp - data + 1) : size;
1753                 if (prefix != '+') {
1754                         ecb->lno_in_preimage++;
1755                         emit_del_line(ecb, data, len);
1756                 } else {
1757                         ecb->lno_in_postimage++;
1758                         emit_add_line(ecb, data, len);
1759                 }
1760                 size -= len;
1761                 data += len;
1762         }
1763         if (!endp)
1764                 emit_diff_symbol(ecb->opt, DIFF_SYMBOL_NO_LF_EOF, NULL, 0, 0);
1765 }
1766
1767 static void emit_rewrite_diff(const char *name_a,
1768                               const char *name_b,
1769                               struct diff_filespec *one,
1770                               struct diff_filespec *two,
1771                               struct userdiff_driver *textconv_one,
1772                               struct userdiff_driver *textconv_two,
1773                               struct diff_options *o)
1774 {
1775         int lc_a, lc_b;
1776         static struct strbuf a_name = STRBUF_INIT, b_name = STRBUF_INIT;
1777         const char *a_prefix, *b_prefix;
1778         char *data_one, *data_two;
1779         size_t size_one, size_two;
1780         struct emit_callback ecbdata;
1781         struct strbuf out = STRBUF_INIT;
1782
1783         if (diff_mnemonic_prefix && o->flags.reverse_diff) {
1784                 a_prefix = o->b_prefix;
1785                 b_prefix = o->a_prefix;
1786         } else {
1787                 a_prefix = o->a_prefix;
1788                 b_prefix = o->b_prefix;
1789         }
1790
1791         name_a += (*name_a == '/');
1792         name_b += (*name_b == '/');
1793
1794         strbuf_reset(&a_name);
1795         strbuf_reset(&b_name);
1796         quote_two_c_style(&a_name, a_prefix, name_a, 0);
1797         quote_two_c_style(&b_name, b_prefix, name_b, 0);
1798
1799         size_one = fill_textconv(o->repo, textconv_one, one, &data_one);
1800         size_two = fill_textconv(o->repo, textconv_two, two, &data_two);
1801
1802         memset(&ecbdata, 0, sizeof(ecbdata));
1803         ecbdata.color_diff = want_color(o->use_color);
1804         ecbdata.ws_rule = whitespace_rule(o->repo->index, name_b);
1805         ecbdata.opt = o;
1806         if (ecbdata.ws_rule & WS_BLANK_AT_EOF) {
1807                 mmfile_t mf1, mf2;
1808                 mf1.ptr = (char *)data_one;
1809                 mf2.ptr = (char *)data_two;
1810                 mf1.size = size_one;
1811                 mf2.size = size_two;
1812                 check_blank_at_eof(&mf1, &mf2, &ecbdata);
1813         }
1814         ecbdata.lno_in_preimage = 1;
1815         ecbdata.lno_in_postimage = 1;
1816
1817         lc_a = count_lines(data_one, size_one);
1818         lc_b = count_lines(data_two, size_two);
1819
1820         emit_diff_symbol(o, DIFF_SYMBOL_FILEPAIR_MINUS,
1821                          a_name.buf, a_name.len, 0);
1822         emit_diff_symbol(o, DIFF_SYMBOL_FILEPAIR_PLUS,
1823                          b_name.buf, b_name.len, 0);
1824
1825         strbuf_addstr(&out, "@@ -");
1826         if (!o->irreversible_delete)
1827                 add_line_count(&out, lc_a);
1828         else
1829                 strbuf_addstr(&out, "?,?");
1830         strbuf_addstr(&out, " +");
1831         add_line_count(&out, lc_b);
1832         strbuf_addstr(&out, " @@\n");
1833         emit_diff_symbol(o, DIFF_SYMBOL_REWRITE_DIFF, out.buf, out.len, 0);
1834         strbuf_release(&out);
1835
1836         if (lc_a && !o->irreversible_delete)
1837                 emit_rewrite_lines(&ecbdata, '-', data_one, size_one);
1838         if (lc_b)
1839                 emit_rewrite_lines(&ecbdata, '+', data_two, size_two);
1840         if (textconv_one)
1841                 free((char *)data_one);
1842         if (textconv_two)
1843                 free((char *)data_two);
1844 }
1845
1846 struct diff_words_buffer {
1847         mmfile_t text;
1848         unsigned long alloc;
1849         struct diff_words_orig {
1850                 const char *begin, *end;
1851         } *orig;
1852         int orig_nr, orig_alloc;
1853 };
1854
1855 static void diff_words_append(char *line, unsigned long len,
1856                 struct diff_words_buffer *buffer)
1857 {
1858         ALLOC_GROW(buffer->text.ptr, buffer->text.size + len, buffer->alloc);
1859         line++;
1860         len--;
1861         memcpy(buffer->text.ptr + buffer->text.size, line, len);
1862         buffer->text.size += len;
1863         buffer->text.ptr[buffer->text.size] = '\0';
1864 }
1865
1866 struct diff_words_style_elem {
1867         const char *prefix;
1868         const char *suffix;
1869         const char *color; /* NULL; filled in by the setup code if
1870                             * color is enabled */
1871 };
1872
1873 struct diff_words_style {
1874         enum diff_words_type type;
1875         struct diff_words_style_elem new_word, old_word, ctx;
1876         const char *newline;
1877 };
1878
1879 static struct diff_words_style diff_words_styles[] = {
1880         { DIFF_WORDS_PORCELAIN, {"+", "\n"}, {"-", "\n"}, {" ", "\n"}, "~\n" },
1881         { DIFF_WORDS_PLAIN, {"{+", "+}"}, {"[-", "-]"}, {"", ""}, "\n" },
1882         { DIFF_WORDS_COLOR, {"", ""}, {"", ""}, {"", ""}, "\n" }
1883 };
1884
1885 struct diff_words_data {
1886         struct diff_words_buffer minus, plus;
1887         const char *current_plus;
1888         int last_minus;
1889         struct diff_options *opt;
1890         regex_t *word_regex;
1891         enum diff_words_type type;
1892         struct diff_words_style *style;
1893 };
1894
1895 static int fn_out_diff_words_write_helper(struct diff_options *o,
1896                                           struct diff_words_style_elem *st_el,
1897                                           const char *newline,
1898                                           size_t count, const char *buf)
1899 {
1900         int print = 0;
1901         struct strbuf sb = STRBUF_INIT;
1902
1903         while (count) {
1904                 char *p = memchr(buf, '\n', count);
1905                 if (print)
1906                         strbuf_addstr(&sb, diff_line_prefix(o));
1907
1908                 if (p != buf) {
1909                         const char *reset = st_el->color && *st_el->color ?
1910                                             GIT_COLOR_RESET : NULL;
1911                         if (st_el->color && *st_el->color)
1912                                 strbuf_addstr(&sb, st_el->color);
1913                         strbuf_addstr(&sb, st_el->prefix);
1914                         strbuf_add(&sb, buf, p ? p - buf : count);
1915                         strbuf_addstr(&sb, st_el->suffix);
1916                         if (reset)
1917                                 strbuf_addstr(&sb, reset);
1918                 }
1919                 if (!p)
1920                         goto out;
1921
1922                 strbuf_addstr(&sb, newline);
1923                 count -= p + 1 - buf;
1924                 buf = p + 1;
1925                 print = 1;
1926                 if (count) {
1927                         emit_diff_symbol(o, DIFF_SYMBOL_WORD_DIFF,
1928                                          sb.buf, sb.len, 0);
1929                         strbuf_reset(&sb);
1930                 }
1931         }
1932
1933 out:
1934         if (sb.len)
1935                 emit_diff_symbol(o, DIFF_SYMBOL_WORD_DIFF,
1936                                  sb.buf, sb.len, 0);
1937         strbuf_release(&sb);
1938         return 0;
1939 }
1940
1941 /*
1942  * '--color-words' algorithm can be described as:
1943  *
1944  *   1. collect the minus/plus lines of a diff hunk, divided into
1945  *      minus-lines and plus-lines;
1946  *
1947  *   2. break both minus-lines and plus-lines into words and
1948  *      place them into two mmfile_t with one word for each line;
1949  *
1950  *   3. use xdiff to run diff on the two mmfile_t to get the words level diff;
1951  *
1952  * And for the common parts of the both file, we output the plus side text.
1953  * diff_words->current_plus is used to trace the current position of the plus file
1954  * which printed. diff_words->last_minus is used to trace the last minus word
1955  * printed.
1956  *
1957  * For '--graph' to work with '--color-words', we need to output the graph prefix
1958  * on each line of color words output. Generally, there are two conditions on
1959  * which we should output the prefix.
1960  *
1961  *   1. diff_words->last_minus == 0 &&
1962  *      diff_words->current_plus == diff_words->plus.text.ptr
1963  *
1964  *      that is: the plus text must start as a new line, and if there is no minus
1965  *      word printed, a graph prefix must be printed.
1966  *
1967  *   2. diff_words->current_plus > diff_words->plus.text.ptr &&
1968  *      *(diff_words->current_plus - 1) == '\n'
1969  *
1970  *      that is: a graph prefix must be printed following a '\n'
1971  */
1972 static int color_words_output_graph_prefix(struct diff_words_data *diff_words)
1973 {
1974         if ((diff_words->last_minus == 0 &&
1975                 diff_words->current_plus == diff_words->plus.text.ptr) ||
1976                 (diff_words->current_plus > diff_words->plus.text.ptr &&
1977                 *(diff_words->current_plus - 1) == '\n')) {
1978                 return 1;
1979         } else {
1980                 return 0;
1981         }
1982 }
1983
1984 static void fn_out_diff_words_aux(void *priv,
1985                                   long minus_first, long minus_len,
1986                                   long plus_first, long plus_len,
1987                                   const char *func, long funclen)
1988 {
1989         struct diff_words_data *diff_words = priv;
1990         struct diff_words_style *style = diff_words->style;
1991         const char *minus_begin, *minus_end, *plus_begin, *plus_end;
1992         struct diff_options *opt = diff_words->opt;
1993         const char *line_prefix;
1994
1995         assert(opt);
1996         line_prefix = diff_line_prefix(opt);
1997
1998         /* POSIX requires that first be decremented by one if len == 0... */
1999         if (minus_len) {
2000                 minus_begin = diff_words->minus.orig[minus_first].begin;
2001                 minus_end =
2002                         diff_words->minus.orig[minus_first + minus_len - 1].end;
2003         } else
2004                 minus_begin = minus_end =
2005                         diff_words->minus.orig[minus_first].end;
2006
2007         if (plus_len) {
2008                 plus_begin = diff_words->plus.orig[plus_first].begin;
2009                 plus_end = diff_words->plus.orig[plus_first + plus_len - 1].end;
2010         } else
2011                 plus_begin = plus_end = diff_words->plus.orig[plus_first].end;
2012
2013         if (color_words_output_graph_prefix(diff_words)) {
2014                 fputs(line_prefix, diff_words->opt->file);
2015         }
2016         if (diff_words->current_plus != plus_begin) {
2017                 fn_out_diff_words_write_helper(diff_words->opt,
2018                                 &style->ctx, style->newline,
2019                                 plus_begin - diff_words->current_plus,
2020                                 diff_words->current_plus);
2021         }
2022         if (minus_begin != minus_end) {
2023                 fn_out_diff_words_write_helper(diff_words->opt,
2024                                 &style->old_word, style->newline,
2025                                 minus_end - minus_begin, minus_begin);
2026         }
2027         if (plus_begin != plus_end) {
2028                 fn_out_diff_words_write_helper(diff_words->opt,
2029                                 &style->new_word, style->newline,
2030                                 plus_end - plus_begin, plus_begin);
2031         }
2032
2033         diff_words->current_plus = plus_end;
2034         diff_words->last_minus = minus_first;
2035 }
2036
2037 /* This function starts looking at *begin, and returns 0 iff a word was found. */
2038 static int find_word_boundaries(mmfile_t *buffer, regex_t *word_regex,
2039                 int *begin, int *end)
2040 {
2041         if (word_regex && *begin < buffer->size) {
2042                 regmatch_t match[1];
2043                 if (!regexec_buf(word_regex, buffer->ptr + *begin,
2044                                  buffer->size - *begin, 1, match, 0)) {
2045                         char *p = memchr(buffer->ptr + *begin + match[0].rm_so,
2046                                         '\n', match[0].rm_eo - match[0].rm_so);
2047                         *end = p ? p - buffer->ptr : match[0].rm_eo + *begin;
2048                         *begin += match[0].rm_so;
2049                         return *begin >= *end;
2050                 }
2051                 return -1;
2052         }
2053
2054         /* find the next word */
2055         while (*begin < buffer->size && isspace(buffer->ptr[*begin]))
2056                 (*begin)++;
2057         if (*begin >= buffer->size)
2058                 return -1;
2059
2060         /* find the end of the word */
2061         *end = *begin + 1;
2062         while (*end < buffer->size && !isspace(buffer->ptr[*end]))
2063                 (*end)++;
2064
2065         return 0;
2066 }
2067
2068 /*
2069  * This function splits the words in buffer->text, stores the list with
2070  * newline separator into out, and saves the offsets of the original words
2071  * in buffer->orig.
2072  */
2073 static void diff_words_fill(struct diff_words_buffer *buffer, mmfile_t *out,
2074                 regex_t *word_regex)
2075 {
2076         int i, j;
2077         long alloc = 0;
2078
2079         out->size = 0;
2080         out->ptr = NULL;
2081
2082         /* fake an empty "0th" word */
2083         ALLOC_GROW(buffer->orig, 1, buffer->orig_alloc);
2084         buffer->orig[0].begin = buffer->orig[0].end = buffer->text.ptr;
2085         buffer->orig_nr = 1;
2086
2087         for (i = 0; i < buffer->text.size; i++) {
2088                 if (find_word_boundaries(&buffer->text, word_regex, &i, &j))
2089                         return;
2090
2091                 /* store original boundaries */
2092                 ALLOC_GROW(buffer->orig, buffer->orig_nr + 1,
2093                                 buffer->orig_alloc);
2094                 buffer->orig[buffer->orig_nr].begin = buffer->text.ptr + i;
2095                 buffer->orig[buffer->orig_nr].end = buffer->text.ptr + j;
2096                 buffer->orig_nr++;
2097
2098                 /* store one word */
2099                 ALLOC_GROW(out->ptr, out->size + j - i + 1, alloc);
2100                 memcpy(out->ptr + out->size, buffer->text.ptr + i, j - i);
2101                 out->ptr[out->size + j - i] = '\n';
2102                 out->size += j - i + 1;
2103
2104                 i = j - 1;
2105         }
2106 }
2107
2108 /* this executes the word diff on the accumulated buffers */
2109 static void diff_words_show(struct diff_words_data *diff_words)
2110 {
2111         xpparam_t xpp;
2112         xdemitconf_t xecfg;
2113         mmfile_t minus, plus;
2114         struct diff_words_style *style = diff_words->style;
2115
2116         struct diff_options *opt = diff_words->opt;
2117         const char *line_prefix;
2118
2119         assert(opt);
2120         line_prefix = diff_line_prefix(opt);
2121
2122         /* special case: only removal */
2123         if (!diff_words->plus.text.size) {
2124                 emit_diff_symbol(diff_words->opt, DIFF_SYMBOL_WORD_DIFF,
2125                                  line_prefix, strlen(line_prefix), 0);
2126                 fn_out_diff_words_write_helper(diff_words->opt,
2127                         &style->old_word, style->newline,
2128                         diff_words->minus.text.size,
2129                         diff_words->minus.text.ptr);
2130                 diff_words->minus.text.size = 0;
2131                 return;
2132         }
2133
2134         diff_words->current_plus = diff_words->plus.text.ptr;
2135         diff_words->last_minus = 0;
2136
2137         memset(&xpp, 0, sizeof(xpp));
2138         memset(&xecfg, 0, sizeof(xecfg));
2139         diff_words_fill(&diff_words->minus, &minus, diff_words->word_regex);
2140         diff_words_fill(&diff_words->plus, &plus, diff_words->word_regex);
2141         xpp.flags = 0;
2142         /* as only the hunk header will be parsed, we need a 0-context */
2143         xecfg.ctxlen = 0;
2144         if (xdi_diff_outf(&minus, &plus, fn_out_diff_words_aux, NULL,
2145                           diff_words, &xpp, &xecfg))
2146                 die("unable to generate word diff");
2147         free(minus.ptr);
2148         free(plus.ptr);
2149         if (diff_words->current_plus != diff_words->plus.text.ptr +
2150                         diff_words->plus.text.size) {
2151                 if (color_words_output_graph_prefix(diff_words))
2152                         emit_diff_symbol(diff_words->opt, DIFF_SYMBOL_WORD_DIFF,
2153                                          line_prefix, strlen(line_prefix), 0);
2154                 fn_out_diff_words_write_helper(diff_words->opt,
2155                         &style->ctx, style->newline,
2156                         diff_words->plus.text.ptr + diff_words->plus.text.size
2157                         - diff_words->current_plus, diff_words->current_plus);
2158         }
2159         diff_words->minus.text.size = diff_words->plus.text.size = 0;
2160 }
2161
2162 /* In "color-words" mode, show word-diff of words accumulated in the buffer */
2163 static void diff_words_flush(struct emit_callback *ecbdata)
2164 {
2165         struct diff_options *wo = ecbdata->diff_words->opt;
2166
2167         if (ecbdata->diff_words->minus.text.size ||
2168             ecbdata->diff_words->plus.text.size)
2169                 diff_words_show(ecbdata->diff_words);
2170
2171         if (wo->emitted_symbols) {
2172                 struct diff_options *o = ecbdata->opt;
2173                 struct emitted_diff_symbols *wol = wo->emitted_symbols;
2174                 int i;
2175
2176                 /*
2177                  * NEEDSWORK:
2178                  * Instead of appending each, concat all words to a line?
2179                  */
2180                 for (i = 0; i < wol->nr; i++)
2181                         append_emitted_diff_symbol(o, &wol->buf[i]);
2182
2183                 for (i = 0; i < wol->nr; i++)
2184                         free((void *)wol->buf[i].line);
2185
2186                 wol->nr = 0;
2187         }
2188 }
2189
2190 static void diff_filespec_load_driver(struct diff_filespec *one,
2191                                       struct index_state *istate)
2192 {
2193         /* Use already-loaded driver */
2194         if (one->driver)
2195                 return;
2196
2197         if (S_ISREG(one->mode))
2198                 one->driver = userdiff_find_by_path(istate, one->path);
2199
2200         /* Fallback to default settings */
2201         if (!one->driver)
2202                 one->driver = userdiff_find_by_name("default");
2203 }
2204
2205 static const char *userdiff_word_regex(struct diff_filespec *one,
2206                                        struct index_state *istate)
2207 {
2208         diff_filespec_load_driver(one, istate);
2209         return one->driver->word_regex;
2210 }
2211
2212 static void init_diff_words_data(struct emit_callback *ecbdata,
2213                                  struct diff_options *orig_opts,
2214                                  struct diff_filespec *one,
2215                                  struct diff_filespec *two)
2216 {
2217         int i;
2218         struct diff_options *o = xmalloc(sizeof(struct diff_options));
2219         memcpy(o, orig_opts, sizeof(struct diff_options));
2220
2221         ecbdata->diff_words =
2222                 xcalloc(1, sizeof(struct diff_words_data));
2223         ecbdata->diff_words->type = o->word_diff;
2224         ecbdata->diff_words->opt = o;
2225
2226         if (orig_opts->emitted_symbols)
2227                 o->emitted_symbols =
2228                         xcalloc(1, sizeof(struct emitted_diff_symbols));
2229
2230         if (!o->word_regex)
2231                 o->word_regex = userdiff_word_regex(one, o->repo->index);
2232         if (!o->word_regex)
2233                 o->word_regex = userdiff_word_regex(two, o->repo->index);
2234         if (!o->word_regex)
2235                 o->word_regex = diff_word_regex_cfg;
2236         if (o->word_regex) {
2237                 ecbdata->diff_words->word_regex = (regex_t *)
2238                         xmalloc(sizeof(regex_t));
2239                 if (regcomp(ecbdata->diff_words->word_regex,
2240                             o->word_regex,
2241                             REG_EXTENDED | REG_NEWLINE))
2242                         die("invalid regular expression: %s",
2243                             o->word_regex);
2244         }
2245         for (i = 0; i < ARRAY_SIZE(diff_words_styles); i++) {
2246                 if (o->word_diff == diff_words_styles[i].type) {
2247                         ecbdata->diff_words->style =
2248                                 &diff_words_styles[i];
2249                         break;
2250                 }
2251         }
2252         if (want_color(o->use_color)) {
2253                 struct diff_words_style *st = ecbdata->diff_words->style;
2254                 st->old_word.color = diff_get_color_opt(o, DIFF_FILE_OLD);
2255                 st->new_word.color = diff_get_color_opt(o, DIFF_FILE_NEW);
2256                 st->ctx.color = diff_get_color_opt(o, DIFF_CONTEXT);
2257         }
2258 }
2259
2260 static void free_diff_words_data(struct emit_callback *ecbdata)
2261 {
2262         if (ecbdata->diff_words) {
2263                 diff_words_flush(ecbdata);
2264                 free (ecbdata->diff_words->opt->emitted_symbols);
2265                 free (ecbdata->diff_words->opt);
2266                 free (ecbdata->diff_words->minus.text.ptr);
2267                 free (ecbdata->diff_words->minus.orig);
2268                 free (ecbdata->diff_words->plus.text.ptr);
2269                 free (ecbdata->diff_words->plus.orig);
2270                 if (ecbdata->diff_words->word_regex) {
2271                         regfree(ecbdata->diff_words->word_regex);
2272                         free(ecbdata->diff_words->word_regex);
2273                 }
2274                 FREE_AND_NULL(ecbdata->diff_words);
2275         }
2276 }
2277
2278 const char *diff_get_color(int diff_use_color, enum color_diff ix)
2279 {
2280         if (want_color(diff_use_color))
2281                 return diff_colors[ix];
2282         return "";
2283 }
2284
2285 const char *diff_line_prefix(struct diff_options *opt)
2286 {
2287         struct strbuf *msgbuf;
2288         if (!opt->output_prefix)
2289                 return "";
2290
2291         msgbuf = opt->output_prefix(opt, opt->output_prefix_data);
2292         return msgbuf->buf;
2293 }
2294
2295 static unsigned long sane_truncate_line(char *line, unsigned long len)
2296 {
2297         const char *cp;
2298         unsigned long allot;
2299         size_t l = len;
2300
2301         cp = line;
2302         allot = l;
2303         while (0 < l) {
2304                 (void) utf8_width(&cp, &l);
2305                 if (!cp)
2306                         break; /* truncated in the middle? */
2307         }
2308         return allot - l;
2309 }
2310
2311 static void find_lno(const char *line, struct emit_callback *ecbdata)
2312 {
2313         const char *p;
2314         ecbdata->lno_in_preimage = 0;
2315         ecbdata->lno_in_postimage = 0;
2316         p = strchr(line, '-');
2317         if (!p)
2318                 return; /* cannot happen */
2319         ecbdata->lno_in_preimage = strtol(p + 1, NULL, 10);
2320         p = strchr(p, '+');
2321         if (!p)
2322                 return; /* cannot happen */
2323         ecbdata->lno_in_postimage = strtol(p + 1, NULL, 10);
2324 }
2325
2326 static void fn_out_consume(void *priv, char *line, unsigned long len)
2327 {
2328         struct emit_callback *ecbdata = priv;
2329         struct diff_options *o = ecbdata->opt;
2330
2331         o->found_changes = 1;
2332
2333         if (ecbdata->header) {
2334                 emit_diff_symbol(o, DIFF_SYMBOL_HEADER,
2335                                  ecbdata->header->buf, ecbdata->header->len, 0);
2336                 strbuf_reset(ecbdata->header);
2337                 ecbdata->header = NULL;
2338         }
2339
2340         if (ecbdata->label_path[0]) {
2341                 emit_diff_symbol(o, DIFF_SYMBOL_FILEPAIR_MINUS,
2342                                  ecbdata->label_path[0],
2343                                  strlen(ecbdata->label_path[0]), 0);
2344                 emit_diff_symbol(o, DIFF_SYMBOL_FILEPAIR_PLUS,
2345                                  ecbdata->label_path[1],
2346                                  strlen(ecbdata->label_path[1]), 0);
2347                 ecbdata->label_path[0] = ecbdata->label_path[1] = NULL;
2348         }
2349
2350         if (diff_suppress_blank_empty
2351             && len == 2 && line[0] == ' ' && line[1] == '\n') {
2352                 line[0] = '\n';
2353                 len = 1;
2354         }
2355
2356         if (line[0] == '@') {
2357                 if (ecbdata->diff_words)
2358                         diff_words_flush(ecbdata);
2359                 len = sane_truncate_line(line, len);
2360                 find_lno(line, ecbdata);
2361                 emit_hunk_header(ecbdata, line, len);
2362                 return;
2363         }
2364
2365         if (ecbdata->diff_words) {
2366                 enum diff_symbol s =
2367                         ecbdata->diff_words->type == DIFF_WORDS_PORCELAIN ?
