git.oblomov.eu Git - git/blob - builtin/blame.c

   1 /*
   2  * Blame
   3  *
   4  * Copyright (c) 2006, 2014 by its authors
   5  * See COPYING for licensing conditions
   6  */
   7
   8 #include "cache.h"
   9 #include "config.h"
  10 #include "color.h"
  11 #include "builtin.h"
  12 #include "repository.h"
  13 #include "commit.h"
  14 #include "diff.h"
  15 #include "revision.h"
  16 #include "quote.h"
  17 #include "string-list.h"
  18 #include "mailmap.h"
  19 #include "parse-options.h"
  20 #include "prio-queue.h"
  21 #include "utf8.h"
  22 #include "userdiff.h"
  23 #include "line-range.h"
  24 #include "line-log.h"
  25 #include "dir.h"
  26 #include "progress.h"
  27 #include "object-store.h"
  28 #include "blame.h"
  29 #include "string-list.h"
  30 #include "refs.h"
  31
  32 static char blame_usage[] = N_("git blame [<options>] [<rev-opts>] [<rev>] [--] <file>");
  33
  34 static const char *blame_opt_usage[] = {
  35         blame_usage,
  36         "",
  37         N_("<rev-opts> are documented in git-rev-list(1)"),
  38         NULL
  39 };
  40
  41 static int longest_file;
  42 static int longest_author;
  43 static int max_orig_digits;
  44 static int max_digits;
  45 static int max_score_digits;
  46 static int show_root;
  47 static int reverse;
  48 static int blank_boundary;
  49 static int incremental;
  50 static int xdl_opts;
  51 static int abbrev = -1;
  52 static int no_whole_file_rename;
  53 static int show_progress;
  54 static char repeated_meta_color[COLOR_MAXLEN];
  55 static int coloring_mode;
  56 static struct string_list ignore_revs_file_list = STRING_LIST_INIT_NODUP;
  57 static int mark_unblamable_lines;
  58 static int mark_ignored_lines;
  59
  60 static struct date_mode blame_date_mode = { DATE_ISO8601 };
  61 static size_t blame_date_width;
  62
  63 static struct string_list mailmap = STRING_LIST_INIT_NODUP;
  64
  65 #ifndef DEBUG_BLAME
  66 #define DEBUG_BLAME 0
  67 #endif
  68
  69 static unsigned blame_move_score;
  70 static unsigned blame_copy_score;
  71
  72 /* Remember to update object flag allocation in object.h */
  73 #define METAINFO_SHOWN          (1u<<12)
  74 #define MORE_THAN_ONE_PATH      (1u<<13)
  75
  76 struct progress_info {
  77         struct progress *progress;
  78         int blamed_lines;
  79 };
  80
  81 static const char *nth_line_cb(void *data, long lno)
  82 {
  83         return blame_nth_line((struct blame_scoreboard *)data, lno);
  84 }
  85
  86 /*
  87  * Information on commits, used for output.
  88  */
  89 struct commit_info {
  90         struct strbuf author;
  91         struct strbuf author_mail;
  92         timestamp_t author_time;
  93         struct strbuf author_tz;
  94
  95         /* filled only when asked for details */
  96         struct strbuf committer;
  97         struct strbuf committer_mail;
  98         timestamp_t committer_time;
  99         struct strbuf committer_tz;
 100
 101         struct strbuf summary;
 102 };
 103
 104 /*
 105  * Parse author/committer line in the commit object buffer
 106  */
 107 static void get_ac_line(const char *inbuf, const char *what,
 108         struct strbuf *name, struct strbuf *mail,
 109         timestamp_t *time, struct strbuf *tz)
 110 {
 111         struct ident_split ident;
 112         size_t len, maillen, namelen;
 113         char *tmp, *endp;
 114         const char *namebuf, *mailbuf;
 115
 116         tmp = strstr(inbuf, what);
 117         if (!tmp)
 118                 goto error_out;
 119         tmp += strlen(what);
 120         endp = strchr(tmp, '\n');
 121         if (!endp)
 122                 len = strlen(tmp);
 123         else
 124                 len = endp - tmp;
 125
 126         if (split_ident_line(&ident, tmp, len)) {
 127         error_out:
 128                 /* Ugh */
 129                 tmp = "(unknown)";
 130                 strbuf_addstr(name, tmp);
 131                 strbuf_addstr(mail, tmp);
 132                 strbuf_addstr(tz, tmp);
 133                 *time = 0;
 134                 return;
 135         }
 136