2368                         DIFF_SYMBOL_WORDS_PORCELAIN : DIFF_SYMBOL_WORDS;
2369                 if (line[0] == '-') {
2370                         diff_words_append(line, len,
2371                                           &ecbdata->diff_words->minus);
2372                         return;
2373                 } else if (line[0] == '+') {
2374                         diff_words_append(line, len,
2375                                           &ecbdata->diff_words->plus);
2376                         return;
2377                 } else if (starts_with(line, "\\ ")) {
2378                         /*
2379                          * Eat the "no newline at eof" marker as if we
2380                          * saw a "+" or "-" line with nothing on it,
2381                          * and return without diff_words_flush() to
2382                          * defer processing. If this is the end of
2383                          * preimage, more "+" lines may come after it.
2384                          */
2385                         return;
2386                 }
2387                 diff_words_flush(ecbdata);
2388                 emit_diff_symbol(o, s, line, len, 0);
2389                 return;
2390         }
2391
2392         switch (line[0]) {
2393         case '+':
2394                 ecbdata->lno_in_postimage++;
2395                 emit_add_line(ecbdata, line + 1, len - 1);
2396                 break;
2397         case '-':
2398                 ecbdata->lno_in_preimage++;
2399                 emit_del_line(ecbdata, line + 1, len - 1);
2400                 break;
2401         case ' ':
2402                 ecbdata->lno_in_postimage++;
2403                 ecbdata->lno_in_preimage++;
2404                 emit_context_line(ecbdata, line + 1, len - 1);
2405                 break;
2406         default:
2407                 /* incomplete line at the end */
2408                 ecbdata->lno_in_preimage++;
2409                 emit_diff_symbol(o, DIFF_SYMBOL_CONTEXT_INCOMPLETE,
2410                                  line, len, 0);
2411                 break;
2412         }
2413 }
2414
2415 static void pprint_rename(struct strbuf *name, const char *a, const char *b)
2416 {
2417         const char *old_name = a;
2418         const char *new_name = b;
2419         int pfx_length, sfx_length;
2420         int pfx_adjust_for_slash;
2421         int len_a = strlen(a);
2422         int len_b = strlen(b);
2423         int a_midlen, b_midlen;
2424         int qlen_a = quote_c_style(a, NULL, NULL, 0);
2425         int qlen_b = quote_c_style(b, NULL, NULL, 0);
2426
2427         if (qlen_a || qlen_b) {
2428                 quote_c_style(a, name, NULL, 0);
2429                 strbuf_addstr(name, " => ");
2430                 quote_c_style(b, name, NULL, 0);
2431                 return;
2432         }
2433
2434         /* Find common prefix */
2435         pfx_length = 0;
2436         while (*old_name && *new_name && *old_name == *new_name) {
2437                 if (*old_name == '/')
2438                         pfx_length = old_name - a + 1;
2439                 old_name++;
2440                 new_name++;
2441         }
2442
2443         /* Find common suffix */
2444         old_name = a + len_a;
2445         new_name = b + len_b;
2446         sfx_length = 0;
2447         /*
2448          * If there is a common prefix, it must end in a slash.  In
2449          * that case we let this loop run 1 into the prefix to see the
2450          * same slash.
2451          *
2452          * If there is no common prefix, we cannot do this as it would
2453          * underrun the input strings.
2454          */
2455         pfx_adjust_for_slash = (pfx_length ? 1 : 0);
2456         while (a + pfx_length - pfx_adjust_for_slash <= old_name &&
2457                b + pfx_length - pfx_adjust_for_slash <= new_name &&
2458                *old_name == *new_name) {
2459                 if (*old_name == '/')
2460                         sfx_length = len_a - (old_name - a);
2461                 old_name--;
2462                 new_name--;
2463         }
2464
2465         /*
2466          * pfx{mid-a => mid-b}sfx
2467          * {pfx-a => pfx-b}sfx
2468          * pfx{sfx-a => sfx-b}
2469          * name-a => name-b
2470          */
2471         a_midlen = len_a - pfx_length - sfx_length;
2472         b_midlen = len_b - pfx_length - sfx_length;
2473         if (a_midlen < 0)
2474                 a_midlen = 0;
2475         if (b_midlen < 0)
2476                 b_midlen = 0;
2477
2478         strbuf_grow(name, pfx_length + a_midlen + b_midlen + sfx_length + 7);
2479         if (pfx_length + sfx_length) {
2480                 strbuf_add(name, a, pfx_length);
2481                 strbuf_addch(name, '{');
2482         }
2483         strbuf_add(name, a + pfx_length, a_midlen);
2484         strbuf_addstr(name, " => ");
2485         strbuf_add(name, b + pfx_length, b_midlen);
2486         if (pfx_length + sfx_length) {
2487                 strbuf_addch(name, '}');
2488                 strbuf_add(name, a + len_a - sfx_length, sfx_length);
2489         }
2490 }
2491
2492 struct diffstat_t {
2493         int nr;
2494         int alloc;
2495         struct diffstat_file {
2496                 char *from_name;
2497                 char *name;
2498                 char *print_name;
2499                 const char *comments;
2500                 unsigned is_unmerged:1;
2501                 unsigned is_binary:1;
2502                 unsigned is_renamed:1;
2503                 unsigned is_interesting:1;
2504                 uintmax_t added, deleted;
2505         } **files;
2506 };
2507
2508 static struct diffstat_file *diffstat_add(struct diffstat_t *diffstat,
2509                                           const char *name_a,
2510                                           const char *name_b)
2511 {
2512         struct diffstat_file *x;
2513         x = xcalloc(1, sizeof(*x));
2514         ALLOC_GROW(diffstat->files, diffstat->nr + 1, diffstat->alloc);
2515         diffstat->files[diffstat->nr++] = x;
2516         if (name_b) {
2517                 x->from_name = xstrdup(name_a);
2518                 x->name = xstrdup(name_b);
2519                 x->is_renamed = 1;
2520         }
2521         else {
2522                 x->from_name = NULL;
2523                 x->name = xstrdup(name_a);
2524         }
2525         return x;
2526 }
2527
2528 static void diffstat_consume(void *priv, char *line, unsigned long len)
2529 {
2530         struct diffstat_t *diffstat = priv;
2531         struct diffstat_file *x = diffstat->files[diffstat->nr - 1];
2532
2533         if (line[0] == '+')
2534                 x->added++;
2535         else if (line[0] == '-')
2536                 x->deleted++;
2537 }
2538
2539 const char mime_boundary_leader[] = "------------";
2540
2541 static int scale_linear(int it, int width, int max_change)
2542 {
2543         if (!it)
2544                 return 0;
2545         /*
2546          * make sure that at least one '-' or '+' is printed if
2547          * there is any change to this path. The easiest way is to
2548          * scale linearly as if the alloted width is one column shorter
2549          * than it is, and then add 1 to the result.
2550          */
2551         return 1 + (it * (width - 1) / max_change);
2552 }
2553
2554 static void show_graph(struct strbuf *out, char ch, int cnt,
2555                        const char *set, const char *reset)
2556 {
2557         if (cnt <= 0)
2558                 return;
2559         strbuf_addstr(out, set);
2560         strbuf_addchars(out, ch, cnt);
2561         strbuf_addstr(out, reset);
2562 }
2563
2564 static void fill_print_name(struct diffstat_file *file)
2565 {
2566         struct strbuf pname = STRBUF_INIT;
2567
2568         if (file->print_name)
2569                 return;
2570
2571         if (file->is_renamed)
2572                 pprint_rename(&pname, file->from_name, file->name);
2573         else
2574                 quote_c_style(file->name, &pname, NULL, 0);
2575
2576         if (file->comments)
2577                 strbuf_addf(&pname, " (%s)", file->comments);
2578
2579         file->print_name = strbuf_detach(&pname, NULL);
2580 }
2581
2582 static void print_stat_summary_inserts_deletes(struct diff_options *options,
2583                 int files, int insertions, int deletions)
2584 {
2585         struct strbuf sb = STRBUF_INIT;
2586
2587         if (!files) {
2588                 assert(insertions == 0 && deletions == 0);
2589                 emit_diff_symbol(options, DIFF_SYMBOL_STATS_SUMMARY_NO_FILES,
2590                                  NULL, 0, 0);
2591                 return;
2592         }
2593
2594         strbuf_addf(&sb,
2595                     (files == 1) ? " %d file changed" : " %d files changed",
2596                     files);
2597
2598         /*
2599          * For binary diff, the caller may want to print "x files
2600          * changed" with insertions == 0 && deletions == 0.
2601          *
2602          * Not omitting "0 insertions(+), 0 deletions(-)" in this case
2603          * is probably less confusing (i.e skip over "2 files changed
2604          * but nothing about added/removed lines? Is this a bug in Git?").
2605          */
2606         if (insertions || deletions == 0) {
2607                 strbuf_addf(&sb,
2608                             (insertions == 1) ? ", %d insertion(+)" : ", %d insertions(+)",
2609                             insertions);
2610         }
2611
2612         if (deletions || insertions == 0) {
2613                 strbuf_addf(&sb,
2614                             (deletions == 1) ? ", %d deletion(-)" : ", %d deletions(-)",
2615                             deletions);
2616         }
2617         strbuf_addch(&sb, '\n');
2618         emit_diff_symbol(options, DIFF_SYMBOL_STATS_SUMMARY_INSERTS_DELETES,
2619                          sb.buf, sb.len, 0);
2620         strbuf_release(&sb);
2621 }
2622
2623 void print_stat_summary(FILE *fp, int files,
2624                         int insertions, int deletions)
2625 {
2626         struct diff_options o;
2627         memset(&o, 0, sizeof(o));
2628         o.file = fp;
2629
2630         print_stat_summary_inserts_deletes(&o, files, insertions, deletions);
2631 }
2632
2633 static void show_stats(struct diffstat_t *data, struct diff_options *options)
2634 {
2635         int i, len, add, del, adds = 0, dels = 0;
2636         uintmax_t max_change = 0, max_len = 0;
2637         int total_files = data->nr, count;
2638         int width, name_width, graph_width, number_width = 0, bin_width = 0;
2639         const char *reset, *add_c, *del_c;
2640         int extra_shown = 0;
2641         const char *line_prefix = diff_line_prefix(options);
2642         struct strbuf out = STRBUF_INIT;
2643
2644         if (data->nr == 0)
2645                 return;
2646
2647         count = options->stat_count ? options->stat_count : data->nr;
2648
2649         reset = diff_get_color_opt(options, DIFF_RESET);
2650         add_c = diff_get_color_opt(options, DIFF_FILE_NEW);
2651         del_c = diff_get_color_opt(options, DIFF_FILE_OLD);
2652
2653         /*
2654          * Find the longest filename and max number of changes
2655          */
2656         for (i = 0; (i < count) && (i < data->nr); i++) {
2657                 struct diffstat_file *file = data->files[i];
2658                 uintmax_t change = file->added + file->deleted;
2659
2660                 if (!file->is_interesting && (change == 0)) {
2661                         count++; /* not shown == room for one more */
2662                         continue;
2663                 }
2664                 fill_print_name(file);
2665                 len = strlen(file->print_name);
2666                 if (max_len < len)
2667                         max_len = len;
2668
2669                 if (file->is_unmerged) {
2670                         /* "Unmerged" is 8 characters */
2671                         bin_width = bin_width < 8 ? 8 : bin_width;
2672                         continue;
2673                 }
2674                 if (file->is_binary) {
2675                         /* "Bin XXX -> YYY bytes" */
2676                         int w = 14 + decimal_width(file->added)
2677                                 + decimal_width(file->deleted);
2678                         bin_width = bin_width < w ? w : bin_width;
2679                         /* Display change counts aligned with "Bin" */
2680                         number_width = 3;
2681                         continue;
2682                 }
2683
2684                 if (max_change < change)
2685                         max_change = change;
2686         }
2687         count = i; /* where we can stop scanning in data->files[] */
2688
2689         /*
2690          * We have width = stat_width or term_columns() columns total.
2691          * We want a maximum of min(max_len, stat_name_width) for the name part.
2692          * We want a maximum of min(max_change, stat_graph_width) for the +- part.
2693          * We also need 1 for " " and 4 + decimal_width(max_change)
2694          * for " | NNNN " and one the empty column at the end, altogether
2695          * 6 + decimal_width(max_change).
2696          *
2697          * If there's not enough space, we will use the smaller of
2698          * stat_name_width (if set) and 5/8*width for the filename,
2699          * and the rest for constant elements + graph part, but no more
2700          * than stat_graph_width for the graph part.
2701          * (5/8 gives 50 for filename and 30 for the constant parts + graph
2702          * for the standard terminal size).
2703          *
2704          * In other words: stat_width limits the maximum width, and
2705          * stat_name_width fixes the maximum width of the filename,
2706          * and is also used to divide available columns if there
2707          * aren't enough.
2708          *
2709          * Binary files are displayed with "Bin XXX -> YYY bytes"
2710          * instead of the change count and graph. This part is treated
2711          * similarly to the graph part, except that it is not
2712          * "scaled". If total width is too small to accommodate the
2713          * guaranteed minimum width of the filename part and the
2714          * separators and this message, this message will "overflow"
2715          * making the line longer than the maximum width.
2716          */
2717
2718         if (options->stat_width == -1)
2719                 width = term_columns() - strlen(line_prefix);
2720         else
2721                 width = options->stat_width ? options->stat_width : 80;
2722         number_width = decimal_width(max_change) > number_width ?
2723                 decimal_width(max_change) : number_width;
2724
2725         if (options->stat_graph_width == -1)
2726                 options->stat_graph_width = diff_stat_graph_width;
2727
2728         /*
2729          * Guarantee 3/8*16==6 for the graph part
2730          * and 5/8*16==10 for the filename part
2731          */
2732         if (width < 16 + 6 + number_width)
2733                 width = 16 + 6 + number_width;
2734
2735         /*
2736          * First assign sizes that are wanted, ignoring available width.
2737          * strlen("Bin XXX -> YYY bytes") == bin_width, and the part
2738          * starting from "XXX" should fit in graph_width.
2739          */
2740         graph_width = max_change + 4 > bin_width ? max_change : bin_width - 4;
2741         if (options->stat_graph_width &&
2742             options->stat_graph_width < graph_width)
2743                 graph_width = options->stat_graph_width;
2744
2745         name_width = (options->stat_name_width > 0 &&
2746                       options->stat_name_width < max_len) ?
2747                 options->stat_name_width : max_len;
2748
2749         /*
2750          * Adjust adjustable widths not to exceed maximum width
2751          */
2752         if (name_width + number_width + 6 + graph_width > width) {
2753                 if (graph_width > width * 3/8 - number_width - 6) {
2754                         graph_width = width * 3/8 - number_width - 6;
2755                         if (graph_width < 6)
2756                                 graph_width = 6;
2757                 }
2758
2759                 if (options->stat_graph_width &&
2760                     graph_width > options->stat_graph_width)
2761                         graph_width = options->stat_graph_width;
2762                 if (name_width > width - number_width - 6 - graph_width)
2763                         name_width = width - number_width - 6 - graph_width;
2764                 else
2765                         graph_width = width - number_width - 6 - name_width;
2766         }
2767
2768         /*
2769          * From here name_width is the width of the name area,
2770          * and graph_width is the width of the graph area.
2771          * max_change is used to scale graph properly.
2772          */
2773         for (i = 0; i < count; i++) {
2774                 const char *prefix = "";
2775                 struct diffstat_file *file = data->files[i];
2776                 char *name = file->print_name;
2777                 uintmax_t added = file->added;
2778                 uintmax_t deleted = file->deleted;
2779                 int name_len;
2780
2781                 if (!file->is_interesting && (added + deleted == 0))
2782                         continue;
2783
2784                 /*
2785                  * "scale" the filename
2786                  */
2787                 len = name_width;
2788                 name_len = strlen(name);
2789                 if (name_width < name_len) {
2790                         char *slash;
2791                         prefix = "...";
2792                         len -= 3;
2793                         name += name_len - len;
2794                         slash = strchr(name, '/');
2795                         if (slash)
2796                                 name = slash;
2797                 }
2798
2799                 if (file->is_binary) {
2800                         strbuf_addf(&out, " %s%-*s |", prefix, len, name);
2801                         strbuf_addf(&out, " %*s", number_width, "Bin");
2802                         if (!added && !deleted) {
2803                                 strbuf_addch(&out, '\n');
2804                                 emit_diff_symbol(options, DIFF_SYMBOL_STATS_LINE,
2805                                                  out.buf, out.len, 0);
2806                                 strbuf_reset(&out);
2807                                 continue;
2808                         }
2809                         strbuf_addf(&out, " %s%"PRIuMAX"%s",
2810                                 del_c, deleted, reset);
2811                         strbuf_addstr(&out, " -> ");
2812                         strbuf_addf(&out, "%s%"PRIuMAX"%s",
2813                                 add_c, added, reset);
2814                         strbuf_addstr(&out, " bytes\n");
2815                         emit_diff_symbol(options, DIFF_SYMBOL_STATS_LINE,
2816                                          out.buf, out.len, 0);
2817                         strbuf_reset(&out);
2818                         continue;
2819                 }
2820                 else if (file->is_unmerged) {
2821                         strbuf_addf(&out, " %s%-*s |", prefix, len, name);
2822                         strbuf_addstr(&out, " Unmerged\n");
2823                         emit_diff_symbol(options, DIFF_SYMBOL_STATS_LINE,
2824                                          out.buf, out.len, 0);
2825                         strbuf_reset(&out);
2826                         continue;
2827                 }
2828
2829                 /*
2830                  * scale the add/delete
2831                  */
2832                 add = added;
2833                 del = deleted;
2834
2835                 if (graph_width <= max_change) {
2836                         int total = scale_linear(add + del, graph_width, max_change);
2837                         if (total < 2 && add && del)
2838                                 /* width >= 2 due to the sanity check */
2839                                 total = 2;
2840                         if (add < del) {
2841                                 add = scale_linear(add, graph_width, max_change);
2842                                 del = total - add;
2843                         } else {
2844                                 del = scale_linear(del, graph_width, max_change);
2845                                 add = total - del;
2846                         }
2847                 }
2848                 strbuf_addf(&out, " %s%-*s |", prefix, len, name);
2849                 strbuf_addf(&out, " %*"PRIuMAX"%s",
2850                         number_width, added + deleted,
2851                         added + deleted ? " " : "");
2852                 show_graph(&out, '+', add, add_c, reset);
2853                 show_graph(&out, '-', del, del_c, reset);
2854                 strbuf_addch(&out, '\n');
2855                 emit_diff_symbol(options, DIFF_SYMBOL_STATS_LINE,
2856                                  out.buf, out.len, 0);
2857                 strbuf_reset(&out);
2858         }
2859
2860         for (i = 0; i < data->nr; i++) {
2861                 struct diffstat_file *file = data->files[i];
2862                 uintmax_t added = file->added;
2863                 uintmax_t deleted = file->deleted;
2864
2865                 if (file->is_unmerged ||
2866                     (!file->is_interesting && (added + deleted == 0))) {
2867                         total_files--;
2868                         continue;
2869                 }
2870
2871                 if (!file->is_binary) {
2872                         adds += added;
2873                         dels += deleted;
2874                 }
2875                 if (i < count)
2876                         continue;
2877                 if (!extra_shown)
2878                         emit_diff_symbol(options,
2879                                          DIFF_SYMBOL_STATS_SUMMARY_ABBREV,
2880                                          NULL, 0, 0);
2881                 extra_shown = 1;
2882         }
2883
2884         print_stat_summary_inserts_deletes(options, total_files, adds, dels);
2885         strbuf_release(&out);
2886 }
2887
2888 static void show_shortstats(struct diffstat_t *data, struct diff_options *options)
2889 {
2890         int i, adds = 0, dels = 0, total_files = data->nr;
2891
2892         if (data->nr == 0)
2893                 return;
2894
2895         for (i = 0; i < data->nr; i++) {
2896                 int added = data->files[i]->added;
2897                 int deleted = data->files[i]->deleted;
2898
2899                 if (data->files[i]->is_unmerged ||
2900                     (!data->files[i]->is_interesting && (added + deleted == 0))) {
2901                         total_files--;
2902                 } else if (!data->files[i]->is_binary) { /* don't count bytes */
2903                         adds += added;
2904                         dels += deleted;
2905                 }
2906         }
2907         print_stat_summary_inserts_deletes(options, total_files, adds, dels);
2908 }
2909
2910 static void show_numstat(struct diffstat_t *data, struct diff_options *options)
2911 {
2912         int i;
2913
2914         if (data->nr == 0)
2915                 return;
2916
2917         for (i = 0; i < data->nr; i++) {
2918                 struct diffstat_file *file = data->files[i];
2919
2920                 fprintf(options->file, "%s", diff_line_prefix(options));
2921
2922                 if (file->is_binary)
2923                         fprintf(options->file, "-\t-\t");
2924                 else
2925                         fprintf(options->file,
2926                                 "%"PRIuMAX"\t%"PRIuMAX"\t",
2927                                 file->added, file->deleted);
2928                 if (options->line_termination) {
2929                         fill_print_name(file);
2930                         if (!file->is_renamed)
2931                                 write_name_quoted(file->name, options->file,
2932                                                   options->line_termination);
2933                         else {
2934                                 fputs(file->print_name, options->file);
2935                                 putc(options->line_termination, options->file);
2936                         }
2937                 } else {
2938                         if (file->is_renamed) {
2939                                 putc('\0', options->file);
2940                                 write_name_quoted(file->from_name, options->file, '\0');
2941                         }
2942                         write_name_quoted(file->name, options->file, '\0');
2943                 }
2944         }
2945 }
2946
2947 struct dirstat_file {
2948         const char *name;
2949         unsigned long changed;
2950 };
2951
2952 struct dirstat_dir {
2953         struct dirstat_file *files;
2954         int alloc, nr, permille, cumulative;
2955 };
2956
2957 static long gather_dirstat(struct diff_options *opt, struct dirstat_dir *dir,
2958                 unsigned long changed, const char *base, int baselen)
2959 {
2960         unsigned long sum_changes = 0;
2961         unsigned int sources = 0;
2962         const char *line_prefix = diff_line_prefix(opt);
2963
2964         while (dir->nr) {
2965                 struct dirstat_file *f = dir->files;
2966                 int namelen = strlen(f->name);
2967                 unsigned long changes;
2968                 char *slash;
2969
2970                 if (namelen < baselen)
2971                         break;
2972                 if (memcmp(f->name, base, baselen))
2973                         break;
2974                 slash = strchr(f->name + baselen, '/');
2975                 if (slash) {
2976                         int newbaselen = slash + 1 - f->name;
2977                         changes = gather_dirstat(opt, dir, changed, f->name, newbaselen);
2978                         sources++;
2979                 } else {
2980                         changes = f->changed;
2981                         dir->files++;
2982                         dir->nr--;
2983                         sources += 2;
2984                 }
2985                 sum_changes += changes;
2986         }
2987
2988         /*
2989          * We don't report dirstat's for
2990          *  - the top level
2991          *  - or cases where everything came from a single directory
2992          *    under this directory (sources == 1).
2993          */
2994         if (baselen && sources != 1) {
2995                 if (sum_changes) {
2996                         int permille = sum_changes * 1000 / changed;
2997                         if (permille >= dir->permille) {
2998                                 fprintf(opt->file, "%s%4d.%01d%% %.*s\n", line_prefix,
2999                                         permille / 10, permille % 10, baselen, base);
3000                                 if (!dir->cumulative)
3001                                         return 0;
3002                         }
3003                 }
3004         }
3005         return sum_changes;
3006 }
3007
3008 static int dirstat_compare(const void *_a, const void *_b)
3009 {
3010         const struct dirstat_file *a = _a;
3011         const struct dirstat_file *b = _b;
3012         return strcmp(a->name, b->name);
3013 }
3014
3015 static void show_dirstat(struct diff_options *options)
3016 {
3017         int i;
3018         unsigned long changed;
3019         struct dirstat_dir dir;
3020         struct diff_queue_struct *q = &diff_queued_diff;
3021
3022         dir.files = NULL;
3023         dir.alloc = 0;
3024         dir.nr = 0;
3025         dir.permille = options->dirstat_permille;
3026         dir.cumulative = options->flags.dirstat_cumulative;
3027
3028         changed = 0;
3029         for (i = 0; i < q->nr; i++) {
3030                 struct diff_filepair *p = q->queue[i];
3031                 const char *name;
3032                 unsigned long copied, added, damage;
3033
3034                 name = p->two->path ? p->two->path : p->one->path;
3035
3036                 if (p->one->oid_valid && p->two->oid_valid &&
3037                     oideq(&p->one->oid, &p->two->oid)) {
3038                         /*
3039                          * The SHA1 has not changed, so pre-/post-content is
3040                          * identical. We can therefore skip looking at the
3041                          * file contents altogether.
3042                          */
3043                         damage = 0;
3044                         goto found_damage;
3045                 }
3046
3047                 if (options->flags.dirstat_by_file) {
3048                         /*
3049                          * In --dirstat-by-file mode, we don't really need to
3050                          * look at the actual file contents at all.
3051                          * The fact that the SHA1 changed is enough for us to
3052                          * add this file to the list of results
3053                          * (with each file contributing equal damage).
3054                          */
3055                         damage = 1;
3056                         goto found_damage;
3057                 }
3058
3059                 if (DIFF_FILE_VALID(p->one) && DIFF_FILE_VALID(p->two)) {
3060                         diff_populate_filespec(options->repo, p->one, 0);
3061                         diff_populate_filespec(options->repo, p->two, 0);
3062                         diffcore_count_changes(options->repo,
3063                                                p->one, p->two, NULL, NULL,
3064                                                &copied, &added);
3065                         diff_free_filespec_data(p->one);
3066                         diff_free_filespec_data(p->two);
3067                 } else if (DIFF_FILE_VALID(p->one)) {
3068                         diff_populate_filespec(options->repo, p->one, CHECK_SIZE_ONLY);
3069                         copied = added = 0;
3070                         diff_free_filespec_data(p->one);
3071                 } else if (DIFF_FILE_VALID(p->two)) {
3072                         diff_populate_filespec(options->repo, p->two, CHECK_SIZE_ONLY);
3073                         copied = 0;
3074                         added = p->two->size;
3075                         diff_free_filespec_data(p->two);
3076                 } else
3077                         continue;
3078
3079                 /*
3080                  * Original minus copied is the removed material,
3081                  * added is the new material.  They are both damages
3082                  * made to the preimage.
3083                  * If the resulting damage is zero, we know that
3084                  * diffcore_count_changes() considers the two entries to
3085                  * be identical, but since the oid changed, we
3086                  * know that there must have been _some_ kind of change,
3087                  * so we force all entries to have damage > 0.
3088                  */
3089                 damage = (p->one->size - copied) + added;
3090                 if (!damage)
3091                         damage = 1;
3092
3093 found_damage:
3094                 ALLOC_GROW(dir.files, dir.nr + 1, dir.alloc);
3095                 dir.files[dir.nr].name = name;
3096                 dir.files[dir.nr].changed = damage;
3097                 changed += damage;
3098                 dir.nr++;
3099         }
3100
3101         /* This can happen even with many files, if everything was renames */
3102         if (!changed)
3103                 return;
3104
3105         /* Show all directories with more than x% of the changes */
3106         QSORT(dir.files, dir.nr, dirstat_compare);
3107         gather_dirstat(options, &dir, changed, "", 0);
3108 }
3109
3110 static void show_dirstat_by_line(struct diffstat_t *data, struct diff_options *options)
3111 {
3112         int i;
3113         unsigned long changed;
3114         struct dirstat_dir dir;
3115
3116         if (data->nr == 0)
3117                 return;
3118
3119         dir.files = NULL;
3120         dir.alloc = 0;
3121         dir.nr = 0;
3122         dir.permille = options->dirstat_permille;
3123         dir.cumulative = options->flags.dirstat_cumulative;
3124
3125         changed = 0;
3126         for (i = 0; i < data->nr; i++) {
3127                 struct diffstat_file *file = data->files[i];
3128                 unsigned long damage = file->added + file->deleted;
3129                 if (file->is_binary)
3130                         /*
3131                          * binary files counts bytes, not lines. Must find some
3132                          * way to normalize binary bytes vs. textual lines.
3133                          * The following heuristic assumes that there are 64
3134                          * bytes per "line".
3135                          * This is stupid and ugly, but very cheap...
3136                          */
3137                         damage = DIV_ROUND_UP(damage, 64);
3138                 ALLOC_GROW(dir.files, dir.nr + 1, dir.alloc);
3139                 dir.files[dir.nr].name = file->name;
3140                 dir.files[dir.nr].changed = damage;
3141                 changed += damage;
3142                 dir.nr++;
3143         }
3144
3145         /* This can happen even with many files, if everything was renames */
3146         if (!changed)
3147                 return;
3148
3149         /* Show all directories with more than x% of the changes */
3150         QSORT(dir.files, dir.nr, dirstat_compare);
3151         gather_dirstat(options, &dir, changed, "", 0);
3152 }
3153
3154 static void free_diffstat_info(struct diffstat_t *diffstat)
3155 {
3156         int i;
3157         for (i = 0; i < diffstat->nr; i++) {
3158                 struct diffstat_file *f = diffstat->files[i];
3159                 free(f->print_name);
3160                 free(f->name);
3161                 free(f->from_name);
3162                 free(f);
3163         }
3164         free(diffstat->files);
3165 }
3166
3167 struct checkdiff_t {
3168         const char *filename;
3169         int lineno;
3170         int conflict_marker_size;
3171         struct diff_options *o;
3172         unsigned ws_rule;
3173         unsigned status;
3174 };
3175
3176 static int is_conflict_marker(const char *line, int marker_size, unsigned long len)
3177 {
3178         char firstchar;
3179         int cnt;
3180
3181         if (len < marker_size + 1)
3182                 return 0;
3183         firstchar = line[0];
3184         switch (firstchar) {
3185         case '=': case '>': case '<': case '|':
3186                 break;
3187         default:
3188                 return 0;
3189         }
3190         for (cnt = 1; cnt < marker_size; cnt++)
3191                 if (line[cnt] != firstchar)
3192                         return 0;
3193         /* line[1] thru line[marker_size-1] are same as firstchar */
3194         if (len < marker_size + 1 || !isspace(line[marker_size]))
3195                 return 0;
3196         return 1;
3197 }
3198
3199 static void checkdiff_consume_hunk(void *priv,
3200                                    long ob, long on, long nb, long nn,
3201                                    const char *func, long funclen)
3202
3203 {
3204         struct checkdiff_t *data = priv;
3205         data->lineno = nb - 1;
3206 }
3207
3208 static void checkdiff_consume(void *priv, char *line, unsigned long len)
3209 {
3210         struct checkdiff_t *data = priv;
3211         int marker_size = data->conflict_marker_size;
3212         const char *ws = diff_get_color(data->o->use_color, DIFF_WHITESPACE);
3213         const char *reset = diff_get_color(data->o->use_color, DIFF_RESET);
3214         const char *set = diff_get_color(data->o->use_color, DIFF_FILE_NEW);
3215         char *err;
3216         const char *line_prefix;
3217
3218         assert(data->o);
3219         line_prefix = diff_line_prefix(data->o);
3220
3221         if (line[0] == '+') {
3222                 unsigned bad;
3223                 data->lineno++;
3224                 if (is_conflict_marker(line + 1, marker_size, len - 1)) {
3225                         data->status |= 1;
3226                         fprintf(data->o->file,
3227                                 "%s%s:%d: leftover conflict marker\n",
3228                                 line_prefix, data->filename, data->lineno);
3229                 }
3230                 bad = ws_check(line + 1, len - 1, data->ws_rule);
3231                 if (!bad)
3232                         return;
3233                 data->status |= bad;
3234                 err = whitespace_error_string(bad);
3235                 fprintf(data->o->file, "%s%s:%d: %s.\n",
3236                         line_prefix, data->filename, data->lineno, err);
3237                 free(err);
3238                 emit_line(data->o, set, reset, line, 1);
3239                 ws_check_emit(line + 1, len - 1, data->ws_rule,
3240                               data->o->file, set, reset, ws);
3241         } else if (line[0] == ' ') {
3242                 data->lineno++;
3243         }
3244 }
3245
3246 static unsigned char *deflate_it(char *data,
3247                                  unsigned long size,
3248                                  unsigned long *result_size)
3249 {
3250         int bound;
3251         unsigned char *deflated;
3252         git_zstream stream;
3253
3254         git_deflate_init(&stream, zlib_compression_level);
3255         bound = git_deflate_bound(&stream, size);
3256         deflated = xmalloc(bound);
3257         stream.next_out = deflated;
3258         stream.avail_out = bound;
3259
3260         stream.next_in = (unsigned char *)data;
3261         stream.avail_in = size;
3262         while (git_deflate(&stream, Z_FINISH) == Z_OK)
3263                 ; /* nothing */
3264         git_deflate_end(&stream);
3265         *result_size = stream.total_out;
3266         return deflated;
3267 }
3268
3269 static void emit_binary_diff_body(struct diff_options *o,
3270                                   mmfile_t *one, mmfile_t *two)
3271 {
3272         void *cp;
3273         void *delta;
3274         void *deflated;
3275         void *data;
3276         unsigned long orig_size;
3277         unsigned long delta_size;
3278         unsigned long deflate_size;
3279         unsigned long data_size;
3280
3281         /* We could do deflated delta, or we could do just deflated two,
3282          * whichever is smaller.
3283          */
3284         delta = NULL;
3285         deflated = deflate_it(two->ptr, two->size, &deflate_size);
3286         if (one->size && two->size) {
3287                 delta = diff_delta(one->ptr, one->size,
3288                                    two->ptr, two->size,
3289                                    &delta_size, deflate_size);
3290                 if (delta) {
3291                         void *to_free = delta;
3292                         orig_size = delta_size;
3293                         delta = deflate_it(delta, delta_size, &delta_size);
3294                         free(to_free);
3295                 }
3296         }
3297
3298         if (delta && delta_size < deflate_size) {
3299                 char *s = xstrfmt("%"PRIuMAX , (uintmax_t)orig_size);
3300                 emit_diff_symbol(o, DIFF_SYMBOL_BINARY_DIFF_HEADER_DELTA,
3301                                  s, strlen(s), 0);
3302                 free(s);
3303                 free(deflated);
3304                 data = delta;
3305                 data_size = delta_size;
3306         } else {
3307                 char *s = xstrfmt("%lu", two->size);
3308                 emit_diff_symbol(o, DIFF_SYMBOL_BINARY_DIFF_HEADER_LITERAL,
3309                                  s, strlen(s), 0);
3310                 free(s);
3311                 free(delta);
3312                 data = deflated;
3313                 data_size = deflate_size;
3314         }
3315
3316         /* emit data encoded in base85 */
3317         cp = data;
3318         while (data_size) {
3319                 int len;
3320                 int bytes = (52 < data_size) ? 52 : data_size;
3321                 char line[71];
3322                 data_size -= bytes;
3323                 if (bytes <= 26)
3324                         line[0] = bytes + 'A' - 1;
3325                 else
3326                         line[0] = bytes - 26 + 'a' - 1;
3327                 encode_85(line + 1, cp, bytes);
3328                 cp = (char *) cp + bytes;
3329
3330                 len = strlen(line);
3331                 line[len++] = '\n';
3332                 line[len] = '\0';
3333
3334                 emit_diff_symbol(o, DIFF_SYMBOL_BINARY_DIFF_BODY,
3335                                  line, len, 0);
3336         }
3337         emit_diff_symbol(o, DIFF_SYMBOL_BINARY_DIFF_FOOTER, NULL, 0, 0);
3338         free(data);
3339 }
3340
3341 static void emit_binary_diff(struct diff_options *o,
3342                              mmfile_t *one, mmfile_t *two)
3343 {
3344         emit_diff_symbol(o, DIFF_SYMBOL_BINARY_DIFF_HEADER, NULL, 0, 0);
3345         emit_binary_diff_body(o, one, two);
3346         emit_binary_diff_body(o, two, one);
3347 }
3348
3349 int diff_filespec_is_binary(struct repository *r,
3350                             struct diff_filespec *one)
3351 {
3352         if (one->is_binary == -1) {
3353                 diff_filespec_load_driver(one, r->index);
3354                 if (one->driver->binary != -1)
3355                         one->is_binary = one->driver->binary;
3356                 else {
3357                         if (!one->data && DIFF_FILE_VALID(one))
3358                                 diff_populate_filespec(r, one, CHECK_BINARY);
3359                         if (one->is_binary == -1 && one->data)
3360                                 one->is_binary = buffer_is_binary(one->data,
3361                                                 one->size);
3362                         if (one->is_binary == -1)
3363                                 one->is_binary = 0;
3364                 }
3365         }
3366         return one->is_binary;
3367 }
3368
3369 static const struct userdiff_funcname *
3370 diff_funcname_pattern(struct diff_options *o, struct diff_filespec *one)
3371 {
3372         diff_filespec_load_driver(one, o->repo->index);
3373         return one->driver->funcname.pattern ? &one->driver->funcname : NULL;
3374 }
3375
3376 void diff_set_mnemonic_prefix(struct diff_options *options, const char *a, const char *b)
3377 {
3378         if (!options->a_prefix)
3379                 options->a_prefix = a;
3380         if (!options->b_prefix)
3381                 options->b_prefix = b;
3382 }
3383
3384 struct userdiff_driver *get_textconv(struct repository *r,
3385                                      struct diff_filespec *one)
3386 {
3387         if (!DIFF_FILE_VALID(one))
3388                 return NULL;
3389
3390         diff_filespec_load_driver(one, r->index);
3391         return userdiff_get_textconv(r, one->driver);
3392 }
3393
3394 static void builtin_diff(const char *name_a,
3395                          const char *name_b,
3396                          struct diff_filespec *one,
3397                          struct diff_filespec *two,
3398                          const char *xfrm_msg,
3399                          int must_show_header,
3400                          struct diff_options *o,
3401                          int complete_rewrite)
3402 {
3403         mmfile_t mf1, mf2;
3404         const char *lbl[2];
3405         char *a_one, *b_two;
3406         const char *meta = diff_get_color_opt(o, DIFF_METAINFO);
3407         const char *reset = diff_get_color_opt(o, DIFF_RESET);
3408         const char *a_prefix, *b_prefix;
3409         struct userdiff_driver *textconv_one = NULL;
3410         struct userdiff_driver *textconv_two = NULL;
3411         struct strbuf header = STRBUF_INIT;
3412         const char *line_prefix = diff_line_prefix(o);
3413
3414         diff_set_mnemonic_prefix(o, "a/", "b/");
3415         if (o->flags.reverse_diff) {
3416                 a_prefix = o->b_prefix;
3417                 b_prefix = o->a_prefix;
3418         } else {
3419                 a_prefix = o->a_prefix;
3420                 b_prefix = o->b_prefix;
3421         }
3422
3423         if (o->submodule_format == DIFF_SUBMODULE_LOG &&
3424             (!one->mode || S_ISGITLINK(one->mode)) &&
3425             (!two->mode || S_ISGITLINK(two->mode))) {
3426                 show_submodule_summary(o, one->path ? one->path : two->path,
3427                                 &one->oid, &two->oid,
3428                                 two->dirty_submodule);
3429                 return;
3430         } else if (o->submodule_format == DIFF_SUBMODULE_INLINE_DIFF &&
3431                    (!one->mode || S_ISGITLINK(one->mode)) &&
3432                    (!two->mode || S_ISGITLINK(two->mode))) {
3433                 show_submodule_inline_diff(o, one->path ? one->path : two->path,
3434                                 &one->oid, &two->oid,
3435                                 two->dirty_submodule);
3436                 return;
3437         }
3438
3439         if (o->flags.allow_textconv) {
3440                 textconv_one = get_textconv(o->repo, one);
3441                 textconv_two = get_textconv(o->repo, two);
3442         }
3443
3444         /* Never use a non-valid filename anywhere if at all possible */
3445         name_a = DIFF_FILE_VALID(one) ? name_a : name_b;
3446         name_b = DIFF_FILE_VALID(two) ? name_b : name_a;
3447
3448         a_one = quote_two(a_prefix, name_a + (*name_a == '/'));
3449         b_two = quote_two(b_prefix, name_b + (*name_b == '/'));
3450         lbl[0] = DIFF_FILE_VALID(one) ? a_one : "/dev/null";
3451         lbl[1] = DIFF_FILE_VALID(two) ? b_two : "/dev/null";
3452         strbuf_addf(&header, "%s%sdiff --git %s %s%s\n", line_prefix, meta, a_one, b_two, reset);
3453         if (lbl[0][0] == '/') {
3454                 /* /dev/null */
3455                 strbuf_addf(&header, "%s%snew file mode %06o%s\n", line_prefix, meta, two->mode, reset);
3456                 if (xfrm_msg)
3457                         strbuf_addstr(&header, xfrm_msg);
3458                 must_show_header = 1;
3459         }
3460         else if (lbl[1][0] == '/') {
3461                 strbuf_addf(&header, "%s%sdeleted file mode %06o%s\n", line_prefix, meta, one->mode, reset);
3462                 if (xfrm_msg)
3463                         strbuf_addstr(&header, xfrm_msg);
3464                 must_show_header = 1;
3465         }
3466         else {
3467                 if (one->mode != two->mode) {
3468                         strbuf_addf(&header, "%s%sold mode %06o%s\n", line_prefix, meta, one->mode, reset);
3469                         strbuf_addf(&header, "%s%snew mode %06o%s\n", line_prefix, meta, two->mode, reset);
3470                         must_show_header = 1;
3471                 }
3472                 if (xfrm_msg)
3473                         strbuf_addstr(&header, xfrm_msg);
3474
3475                 /*
3476                  * we do not run diff between different kind
3477                  * of objects.
3478                  */
3479                 if ((one->mode ^ two->mode) & S_IFMT)
3480                         goto free_ab_and_return;
3481                 if (complete_rewrite &&
3482                     (textconv_one || !diff_filespec_is_binary(o->repo, one)) &&
3483                     (textconv_two || !diff_filespec_is_binary(o->repo, two))) {
3484                         emit_diff_symbol(o, DIFF_SYMBOL_HEADER,
3485                                          header.buf, header.len, 0);
3486                         strbuf_reset(&header);
3487                         emit_rewrite_diff(name_a, name_b, one, two,
3488                                           textconv_one, textconv_two, o);
3489                         o->found_changes = 1;
3490                         goto free_ab_and_return;
3491                 }
3492         }
3493
3494         if (o->irreversible_delete && lbl[1][0] == '/') {
3495                 emit_diff_symbol(o, DIFF_SYMBOL_HEADER, header.buf,
3496                                  header.len, 0);
3497                 strbuf_reset(&header);
3498                 goto free_ab_and_return;
3499         } else if (!o->flags.text &&
3500                    ( (!textconv_one && diff_filespec_is_binary(o->repo, one)) ||
3501                      (!textconv_two && diff_filespec_is_binary(o->repo, two)) )) {
3502                 struct strbuf sb = STRBUF_INIT;
3503                 if (!one->data && !two->data &&
3504                     S_ISREG(one->mode) && S_ISREG(two->mode) &&
3505                     !o->flags.binary) {
3506                         if (oideq(&one->oid, &two->oid)) {
3507                                 if (must_show_header)
3508                                         emit_diff_symbol(o, DIFF_SYMBOL_HEADER,
3509                                                          header.buf, header.len,
3510                                                          0);
3511                                 goto free_ab_and_return;
3512                         }
3513                         emit_diff_symbol(o, DIFF_SYMBOL_HEADER,
3514                                          header.buf, header.len, 0);
3515                         strbuf_addf(&sb, "%sBinary files %s and %s differ\n",
3516                                     diff_line_prefix(o), lbl[0], lbl[1]);
3517                         emit_diff_symbol(o, DIFF_SYMBOL_BINARY_FILES,
3518                                          sb.buf, sb.len, 0);
3519                         strbuf_release(&sb);
3520                         goto free_ab_and_return;
3521                 }
3522                 if (fill_mmfile(o->repo, &mf1, one) < 0 ||
3523                     fill_mmfile(o->repo, &mf2, two) < 0)
3524                         die("unable to read files to diff");
3525                 /* Quite common confusing case */
3526                 if (mf1.size == mf2.size &&
3527                     !memcmp(mf1.ptr, mf2.ptr, mf1.size)) {
3528                         if (must_show_header)
3529                                 emit_diff_symbol(o, DIFF_SYMBOL_HEADER,
3530                                                  header.buf, header.len, 0);
3531                         goto free_ab_and_return;
3532                 }
3533                 emit_diff_symbol(o, DIFF_SYMBOL_HEADER, header.buf, header.len, 0);
3534                 strbuf_reset(&header);
3535                 if (o->flags.binary)
3536                         emit_binary_diff(o, &mf1, &mf2);
3537                 else {
3538                         strbuf_addf(&sb, "%sBinary files %s and %s differ\n",
3539                                     diff_line_prefix(o), lbl[0], lbl[1]);
3540                         emit_diff_symbol(o, DIFF_SYMBOL_BINARY_FILES,
3541                                          sb.buf, sb.len, 0);
3542                         strbuf_release(&sb);
3543                 }
3544                 o->found_changes = 1;
3545         } else {
3546                 /* Crazy xdl interfaces.. */
3547                 const char *diffopts;
3548                 const char *v;
3549                 xpparam_t xpp;
3550                 xdemitconf_t xecfg;
3551                 struct emit_callback ecbdata;
3552                 const struct userdiff_funcname *pe;
3553
3554                 if (must_show_header) {
3555                         emit_diff_symbol(o, DIFF_SYMBOL_HEADER,
3556                                          header.buf, header.len, 0);
3557                         strbuf_reset(&header);
3558                 }
3559
3560                 mf1.size = fill_textconv(o->repo, textconv_one, one, &mf1.ptr);
3561                 mf2.size = fill_textconv(o->repo, textconv_two, two, &mf2.ptr);
3562
3563                 pe = diff_funcname_pattern(o, one);
3564                 if (!pe)
3565                         pe = diff_funcname_pattern(o, two);
3566
3567                 memset(&xpp, 0, sizeof(xpp));
3568                 memset(&xecfg, 0, sizeof(xecfg));
3569                 memset(&ecbdata, 0, sizeof(ecbdata));
3570                 if (o->flags.suppress_diff_headers)
3571                         lbl[0] = NULL;
3572                 ecbdata.label_path = lbl;
3573                 ecbdata.color_diff = want_color(o->use_color);
3574                 ecbdata.ws_rule = whitespace_rule(o->repo->index, name_b);
3575                 if (ecbdata.ws_rule & WS_BLANK_AT_EOF)
3576                         check_blank_at_eof(&mf1, &mf2, &ecbdata);
3577                 ecbdata.opt = o;
3578                 if (header.len && !o->flags.suppress_diff_headers)
3579                         ecbdata.header = &header;
3580                 xpp.flags = o->xdl_opts;
3581                 xpp.anchors = o->anchors;
3582                 xpp.anchors_nr = o->anchors_nr;
3583                 xecfg.ctxlen = o->context;
3584                 xecfg.interhunkctxlen = o->interhunkcontext;
3585                 xecfg.flags = XDL_EMIT_FUNCNAMES;
3586                 if (o->flags.funccontext)
3587                         xecfg.flags |= XDL_EMIT_FUNCCONTEXT;
3588                 if (pe)
3589                         xdiff_set_find_func(&xecfg, pe->pattern, pe->cflags);
3590
3591                 diffopts = getenv("GIT_DIFF_OPTS");
3592                 if (!diffopts)
3593                         ;
3594                 else if (skip_prefix(diffopts, "--unified=", &v))
3595                         xecfg.ctxlen = strtoul(v, NULL, 10);
3596                 else if (skip_prefix(diffopts, "-u", &v))
3597                         xecfg.ctxlen = strtoul(v, NULL, 10);
3598
3599                 if (o->word_diff)
3600                         init_diff_words_data(&ecbdata, o, one, two);
3601                 if (xdi_diff_outf(&mf1, &mf2, NULL, fn_out_consume,
3602                                   &ecbdata, &xpp, &xecfg))
3603                         die("unable to generate diff for %s", one->path);
3604                 if (o->word_diff)
3605                         free_diff_words_data(&ecbdata);
3606                 if (textconv_one)
3607                         free(mf1.ptr);
3608                 if (textconv_two)
3609                         free(mf2.ptr);
3610                 xdiff_clear_find_func(&xecfg);
3611         }
3612
3613  free_ab_and_return:
3614         strbuf_release(&header);
3615         diff_free_filespec_data(one);
3616         diff_free_filespec_data(two);
3617         free(a_one);
3618         free(b_two);
3619         return;
3620 }
3621
3622 static char *get_compact_summary(const struct diff_filepair *p, int is_renamed)
3623 {
3624         if (!is_renamed) {
3625                 if (p->status == DIFF_STATUS_ADDED) {
3626                         if (S_ISLNK(p->two->mode))
3627                                 return "new +l";
3628                         else if ((p->two->mode & 0777) == 0755)
3629                                 return "new +x";
3630                         else
3631                                 return "new";
3632                 } else if (p->status == DIFF_STATUS_DELETED)
3633                         return "gone";
3634         }
3635         if (S_ISLNK(p->one->mode) && !S_ISLNK(p->two->mode))
3636                 return "mode -l";
3637         else if (!S_ISLNK(p->one->mode) && S_ISLNK(p->two->mode))
3638                 return "mode +l";
3639         else if ((p->one->mode & 0777) == 0644 &&
3640                  (p->two->mode & 0777) == 0755)
3641                 return "mode +x";
3642         else if ((p->one->mode & 0777) == 0755 &&
3643                  (p->two->mode & 0777) == 0644)
3644                 return "mode -x";
3645         return NULL;
3646 }
3647
3648 static void builtin_diffstat(const char *name_a, const char *name_b,
3649                              struct diff_filespec *one,
3650                              struct diff_filespec *two,
3651                              struct diffstat_t *diffstat,
3652                              struct diff_options *o,
3653                              struct diff_filepair *p)
3654 {
3655         mmfile_t mf1, mf2;
3656         struct diffstat_file *data;
3657         int same_contents;
3658         int complete_rewrite = 0;
3659
3660         if (!DIFF_PAIR_UNMERGED(p)) {
3661                 if (p->status == DIFF_STATUS_MODIFIED && p->score)
3662                         complete_rewrite = 1;
3663         }
3664
3665         data = diffstat_add(diffstat, name_a, name_b);
3666         data->is_interesting = p->status != DIFF_STATUS_UNKNOWN;
3667         if (o->flags.stat_with_summary)
3668                 data->comments = get_compact_summary(p, data->is_renamed);
3669
3670         if (!one || !two) {
3671                 data->is_unmerged = 1;
3672                 return;
3673         }
3674
3675         same_contents = oideq(&one->oid, &two->oid);
3676
3677         if (diff_filespec_is_binary(o->repo, one) ||
3678             diff_filespec_is_binary(o->repo, two)) {
3679                 data->is_binary = 1;
3680                 if (same_contents) {
3681                         data->added = 0;
3682                         data->deleted = 0;
3683                 } else {
3684                         data->added = diff_filespec_size(o->repo, two);
3685                         data->deleted = diff_filespec_size(o->repo, one);
3686                 }
3687         }
3688
3689         else if (complete_rewrite) {
3690                 diff_populate_filespec(o->repo, one, 0);
3691                 diff_populate_filespec(o->repo, two, 0);
3692                 data->deleted = count_lines(one->data, one->size);
3693                 data->added = count_lines(two->data, two->size);
3694         }
3695
3696         else if (!same_contents) {
3697                 /* Crazy xdl interfaces.. */
3698                 xpparam_t xpp;
3699                 xdemitconf_t xecfg;
3700
3701                 if (fill_mmfile(o->repo, &mf1, one) < 0 ||
3702                     fill_mmfile(o->repo, &mf2, two) < 0)
3703                         die("unable to read files to diff");
3704
3705                 memset(&xpp, 0, sizeof(xpp));
3706                 memset(&xecfg, 0, sizeof(xecfg));
3707                 xpp.flags = o->xdl_opts;
3708                 xpp.anchors = o->anchors;
3709                 xpp.anchors_nr = o->anchors_nr;
3710                 xecfg.ctxlen = o->context;
3711                 xecfg.interhunkctxlen = o->interhunkcontext;
3712                 if (xdi_diff_outf(&mf1, &mf2, discard_hunk_line,
3713                                   diffstat_consume, diffstat, &xpp, &xecfg))
3714                         die("unable to generate diffstat for %s", one->path);
3715         }
3716
3717         diff_free_filespec_data(one);
3718         diff_free_filespec_data(two);
3719 }
3720
3721 static void builtin_checkdiff(const char *name_a, const char *name_b,
3722                               const char *attr_path,
3723                               struct diff_filespec *one,
3724                               struct diff_filespec *two,
3725                               struct diff_options *o)
3726 {
3727         mmfile_t mf1, mf2;
3728         struct checkdiff_t data;
3729
3730         if (!two)
3731                 return;
3732
3733         memset(&data, 0, sizeof(data));
3734         data.filename = name_b ? name_b : name_a;
3735         data.lineno = 0;
3736         data.o = o;
3737         data.ws_rule = whitespace_rule(o->repo->index, attr_path);
3738         data.conflict_marker_size = ll_merge_marker_size(o->repo->index, attr_path);
3739
3740         if (fill_mmfile(o->repo, &mf1, one) < 0 ||
3741             fill_mmfile(o->repo, &mf2, two) < 0)
3742                 die("unable to read files to diff");
3743
3744         /*
3745          * All the other codepaths check both sides, but not checking
3746          * the "old" side here is deliberate.  We are checking the newly
3747          * introduced changes, and as long as the "new" side is text, we
3748          * can and should check what it introduces.