 137         namelen = ident.name_end - ident.name_begin;
 138         namebuf = ident.name_begin;
 139
 140         maillen = ident.mail_end - ident.mail_begin;
 141         mailbuf = ident.mail_begin;
 142
 143         if (ident.date_begin && ident.date_end)
 144                 *time = strtoul(ident.date_begin, NULL, 10);
 145         else
 146                 *time = 0;
 147
 148         if (ident.tz_begin && ident.tz_end)
 149                 strbuf_add(tz, ident.tz_begin, ident.tz_end - ident.tz_begin);
 150         else
 151                 strbuf_addstr(tz, "(unknown)");
 152
 153         /*
 154          * Now, convert both name and e-mail using mailmap
 155          */
 156         map_user(&mailmap, &mailbuf, &maillen,
 157                  &namebuf, &namelen);
 158
 159         strbuf_addf(mail, "<%.*s>", (int)maillen, mailbuf);
 160         strbuf_add(name, namebuf, namelen);
 161 }
 162
 163 static void commit_info_init(struct commit_info *ci)
 164 {
 165
 166         strbuf_init(&ci->author, 0);
 167         strbuf_init(&ci->author_mail, 0);
 168         strbuf_init(&ci->author_tz, 0);
 169         strbuf_init(&ci->committer, 0);
 170         strbuf_init(&ci->committer_mail, 0);
 171         strbuf_init(&ci->committer_tz, 0);
 172         strbuf_init(&ci->summary, 0);
 173 }
 174
 175 static void commit_info_destroy(struct commit_info *ci)
 176 {
 177
 178         strbuf_release(&ci->author);
 179         strbuf_release(&ci->author_mail);
 180         strbuf_release(&ci->author_tz);
 181         strbuf_release(&ci->committer);
 182         strbuf_release(&ci->committer_mail);
 183         strbuf_release(&ci->committer_tz);
 184         strbuf_release(&ci->summary);
 185 }
 186
 187 static void get_commit_info(struct commit *commit,
 188                             struct commit_info *ret,
 189                             int detailed)
 190 {
 191         int len;
 192         const char *subject, *encoding;
 193         const char *message;
 194
 195         commit_info_init(ret);
 196
 197         encoding = get_log_output_encoding();
 198         message = logmsg_reencode(commit, NULL, encoding);
 199         get_ac_line(message, "\nauthor ",
 200                     &ret->author, &ret->author_mail,
 201                     &ret->author_time, &ret->author_tz);
 202
 203         if (!detailed) {
 204                 unuse_commit_buffer(commit, message);
 205                 return;
 206         }
 207
 208         get_ac_line(message, "\ncommitter ",
 209                     &ret->committer, &ret->committer_mail,
 210                     &ret->committer_time, &ret->committer_tz);
 211
 212         len = find_commit_subject(message, &subject);
 213         if (len)
 214                 strbuf_add(&ret->summary, subject, len);
 215         else
 216                 strbuf_addf(&ret->summary, "(%s)", oid_to_hex(&commit->object.oid));
 217
 218         unuse_commit_buffer(commit, message);
 219 }
 220
 221 /*
 222  * Write out any suspect information which depends on the path. This must be
 223  * handled separately from emit_one_suspect_detail(), because a given commit
 224  * may have changes in multiple paths. So this needs to appear each time
 225  * we mention a new group.
 226  *
 227  * To allow LF and other nonportable characters in pathnames,
 228  * they are c-style quoted as needed.
 229  */
 230 static void write_filename_info(struct blame_origin *suspect)
 231 {
 232         if (suspect->previous) {
 233                 struct blame_origin *prev = suspect->previous;
 234                 printf("previous %s ", oid_to_hex(&prev->commit->object.oid));
 235                 write_name_quoted(prev->path, stdout, '\n');
 236         }
 237         printf("filename ");
 238         write_name_quoted(suspect->path, stdout, '\n');
 239 }
 240
 241 /*
 242  * Porcelain/Incremental format wants to show a lot of details per
 243  * commit.  Instead of repeating this every line, emit it only once,
 244  * the first time each commit appears in the output (unless the
 245  * user has specifically asked for us to repeat).