3749          */
3750         if (diff_filespec_is_binary(o->repo, two))
3751                 goto free_and_return;
3752         else {
3753                 /* Crazy xdl interfaces.. */
3754                 xpparam_t xpp;
3755                 xdemitconf_t xecfg;
3756
3757                 memset(&xpp, 0, sizeof(xpp));
3758                 memset(&xecfg, 0, sizeof(xecfg));
3759                 xecfg.ctxlen = 1; /* at least one context line */
3760                 xpp.flags = 0;
3761                 if (xdi_diff_outf(&mf1, &mf2, checkdiff_consume_hunk,
3762                                   checkdiff_consume, &data,
3763                                   &xpp, &xecfg))
3764                         die("unable to generate checkdiff for %s", one->path);
3765
3766                 if (data.ws_rule & WS_BLANK_AT_EOF) {
3767                         struct emit_callback ecbdata;
3768                         int blank_at_eof;
3769
3770                         ecbdata.ws_rule = data.ws_rule;
3771                         check_blank_at_eof(&mf1, &mf2, &ecbdata);
3772                         blank_at_eof = ecbdata.blank_at_eof_in_postimage;
3773
3774                         if (blank_at_eof) {
3775                                 static char *err;
3776                                 if (!err)
3777                                         err = whitespace_error_string(WS_BLANK_AT_EOF);
3778                                 fprintf(o->file, "%s:%d: %s.\n",
3779                                         data.filename, blank_at_eof, err);
3780                                 data.status = 1; /* report errors */
3781                         }
3782                 }
3783         }
3784  free_and_return:
3785         diff_free_filespec_data(one);
3786         diff_free_filespec_data(two);
3787         if (data.status)
3788                 o->flags.check_failed = 1;
3789 }
3790
3791 struct diff_filespec *alloc_filespec(const char *path)
3792 {
3793         struct diff_filespec *spec;
3794
3795         FLEXPTR_ALLOC_STR(spec, path, path);
3796         spec->count = 1;
3797         spec->is_binary = -1;
3798         return spec;
3799 }
3800
3801 void free_filespec(struct diff_filespec *spec)
3802 {
3803         if (!--spec->count) {
3804                 diff_free_filespec_data(spec);
3805                 free(spec);
3806         }
3807 }
3808
3809 void fill_filespec(struct diff_filespec *spec, const struct object_id *oid,
3810                    int oid_valid, unsigned short mode)
3811 {
3812         if (mode) {
3813                 spec->mode = canon_mode(mode);
3814                 oidcpy(&spec->oid, oid);
3815                 spec->oid_valid = oid_valid;
3816         }
3817 }
3818
3819 /*
3820  * Given a name and sha1 pair, if the index tells us the file in
3821  * the work tree has that object contents, return true, so that
3822  * prepare_temp_file() does not have to inflate and extract.
3823  */
3824 static int reuse_worktree_file(struct index_state *istate,
3825                                const char *name,
3826                                const struct object_id *oid,
3827                                int want_file)
3828 {
3829         const struct cache_entry *ce;
3830         struct stat st;
3831         int pos, len;
3832
3833         /*
3834          * We do not read the cache ourselves here, because the
3835          * benchmark with my previous version that always reads cache
3836          * shows that it makes things worse for diff-tree comparing
3837          * two linux-2.6 kernel trees in an already checked out work
3838          * tree.  This is because most diff-tree comparisons deal with
3839          * only a small number of files, while reading the cache is
3840          * expensive for a large project, and its cost outweighs the
3841          * savings we get by not inflating the object to a temporary
3842          * file.  Practically, this code only helps when we are used
3843          * by diff-cache --cached, which does read the cache before
3844          * calling us.
3845          */
3846         if (!istate->cache)
3847                 return 0;
3848
3849         /* We want to avoid the working directory if our caller
3850          * doesn't need the data in a normal file, this system
3851          * is rather slow with its stat/open/mmap/close syscalls,
3852          * and the object is contained in a pack file.  The pack
3853          * is probably already open and will be faster to obtain
3854          * the data through than the working directory.  Loose
3855          * objects however would tend to be slower as they need
3856          * to be individually opened and inflated.
3857          */
3858         if (!FAST_WORKING_DIRECTORY && !want_file && has_object_pack(oid))
3859                 return 0;
3860
3861         /*
3862          * Similarly, if we'd have to convert the file contents anyway, that
3863          * makes the optimization not worthwhile.
3864          */
3865         if (!want_file && would_convert_to_git(istate, name))
3866                 return 0;
3867
3868         len = strlen(name);
3869         pos = index_name_pos(istate, name, len);
3870         if (pos < 0)
3871                 return 0;
3872         ce = istate->cache[pos];
3873
3874         /*
3875          * This is not the sha1 we are looking for, or
3876          * unreusable because it is not a regular file.
3877          */
3878         if (!oideq(oid, &ce->oid) || !S_ISREG(ce->ce_mode))
3879                 return 0;
3880
3881         /*
3882          * If ce is marked as "assume unchanged", there is no
3883          * guarantee that work tree matches what we are looking for.
3884          */
3885         if ((ce->ce_flags & CE_VALID) || ce_skip_worktree(ce))
3886                 return 0;
3887
3888         /*
3889          * If ce matches the file in the work tree, we can reuse it.
3890          */
3891         if (ce_uptodate(ce) ||
3892             (!lstat(name, &st) && !ie_match_stat(istate, ce, &st, 0)))
3893                 return 1;
3894
3895         return 0;
3896 }
3897
3898 static int diff_populate_gitlink(struct diff_filespec *s, int size_only)
3899 {
3900         struct strbuf buf = STRBUF_INIT;
3901         char *dirty = "";
3902
3903         /* Are we looking at the work tree? */
3904         if (s->dirty_submodule)
3905                 dirty = "-dirty";
3906
3907         strbuf_addf(&buf, "Subproject commit %s%s\n",
3908                     oid_to_hex(&s->oid), dirty);
3909         s->size = buf.len;
3910         if (size_only) {
3911                 s->data = NULL;
3912                 strbuf_release(&buf);
3913         } else {
3914                 s->data = strbuf_detach(&buf, NULL);
3915                 s->should_free = 1;
3916         }
3917         return 0;
3918 }
3919
3920 /*
3921  * While doing rename detection and pickaxe operation, we may need to
3922  * grab the data for the blob (or file) for our own in-core comparison.
3923  * diff_filespec has data and size fields for this purpose.
3924  */
3925 int diff_populate_filespec(struct repository *r,
3926                            struct diff_filespec *s,
3927                            unsigned int flags)
3928 {
3929         int size_only = flags & CHECK_SIZE_ONLY;
3930         int err = 0;
3931         int conv_flags = global_conv_flags_eol;
3932         /*
3933          * demote FAIL to WARN to allow inspecting the situation
3934          * instead of refusing.
3935          */
3936         if (conv_flags & CONV_EOL_RNDTRP_DIE)
3937                 conv_flags = CONV_EOL_RNDTRP_WARN;
3938
3939         if (!DIFF_FILE_VALID(s))
3940                 die("internal error: asking to populate invalid file.");
3941         if (S_ISDIR(s->mode))
3942                 return -1;
3943
3944         if (s->data)
3945                 return 0;
3946
3947         if (size_only && 0 < s->size)
3948                 return 0;
3949
3950         if (S_ISGITLINK(s->mode))
3951                 return diff_populate_gitlink(s, size_only);
3952
3953         if (!s->oid_valid ||
3954             reuse_worktree_file(r->index, s->path, &s->oid, 0)) {
3955                 struct strbuf buf = STRBUF_INIT;
3956                 struct stat st;
3957                 int fd;
3958
3959                 if (lstat(s->path, &st) < 0) {
3960                 err_empty:
3961                         err = -1;
3962                 empty:
3963                         s->data = (char *)"";
3964                         s->size = 0;
3965                         return err;
3966                 }
3967                 s->size = xsize_t(st.st_size);
3968                 if (!s->size)
3969                         goto empty;
3970                 if (S_ISLNK(st.st_mode)) {
3971                         struct strbuf sb = STRBUF_INIT;
3972
3973                         if (strbuf_readlink(&sb, s->path, s->size))
3974                                 goto err_empty;
3975                         s->size = sb.len;
3976                         s->data = strbuf_detach(&sb, NULL);
3977                         s->should_free = 1;
3978                         return 0;
3979                 }
3980
3981                 /*
3982                  * Even if the caller would be happy with getting
3983                  * only the size, we cannot return early at this
3984                  * point if the path requires us to run the content
3985                  * conversion.
3986                  */
3987                 if (size_only && !would_convert_to_git(r->index, s->path))
3988                         return 0;
3989
3990                 /*
3991                  * Note: this check uses xsize_t(st.st_size) that may
3992                  * not be the true size of the blob after it goes
3993                  * through convert_to_git().  This may not strictly be
3994                  * correct, but the whole point of big_file_threshold
3995                  * and is_binary check being that we want to avoid
3996                  * opening the file and inspecting the contents, this
3997                  * is probably fine.
3998                  */
3999                 if ((flags & CHECK_BINARY) &&
4000                     s->size > big_file_threshold && s->is_binary == -1) {
4001                         s->is_binary = 1;
4002                         return 0;
4003                 }
4004                 fd = open(s->path, O_RDONLY);
4005                 if (fd < 0)
4006                         goto err_empty;
4007                 s->data = xmmap(NULL, s->size, PROT_READ, MAP_PRIVATE, fd, 0);
4008                 close(fd);
4009                 s->should_munmap = 1;
4010
4011                 /*
4012                  * Convert from working tree format to canonical git format
4013                  */
4014                 if (convert_to_git(r->index, s->path, s->data, s->size, &buf, conv_flags)) {
4015                         size_t size = 0;
4016                         munmap(s->data, s->size);
4017                         s->should_munmap = 0;
4018                         s->data = strbuf_detach(&buf, &size);
4019                         s->size = size;
4020                         s->should_free = 1;
4021                 }
4022         }
4023         else {
4024                 enum object_type type;
4025                 if (size_only || (flags & CHECK_BINARY)) {
4026                         type = oid_object_info(r, &s->oid, &s->size);
4027                         if (type < 0)
4028                                 die("unable to read %s",
4029                                     oid_to_hex(&s->oid));
4030                         if (size_only)
4031                                 return 0;
4032                         if (s->size > big_file_threshold && s->is_binary == -1) {
4033                                 s->is_binary = 1;
4034                                 return 0;
4035                         }
4036                 }
4037                 s->data = read_object_file(&s->oid, &type, &s->size);
4038                 if (!s->data)
4039                         die("unable to read %s", oid_to_hex(&s->oid));
4040                 s->should_free = 1;
4041         }
4042         return 0;
4043 }
4044
4045 void diff_free_filespec_blob(struct diff_filespec *s)
4046 {
4047         if (s->should_free)
4048                 free(s->data);
4049         else if (s->should_munmap)
4050                 munmap(s->data, s->size);
4051
4052         if (s->should_free || s->should_munmap) {
4053                 s->should_free = s->should_munmap = 0;
4054                 s->data = NULL;
4055         }
4056 }
4057
4058 void diff_free_filespec_data(struct diff_filespec *s)
4059 {
4060         diff_free_filespec_blob(s);
4061         FREE_AND_NULL(s->cnt_data);
4062 }
4063
4064 static void prep_temp_blob(struct index_state *istate,
4065                            const char *path, struct diff_tempfile *temp,
4066                            void *blob,
4067                            unsigned long size,
4068                            const struct object_id *oid,
4069                            int mode)
4070 {
4071         struct strbuf buf = STRBUF_INIT;
4072         struct strbuf tempfile = STRBUF_INIT;
4073         char *path_dup = xstrdup(path);
4074         const char *base = basename(path_dup);
4075
4076         /* Generate "XXXXXX_basename.ext" */
4077         strbuf_addstr(&tempfile, "XXXXXX_");
4078         strbuf_addstr(&tempfile, base);
4079
4080         temp->tempfile = mks_tempfile_ts(tempfile.buf, strlen(base) + 1);
4081         if (!temp->tempfile)
4082                 die_errno("unable to create temp-file");
4083         if (convert_to_working_tree(istate, path,
4084                         (const char *)blob, (size_t)size, &buf)) {
4085                 blob = buf.buf;
4086                 size = buf.len;
4087         }
4088         if (write_in_full(temp->tempfile->fd, blob, size) < 0 ||
4089             close_tempfile_gently(temp->tempfile))
4090                 die_errno("unable to write temp-file");
4091         temp->name = get_tempfile_path(temp->tempfile);
4092         oid_to_hex_r(temp->hex, oid);
4093         xsnprintf(temp->mode, sizeof(temp->mode), "%06o", mode);
4094         strbuf_release(&buf);
4095         strbuf_release(&tempfile);
4096         free(path_dup);
4097 }
4098
4099 static struct diff_tempfile *prepare_temp_file(struct repository *r,
4100                                                const char *name,
4101                                                struct diff_filespec *one)
4102 {
4103         struct diff_tempfile *temp = claim_diff_tempfile();
4104
4105         if (!DIFF_FILE_VALID(one)) {
4106         not_a_valid_file:
4107                 /* A '-' entry produces this for file-2, and
4108                  * a '+' entry produces this for file-1.
4109                  */
4110                 temp->name = "/dev/null";
4111                 xsnprintf(temp->hex, sizeof(temp->hex), ".");
4112                 xsnprintf(temp->mode, sizeof(temp->mode), ".");
4113                 return temp;
4114         }
4115
4116         if (!S_ISGITLINK(one->mode) &&
4117             (!one->oid_valid ||
4118              reuse_worktree_file(r->index, name, &one->oid, 1))) {
4119                 struct stat st;
4120                 if (lstat(name, &st) < 0) {
4121                         if (errno == ENOENT)
4122                                 goto not_a_valid_file;
4123                         die_errno("stat(%s)", name);
4124                 }
4125                 if (S_ISLNK(st.st_mode)) {
4126                         struct strbuf sb = STRBUF_INIT;
4127                         if (strbuf_readlink(&sb, name, st.st_size) < 0)
4128                                 die_errno("readlink(%s)", name);
4129                         prep_temp_blob(r->index, name, temp, sb.buf, sb.len,
4130                                        (one->oid_valid ?
4131                                         &one->oid : &null_oid),
4132                                        (one->oid_valid ?
4133                                         one->mode : S_IFLNK));
4134                         strbuf_release(&sb);
4135                 }
4136                 else {
4137                         /* we can borrow from the file in the work tree */
4138                         temp->name = name;
4139                         if (!one->oid_valid)
4140                                 oid_to_hex_r(temp->hex, &null_oid);
4141                         else
4142                                 oid_to_hex_r(temp->hex, &one->oid);
4143                         /* Even though we may sometimes borrow the
4144                          * contents from the work tree, we always want
4145                          * one->mode.  mode is trustworthy even when
4146                          * !(one->oid_valid), as long as
4147                          * DIFF_FILE_VALID(one).
4148                          */
4149                         xsnprintf(temp->mode, sizeof(temp->mode), "%06o", one->mode);
4150                 }
4151                 return temp;
4152         }
4153         else {
4154                 if (diff_populate_filespec(r, one, 0))
4155                         die("cannot read data blob for %s", one->path);
4156                 prep_temp_blob(r->index, name, temp,
4157                                one->data, one->size,
4158                                &one->oid, one->mode);
4159         }
4160         return temp;
4161 }
4162
4163 static void add_external_diff_name(struct repository *r,
4164                                    struct argv_array *argv,
4165                                    const char *name,
4166                                    struct diff_filespec *df)
4167 {
4168         struct diff_tempfile *temp = prepare_temp_file(r, name, df);
4169         argv_array_push(argv, temp->name);
4170         argv_array_push(argv, temp->hex);
4171         argv_array_push(argv, temp->mode);
4172 }
4173
4174 /* An external diff command takes:
4175  *
4176  * diff-cmd name infile1 infile1-sha1 infile1-mode \
4177  *               infile2 infile2-sha1 infile2-mode [ rename-to ]
4178  *
4179  */
4180 static void run_external_diff(const char *pgm,
4181                               const char *name,
4182                               const char *other,
4183                               struct diff_filespec *one,
4184                               struct diff_filespec *two,
4185                               const char *xfrm_msg,
4186                               struct diff_options *o)
4187 {
4188         struct argv_array argv = ARGV_ARRAY_INIT;
4189         struct argv_array env = ARGV_ARRAY_INIT;
4190         struct diff_queue_struct *q = &diff_queued_diff;
4191
4192         argv_array_push(&argv, pgm);
4193         argv_array_push(&argv, name);
4194
4195         if (one && two) {
4196                 add_external_diff_name(o->repo, &argv, name, one);
4197                 if (!other)
4198                         add_external_diff_name(o->repo, &argv, name, two);
4199                 else {
4200                         add_external_diff_name(o->repo, &argv, other, two);
4201                         argv_array_push(&argv, other);
4202                         argv_array_push(&argv, xfrm_msg);
4203                 }
4204         }
4205
4206         argv_array_pushf(&env, "GIT_DIFF_PATH_COUNTER=%d", ++o->diff_path_counter);
4207         argv_array_pushf(&env, "GIT_DIFF_PATH_TOTAL=%d", q->nr);
4208
4209         if (run_command_v_opt_cd_env(argv.argv, RUN_USING_SHELL, NULL, env.argv))
4210                 die(_("external diff died, stopping at %s"), name);
4211
4212         remove_tempfile();
4213         argv_array_clear(&argv);
4214         argv_array_clear(&env);
4215 }
4216
4217 static int similarity_index(struct diff_filepair *p)
4218 {
4219         return p->score * 100 / MAX_SCORE;
4220 }
4221
4222 static const char *diff_abbrev_oid(const struct object_id *oid, int abbrev)
4223 {
4224         if (startup_info->have_repository)
4225                 return find_unique_abbrev(oid, abbrev);
4226         else {
4227                 char *hex = oid_to_hex(oid);
4228                 if (abbrev < 0)
4229                         abbrev = FALLBACK_DEFAULT_ABBREV;
4230                 if (abbrev > the_hash_algo->hexsz)
4231                         BUG("oid abbreviation out of range: %d", abbrev);
4232                 if (abbrev)
4233                         hex[abbrev] = '\0';
4234                 return hex;
4235         }
4236 }
4237
4238 static void fill_metainfo(struct strbuf *msg,
4239                           const char *name,
4240                           const char *other,
4241                           struct diff_filespec *one,
4242                           struct diff_filespec *two,
4243                           struct diff_options *o,
4244                           struct diff_filepair *p,
4245                           int *must_show_header,
4246                           int use_color)
4247 {
4248         const char *set = diff_get_color(use_color, DIFF_METAINFO);
4249         const char *reset = diff_get_color(use_color, DIFF_RESET);
4250         const char *line_prefix = diff_line_prefix(o);
4251
4252         *must_show_header = 1;
4253         strbuf_init(msg, PATH_MAX * 2 + 300);
4254         switch (p->status) {
4255         case DIFF_STATUS_COPIED:
4256                 strbuf_addf(msg, "%s%ssimilarity index %d%%",
4257                             line_prefix, set, similarity_index(p));
4258                 strbuf_addf(msg, "%s\n%s%scopy from ",
4259                             reset,  line_prefix, set);
4260                 quote_c_style(name, msg, NULL, 0);
4261                 strbuf_addf(msg, "%s\n%s%scopy to ", reset, line_prefix, set);
4262                 quote_c_style(other, msg, NULL, 0);
4263                 strbuf_addf(msg, "%s\n", reset);
4264                 break;
4265         case DIFF_STATUS_RENAMED:
4266                 strbuf_addf(msg, "%s%ssimilarity index %d%%",
4267                             line_prefix, set, similarity_index(p));
4268                 strbuf_addf(msg, "%s\n%s%srename from ",
4269                             reset, line_prefix, set);
4270                 quote_c_style(name, msg, NULL, 0);
4271                 strbuf_addf(msg, "%s\n%s%srename to ",
4272                             reset, line_prefix, set);
4273                 quote_c_style(other, msg, NULL, 0);
4274                 strbuf_addf(msg, "%s\n", reset);
4275                 break;
4276         case DIFF_STATUS_MODIFIED:
4277                 if (p->score) {
4278                         strbuf_addf(msg, "%s%sdissimilarity index %d%%%s\n",
4279                                     line_prefix,
4280                                     set, similarity_index(p), reset);
4281                         break;
4282                 }
4283                 /* fallthru */
4284         default:
4285                 *must_show_header = 0;
4286         }
4287         if (one && two && !oideq(&one->oid, &two->oid)) {
4288                 const unsigned hexsz = the_hash_algo->hexsz;
4289                 int abbrev = o->flags.full_index ? hexsz : DEFAULT_ABBREV;
4290
4291                 if (o->flags.binary) {
4292                         mmfile_t mf;
4293                         if ((!fill_mmfile(o->repo, &mf, one) &&
4294                              diff_filespec_is_binary(o->repo, one)) ||
4295                             (!fill_mmfile(o->repo, &mf, two) &&
4296                              diff_filespec_is_binary(o->repo, two)))
4297                                 abbrev = hexsz;
4298                 }
4299                 strbuf_addf(msg, "%s%sindex %s..%s", line_prefix, set,
4300                             diff_abbrev_oid(&one->oid, abbrev),
4301                             diff_abbrev_oid(&two->oid, abbrev));
4302                 if (one->mode == two->mode)
4303                         strbuf_addf(msg, " %06o", one->mode);
4304                 strbuf_addf(msg, "%s\n", reset);
4305         }
4306 }
4307
4308 static void run_diff_cmd(const char *pgm,
4309                          const char *name,
4310                          const char *other,
4311                          const char *attr_path,
4312                          struct diff_filespec *one,
4313                          struct diff_filespec *two,
4314                          struct strbuf *msg,
4315                          struct diff_options *o,
4316                          struct diff_filepair *p)
4317 {
4318         const char *xfrm_msg = NULL;
4319         int complete_rewrite = (p->status == DIFF_STATUS_MODIFIED) && p->score;
4320         int must_show_header = 0;
4321
4322
4323         if (o->flags.allow_external) {
4324                 struct userdiff_driver *drv;
4325
4326                 drv = userdiff_find_by_path(o->repo->index, attr_path);
4327                 if (drv && drv->external)
4328                         pgm = drv->external;
4329         }
4330
4331         if (msg) {
4332                 /*
4333                  * don't use colors when the header is intended for an
4334                  * external diff driver
4335                  */
4336                 fill_metainfo(msg, name, other, one, two, o, p,
4337                               &must_show_header,
4338                               want_color(o->use_color) && !pgm);
4339                 xfrm_msg = msg->len ? msg->buf : NULL;
4340         }
4341
4342         if (pgm) {
4343                 run_external_diff(pgm, name, other, one, two, xfrm_msg, o);
4344                 return;
4345         }
4346         if (one && two)
4347                 builtin_diff(name, other ? other : name,
4348                              one, two, xfrm_msg, must_show_header,
4349                              o, complete_rewrite);
4350         else
4351                 fprintf(o->file, "* Unmerged path %s\n", name);
4352 }
4353
4354 static void diff_fill_oid_info(struct diff_filespec *one, struct index_state *istate)
4355 {
4356         if (DIFF_FILE_VALID(one)) {
4357                 if (!one->oid_valid) {
4358                         struct stat st;
4359                         if (one->is_stdin) {
4360                                 oidclr(&one->oid);
4361                                 return;
4362                         }
4363                         if (lstat(one->path, &st) < 0)
4364                                 die_errno("stat '%s'", one->path);
4365                         if (index_path(istate, &one->oid, one->path, &st, 0))
4366                                 die("cannot hash %s", one->path);
4367                 }
4368         }
4369         else
4370                 oidclr(&one->oid);
4371 }
4372
4373 static void strip_prefix(int prefix_length, const char **namep, const char **otherp)
4374 {
4375         /* Strip the prefix but do not molest /dev/null and absolute paths */
4376         if (*namep && !is_absolute_path(*namep)) {
4377                 *namep += prefix_length;
4378                 if (**namep == '/')
4379                         ++*namep;
4380         }
4381         if (*otherp && !is_absolute_path(*otherp)) {
4382                 *otherp += prefix_length;
4383                 if (**otherp == '/')
4384                         ++*otherp;
4385         }
4386 }
4387
4388 static void run_diff(struct diff_filepair *p, struct diff_options *o)
4389 {
4390         const char *pgm = external_diff();
4391         struct strbuf msg;
4392         struct diff_filespec *one = p->one;
4393         struct diff_filespec *two = p->two;
4394         const char *name;
4395         const char *other;
4396         const char *attr_path;
4397
4398         name  = one->path;
4399         other = (strcmp(name, two->path) ? two->path : NULL);
4400         attr_path = name;
4401         if (o->prefix_length)
4402                 strip_prefix(o->prefix_length, &name, &other);
4403
4404         if (!o->flags.allow_external)
4405                 pgm = NULL;
4406
4407         if (DIFF_PAIR_UNMERGED(p)) {
4408                 run_diff_cmd(pgm, name, NULL, attr_path,
4409                              NULL, NULL, NULL, o, p);
4410                 return;
4411         }
4412
4413         diff_fill_oid_info(one, o->repo->index);
4414         diff_fill_oid_info(two, o->repo->index);
4415
4416         if (!pgm &&
4417             DIFF_FILE_VALID(one) && DIFF_FILE_VALID(two) &&
4418             (S_IFMT & one->mode) != (S_IFMT & two->mode)) {
4419                 /*
4420                  * a filepair that changes between file and symlink
4421                  * needs to be split into deletion and creation.
4422                  */
4423                 struct diff_filespec *null = alloc_filespec(two->path);
4424                 run_diff_cmd(NULL, name, other, attr_path,
4425                              one, null, &msg,
4426                              o, p);
4427                 free(null);
4428                 strbuf_release(&msg);
4429
4430                 null = alloc_filespec(one->path);
4431                 run_diff_cmd(NULL, name, other, attr_path,
4432                              null, two, &msg, o, p);
4433                 free(null);
4434         }
4435         else
4436                 run_diff_cmd(pgm, name, other, attr_path,
4437                              one, two, &msg, o, p);
4438
4439         strbuf_release(&msg);
4440 }
4441
4442 static void run_diffstat(struct diff_filepair *p, struct diff_options *o,
4443                          struct diffstat_t *diffstat)
4444 {
4445         const char *name;
4446         const char *other;
4447
4448         if (DIFF_PAIR_UNMERGED(p)) {
4449                 /* unmerged */
4450                 builtin_diffstat(p->one->path, NULL, NULL, NULL,
4451                                  diffstat, o, p);
4452                 return;
4453         }
4454
4455         name = p->one->path;
4456         other = (strcmp(name, p->two->path) ? p->two->path : NULL);
4457
4458         if (o->prefix_length)
4459                 strip_prefix(o->prefix_length, &name, &other);
4460
4461         diff_fill_oid_info(p->one, o->repo->index);
4462         diff_fill_oid_info(p->two, o->repo->index);
4463
4464         builtin_diffstat(name, other, p->one, p->two,
4465                          diffstat, o, p);
4466 }
4467
4468 static void run_checkdiff(struct diff_filepair *p, struct diff_options *o)
4469 {
4470         const char *name;
4471         const char *other;
4472         const char *attr_path;
4473
4474         if (DIFF_PAIR_UNMERGED(p)) {
4475                 /* unmerged */
4476                 return;
4477         }
4478
4479         name = p->one->path;
4480         other = (strcmp(name, p->two->path) ? p->two->path : NULL);
4481         attr_path = other ? other : name;
4482
4483         if (o->prefix_length)
4484                 strip_prefix(o->prefix_length, &name, &other);
4485
4486         diff_fill_oid_info(p->one, o->repo->index);
4487         diff_fill_oid_info(p->two, o->repo->index);
4488
4489         builtin_checkdiff(name, other, attr_path, p->one, p->two, o);
4490 }
4491
4492 static void prep_parse_options(struct diff_options *options);
4493
4494 void repo_diff_setup(struct repository *r, struct diff_options *options)
4495 {
4496         memcpy(options, &default_diff_options, sizeof(*options));
4497
4498         options->file = stdout;
4499         options->repo = r;
4500
4501         options->output_indicators[OUTPUT_INDICATOR_NEW] = '+';
4502         options->output_indicators[OUTPUT_INDICATOR_OLD] = '-';
4503         options->output_indicators[OUTPUT_INDICATOR_CONTEXT] = ' ';
4504         options->abbrev = DEFAULT_ABBREV;
4505         options->line_termination = '\n';
4506         options->break_opt = -1;
4507         options->rename_limit = -1;
4508         options->dirstat_permille = diff_dirstat_permille_default;
4509         options->context = diff_context_default;
4510         options->interhunkcontext = diff_interhunk_context_default;
4511         options->ws_error_highlight = ws_error_highlight_default;
4512         options->flags.rename_empty = 1;
4513         options->objfind = NULL;
4514
4515         /* pathchange left =NULL by default */
4516         options->change = diff_change;
4517         options->add_remove = diff_addremove;
4518         options->use_color = diff_use_color_default;
4519         options->detect_rename = diff_detect_rename_default;
4520         options->xdl_opts |= diff_algorithm;
4521         if (diff_indent_heuristic)
4522                 DIFF_XDL_SET(options, INDENT_HEURISTIC);
4523
4524         options->orderfile = diff_order_file_cfg;
4525
4526         if (diff_no_prefix) {
4527                 options->a_prefix = options->b_prefix = "";
4528         } else if (!diff_mnemonic_prefix) {
4529                 options->a_prefix = "a/";
4530                 options->b_prefix = "b/";
4531         }
4532
4533         options->color_moved = diff_color_moved_default;
4534         options->color_moved_ws_handling = diff_color_moved_ws_default;
4535
4536         prep_parse_options(options);
4537 }
4538
4539 void diff_setup_done(struct diff_options *options)
4540 {
4541         unsigned check_mask = DIFF_FORMAT_NAME |
4542                               DIFF_FORMAT_NAME_STATUS |
4543                               DIFF_FORMAT_CHECKDIFF |
4544                               DIFF_FORMAT_NO_OUTPUT;
4545         /*
4546          * This must be signed because we're comparing against a potentially
4547          * negative value.
4548          */
4549         const int hexsz = the_hash_algo->hexsz;
4550
4551         if (options->set_default)
4552                 options->set_default(options);
4553
4554         if (HAS_MULTI_BITS(options->output_format & check_mask))
4555                 die(_("--name-only, --name-status, --check and -s are mutually exclusive"));
4556
4557         if (HAS_MULTI_BITS(options->pickaxe_opts & DIFF_PICKAXE_KINDS_MASK))
4558                 die(_("-G, -S and --find-object are mutually exclusive"));
4559
4560         /*
4561          * Most of the time we can say "there are changes"
4562          * only by checking if there are changed paths, but
4563          * --ignore-whitespace* options force us to look
4564          * inside contents.
4565          */
4566
4567         if ((options->xdl_opts & XDF_WHITESPACE_FLAGS))
4568                 options->flags.diff_from_contents = 1;
4569         else
4570                 options->flags.diff_from_contents = 0;
4571
4572         if (options->flags.find_copies_harder)
4573                 options->detect_rename = DIFF_DETECT_COPY;
4574
4575         if (!options->flags.relative_name)
4576                 options->prefix = NULL;
4577         if (options->prefix)
4578                 options->prefix_length = strlen(options->prefix);
4579         else
4580                 options->prefix_length = 0;
4581
4582         if (options->output_format & (DIFF_FORMAT_NAME |
4583                                       DIFF_FORMAT_NAME_STATUS |
4584                                       DIFF_FORMAT_CHECKDIFF |
4585                                       DIFF_FORMAT_NO_OUTPUT))
4586                 options->output_format &= ~(DIFF_FORMAT_RAW |
4587                                             DIFF_FORMAT_NUMSTAT |
4588                                             DIFF_FORMAT_DIFFSTAT |
4589                                             DIFF_FORMAT_SHORTSTAT |
4590                                             DIFF_FORMAT_DIRSTAT |
4591                                             DIFF_FORMAT_SUMMARY |
4592                                             DIFF_FORMAT_PATCH);
4593
4594         /*
4595          * These cases always need recursive; we do not drop caller-supplied
4596          * recursive bits for other formats here.
4597          */
4598         if (options->output_format & (DIFF_FORMAT_PATCH |
4599                                       DIFF_FORMAT_NUMSTAT |
4600                                       DIFF_FORMAT_DIFFSTAT |
4601                                       DIFF_FORMAT_SHORTSTAT |
4602                                       DIFF_FORMAT_DIRSTAT |
4603                                       DIFF_FORMAT_SUMMARY |
4604                                       DIFF_FORMAT_CHECKDIFF))
4605                 options->flags.recursive = 1;
4606         /*
4607          * Also pickaxe would not work very well if you do not say recursive
4608          */
4609         if (options->pickaxe_opts & DIFF_PICKAXE_KINDS_MASK)
4610                 options->flags.recursive = 1;
4611         /*
4612          * When patches are generated, submodules diffed against the work tree
4613          * must be checked for dirtiness too so it can be shown in the output
4614          */
4615         if (options->output_format & DIFF_FORMAT_PATCH)
4616                 options->flags.dirty_submodules = 1;
4617
4618         if (options->detect_rename && options->rename_limit < 0)
4619                 options->rename_limit = diff_rename_limit_default;
4620         if (hexsz < options->abbrev)
4621                 options->abbrev = hexsz; /* full */
4622
4623         /*
4624          * It does not make sense to show the first hit we happened
4625          * to have found.  It does not make sense not to return with
4626          * exit code in such a case either.
4627          */
4628         if (options->flags.quick) {
4629                 options->output_format = DIFF_FORMAT_NO_OUTPUT;
4630                 options->flags.exit_with_status = 1;
4631         }
4632
4633         options->diff_path_counter = 0;
4634
4635         if (options->flags.follow_renames && options->pathspec.nr != 1)
4636                 die(_("--follow requires exactly one pathspec"));
4637
4638         if (!options->use_color || external_diff())
4639                 options->color_moved = 0;
4640
4641         FREE_AND_NULL(options->parseopts);
4642 }
4643
4644 static int opt_arg(const char *arg, int arg_short, const char *arg_long, int *val)
4645 {
4646         char c, *eq;
4647         int len;
4648
4649         if (*arg != '-')
4650                 return 0;
4651         c = *++arg;
4652         if (!c)
4653                 return 0;
4654         if (c == arg_short) {
4655                 c = *++arg;
4656                 if (!c)
4657                         return 1;
4658                 if (val && isdigit(c)) {
4659                         char *end;
4660                         int n = strtoul(arg, &end, 10);
4661                         if (*end)
4662                                 return 0;
4663                         *val = n;
4664                         return 1;
4665                 }
4666                 return 0;
4667         }
4668         if (c != '-')
4669                 return 0;
4670         arg++;
4671         eq = strchrnul(arg, '=');
4672         len = eq - arg;
4673         if (!len || strncmp(arg, arg_long, len))
4674                 return 0;
4675         if (*eq) {
4676                 int n;
4677                 char *end;
4678                 if (!isdigit(*++eq))
4679                         return 0;
4680                 n = strtoul(eq, &end, 10);
4681                 if (*end)
4682                         return 0;
4683                 *val = n;
4684         }
4685         return 1;
4686 }
4687
4688 static inline int short_opt(char opt, const char **argv,
4689                             const char **optarg)
4690 {
4691         const char *arg = argv[0];
4692         if (arg[0] != '-' || arg[1] != opt)
4693                 return 0;
4694         if (arg[2] != '\0') {
4695                 *optarg = arg + 2;
4696                 return 1;
4697         }
4698         if (!argv[1])
4699                 die("Option '%c' requires a value", opt);
4700         *optarg = argv[1];
4701         return 2;
4702 }
4703
4704 int parse_long_opt(const char *opt, const char **argv,
4705                    const char **optarg)
4706 {
4707         const char *arg = argv[0];
4708         if (!skip_prefix(arg, "--", &arg))
4709                 return 0;
4710         if (!skip_prefix(arg, opt, &arg))
4711                 return 0;
4712         if (*arg == '=') { /* stuck form: --option=value */
4713                 *optarg = arg + 1;
4714                 return 1;
4715         }
4716         if (*arg != '\0')
4717                 return 0;
4718         /* separate form: --option value */
4719         if (!argv[1])
4720                 die("Option '--%s' requires a value", opt);
4721         *optarg = argv[1];
4722         return 2;
4723 }
4724
4725 static int diff_opt_stat(const struct option *opt, const char *value, int unset)
4726 {
4727         struct diff_options *options = opt->value;
4728         int width = options->stat_width;
4729         int name_width = options->stat_name_width;
4730         int graph_width = options->stat_graph_width;
4731         int count = options->stat_count;
4732         char *end;
4733
4734         BUG_ON_OPT_NEG(unset);
4735
4736         if (!strcmp(opt->long_name, "stat")) {
4737                 if (value) {
4738                         width = strtoul(value, &end, 10);
4739                         if (*end == ',')
4740                                 name_width = strtoul(end+1, &end, 10);
4741                         if (*end == ',')
4742                                 count = strtoul(end+1, &end, 10);
4743                         if (*end)
4744                                 return error(_("invalid --stat value: %s"), value);
4745                 }
4746         } else if (!strcmp(opt->long_name, "stat-width")) {
4747                 width = strtoul(value, &end, 10);
4748                 if (*end)
4749                         return error(_("%s expects a numerical value"),
4750                                      opt->long_name);
4751         } else if (!strcmp(opt->long_name, "stat-name-width")) {
4752                 name_width = strtoul(value, &end, 10);
4753                 if (*end)
4754                         return error(_("%s expects a numerical value"),
4755                                      opt->long_name);
4756         } else if (!strcmp(opt->long_name, "stat-graph-width")) {
4757                 graph_width = strtoul(value, &end, 10);
4758                 if (*end)
4759                         return error(_("%s expects a numerical value"),
4760                                      opt->long_name);
4761         } else if (!strcmp(opt->long_name, "stat-count")) {
4762                 count = strtoul(value, &end, 10);
4763                 if (*end)
4764                         return error(_("%s expects a numerical value"),
4765                                      opt->long_name);
4766         } else
4767                 BUG("%s should not get here", opt->long_name);
4768
4769         options->output_format |= DIFF_FORMAT_DIFFSTAT;
4770         options->stat_name_width = name_width;
4771         options->stat_graph_width = graph_width;
4772         options->stat_width = width;
4773         options->stat_count = count;
4774         return 0;
4775 }
4776
4777 static int parse_dirstat_opt(struct diff_options *options, const char *params)
4778 {
4779         struct strbuf errmsg = STRBUF_INIT;
4780         if (parse_dirstat_params(options, params, &errmsg))
4781                 die(_("Failed to parse --dirstat/-X option parameter:\n%s"),
4782                     errmsg.buf);
4783         strbuf_release(&errmsg);
4784         /*
4785          * The caller knows a dirstat-related option is given from the command
4786          * line; allow it to say "return this_function();"
4787          */
4788         options->output_format |= DIFF_FORMAT_DIRSTAT;
4789         return 1;
4790 }
4791
4792 static int parse_submodule_opt(struct diff_options *options, const char *value)
4793 {
4794         if (parse_submodule_params(options, value))
4795                 die(_("Failed to parse --submodule option parameter: '%s'"),
4796                         value);
4797         return 1;
4798 }
4799
4800 static const char diff_status_letters[] = {
4801         DIFF_STATUS_ADDED,
4802         DIFF_STATUS_COPIED,
4803         DIFF_STATUS_DELETED,
4804         DIFF_STATUS_MODIFIED,
4805         DIFF_STATUS_RENAMED,
4806         DIFF_STATUS_TYPE_CHANGED,
4807         DIFF_STATUS_UNKNOWN,
4808         DIFF_STATUS_UNMERGED,
4809         DIFF_STATUS_FILTER_AON,
4810         DIFF_STATUS_FILTER_BROKEN,
4811         '\0',
4812 };
4813
4814 static unsigned int filter_bit['Z' + 1];
4815
4816 static void prepare_filter_bits(void)
4817 {
4818         int i;
4819
4820         if (!filter_bit[DIFF_STATUS_ADDED]) {
4821                 for (i = 0; diff_status_letters[i]; i++)
4822                         filter_bit[(int) diff_status_letters[i]] = (1 << i);
4823         }
4824 }
4825
4826 static unsigned filter_bit_tst(char status, const struct diff_options *opt)
4827 {
4828         return opt->filter & filter_bit[(int) status];
4829 }
4830
4831 static int parse_diff_filter_opt(const char *optarg, struct diff_options *opt)
4832 {
4833         int i, optch;
4834
4835         prepare_filter_bits();
4836
4837         /*
4838          * If there is a negation e.g. 'd' in the input, and we haven't
4839          * initialized the filter field with another --diff-filter, start
4840          * from full set of bits, except for AON.