 246  */
 247 static int emit_one_suspect_detail(struct blame_origin *suspect, int repeat)
 248 {
 249         struct commit_info ci;
 250
 251         if (!repeat && (suspect->commit->object.flags & METAINFO_SHOWN))
 252                 return 0;
 253
 254         suspect->commit->object.flags |= METAINFO_SHOWN;
 255         get_commit_info(suspect->commit, &ci, 1);
 256         printf("author %s\n", ci.author.buf);
 257         printf("author-mail %s\n", ci.author_mail.buf);
 258         printf("author-time %"PRItime"\n", ci.author_time);
 259         printf("author-tz %s\n", ci.author_tz.buf);
 260         printf("committer %s\n", ci.committer.buf);
 261         printf("committer-mail %s\n", ci.committer_mail.buf);
 262         printf("committer-time %"PRItime"\n", ci.committer_time);
 263         printf("committer-tz %s\n", ci.committer_tz.buf);
 264         printf("summary %s\n", ci.summary.buf);
 265         if (suspect->commit->object.flags & UNINTERESTING)
 266                 printf("boundary\n");
 267
 268         commit_info_destroy(&ci);
 269
 270         return 1;
 271 }
 272
 273 /*
 274  * The blame_entry is found to be guilty for the range.
 275  * Show it in incremental output.
 276  */
 277 static void found_guilty_entry(struct blame_entry *ent, void *data)
 278 {
 279         struct progress_info *pi = (struct progress_info *)data;
 280
 281         if (incremental) {
 282                 struct blame_origin *suspect = ent->suspect;
 283
 284                 printf("%s %d %d %d\n",
 285                        oid_to_hex(&suspect->commit->object.oid),
 286                        ent->s_lno + 1, ent->lno + 1, ent->num_lines);
 287                 emit_one_suspect_detail(suspect, 0);
 288                 write_filename_info(suspect);
 289                 maybe_flush_or_die(stdout, "stdout");
 290         }
 291         pi->blamed_lines += ent->num_lines;
 292         display_progress(pi->progress, pi->blamed_lines);
 293 }
 294
 295 static const char *format_time(timestamp_t time, const char *tz_str,
 296                                int show_raw_time)
 297 {
 298         static struct strbuf time_buf = STRBUF_INIT;
 299
 300         strbuf_reset(&time_buf);
 301         if (show_raw_time) {
 302                 strbuf_addf(&time_buf, "%"PRItime" %s", time, tz_str);
 303         }
 304         else {
 305                 const char *time_str;
 306                 size_t time_width;
 307                 int tz;
 308                 tz = atoi(tz_str);
 309                 time_str = show_date(time, tz, &blame_date_mode);
 310                 strbuf_addstr(&time_buf, time_str);
 311                 /*
 312                  * Add space paddings to time_buf to display a fixed width
 313                  * string, and use time_width for display width calibration.
 314                  */
 315                 for (time_width = utf8_strwidth(time_str);
 316                      time_width < blame_date_width;
 317                      time_width++)
 318                         strbuf_addch(&time_buf, ' ');
 319         }
 320         return time_buf.buf;
 321 }
 322
 323 #define OUTPUT_ANNOTATE_COMPAT  001
 324 #define OUTPUT_LONG_OBJECT_NAME 002
 325 #define OUTPUT_RAW_TIMESTAMP    004
 326 #define OUTPUT_PORCELAIN        010
 327 #define OUTPUT_SHOW_NAME        020
 328 #define OUTPUT_SHOW_NUMBER      040
 329 #define OUTPUT_SHOW_SCORE       0100
 330 #define OUTPUT_NO_AUTHOR        0200
 331 #define OUTPUT_SHOW_EMAIL       0400
 332 #define OUTPUT_LINE_PORCELAIN   01000
 333 #define OUTPUT_COLOR_LINE       02000
 334 #define OUTPUT_SHOW_AGE_WITH_COLOR      04000
 335
 336 static void emit_porcelain_details(struct blame_origin *suspect, int repeat)
 337 {
 338         if (emit_one_suspect_detail(suspect, repeat) ||
 339             (suspect->commit->object.flags & MORE_THAN_ONE_PATH))
 340                 write_filename_info(suspect);
 341 }
 342
 343 static void emit_porcelain(struct blame_scoreboard *sb, struct blame_entry *ent,
 344                            int opt)
 345 {
 346         int repeat = opt & OUTPUT_LINE_PORCELAIN;
 347         int cnt;
 348         const char *cp;
 349         struct blame_origin *suspect = ent->suspect;
 350         char hex[GIT_MAX_HEXSZ + 1];
 351
 352         oid_to_hex_r(hex, &suspect->commit->object.oid);
 353         printf("%s %d %d %d\n",
 354                hex,
 355                ent->s_lno + 1,
 356                ent->lno + 1,
 357                ent->num_lines);
 358         emit_porcelain_details(suspect, repeat);