4841          */
4842         if (!opt->filter) {
4843                 for (i = 0; (optch = optarg[i]) != '\0'; i++) {
4844                         if (optch < 'a' || 'z' < optch)
4845                                 continue;
4846                         opt->filter = (1 << (ARRAY_SIZE(diff_status_letters) - 1)) - 1;
4847                         opt->filter &= ~filter_bit[DIFF_STATUS_FILTER_AON];
4848                         break;
4849                 }
4850         }
4851
4852         for (i = 0; (optch = optarg[i]) != '\0'; i++) {
4853                 unsigned int bit;
4854                 int negate;
4855
4856                 if ('a' <= optch && optch <= 'z') {
4857                         negate = 1;
4858                         optch = toupper(optch);
4859                 } else {
4860                         negate = 0;
4861                 }
4862
4863                 bit = (0 <= optch && optch <= 'Z') ? filter_bit[optch] : 0;
4864                 if (!bit)
4865                         return optarg[i];
4866                 if (negate)
4867                         opt->filter &= ~bit;
4868                 else
4869                         opt->filter |= bit;
4870         }
4871         return 0;
4872 }
4873
4874 static void enable_patch_output(int *fmt)
4875 {
4876         *fmt &= ~DIFF_FORMAT_NO_OUTPUT;
4877         *fmt |= DIFF_FORMAT_PATCH;
4878 }
4879
4880 static int parse_ws_error_highlight_opt(struct diff_options *opt, const char *arg)
4881 {
4882         int val = parse_ws_error_highlight(arg);
4883
4884         if (val < 0) {
4885                 error("unknown value after ws-error-highlight=%.*s",
4886                       -1 - val, arg);
4887                 return 0;
4888         }
4889         opt->ws_error_highlight = val;
4890         return 1;
4891 }
4892
4893 static int parse_objfind_opt(struct diff_options *opt, const char *arg)
4894 {
4895         struct object_id oid;
4896
4897         if (get_oid(arg, &oid))
4898                 return error("unable to resolve '%s'", arg);
4899
4900         if (!opt->objfind)
4901                 opt->objfind = xcalloc(1, sizeof(*opt->objfind));
4902
4903         opt->pickaxe_opts |= DIFF_PICKAXE_KIND_OBJFIND;
4904         opt->flags.recursive = 1;
4905         opt->flags.tree_in_recursive = 1;
4906         oidset_insert(opt->objfind, &oid);
4907         return 1;
4908 }
4909
4910 static int diff_opt_break_rewrites(const struct option *opt,
4911                                    const char *arg, int unset)
4912 {
4913         int *break_opt = opt->value;
4914         int opt1, opt2;
4915
4916         BUG_ON_OPT_NEG(unset);
4917         if (!arg)
4918                 arg = "";
4919         opt1 = parse_rename_score(&arg);
4920         if (*arg == 0)
4921                 opt2 = 0;
4922         else if (*arg != '/')
4923                 return error(_("%s expects <n>/<m> form"), opt->long_name);
4924         else {
4925                 arg++;
4926                 opt2 = parse_rename_score(&arg);
4927         }
4928         if (*arg != 0)
4929                 return error(_("%s expects <n>/<m> form"), opt->long_name);
4930         *break_opt = opt1 | (opt2 << 16);
4931         return 0;
4932 }
4933
4934 static int diff_opt_char(const struct option *opt,
4935                          const char *arg, int unset)
4936 {
4937         char *value = opt->value;
4938
4939         BUG_ON_OPT_NEG(unset);
4940         if (arg[1])
4941                 return error(_("%s expects a character, got '%s'"),
4942                              opt->long_name, arg);
4943         *value = arg[0];
4944         return 0;
4945 }
4946
4947 static int diff_opt_compact_summary(const struct option *opt,
4948                                     const char *arg, int unset)
4949 {
4950         struct diff_options *options = opt->value;
4951
4952         BUG_ON_OPT_ARG(arg);
4953         if (unset) {
4954                 options->flags.stat_with_summary = 0;
4955         } else {
4956                 options->flags.stat_with_summary = 1;
4957                 options->output_format |= DIFF_FORMAT_DIFFSTAT;
4958         }
4959         return 0;
4960 }
4961
4962 static int diff_opt_dirstat(const struct option *opt,
4963                             const char *arg, int unset)
4964 {
4965         struct diff_options *options = opt->value;
4966
4967         BUG_ON_OPT_NEG(unset);
4968         if (!strcmp(opt->long_name, "cumulative")) {
4969                 if (arg)
4970                         BUG("how come --cumulative take a value?");
4971                 arg = "cumulative";
4972         } else if (!strcmp(opt->long_name, "dirstat-by-file"))
4973                 parse_dirstat_opt(options, "files");
4974         parse_dirstat_opt(options, arg ? arg : "");
4975         return 0;
4976 }
4977
4978 static int diff_opt_find_copies(const struct option *opt,
4979                                 const char *arg, int unset)
4980 {
4981         struct diff_options *options = opt->value;
4982
4983         BUG_ON_OPT_NEG(unset);
4984         if (!arg)
4985                 arg = "";
4986         options->rename_score = parse_rename_score(&arg);
4987         if (*arg != 0)
4988                 return error(_("invalid argument to %s"), opt->long_name);
4989
4990         if (options->detect_rename == DIFF_DETECT_COPY)
4991                 options->flags.find_copies_harder = 1;
4992         else
4993                 options->detect_rename = DIFF_DETECT_COPY;
4994
4995         return 0;
4996 }
4997
4998 static int diff_opt_find_renames(const struct option *opt,
4999                                  const char *arg, int unset)
5000 {
5001         struct diff_options *options = opt->value;
5002
5003         BUG_ON_OPT_NEG(unset);
5004         if (!arg)
5005                 arg = "";
5006         options->rename_score = parse_rename_score(&arg);
5007         if (*arg != 0)
5008                 return error(_("invalid argument to %s"), opt->long_name);
5009
5010         options->detect_rename = DIFF_DETECT_RENAME;
5011         return 0;
5012 }
5013
5014 static enum parse_opt_result diff_opt_output(struct parse_opt_ctx_t *ctx,
5015                                              const struct option *opt,
5016                                              const char *arg, int unset)
5017 {
5018         struct diff_options *options = opt->value;
5019         char *path;
5020
5021         BUG_ON_OPT_NEG(unset);
5022         path = prefix_filename(ctx->prefix, arg);
5023         options->file = xfopen(path, "w");
5024         options->close_file = 1;
5025         if (options->use_color != GIT_COLOR_ALWAYS)
5026                 options->use_color = GIT_COLOR_NEVER;
5027         free(path);
5028         return 0;
5029 }
5030
5031 static int diff_opt_relative(const struct option *opt,
5032                              const char *arg, int unset)
5033 {
5034         struct diff_options *options = opt->value;
5035
5036         BUG_ON_OPT_NEG(unset);
5037         options->flags.relative_name = 1;
5038         if (arg)
5039                 options->prefix = arg;
5040         return 0;
5041 }
5042
5043 static int diff_opt_unified(const struct option *opt,
5044                             const char *arg, int unset)
5045 {
5046         struct diff_options *options = opt->value;
5047         char *s;
5048
5049         BUG_ON_OPT_NEG(unset);
5050
5051         options->context = strtol(arg, &s, 10);
5052         if (*s)
5053                 return error(_("%s expects a numerical value"), "--unified");
5054         enable_patch_output(&options->output_format);
5055
5056         return 0;
5057 }
5058
5059 static void prep_parse_options(struct diff_options *options)
5060 {
5061         struct option parseopts[] = {
5062                 OPT_GROUP(N_("Diff output format options")),
5063                 OPT_BITOP('p', "patch", &options->output_format,
5064                           N_("generate patch"),
5065                           DIFF_FORMAT_PATCH, DIFF_FORMAT_NO_OUTPUT),
5066                 OPT_BIT_F('s', "no-patch", &options->output_format,
5067                           N_("suppress diff output"),
5068                           DIFF_FORMAT_NO_OUTPUT, PARSE_OPT_NONEG),
5069                 OPT_BITOP('u', NULL, &options->output_format,
5070                           N_("generate patch"),
5071                           DIFF_FORMAT_PATCH, DIFF_FORMAT_NO_OUTPUT),
5072                 OPT_CALLBACK_F('U', "unified", options, N_("<n>"),
5073                                N_("generate diffs with <n> lines context"),
5074                                PARSE_OPT_NONEG, diff_opt_unified),
5075                 OPT_BOOL('W', "function-context", &options->flags.funccontext,
5076                          N_("generate diffs with <n> lines context")),
5077                 OPT_BIT_F(0, "raw", &options->output_format,
5078                           N_("generate the diff in raw format"),
5079                           DIFF_FORMAT_RAW, PARSE_OPT_NONEG),
5080                 OPT_BITOP(0, "patch-with-raw", &options->output_format,
5081                           N_("synonym for '-p --raw'"),
5082                           DIFF_FORMAT_PATCH | DIFF_FORMAT_RAW,
5083                           DIFF_FORMAT_NO_OUTPUT),
5084                 OPT_BITOP(0, "patch-with-stat", &options->output_format,
5085                           N_("synonym for '-p --stat'"),
5086                           DIFF_FORMAT_PATCH | DIFF_FORMAT_DIFFSTAT,
5087                           DIFF_FORMAT_NO_OUTPUT),
5088                 OPT_BIT_F(0, "numstat", &options->output_format,
5089                           N_("machine friendly --stat"),
5090                           DIFF_FORMAT_NUMSTAT, PARSE_OPT_NONEG),
5091                 OPT_BIT_F(0, "shortstat", &options->output_format,
5092                           N_("output only the last line of --stat"),
5093                           DIFF_FORMAT_SHORTSTAT, PARSE_OPT_NONEG),
5094                 OPT_CALLBACK_F('X', "dirstat", options, N_("<param1,param2>..."),
5095                                N_("output the distribution of relative amount of changes for each sub-directory"),
5096                                PARSE_OPT_NONEG | PARSE_OPT_OPTARG,
5097                                diff_opt_dirstat),
5098                 OPT_CALLBACK_F(0, "cumulative", options, NULL,
5099                                N_("synonym for --dirstat=cumulative"),
5100                                PARSE_OPT_NONEG | PARSE_OPT_NOARG,
5101                                diff_opt_dirstat),
5102                 OPT_CALLBACK_F(0, "dirstat-by-file", options, N_("<param1,param2>..."),
5103                                N_("synonym for --dirstat=files,param1,param2..."),
5104                                PARSE_OPT_NONEG | PARSE_OPT_OPTARG,
5105                                diff_opt_dirstat),
5106                 OPT_BIT_F(0, "check", &options->output_format,
5107                           N_("warn if changes introduce conflict markers or whitespace errors"),
5108                           DIFF_FORMAT_CHECKDIFF, PARSE_OPT_NONEG),
5109                 OPT_BIT_F(0, "summary", &options->output_format,
5110                           N_("condensed summary such as creations, renames and mode changes"),
5111                           DIFF_FORMAT_SUMMARY, PARSE_OPT_NONEG),
5112                 OPT_BIT_F(0, "name-only", &options->output_format,
5113                           N_("show only names of changed files"),
5114                           DIFF_FORMAT_NAME, PARSE_OPT_NONEG),
5115                 OPT_BIT_F(0, "name-status", &options->output_format,
5116                           N_("show only names and status of changed files"),
5117                           DIFF_FORMAT_NAME_STATUS, PARSE_OPT_NONEG),
5118                 OPT_CALLBACK_F(0, "stat", options, N_("<width>[,<name-width>[,<count>]]"),
5119                                N_("generate diffstat"),
5120                                PARSE_OPT_NONEG | PARSE_OPT_OPTARG, diff_opt_stat),
5121                 OPT_CALLBACK_F(0, "stat-width", options, N_("<width>"),
5122                                N_("generate diffstat with a given width"),
5123                                PARSE_OPT_NONEG, diff_opt_stat),
5124                 OPT_CALLBACK_F(0, "stat-name-width", options, N_("<width>"),
5125                                N_("generate diffstat with a given name width"),
5126                                PARSE_OPT_NONEG, diff_opt_stat),
5127                 OPT_CALLBACK_F(0, "stat-graph-width", options, N_("<width>"),
5128                                N_("generate diffstat with a given graph width"),
5129                                PARSE_OPT_NONEG, diff_opt_stat),
5130                 OPT_CALLBACK_F(0, "stat-count", options, N_("<count>"),
5131                                N_("generate diffstat with limited lines"),
5132                                PARSE_OPT_NONEG, diff_opt_stat),
5133                 OPT_CALLBACK_F(0, "compact-summary", options, NULL,
5134                                N_("generate compact summary in diffstat"),
5135                                PARSE_OPT_NOARG, diff_opt_compact_summary),
5136                 OPT_CALLBACK_F(0, "output-indicator-new",
5137                                &options->output_indicators[OUTPUT_INDICATOR_NEW],
5138                                N_("<char>"),
5139                                N_("specify the character to indicate a new line instead of '+'"),
5140                                PARSE_OPT_NONEG, diff_opt_char),
5141                 OPT_CALLBACK_F(0, "output-indicator-old",
5142                                &options->output_indicators[OUTPUT_INDICATOR_OLD],
5143                                N_("<char>"),
5144                                N_("specify the character to indicate an old line instead of '-'"),
5145                                PARSE_OPT_NONEG, diff_opt_char),
5146                 OPT_CALLBACK_F(0, "output-indicator-context",
5147                                &options->output_indicators[OUTPUT_INDICATOR_CONTEXT],
5148                                N_("<char>"),
5149                                N_("specify the character to indicate a context instead of ' '"),
5150                                PARSE_OPT_NONEG, diff_opt_char),
5151
5152                 OPT_GROUP(N_("Diff rename options")),
5153                 OPT_CALLBACK_F('B', "break-rewrites", &options->break_opt, N_("<n>[/<m>]"),
5154                                N_("break complete rewrite changes into pairs of delete and create"),
5155                                PARSE_OPT_NONEG | PARSE_OPT_OPTARG,
5156                                diff_opt_break_rewrites),
5157                 OPT_CALLBACK_F('M', "find-renames", options, N_("<n>"),
5158                                N_("detect renames"),
5159                                PARSE_OPT_NONEG | PARSE_OPT_OPTARG,
5160                                diff_opt_find_renames),
5161                 OPT_SET_INT_F('D', "irreversible-delete", &options->irreversible_delete,
5162                               N_("omit the preimage for deletes"),
5163                               1, PARSE_OPT_NONEG),
5164                 OPT_CALLBACK_F('C', "find-copies", options, N_("<n>"),
5165                                N_("detect copies"),
5166                                PARSE_OPT_NONEG | PARSE_OPT_OPTARG,
5167                                diff_opt_find_copies),
5168                 OPT_BOOL(0, "find-copies-harder", &options->flags.find_copies_harder,
5169                          N_("use unmodified files as source to find copies")),
5170                 OPT_SET_INT_F(0, "no-renames", &options->detect_rename,
5171                               N_("disable rename detection"),
5172                               0, PARSE_OPT_NONEG),
5173                 OPT_BOOL(0, "rename-empty", &options->flags.rename_empty,
5174                          N_("use empty blobs as rename source")),
5175
5176                 OPT_GROUP(N_("Diff algorithm options")),
5177                 OPT_BIT(0, "minimal", &options->xdl_opts,
5178                         N_("produce the smallest possible diff"),
5179                         XDF_NEED_MINIMAL),
5180                 OPT_BIT_F('w', "ignore-all-space", &options->xdl_opts,
5181                           N_("ignore whitespace when comparing lines"),
5182                           XDF_IGNORE_WHITESPACE, PARSE_OPT_NONEG),
5183                 OPT_BIT_F('b', "ignore-space-change", &options->xdl_opts,
5184                           N_("ignore changes in amount of whitespace"),
5185                           XDF_IGNORE_WHITESPACE_CHANGE, PARSE_OPT_NONEG),
5186                 OPT_BIT_F(0, "ignore-space-at-eol", &options->xdl_opts,
5187                           N_("ignore changes in whitespace at EOL"),
5188                           XDF_IGNORE_WHITESPACE_AT_EOL, PARSE_OPT_NONEG),
5189                 OPT_BIT_F(0, "ignore-cr-at-eol", &options->xdl_opts,
5190                           N_("ignore carrier-return at the end of line"),
5191                           XDF_IGNORE_CR_AT_EOL, PARSE_OPT_NONEG),
5192                 OPT_BIT_F(0, "ignore-blank-lines", &options->xdl_opts,
5193                           N_("ignore changes whose lines are all blank"),
5194                           XDF_IGNORE_BLANK_LINES, PARSE_OPT_NONEG),
5195
5196                 OPT_GROUP(N_("Diff other options")),
5197                 OPT_CALLBACK_F(0, "relative", options, N_("<prefix>"),
5198                                N_("when run from subdir, exclude changes outside and show relative paths"),
5199                                PARSE_OPT_NONEG | PARSE_OPT_OPTARG,
5200                                diff_opt_relative),
5201                 { OPTION_CALLBACK, 0, "output", options, N_("<file>"),
5202                   N_("Output to a specific file"),
5203                   PARSE_OPT_NONEG, NULL, 0, diff_opt_output },
5204
5205                 OPT_END()
5206         };
5207
5208         ALLOC_ARRAY(options->parseopts, ARRAY_SIZE(parseopts));
5209         memcpy(options->parseopts, parseopts, sizeof(parseopts));
5210 }
5211
5212 int diff_opt_parse(struct diff_options *options,
5213                    const char **av, int ac, const char *prefix)
5214 {
5215         const char *arg = av[0];
5216         const char *optarg;
5217         int argcount;
5218
5219         if (!prefix)
5220                 prefix = "";
5221
5222         ac = parse_options(ac, av, prefix, options->parseopts, NULL,
5223                            PARSE_OPT_KEEP_DASHDASH |
5224                            PARSE_OPT_KEEP_UNKNOWN |
5225                            PARSE_OPT_NO_INTERNAL_HELP |
5226                            PARSE_OPT_ONE_SHOT |
5227                            PARSE_OPT_STOP_AT_NON_OPTION);
5228
5229         if (ac)
5230                 return ac;
5231
5232         /* xdiff options */
5233         if (!strcmp(arg, "--indent-heuristic"))
5234                 DIFF_XDL_SET(options, INDENT_HEURISTIC);
5235         else if (!strcmp(arg, "--no-indent-heuristic"))
5236                 DIFF_XDL_CLR(options, INDENT_HEURISTIC);
5237         else if (!strcmp(arg, "--patience")) {
5238                 int i;
5239                 options->xdl_opts = DIFF_WITH_ALG(options, PATIENCE_DIFF);
5240                 /*
5241                  * Both --patience and --anchored use PATIENCE_DIFF
5242                  * internally, so remove any anchors previously
5243                  * specified.
5244                  */
5245                 for (i = 0; i < options->anchors_nr; i++)
5246                         free(options->anchors[i]);
5247                 options->anchors_nr = 0;
5248         } else if (!strcmp(arg, "--histogram"))
5249                 options->xdl_opts = DIFF_WITH_ALG(options, HISTOGRAM_DIFF);
5250         else if ((argcount = parse_long_opt("diff-algorithm", av, &optarg))) {
5251                 long value = parse_algorithm_value(optarg);
5252                 if (value < 0)
5253                         return error("option diff-algorithm accepts \"myers\", "
5254                                      "\"minimal\", \"patience\" and \"histogram\"");
5255                 /* clear out previous settings */
5256                 DIFF_XDL_CLR(options, NEED_MINIMAL);
5257                 options->xdl_opts &= ~XDF_DIFF_ALGORITHM_MASK;
5258                 options->xdl_opts |= value;
5259                 return argcount;
5260         } else if (skip_prefix(arg, "--anchored=", &arg)) {
5261                 options->xdl_opts = DIFF_WITH_ALG(options, PATIENCE_DIFF);
5262                 ALLOC_GROW(options->anchors, options->anchors_nr + 1,
5263                            options->anchors_alloc);
5264                 options->anchors[options->anchors_nr++] = xstrdup(arg);
5265         }
5266
5267         /* flags options */
5268         else if (!strcmp(arg, "--binary")) {
5269                 enable_patch_output(&options->output_format);
5270                 options->flags.binary = 1;
5271         }
5272         else if (!strcmp(arg, "--full-index"))
5273                 options->flags.full_index = 1;
5274         else if (!strcmp(arg, "-a") || !strcmp(arg, "--text"))
5275                 options->flags.text = 1;
5276         else if (!strcmp(arg, "-R"))
5277                 options->flags.reverse_diff = 1;
5278         else if (!strcmp(arg, "--follow"))
5279                 options->flags.follow_renames = 1;
5280         else if (!strcmp(arg, "--no-follow")) {
5281                 options->flags.follow_renames = 0;
5282                 options->flags.default_follow_renames = 0;
5283         } else if (skip_to_optional_arg_default(arg, "--color", &arg, "always")) {
5284                 int value = git_config_colorbool(NULL, arg);
5285                 if (value < 0)
5286                         return error("option `color' expects \"always\", \"auto\", or \"never\"");
5287                 options->use_color = value;
5288         }
5289         else if (!strcmp(arg, "--no-color"))
5290                 options->use_color = 0;
5291         else if (!strcmp(arg, "--color-moved")) {
5292                 if (diff_color_moved_default)
5293                         options->color_moved = diff_color_moved_default;
5294                 if (options->color_moved == COLOR_MOVED_NO)
5295                         options->color_moved = COLOR_MOVED_DEFAULT;
5296         } else if (!strcmp(arg, "--no-color-moved"))
5297                 options->color_moved = COLOR_MOVED_NO;
5298         else if (skip_prefix(arg, "--color-moved=", &arg)) {
5299                 int cm = parse_color_moved(arg);
5300                 if (cm < 0)
5301                         return error("bad --color-moved argument: %s", arg);
5302                 options->color_moved = cm;
5303         } else if (!strcmp(arg, "--no-color-moved-ws")) {
5304                 options->color_moved_ws_handling = 0;
5305         } else if (skip_prefix(arg, "--color-moved-ws=", &arg)) {
5306                 unsigned cm = parse_color_moved_ws(arg);
5307                 if (cm & COLOR_MOVED_WS_ERROR)
5308                         return -1;
5309                 options->color_moved_ws_handling = cm;
5310         } else if (skip_to_optional_arg_default(arg, "--color-words", &options->word_regex, NULL)) {
5311                 options->use_color = 1;
5312                 options->word_diff = DIFF_WORDS_COLOR;
5313         }
5314         else if (!strcmp(arg, "--word-diff")) {
5315                 if (options->word_diff == DIFF_WORDS_NONE)
5316                         options->word_diff = DIFF_WORDS_PLAIN;
5317         }
5318         else if (skip_prefix(arg, "--word-diff=", &arg)) {
5319                 if (!strcmp(arg, "plain"))
5320                         options->word_diff = DIFF_WORDS_PLAIN;
5321                 else if (!strcmp(arg, "color")) {
5322                         options->use_color = 1;
5323                         options->word_diff = DIFF_WORDS_COLOR;
5324                 }
5325                 else if (!strcmp(arg, "porcelain"))
5326                         options->word_diff = DIFF_WORDS_PORCELAIN;
5327                 else if (!strcmp(arg, "none"))
5328                         options->word_diff = DIFF_WORDS_NONE;
5329                 else
5330                         die("bad --word-diff argument: %s", arg);
5331         }
5332         else if ((argcount = parse_long_opt("word-diff-regex", av, &optarg))) {
5333                 if (options->word_diff == DIFF_WORDS_NONE)
5334                         options->word_diff = DIFF_WORDS_PLAIN;
5335                 options->word_regex = optarg;
5336                 return argcount;
5337         }
5338         else if (!strcmp(arg, "--exit-code"))
5339                 options->flags.exit_with_status = 1;
5340         else if (!strcmp(arg, "--quiet"))
5341                 options->flags.quick = 1;
5342         else if (!strcmp(arg, "--ext-diff"))
5343                 options->flags.allow_external = 1;
5344         else if (!strcmp(arg, "--no-ext-diff"))
5345                 options->flags.allow_external = 0;
5346         else if (!strcmp(arg, "--textconv")) {
5347                 options->flags.allow_textconv = 1;
5348                 options->flags.textconv_set_via_cmdline = 1;
5349         } else if (!strcmp(arg, "--no-textconv"))
5350                 options->flags.allow_textconv = 0;
5351         else if (skip_to_optional_arg_default(arg, "--ignore-submodules", &arg, "all")) {
5352                 options->flags.override_submodule_config = 1;
5353                 handle_ignore_submodules_arg(options, arg);
5354         } else if (skip_to_optional_arg_default(arg, "--submodule", &arg, "log"))
5355                 return parse_submodule_opt(options, arg);
5356         else if (skip_prefix(arg, "--ws-error-highlight=", &arg))
5357                 return parse_ws_error_highlight_opt(options, arg);
5358         else if (!strcmp(arg, "--ita-invisible-in-index"))
5359                 options->ita_invisible_in_index = 1;
5360         else if (!strcmp(arg, "--ita-visible-in-index"))
5361                 options->ita_invisible_in_index = 0;
5362
5363         /* misc options */
5364         else if (!strcmp(arg, "-z"))
5365                 options->line_termination = 0;
5366         else if ((argcount = short_opt('l', av, &optarg))) {
5367                 options->rename_limit = strtoul(optarg, NULL, 10);
5368                 return argcount;
5369         }
5370         else if ((argcount = short_opt('S', av, &optarg))) {
5371                 options->pickaxe = optarg;
5372                 options->pickaxe_opts |= DIFF_PICKAXE_KIND_S;
5373                 return argcount;
5374         } else if ((argcount = short_opt('G', av, &optarg))) {
5375                 options->pickaxe = optarg;
5376                 options->pickaxe_opts |= DIFF_PICKAXE_KIND_G;
5377                 return argcount;
5378         }
5379         else if (!strcmp(arg, "--pickaxe-all"))
5380                 options->pickaxe_opts |= DIFF_PICKAXE_ALL;
5381         else if (!strcmp(arg, "--pickaxe-regex"))
5382                 options->pickaxe_opts |= DIFF_PICKAXE_REGEX;
5383         else if ((argcount = short_opt('O', av, &optarg))) {
5384                 options->orderfile = prefix_filename(prefix, optarg);
5385                 return argcount;
5386         } else if (skip_prefix(arg, "--find-object=", &arg))
5387                 return parse_objfind_opt(options, arg);
5388         else if ((argcount = parse_long_opt("diff-filter", av, &optarg))) {
5389                 int offending = parse_diff_filter_opt(optarg, options);
5390                 if (offending)
5391                         die("unknown change class '%c' in --diff-filter=%s",
5392                             offending, optarg);
5393                 return argcount;
5394         }
5395         else if (!strcmp(arg, "--no-abbrev"))
5396                 options->abbrev = 0;
5397         else if (!strcmp(arg, "--abbrev"))
5398                 options->abbrev = DEFAULT_ABBREV;
5399         else if (skip_prefix(arg, "--abbrev=", &arg)) {
5400                 options->abbrev = strtoul(arg, NULL, 10);
5401                 if (options->abbrev < MINIMUM_ABBREV)
5402                         options->abbrev = MINIMUM_ABBREV;
5403                 else if (the_hash_algo->hexsz < options->abbrev)
5404                         options->abbrev = the_hash_algo->hexsz;
5405         }
5406         else if ((argcount = parse_long_opt("src-prefix", av, &optarg))) {
5407                 options->a_prefix = optarg;
5408                 return argcount;
5409         }
5410         else if ((argcount = parse_long_opt("line-prefix", av, &optarg))) {
5411                 options->line_prefix = optarg;
5412                 options->line_prefix_length = strlen(options->line_prefix);
5413                 graph_setup_line_prefix(options);
5414                 return argcount;
5415         }
5416         else if ((argcount = parse_long_opt("dst-prefix", av, &optarg))) {
5417                 options->b_prefix = optarg;
5418                 return argcount;
5419         }
5420         else if (!strcmp(arg, "--no-prefix"))
5421                 options->a_prefix = options->b_prefix = "";
5422         else if (opt_arg(arg, '\0', "inter-hunk-context",
5423                          &options->interhunkcontext))
5424                 ;
5425         else
5426                 return 0;
5427         return 1;
5428 }
5429
5430 int parse_rename_score(const char **cp_p)
5431 {
5432         unsigned long num, scale;
5433         int ch, dot;
5434         const char *cp = *cp_p;
5435
5436         num = 0;
5437         scale = 1;
5438         dot = 0;
5439         for (;;) {
5440                 ch = *cp;
5441                 if ( !dot && ch == '.' ) {
5442                         scale = 1;
5443                         dot = 1;
5444                 } else if ( ch == '%' ) {
5445                         scale = dot ? scale*100 : 100;
5446                         cp++;   /* % is always at the end */
5447                         break;
5448                 } else if ( ch >= '0' && ch <= '9' ) {
5449                         if ( scale < 100000 ) {
5450                                 scale *= 10;
5451                                 num = (num*10) + (ch-'0');
5452                         }
5453                 } else {
5454                         break;
5455                 }
5456                 cp++;
5457         }
5458         *cp_p = cp;
5459
5460         /* user says num divided by scale and we say internally that
5461          * is MAX_SCORE * num / scale.
5462          */
5463         return (int)((num >= scale) ? MAX_SCORE : (MAX_SCORE * num / scale));
5464 }
5465
5466 struct diff_queue_struct diff_queued_diff;
5467
5468 void diff_q(struct diff_queue_struct *queue, struct diff_filepair *dp)
5469 {
5470         ALLOC_GROW(queue->queue, queue->nr + 1, queue->alloc);
5471         queue->queue[queue->nr++] = dp;
5472 }
5473
5474 struct diff_filepair *diff_queue(struct diff_queue_struct *queue,
5475                                  struct diff_filespec *one,
5476                                  struct diff_filespec *two)
5477 {
5478         struct diff_filepair *dp = xcalloc(1, sizeof(*dp));
5479         dp->one = one;
5480         dp->two = two;
5481         if (queue)
5482                 diff_q(queue, dp);
5483         return dp;
5484 }
5485
5486 void diff_free_filepair(struct diff_filepair *p)
5487 {
5488         free_filespec(p->one);
5489         free_filespec(p->two);
5490         free(p);
5491 }
5492
5493 const char *diff_aligned_abbrev(const struct object_id *oid, int len)
5494 {
5495         int abblen;
5496         const char *abbrev;
5497
5498         /* Do we want all 40 hex characters? */
5499         if (len == the_hash_algo->hexsz)
5500                 return oid_to_hex(oid);
5501
5502         /* An abbreviated value is fine, possibly followed by an ellipsis. */
5503         abbrev = diff_abbrev_oid(oid, len);
5504
5505         if (!print_sha1_ellipsis())
5506                 return abbrev;
5507
5508         abblen = strlen(abbrev);
5509
5510         /*
5511          * In well-behaved cases, where the abbreviated result is the
5512          * same as the requested length, append three dots after the
5513          * abbreviation (hence the whole logic is limited to the case
5514          * where abblen < 37); when the actual abbreviated result is a
5515          * bit longer than the requested length, we reduce the number
5516          * of dots so that they match the well-behaved ones.  However,
5517          * if the actual abbreviation is longer than the requested
5518          * length by more than three, we give up on aligning, and add
5519          * three dots anyway, to indicate that the output is not the
5520          * full object name.  Yes, this may be suboptimal, but this
5521          * appears only in "diff --raw --abbrev" output and it is not
5522          * worth the effort to change it now.  Note that this would
5523          * likely to work fine when the automatic sizing of default
5524          * abbreviation length is used--we would be fed -1 in "len" in
5525          * that case, and will end up always appending three-dots, but
5526          * the automatic sizing is supposed to give abblen that ensures
5527          * uniqueness across all objects (statistically speaking).
5528          */
5529         if (abblen < the_hash_algo->hexsz - 3) {
5530                 static char hex[GIT_MAX_HEXSZ + 1];
5531                 if (len < abblen && abblen <= len + 2)
5532                         xsnprintf(hex, sizeof(hex), "%s%.*s", abbrev, len+3-abblen, "..");
5533                 else
5534                         xsnprintf(hex, sizeof(hex), "%s...", abbrev);
5535                 return hex;
5536         }
5537
5538         return oid_to_hex(oid);
5539 }
5540
5541 static void diff_flush_raw(struct diff_filepair *p, struct diff_options *opt)
5542 {
5543         int line_termination = opt->line_termination;
5544         int inter_name_termination = line_termination ? '\t' : '\0';
5545
5546         fprintf(opt->file, "%s", diff_line_prefix(opt));
5547         if (!(opt->output_format & DIFF_FORMAT_NAME_STATUS)) {
5548                 fprintf(opt->file, ":%06o %06o %s ", p->one->mode, p->two->mode,
5549                         diff_aligned_abbrev(&p->one->oid, opt->abbrev));
5550                 fprintf(opt->file, "%s ",
5551                         diff_aligned_abbrev(&p->two->oid, opt->abbrev));
5552         }
5553         if (p->score) {
5554                 fprintf(opt->file, "%c%03d%c", p->status, similarity_index(p),
5555                         inter_name_termination);
5556         } else {
5557                 fprintf(opt->file, "%c%c", p->status, inter_name_termination);
5558         }
5559
5560         if (p->status == DIFF_STATUS_COPIED ||
5561             p->status == DIFF_STATUS_RENAMED) {
5562                 const char *name_a, *name_b;
5563                 name_a = p->one->path;
5564                 name_b = p->two->path;
5565                 strip_prefix(opt->prefix_length, &name_a, &name_b);
5566                 write_name_quoted(name_a, opt->file, inter_name_termination);
5567                 write_name_quoted(name_b, opt->file, line_termination);
5568         } else {
5569                 const char *name_a, *name_b;
5570                 name_a = p->one->mode ? p->one->path : p->two->path;
5571                 name_b = NULL;
5572                 strip_prefix(opt->prefix_length, &name_a, &name_b);
5573                 write_name_quoted(name_a, opt->file, line_termination);
5574         }
5575 }
5576
5577 int diff_unmodified_pair(struct diff_filepair *p)
5578 {
5579         /* This function is written stricter than necessary to support
5580          * the currently implemented transformers, but the idea is to
5581          * let transformers to produce diff_filepairs any way they want,
5582          * and filter and clean them up here before producing the output.
5583          */
5584         struct diff_filespec *one = p->one, *two = p->two;
5585
5586         if (DIFF_PAIR_UNMERGED(p))
5587                 return 0; /* unmerged is interesting */
5588
5589         /* deletion, addition, mode or type change
5590          * and rename are all interesting.
5591          */
5592         if (DIFF_FILE_VALID(one) != DIFF_FILE_VALID(two) ||
5593             DIFF_PAIR_MODE_CHANGED(p) ||
5594             strcmp(one->path, two->path))
5595                 return 0;
5596
5597         /* both are valid and point at the same path.  that is, we are
5598          * dealing with a change.
5599          */
5600         if (one->oid_valid && two->oid_valid &&
5601             oideq(&one->oid, &two->oid) &&
5602             !one->dirty_submodule && !two->dirty_submodule)
5603                 return 1; /* no change */
5604         if (!one->oid_valid && !two->oid_valid)
5605                 return 1; /* both look at the same file on the filesystem. */
5606         return 0;
5607 }
5608
5609 static void diff_flush_patch(struct diff_filepair *p, struct diff_options *o)
5610 {
5611         if (diff_unmodified_pair(p))
5612                 return;
5613
5614         if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
5615             (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
5616                 return; /* no tree diffs in patch format */
5617
5618         run_diff(p, o);
5619 }
5620
5621 static void diff_flush_stat(struct diff_filepair *p, struct diff_options *o,
5622                             struct diffstat_t *diffstat)
5623 {
5624         if (diff_unmodified_pair(p))
5625                 return;
5626
5627         if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
5628             (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
5629                 return; /* no useful stat for tree diffs */
5630
5631         run_diffstat(p, o, diffstat);
5632 }
5633
5634 static void diff_flush_checkdiff(struct diff_filepair *p,
5635                 struct diff_options *o)
5636 {
5637         if (diff_unmodified_pair(p))
5638                 return;
5639
5640         if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
5641             (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
5642                 return; /* nothing to check in tree diffs */
5643
5644         run_checkdiff(p, o);
5645 }
5646
5647 int diff_queue_is_empty(void)
5648 {
5649         struct diff_queue_struct *q = &diff_queued_diff;
5650         int i;
5651         for (i = 0; i < q->nr; i++)
5652                 if (!diff_unmodified_pair(q->queue[i]))
5653                         return 0;
5654         return 1;
5655 }
5656
5657 #if DIFF_DEBUG
5658 void diff_debug_filespec(struct diff_filespec *s, int x, const char *one)
5659 {
5660         fprintf(stderr, "queue[%d] %s (%s) %s %06o %s\n",
5661                 x, one ? one : "",
5662                 s->path,
5663                 DIFF_FILE_VALID(s) ? "valid" : "invalid",
5664                 s->mode,
5665                 s->oid_valid ? oid_to_hex(&s->oid) : "");
5666         fprintf(stderr, "queue[%d] %s size %lu\n",
5667                 x, one ? one : "",
5668                 s->size);
5669 }
5670
5671 void diff_debug_filepair(const struct diff_filepair *p, int i)
5672 {
5673         diff_debug_filespec(p->one, i, "one");
5674         diff_debug_filespec(p->two, i, "two");
5675         fprintf(stderr, "score %d, status %c rename_used %d broken %d\n",
5676                 p->score, p->status ? p->status : '?',
5677                 p->one->rename_used, p->broken_pair);
5678 }
5679
5680 void diff_debug_queue(const char *msg, struct diff_queue_struct *q)
5681 {
5682         int i;
5683         if (msg)
5684                 fprintf(stderr, "%s\n", msg);
5685         fprintf(stderr, "q->nr = %d\n", q->nr);
5686         for (i = 0; i < q->nr; i++) {
5687                 struct diff_filepair *p = q->queue[i];
5688                 diff_debug_filepair(p, i);
5689         }
5690 }
5691 #endif
5692
5693 static void diff_resolve_rename_copy(void)
5694 {
5695         int i;
5696         struct diff_filepair *p;
5697         struct diff_queue_struct *q = &diff_queued_diff;
5698
5699         diff_debug_queue("resolve-rename-copy", q);
5700
5701         for (i = 0; i < q->nr; i++) {
5702                 p = q->queue[i];
5703                 p->status = 0; /* undecided */
5704                 if (DIFF_PAIR_UNMERGED(p))
5705                         p->status = DIFF_STATUS_UNMERGED;
5706                 else if (!DIFF_FILE_VALID(p->one))
5707                         p->status = DIFF_STATUS_ADDED;
5708                 else if (!DIFF_FILE_VALID(p->two))
5709                         p->status = DIFF_STATUS_DELETED;
5710                 else if (DIFF_PAIR_TYPE_CHANGED(p))
5711                         p->status = DIFF_STATUS_TYPE_CHANGED;
5712
5713                 /* from this point on, we are dealing with a pair
5714                  * whose both sides are valid and of the same type, i.e.
5715                  * either in-place edit or rename/copy edit.
5716                  */
5717                 else if (DIFF_PAIR_RENAME(p)) {
5718                         /*
5719                          * A rename might have re-connected a broken
5720                          * pair up, causing the pathnames to be the
5721                          * same again. If so, that's not a rename at
5722                          * all, just a modification..
5723                          *
5724                          * Otherwise, see if this source was used for
5725                          * multiple renames, in which case we decrement
5726                          * the count, and call it a copy.
5727                          */
5728                         if (!strcmp(p->one->path, p->two->path))
5729                                 p->status = DIFF_STATUS_MODIFIED;
5730                         else if (--p->one->rename_used > 0)
5731                                 p->status = DIFF_STATUS_COPIED;
5732                         else
5733                                 p->status = DIFF_STATUS_RENAMED;
5734                 }
5735                 else if (!oideq(&p->one->oid, &p->two->oid) ||
5736                          p->one->mode != p->two->mode ||
5737                          p->one->dirty_submodule ||
5738                          p->two->dirty_submodule ||
5739                          is_null_oid(&p->one->oid))
5740                         p->status = DIFF_STATUS_MODIFIED;
5741                 else {
5742                         /* This is a "no-change" entry and should not
5743                          * happen anymore, but prepare for broken callers.
5744                          */
5745                         error("feeding unmodified %s to diffcore",
5746                               p->one->path);
5747                         p->status = DIFF_STATUS_UNKNOWN;
5748                 }
5749         }
5750         diff_debug_queue("resolve-rename-copy done", q);
5751 }
5752
5753 static int check_pair_status(struct diff_filepair *p)
5754 {
5755         switch (p->status) {
5756         case DIFF_STATUS_UNKNOWN:
5757                 return 0;
5758         case 0:
5759                 die("internal error in diff-resolve-rename-copy");
5760         default:
5761                 return 1;
5762         }
5763 }
5764
5765 static void flush_one_pair(struct diff_filepair *p, struct diff_options *opt)
5766 {
5767         int fmt = opt->output_format;
5768
5769         if (fmt & DIFF_FORMAT_CHECKDIFF)
5770                 diff_flush_checkdiff(p, opt);
5771         else if (fmt & (DIFF_FORMAT_RAW | DIFF_FORMAT_NAME_STATUS))
5772                 diff_flush_raw(p, opt);
5773         else if (fmt & DIFF_FORMAT_NAME) {
5774                 const char *name_a, *name_b;
5775                 name_a = p->two->path;
5776                 name_b = NULL;
5777                 strip_prefix(opt->prefix_length, &name_a, &name_b);
5778                 fprintf(opt->file, "%s", diff_line_prefix(opt));
5779                 write_name_quoted(name_a, opt->file, opt->line_termination);
5780         }
5781 }
5782
5783 static void show_file_mode_name(struct diff_options *opt, const char *newdelete, struct diff_filespec *fs)
5784 {
5785         struct strbuf sb = STRBUF_INIT;
5786         if (fs->mode)
5787                 strbuf_addf(&sb, " %s mode %06o ", newdelete, fs->mode);
5788         else
5789                 strbuf_addf(&sb, " %s ", newdelete);
5790
5791         quote_c_style(fs->path, &sb, NULL, 0);
5792         strbuf_addch(&sb, '\n');
5793         emit_diff_symbol(opt, DIFF_SYMBOL_SUMMARY,
5794                          sb.buf, sb.len, 0);
5795         strbuf_release(&sb);
5796 }
5797
5798 static void show_mode_change(struct diff_options *opt, struct diff_filepair *p,
5799                 int show_name)
5800 {
5801         if (p->one->mode && p->two->mode && p->one->mode != p->two->mode) {
5802                 struct strbuf sb = STRBUF_INIT;
5803                 strbuf_addf(&sb, " mode change %06o => %06o",
5804                             p->one->mode, p->two->mode);
5805                 if (show_name) {
5806                         strbuf_addch(&sb, ' ');
5807                         quote_c_style(p->two->path, &sb, NULL, 0);
5808                 }
5809                 strbuf_addch(&sb, '\n');
5810                 emit_diff_symbol(opt, DIFF_SYMBOL_SUMMARY,
5811                                  sb.buf, sb.len, 0);
5812                 strbuf_release(&sb);
5813         }
5814 }
5815
5816 static void show_rename_copy(struct diff_options *opt, const char *renamecopy,
5817                 struct diff_filepair *p)
5818 {
5819         struct strbuf sb = STRBUF_INIT;
5820         struct strbuf names = STRBUF_INIT;
5821
5822         pprint_rename(&names, p->one->path, p->two->path);
5823         strbuf_addf(&sb, " %s %s (%d%%)\n",
5824                     renamecopy, names.buf, similarity_index(p));
5825         strbuf_release(&names);
5826         emit_diff_symbol(opt, DIFF_SYMBOL_SUMMARY,
5827                                  sb.buf, sb.len, 0);
5828         show_mode_change(opt, p, 0);
5829         strbuf_release(&sb);
5830 }
5831
5832 static void diff_summary(struct diff_options *opt, struct diff_filepair *p)
5833 {
5834         switch(p->status) {
5835         case DIFF_STATUS_DELETED:
5836                 show_file_mode_name(opt, "delete", p->one);
5837                 break;
5838         case DIFF_STATUS_ADDED:
5839                 show_file_mode_name(opt, "create", p->two);
5840                 break;
5841         case DIFF_STATUS_COPIED:
5842                 show_rename_copy(opt, "copy", p);
5843                 break;
5844         case DIFF_STATUS_RENAMED:
5845                 show_rename_copy(opt, "rename", p);
5846                 break;
5847         default:
5848                 if (p->score) {
5849                         struct strbuf sb = STRBUF_INIT;
5850                         strbuf_addstr(&sb, " rewrite ");
5851                         quote_c_style(p->two->path, &sb, NULL, 0);
5852                         strbuf_addf(&sb, " (%d%%)\n", similarity_index(p));
5853                         emit_diff_symbol(opt, DIFF_SYMBOL_SUMMARY,
5854                                          sb.buf, sb.len, 0);
5855                         strbuf_release(&sb);
5856                 }
5857                 show_mode_change(opt, p, !p->score);
5858                 break;
5859         }
5860 }
5861
5862 struct patch_id_t {
5863         git_SHA_CTX *ctx;
5864         int patchlen;
5865 };
5866
5867 static int remove_space(char *line, int len)
5868 {
5869         int i;
5870         char *dst = line;
5871         unsigned char c;
5872
5873         for (i = 0; i < len; i++)
5874                 if (!isspace((c = line[i])))
5875                         *dst++ = c;
5876
5877         return dst - line;
5878 }
5879
5880 static void patch_id_consume(void *priv, char *line, unsigned long len)
5881 {
5882         struct patch_id_t *data = priv;
5883         int new_len;
5884
5885         new_len = remove_space(line, len);
5886
5887         git_SHA1_Update(data->ctx, line, new_len);
5888         data->patchlen += new_len;
5889 }
5890
5891 static void patch_id_add_string(git_SHA_CTX *ctx, const char *str)
5892 {
5893         git_SHA1_Update(ctx, str, strlen(str));
5894 }
5895
5896 static void patch_id_add_mode(git_SHA_CTX *ctx, unsigned mode)
5897 {
5898         /* large enough for 2^32 in octal */
5899         char buf[12];
5900         int len = xsnprintf(buf, sizeof(buf), "%06o", mode);
5901         git_SHA1_Update(ctx, buf, len);
5902 }
5903
5904 /* returns 0 upon success, and writes result into sha1 */
5905 static int diff_get_patch_id(struct diff_options *options, struct object_id *oid, int diff_header_only)
5906 {
5907         struct diff_queue_struct *q = &diff_queued_diff;
5908         int i;
5909         git_SHA_CTX ctx;
5910         struct patch_id_t data;
5911
5912         git_SHA1_Init(&ctx);
5913         memset(&data, 0, sizeof(struct patch_id_t));
5914         data.ctx = &ctx;
5915
5916         for (i = 0; i < q->nr; i++) {
5917                 xpparam_t xpp;
5918                 xdemitconf_t xecfg;
5919                 mmfile_t mf1, mf2;
5920                 struct diff_filepair *p = q->queue[i];
5921                 int len1, len2;
5922
5923                 memset(&xpp, 0, sizeof(xpp));
5924                 memset(&xecfg, 0, sizeof(xecfg));
5925                 if (p->status == 0)
5926                         return error("internal diff status error");
5927                 if (p->status == DIFF_STATUS_UNKNOWN)
5928                         continue;
5929                 if (diff_unmodified_pair(p))
5930                         continue;
5931                 if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
5932                     (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
5933                         continue;
5934                 if (DIFF_PAIR_UNMERGED(p))
5935                         continue;
5936
5937                 diff_fill_oid_info(p->one, options->repo->index);
5938                 diff_fill_oid_info(p->two, options->repo->index);
5939
5940                 len1 = remove_space(p->one->path, strlen(p->one->path));
5941                 len2 = remove_space(p->two->path, strlen(p->two->path));
5942                 patch_id_add_string(&ctx, "diff--git");
5943                 patch_id_add_string(&ctx, "a/");
5944                 git_SHA1_Update(&ctx, p->one->path, len1);
5945                 patch_id_add_string(&ctx, "b/");
5946                 git_SHA1_Update(&ctx, p->two->path, len2);
5947
5948                 if (p->one->mode == 0) {
5949                         patch_id_add_string(&ctx, "newfilemode");
5950                         patch_id_add_mode(&ctx, p->two->mode);
5951                         patch_id_add_string(&ctx, "---/dev/null");
5952                         patch_id_add_string(&ctx, "+++b/");
5953                         git_SHA1_Update(&ctx, p->two->path, len2);
5954                 } else if (p->two->mode == 0) {
5955                         patch_id_add_string(&ctx, "deletedfilemode");
5956                         patch_id_add_mode(&ctx, p->one->mode);
5957                         patch_id_add_string(&ctx, "---a/");
5958                         git_SHA1_Update(&ctx, p->one->path, len1);
5959                         patch_id_add_string(&ctx, "+++/dev/null");
5960                 } else {
5961                         patch_id_add_string(&ctx, "---a/");
5962                         git_SHA1_Update(&ctx, p->one->path, len1);
5963                         patch_id_add_string(&ctx, "+++b/");
5964                         git_SHA1_Update(&ctx, p->two->path, len2);
5965                 }
5966
5967                 if (diff_header_only)
5968                         continue;
5969
5970                 if (fill_mmfile(options->repo, &mf1, p->one) < 0 ||
5971                     fill_mmfile(options->repo, &mf2, p->two) < 0)
5972                         return error("unable to read files to diff");
5973
5974                 if (diff_filespec_is_binary(options->repo, p->one) ||
5975                     diff_filespec_is_binary(options->repo, p->two)) {
5976                         git_SHA1_Update(&ctx, oid_to_hex(&p->one->oid),
5977                                         GIT_SHA1_HEXSZ);
5978                         git_SHA1_Update(&ctx, oid_to_hex(&p->two->oid),
5979                                         GIT_SHA1_HEXSZ);
5980                         continue;
5981                 }
5982
5983                 xpp.flags = 0;
5984                 xecfg.ctxlen = 3;
5985                 xecfg.flags = 0;
5986                 if (xdi_diff_outf(&mf1, &mf2, discard_hunk_line,
5987                                   patch_id_consume, &data, &xpp, &xecfg))
5988                         return error("unable to generate patch-id diff for %s",
5989                                      p->one->path);
5990         }
5991
5992         git_SHA1_Final(oid->hash, &ctx);
5993         return 0;
5994 }
5995
5996 int diff_flush_patch_id(struct diff_options *options, struct object_id *oid, int diff_header_only)
5997 {
5998         struct diff_queue_struct *q = &diff_queued_diff;
5999         int i;
6000         int result = diff_get_patch_id(options, oid, diff_header_only);
6001
6002         for (i = 0; i < q->nr; i++)
6003                 diff_free_filepair(q->queue[i]);
6004
6005         free(q->queue);
6006         DIFF_QUEUE_CLEAR(q);
6007
6008         return result;
6009 }
6010
6011 static int is_summary_empty(const struct diff_queue_struct *q)
6012 {
6013         int i;
6014
6015         for (i = 0; i < q->nr; i++) {
6016                 const struct diff_filepair *p = q->queue[i];
6017
6018                 switch (p->status) {
6019                 case DIFF_STATUS_DELETED:
6020                 case DIFF_STATUS_ADDED:
6021                 case DIFF_STATUS_COPIED:
6022                 case DIFF_STATUS_RENAMED:
6023                         return 0;
6024                 default:
6025                         if (p->score)
6026                                 return 0;
6027                         if (p->one->mode && p->two->mode &&
6028                             p->one->mode != p->two->mode)
6029                                 return 0;
6030                         break;
6031                 }
6032         }
6033         return 1;
6034 }
6035
6036 static const char rename_limit_warning[] =
6037 N_("inexact rename detection was skipped due to too many files.");
6038
6039 static const char degrade_cc_to_c_warning[] =
6040 N_("only found copies from modified paths due to too many files.");
6041
6042 static const char rename_limit_advice[] =
6043 N_("you may want to set your %s variable to at least "
6044    "%d and retry the command.");
6045
6046 void diff_warn_rename_limit(const char *varname, int needed, int degraded_cc)
6047 {
6048         fflush(stdout);
6049         if (degraded_cc)
6050                 warning(_(degrade_cc_to_c_warning));
6051         else if (needed)
6052                 warning(_(rename_limit_warning));
6053         else
6054                 return;
6055         if (0 < needed)
6056                 warning(_(rename_limit_advice), varname, needed);
6057 }
6058
6059 static void diff_flush_patch_all_file_pairs(struct diff_options *o)
6060 {
6061         int i;
6062         static struct emitted_diff_symbols esm = EMITTED_DIFF_SYMBOLS_INIT;
6063         struct diff_queue_struct *q = &diff_queued_diff;
6064
6065         if (WSEH_NEW & WS_RULE_MASK)
6066                 BUG("WS rules bit mask overlaps with diff symbol flags");
6067
6068         if (o->color_moved)
6069                 o->emitted_symbols = &esm;
6070
6071         for (i = 0; i < q->nr; i++) {
6072                 struct diff_filepair *p = q->queue[i];
6073                 if (check_pair_status(p))
6074                         diff_flush_patch(p, o);
6075         }
6076
6077         if (o->emitted_symbols) {
6078                 if (o->color_moved) {
6079                         struct hashmap add_lines, del_lines;
6080
6081                         if (o->color_moved_ws_handling &
6082                             COLOR_MOVED_WS_ALLOW_INDENTATION_CHANGE)
6083                                 o->color_moved_ws_handling |= XDF_IGNORE_WHITESPACE;
6084
6085                         hashmap_init(&del_lines, moved_entry_cmp, o, 0);
6086                         hashmap_init(&add_lines, moved_entry_cmp, o, 0);
6087
6088                         add_lines_to_move_detection(o, &add_lines, &del_lines);
6089                         mark_color_as_moved(o, &add_lines, &del_lines);
6090                         if (o->color_moved == COLOR_MOVED_ZEBRA_DIM)
6091                                 dim_moved_lines(o);
6092
6093                         hashmap_free(&add_lines, 1);
6094                         hashmap_free(&del_lines, 1);
6095                 }
6096
6097                 for (i = 0; i < esm.nr; i++)
6098                         emit_diff_symbol_from_struct(o, &esm.buf[i]);
6099
6100                 for (i = 0; i < esm.nr; i++)
6101                         free((void *)esm.buf[i].line);
6102                 esm.nr = 0;
6103
6104                 o->emitted_symbols = NULL;
6105         }
6106 }
6107
6108 void diff_flush(struct diff_options *options)
6109 {
6110         struct diff_queue_struct *q = &diff_queued_diff;
6111         int i, output_format = options->output_format;
6112         int separator = 0;
6113         int dirstat_by_line = 0;
6114
6115         /*
6116          * Order: raw, stat, summary, patch
6117          * or:    name/name-status/checkdiff (other bits clear)
6118          */
6119         if (!q->nr)
6120                 goto free_queue;
6121
6122         if (output_format & (DIFF_FORMAT_RAW |
6123                              DIFF_FORMAT_NAME |
6124                              DIFF_FORMAT_NAME_STATUS |
6125                              DIFF_FORMAT_CHECKDIFF)) {
6126                 for (i = 0; i < q->nr; i++) {
6127                         struct diff_filepair *p = q->queue[i];
6128                         if (check_pair_status(p))
6129                                 flush_one_pair(p, options);
6130                 }
6131                 separator++;
6132         }
6133
6134         if (output_format & DIFF_FORMAT_DIRSTAT && options->flags.dirstat_by_line)
6135                 dirstat_by_line = 1;
6136
6137         if (output_format & (DIFF_FORMAT_DIFFSTAT|DIFF_FORMAT_SHORTSTAT|DIFF_FORMAT_NUMSTAT) ||
6138             dirstat_by_line) {
6139                 struct diffstat_t diffstat;
6140
6141                 memset(&diffstat, 0, sizeof(struct diffstat_t));
6142                 for (i = 0; i < q->nr; i++) {
6143                         struct diff_filepair *p = q->queue[i];
6144                         if (check_pair_status(p))
6145                                 diff_flush_stat(p, options, &diffstat);
6146                 }
6147                 if (output_format & DIFF_FORMAT_NUMSTAT)
6148                         show_numstat(&diffstat, options);
6149                 if (output_format & DIFF_FORMAT_DIFFSTAT)
6150                         show_stats(&diffstat, options);
6151                 if (output_format & DIFF_FORMAT_SHORTSTAT)
6152                         show_shortstats(&diffstat, options);
6153                 if (output_format & DIFF_FORMAT_DIRSTAT && dirstat_by_line)
6154                         show_dirstat_by_line(&diffstat, options);
6155                 free_diffstat_info(&diffstat);
6156                 separator++;
6157         }
6158         if ((output_format & DIFF_FORMAT_DIRSTAT) && !dirstat_by_line)
6159                 show_dirstat(options);
6160
6161         if (output_format & DIFF_FORMAT_SUMMARY && !is_summary_empty(q)) {
6162                 for (i = 0; i < q->nr; i++) {
6163                         diff_summary(options, q->queue[i]);
6164                 }
6165                 separator++;
6166         }
6167
6168         if (output_format & DIFF_FORMAT_NO_OUTPUT &&
6169             options->flags.exit_with_status &&
6170             options->flags.diff_from_contents) {
6171                 /*
6172                  * run diff_flush_patch for the exit status. setting
6173                  * options->file to /dev/null should be safe, because we
6174                  * aren't supposed to produce any output anyway.