 359
 360         cp = blame_nth_line(sb, ent->lno);
 361         for (cnt = 0; cnt < ent->num_lines; cnt++) {
 362                 char ch;
 363                 if (cnt) {
 364                         printf("%s %d %d\n", hex,
 365                                ent->s_lno + 1 + cnt,
 366                                ent->lno + 1 + cnt);
 367                         if (repeat)
 368                                 emit_porcelain_details(suspect, 1);
 369                 }
 370                 putchar('\t');
 371                 do {
 372                         ch = *cp++;
 373                         putchar(ch);
 374                 } while (ch != '\n' &&
 375                          cp < sb->final_buf + sb->final_buf_size);
 376         }
 377
 378         if (sb->final_buf_size && cp[-1] != '\n')
 379                 putchar('\n');
 380 }
 381
 382 static struct color_field {
 383         timestamp_t hop;
 384         char col[COLOR_MAXLEN];
 385 } *colorfield;
 386 static int colorfield_nr, colorfield_alloc;
 387
 388 static void parse_color_fields(const char *s)
 389 {
 390         struct string_list l = STRING_LIST_INIT_DUP;
 391         struct string_list_item *item;
 392         enum { EXPECT_DATE, EXPECT_COLOR } next = EXPECT_COLOR;
 393
 394         colorfield_nr = 0;
 395
 396         /* Ideally this would be stripped and split at the same time? */
 397         string_list_split(&l, s, ',', -1);
 398         ALLOC_GROW(colorfield, colorfield_nr + 1, colorfield_alloc);
 399
 400         for_each_string_list_item(item, &l) {
 401                 switch (next) {
 402                 case EXPECT_DATE:
 403                         colorfield[colorfield_nr].hop = approxidate(item->string);
 404                         next = EXPECT_COLOR;
 405                         colorfield_nr++;
 406                         ALLOC_GROW(colorfield, colorfield_nr + 1, colorfield_alloc);
 407                         break;
 408                 case EXPECT_COLOR:
 409                         if (color_parse(item->string, colorfield[colorfield_nr].col))
 410                                 die(_("expecting a color: %s"), item->string);
 411                         next = EXPECT_DATE;
 412                         break;
 413                 }
 414         }
 415
 416         if (next == EXPECT_COLOR)
 417                 die(_("must end with a color"));
 418
 419         colorfield[colorfield_nr].hop = TIME_MAX;
 420         string_list_clear(&l, 0);
 421 }
 422
 423 static void setup_default_color_by_age(void)
 424 {
 425         parse_color_fields("blue,12 month ago,white,1 month ago,red");
 426 }
 427
 428 static void determine_line_heat(struct blame_entry *ent, const char **dest_color)
 429 {
 430         int i = 0;
 431         struct commit_info ci;
 432         get_commit_info(ent->suspect->commit, &ci, 1);
 433
 434         while (i < colorfield_nr && ci.author_time > colorfield[i].hop)
 435                 i++;
 436
 437         *dest_color = colorfield[i].col;
 438 }
 439
 440 static void emit_other(struct blame_scoreboard *sb, struct blame_entry *ent, int opt)
 441 {
 442         int cnt;
 443         const char *cp;
 444         struct blame_origin *suspect = ent->suspect;
 445         struct commit_info ci;
 446         char hex[GIT_MAX_HEXSZ + 1];
 447         int show_raw_time = !!(opt & OUTPUT_RAW_TIMESTAMP);
 448         const char *default_color = NULL, *color = NULL, *reset = NULL;
 449
 450         get_commit_info(suspect->commit, &ci, 1);
 451         oid_to_hex_r(hex, &suspect->commit->object.oid);
 452
 453         cp = blame_nth_line(sb, ent->lno);
 454
 455         if (opt & OUTPUT_SHOW_AGE_WITH_COLOR) {
 456                 determine_line_heat(ent, &default_color);
 457                 color = default_color;
 458                 reset = GIT_COLOR_RESET;
 459         }
 460
 461         for (cnt = 0; cnt < ent->num_lines; cnt++) {
 462                 char ch;
 463                 int length = (opt & OUTPUT_LONG_OBJECT_NAME) ? GIT_SHA1_HEXSZ : abbrev;
 464
 465                 if (opt & OUTPUT_COLOR_LINE) {
 466                         if (cnt > 0) {
 467                                 color = repeated_meta_color;
 468                                 reset = GIT_COLOR_RESET;
 469                         } else  {
 470                                 color = default_color ? default_color : NULL;
 471                                 reset = default_color ? GIT_COLOR_RESET : NULL;
 472                         }
 473                 }