6175                  */
6176                 if (options->close_file)
6177                         fclose(options->file);
6178                 options->file = xfopen("/dev/null", "w");
6179                 options->close_file = 1;
6180                 options->color_moved = 0;
6181                 for (i = 0; i < q->nr; i++) {
6182                         struct diff_filepair *p = q->queue[i];
6183                         if (check_pair_status(p))
6184                                 diff_flush_patch(p, options);
6185                         if (options->found_changes)
6186                                 break;
6187                 }
6188         }
6189
6190         if (output_format & DIFF_FORMAT_PATCH) {
6191                 if (separator) {
6192                         emit_diff_symbol(options, DIFF_SYMBOL_SEPARATOR, NULL, 0, 0);
6193                         if (options->stat_sep)
6194                                 /* attach patch instead of inline */
6195                                 emit_diff_symbol(options, DIFF_SYMBOL_STAT_SEP,
6196                                                  NULL, 0, 0);
6197                 }
6198
6199                 diff_flush_patch_all_file_pairs(options);
6200         }
6201
6202         if (output_format & DIFF_FORMAT_CALLBACK)
6203                 options->format_callback(q, options, options->format_callback_data);
6204
6205         for (i = 0; i < q->nr; i++)
6206                 diff_free_filepair(q->queue[i]);
6207 free_queue:
6208         free(q->queue);
6209         DIFF_QUEUE_CLEAR(q);
6210         if (options->close_file)
6211                 fclose(options->file);
6212
6213         /*
6214          * Report the content-level differences with HAS_CHANGES;
6215          * diff_addremove/diff_change does not set the bit when
6216          * DIFF_FROM_CONTENTS is in effect (e.g. with -w).
6217          */
6218         if (options->flags.diff_from_contents) {
6219                 if (options->found_changes)
6220                         options->flags.has_changes = 1;
6221                 else
6222                         options->flags.has_changes = 0;
6223         }
6224 }
6225
6226 static int match_filter(const struct diff_options *options, const struct diff_filepair *p)
6227 {
6228         return (((p->status == DIFF_STATUS_MODIFIED) &&
6229                  ((p->score &&
6230                    filter_bit_tst(DIFF_STATUS_FILTER_BROKEN, options)) ||
6231                   (!p->score &&
6232                    filter_bit_tst(DIFF_STATUS_MODIFIED, options)))) ||
6233                 ((p->status != DIFF_STATUS_MODIFIED) &&
6234                  filter_bit_tst(p->status, options)));
6235 }
6236
6237 static void diffcore_apply_filter(struct diff_options *options)
6238 {
6239         int i;
6240         struct diff_queue_struct *q = &diff_queued_diff;
6241         struct diff_queue_struct outq;
6242
6243         DIFF_QUEUE_CLEAR(&outq);
6244
6245         if (!options->filter)
6246                 return;
6247
6248         if (filter_bit_tst(DIFF_STATUS_FILTER_AON, options)) {
6249                 int found;
6250                 for (i = found = 0; !found && i < q->nr; i++) {
6251                         if (match_filter(options, q->queue[i]))
6252                                 found++;
6253                 }
6254                 if (found)
6255                         return;
6256
6257                 /* otherwise we will clear the whole queue
6258                  * by copying the empty outq at the end of this
6259                  * function, but first clear the current entries
6260                  * in the queue.
6261                  */
6262                 for (i = 0; i < q->nr; i++)
6263                         diff_free_filepair(q->queue[i]);
6264         }
6265         else {
6266                 /* Only the matching ones */
6267                 for (i = 0; i < q->nr; i++) {
6268                         struct diff_filepair *p = q->queue[i];
6269                         if (match_filter(options, p))
6270                                 diff_q(&outq, p);
6271                         else
6272                                 diff_free_filepair(p);
6273                 }
6274         }
6275         free(q->queue);
6276         *q = outq;
6277 }
6278
6279 /* Check whether two filespecs with the same mode and size are identical */
6280 static int diff_filespec_is_identical(struct repository *r,
6281                                       struct diff_filespec *one,
6282                                       struct diff_filespec *two)
6283 {
6284         if (S_ISGITLINK(one->mode))
6285                 return 0;
6286         if (diff_populate_filespec(r, one, 0))
6287                 return 0;
6288         if (diff_populate_filespec(r, two, 0))
6289                 return 0;
6290         return !memcmp(one->data, two->data, one->size);
6291 }
6292
6293 static int diff_filespec_check_stat_unmatch(struct repository *r,
6294                                             struct diff_filepair *p)
6295 {
6296         if (p->done_skip_stat_unmatch)
6297                 return p->skip_stat_unmatch_result;
6298
6299         p->done_skip_stat_unmatch = 1;
6300         p->skip_stat_unmatch_result = 0;
6301         /*
6302          * 1. Entries that come from stat info dirtiness
6303          *    always have both sides (iow, not create/delete),
6304          *    one side of the object name is unknown, with
6305          *    the same mode and size.  Keep the ones that
6306          *    do not match these criteria.  They have real
6307          *    differences.
6308          *
6309          * 2. At this point, the file is known to be modified,
6310          *    with the same mode and size, and the object
6311          *    name of one side is unknown.  Need to inspect
6312          *    the identical contents.
6313          */
6314         if (!DIFF_FILE_VALID(p->one) || /* (1) */
6315             !DIFF_FILE_VALID(p->two) ||
6316             (p->one->oid_valid && p->two->oid_valid) ||
6317             (p->one->mode != p->two->mode) ||
6318             diff_populate_filespec(r, p->one, CHECK_SIZE_ONLY) ||
6319             diff_populate_filespec(r, p->two, CHECK_SIZE_ONLY) ||
6320             (p->one->size != p->two->size) ||
6321             !diff_filespec_is_identical(r, p->one, p->two)) /* (2) */
6322                 p->skip_stat_unmatch_result = 1;
6323         return p->skip_stat_unmatch_result;
6324 }
6325
6326 static void diffcore_skip_stat_unmatch(struct diff_options *diffopt)
6327 {
6328         int i;
6329         struct diff_queue_struct *q = &diff_queued_diff;
6330         struct diff_queue_struct outq;
6331         DIFF_QUEUE_CLEAR(&outq);
6332
6333         for (i = 0; i < q->nr; i++) {
6334                 struct diff_filepair *p = q->queue[i];
6335
6336                 if (diff_filespec_check_stat_unmatch(diffopt->repo, p))
6337                         diff_q(&outq, p);
6338                 else {
6339                         /*
6340                          * The caller can subtract 1 from skip_stat_unmatch
6341                          * to determine how many paths were dirty only
6342                          * due to stat info mismatch.
6343                          */
6344                         if (!diffopt->flags.no_index)
6345                                 diffopt->skip_stat_unmatch++;
6346                         diff_free_filepair(p);
6347                 }
6348         }
6349         free(q->queue);
6350         *q = outq;
6351 }
6352
6353 static int diffnamecmp(const void *a_, const void *b_)
6354 {
6355         const struct diff_filepair *a = *((const struct diff_filepair **)a_);
6356         const struct diff_filepair *b = *((const struct diff_filepair **)b_);
6357         const char *name_a, *name_b;
6358
6359         name_a = a->one ? a->one->path : a->two->path;
6360         name_b = b->one ? b->one->path : b->two->path;
6361         return strcmp(name_a, name_b);
6362 }
6363
6364 void diffcore_fix_diff_index(void)
6365 {
6366         struct diff_queue_struct *q = &diff_queued_diff;
6367         QSORT(q->queue, q->nr, diffnamecmp);
6368 }
6369
6370 static void add_if_missing(struct repository *r,
6371                            struct oid_array *to_fetch,
6372                            const struct diff_filespec *filespec)
6373 {
6374         if (filespec && filespec->oid_valid &&
6375             oid_object_info_extended(r, &filespec->oid, NULL,
6376                                      OBJECT_INFO_FOR_PREFETCH))
6377                 oid_array_append(to_fetch, &filespec->oid);
6378 }
6379
6380 void diffcore_std(struct diff_options *options)
6381 {
6382         if (options->repo == the_repository &&
6383             repository_format_partial_clone) {
6384                 /*
6385                  * Prefetch the diff pairs that are about to be flushed.
6386                  */
6387                 int i;
6388                 struct diff_queue_struct *q = &diff_queued_diff;
6389                 struct oid_array to_fetch = OID_ARRAY_INIT;
6390
6391                 for (i = 0; i < q->nr; i++) {
6392                         struct diff_filepair *p = q->queue[i];
6393                         add_if_missing(options->repo, &to_fetch, p->one);
6394                         add_if_missing(options->repo, &to_fetch, p->two);
6395                 }
6396                 if (to_fetch.nr)
6397                         /*
6398                          * NEEDSWORK: Consider deduplicating the OIDs sent.
6399                          */
6400                         fetch_objects(repository_format_partial_clone,
6401                                       to_fetch.oid, to_fetch.nr);
6402                 oid_array_clear(&to_fetch);
6403         }
6404
6405         /* NOTE please keep the following in sync with diff_tree_combined() */
6406         if (options->skip_stat_unmatch)
6407                 diffcore_skip_stat_unmatch(options);
6408         if (!options->found_follow) {
6409                 /* See try_to_follow_renames() in tree-diff.c */
6410                 if (options->break_opt != -1)
6411                         diffcore_break(options->repo,
6412                                        options->break_opt);
6413                 if (options->detect_rename)
6414                         diffcore_rename(options);
6415                 if (options->break_opt != -1)
6416                         diffcore_merge_broken();
6417         }
6418         if (options->pickaxe_opts & DIFF_PICKAXE_KINDS_MASK)
6419                 diffcore_pickaxe(options);
6420         if (options->orderfile)
6421                 diffcore_order(options->orderfile);
6422         if (!options->found_follow)
6423                 /* See try_to_follow_renames() in tree-diff.c */
6424                 diff_resolve_rename_copy();
6425         diffcore_apply_filter(options);
6426
6427         if (diff_queued_diff.nr && !options->flags.diff_from_contents)
6428                 options->flags.has_changes = 1;
6429         else
6430                 options->flags.has_changes = 0;
6431
6432         options->found_follow = 0;
6433 }
6434
6435 int diff_result_code(struct diff_options *opt, int status)
6436 {
6437         int result = 0;
6438
6439         diff_warn_rename_limit("diff.renameLimit",
6440                                opt->needed_rename_limit,
6441                                opt->degraded_cc_to_c);
6442         if (!opt->flags.exit_with_status &&
6443             !(opt->output_format & DIFF_FORMAT_CHECKDIFF))
6444                 return status;
6445         if (opt->flags.exit_with_status &&
6446             opt->flags.has_changes)
6447                 result |= 01;
6448         if ((opt->output_format & DIFF_FORMAT_CHECKDIFF) &&
6449             opt->flags.check_failed)
6450                 result |= 02;
6451         return result;
6452 }
6453
6454 int diff_can_quit_early(struct diff_options *opt)
6455 {
6456         return (opt->flags.quick &&
6457                 !opt->filter &&
6458                 opt->flags.has_changes);
6459 }
6460
6461 /*
6462  * Shall changes to this submodule be ignored?
6463  *
6464  * Submodule changes can be configured to be ignored separately for each path,
6465  * but that configuration can be overridden from the command line.
6466  */
6467 static int is_submodule_ignored(const char *path, struct diff_options *options)
6468 {
6469         int ignored = 0;
6470         struct diff_flags orig_flags = options->flags;
6471         if (!options->flags.override_submodule_config)
6472                 set_diffopt_flags_from_submodule_config(options, path);
6473         if (options->flags.ignore_submodules)
6474                 ignored = 1;
6475         options->flags = orig_flags;
6476         return ignored;
6477 }
6478
6479 void diff_addremove(struct diff_options *options,
6480                     int addremove, unsigned mode,
6481                     const struct object_id *oid,
6482                     int oid_valid,
6483                     const char *concatpath, unsigned dirty_submodule)
6484 {
6485         struct diff_filespec *one, *two;
6486
6487         if (S_ISGITLINK(mode) && is_submodule_ignored(concatpath, options))
6488                 return;
6489
6490         /* This may look odd, but it is a preparation for
6491          * feeding "there are unchanged files which should
6492          * not produce diffs, but when you are doing copy
6493          * detection you would need them, so here they are"
6494          * entries to the diff-core.  They will be prefixed
6495          * with something like '=' or '*' (I haven't decided
6496          * which but should not make any difference).
6497          * Feeding the same new and old to diff_change()
6498          * also has the same effect.
6499          * Before the final output happens, they are pruned after
6500          * merged into rename/copy pairs as appropriate.
6501          */
6502         if (options->flags.reverse_diff)
6503                 addremove = (addremove == '+' ? '-' :
6504                              addremove == '-' ? '+' : addremove);
6505
6506         if (options->prefix &&
6507             strncmp(concatpath, options->prefix, options->prefix_length))
6508                 return;
6509
6510         one = alloc_filespec(concatpath);
6511         two = alloc_filespec(concatpath);
6512
6513         if (addremove != '+')
6514                 fill_filespec(one, oid, oid_valid, mode);
6515         if (addremove != '-') {
6516                 fill_filespec(two, oid, oid_valid, mode);
6517                 two->dirty_submodule = dirty_submodule;
6518         }
6519
6520         diff_queue(&diff_queued_diff, one, two);
6521         if (!options->flags.diff_from_contents)
6522                 options->flags.has_changes = 1;
6523 }
6524
6525 void diff_change(struct diff_options *options,
6526                  unsigned old_mode, unsigned new_mode,
6527                  const struct object_id *old_oid,
6528                  const struct object_id *new_oid,
6529                  int old_oid_valid, int new_oid_valid,
6530                  const char *concatpath,
6531                  unsigned old_dirty_submodule, unsigned new_dirty_submodule)
6532 {
6533         struct diff_filespec *one, *two;
6534         struct diff_filepair *p;
6535
6536         if (S_ISGITLINK(old_mode) && S_ISGITLINK(new_mode) &&
6537             is_submodule_ignored(concatpath, options))
6538                 return;
6539
6540         if (options->flags.reverse_diff) {
6541                 SWAP(old_mode, new_mode);
6542                 SWAP(old_oid, new_oid);
6543                 SWAP(old_oid_valid, new_oid_valid);
6544                 SWAP(old_dirty_submodule, new_dirty_submodule);
6545         }
6546
6547         if (options->prefix &&
6548             strncmp(concatpath, options->prefix, options->prefix_length))
6549                 return;
6550
6551         one = alloc_filespec(concatpath);
6552         two = alloc_filespec(concatpath);
6553         fill_filespec(one, old_oid, old_oid_valid, old_mode);
6554         fill_filespec(two, new_oid, new_oid_valid, new_mode);
6555         one->dirty_submodule = old_dirty_submodule;
6556         two->dirty_submodule = new_dirty_submodule;
6557         p = diff_queue(&diff_queued_diff, one, two);
6558
6559         if (options->flags.diff_from_contents)
6560                 return;
6561
6562         if (options->flags.quick && options->skip_stat_unmatch &&
6563             !diff_filespec_check_stat_unmatch(options->repo, p))
6564                 return;
6565
6566         options->flags.has_changes = 1;
6567 }
6568
6569 struct diff_filepair *diff_unmerge(struct diff_options *options, const char *path)
6570 {
6571         struct diff_filepair *pair;
6572         struct diff_filespec *one, *two;
6573
6574         if (options->prefix &&
6575             strncmp(path, options->prefix, options->prefix_length))
6576                 return NULL;
6577
6578         one = alloc_filespec(path);
6579         two = alloc_filespec(path);
6580         pair = diff_queue(&diff_queued_diff, one, two);
6581         pair->is_unmerged = 1;
6582         return pair;
6583 }
6584
6585 static char *run_textconv(struct repository *r,
6586                           const char *pgm,
6587                           struct diff_filespec *spec,
6588                           size_t *outsize)
6589 {
6590         struct diff_tempfile *temp;
6591         const char *argv[3];
6592         const char **arg = argv;
6593         struct child_process child = CHILD_PROCESS_INIT;
6594         struct strbuf buf = STRBUF_INIT;
6595         int err = 0;
6596
6597         temp = prepare_temp_file(r, spec->path, spec);
6598         *arg++ = pgm;
6599         *arg++ = temp->name;
6600         *arg = NULL;
6601
6602         child.use_shell = 1;
6603         child.argv = argv;
6604         child.out = -1;
6605         if (start_command(&child)) {
6606                 remove_tempfile();
6607                 return NULL;
6608         }
6609
6610         if (strbuf_read(&buf, child.out, 0) < 0)
6611                 err = error("error reading from textconv command '%s'", pgm);
6612         close(child.out);
6613
6614         if (finish_command(&child) || err) {
6615                 strbuf_release(&buf);
6616                 remove_tempfile();
6617                 return NULL;
6618         }
6619         remove_tempfile();
6620
6621         return strbuf_detach(&buf, outsize);
6622 }
6623
6624 size_t fill_textconv(struct repository *r,
6625                      struct userdiff_driver *driver,
6626                      struct diff_filespec *df,
6627                      char **outbuf)
6628 {
6629         size_t size;
6630
6631         if (!driver) {
6632                 if (!DIFF_FILE_VALID(df)) {
6633                         *outbuf = "";
6634                         return 0;
6635                 }
6636                 if (diff_populate_filespec(r, df, 0))
6637                         die("unable to read files to diff");
6638                 *outbuf = df->data;
6639                 return df->size;
6640         }
6641
6642         if (!driver->textconv)
6643                 BUG("fill_textconv called with non-textconv driver");
6644
6645         if (driver->textconv_cache && df->oid_valid) {
6646                 *outbuf = notes_cache_get(driver->textconv_cache,
6647                                           &df->oid,
6648                                           &size);
6649                 if (*outbuf)
6650                         return size;
6651         }
6652
6653         *outbuf = run_textconv(r, driver->textconv, df, &size);
6654         if (!*outbuf)
6655                 die("unable to read files to diff");
6656
6657         if (driver->textconv_cache && df->oid_valid) {
6658                 /* ignore errors, as we might be in a readonly repository */
6659                 notes_cache_put(driver->textconv_cache, &df->oid, *outbuf,
6660                                 size);
6661                 /*
6662                  * we could save up changes and flush them all at the end,
6663                  * but we would need an extra call after all diffing is done.
6664                  * Since generating a cache entry is the slow path anyway,
6665                  * this extra overhead probably isn't a big deal.
6666                  */
6667                 notes_cache_write(driver->textconv_cache);
6668         }
6669
6670         return size;
6671 }
6672
6673 int textconv_object(struct repository *r,
6674                     const char *path,
6675                     unsigned mode,
6676                     const struct object_id *oid,
6677                     int oid_valid,
6678                     char **buf,
6679                     unsigned long *buf_size)
6680 {
6681         struct diff_filespec *df;
6682         struct userdiff_driver *textconv;
6683
6684         df = alloc_filespec(path);
6685         fill_filespec(df, oid, oid_valid, mode);
6686         textconv = get_textconv(r, df);
6687         if (!textconv) {
6688                 free_filespec(df);
6689                 return 0;
6690         }
6691
6692         *buf_size = fill_textconv(r, textconv, df, buf);
6693         free_filespec(df);
6694         return 1;
6695 }
6696
6697 void setup_diff_pager(struct diff_options *opt)
6698 {
6699         /*
6700          * If the user asked for our exit code, then either they want --quiet
6701          * or --exit-code. We should definitely not bother with a pager in the
6702          * former case, as we will generate no output. Since we still properly
6703          * report our exit code even when a pager is run, we _could_ run a
6704          * pager with --exit-code. But since we have not done so historically,
6705          * and because it is easy to find people oneline advising "git diff
6706          * --exit-code" in hooks and other scripts, we do not do so.
6707          */
6708         if (!opt->flags.exit_with_status &&
6709             check_pager_config("diff") != 0)
6710                 setup_pager();
6711 }