 474                 if (color)
 475                         fputs(color, stdout);
 476
 477                 if (suspect->commit->object.flags & UNINTERESTING) {
 478                         if (blank_boundary)
 479                                 memset(hex, ' ', length);
 480                         else if (!(opt & OUTPUT_ANNOTATE_COMPAT)) {
 481                                 length--;
 482                                 putchar('^');
 483                         }
 484                 }
 485
 486                 if (mark_unblamable_lines && ent->unblamable) {
 487                         length--;
 488                         putchar('*');
 489                 }
 490                 if (mark_ignored_lines && ent->ignored) {
 491                         length--;
 492                         putchar('?');
 493                 }
 494                 printf("%.*s", length, hex);
 495                 if (opt & OUTPUT_ANNOTATE_COMPAT) {
 496                         const char *name;
 497                         if (opt & OUTPUT_SHOW_EMAIL)
 498                                 name = ci.author_mail.buf;
 499                         else
 500                                 name = ci.author.buf;
 501                         printf("\t(%10s\t%10s\t%d)", name,
 502                                format_time(ci.author_time, ci.author_tz.buf,
 503                                            show_raw_time),
 504                                ent->lno + 1 + cnt);
 505                 } else {
 506                         if (opt & OUTPUT_SHOW_SCORE)
 507                                 printf(" %*d %02d",
 508                                        max_score_digits, ent->score,
 509                                        ent->suspect->refcnt);
 510                         if (opt & OUTPUT_SHOW_NAME)
 511                                 printf(" %-*.*s", longest_file, longest_file,
 512                                        suspect->path);
 513                         if (opt & OUTPUT_SHOW_NUMBER)
 514                                 printf(" %*d", max_orig_digits,
 515                                        ent->s_lno + 1 + cnt);
 516
 517                         if (!(opt & OUTPUT_NO_AUTHOR)) {
 518                                 const char *name;
 519                                 int pad;
 520                                 if (opt & OUTPUT_SHOW_EMAIL)
 521                                         name = ci.author_mail.buf;
 522                                 else
 523                                         name = ci.author.buf;
 524                                 pad = longest_author - utf8_strwidth(name);
 525                                 printf(" (%s%*s %10s",
 526                                        name, pad, "",
 527                                        format_time(ci.author_time,
 528                                                    ci.author_tz.buf,
 529                                                    show_raw_time));
 530                         }
 531                         printf(" %*d) ",
 532                                max_digits, ent->lno + 1 + cnt);
 533                 }
 534                 if (reset)
 535                         fputs(reset, stdout);
 536                 do {
 537                         ch = *cp++;
 538                         putchar(ch);
 539                 } while (ch != '\n' &&
 540                          cp < sb->final_buf + sb->final_buf_size);
 541         }
 542
 543         if (sb->final_buf_size && cp[-1] != '\n')
 544                 putchar('\n');
 545
 546         commit_info_destroy(&ci);
 547 }
 548
 549 static void output(struct blame_scoreboard *sb, int option)
 550 {
 551         struct blame_entry *ent;
 552
 553         if (option & OUTPUT_PORCELAIN) {
 554                 for (ent = sb->ent; ent; ent = ent->next) {
 555                         int count = 0;
 556                         struct blame_origin *suspect;
 557                         struct commit *commit = ent->suspect->commit;
 558                         if (commit->object.flags & MORE_THAN_ONE_PATH)
 559                                 continue;
 560                         for (suspect = get_blame_suspects(commit); suspect; suspect = suspect->next) {
 561                                 if (suspect->guilty && count++) {
 562                                         commit->object.flags |= MORE_THAN_ONE_PATH;
 563                                         break;
 564                                 }
 565                         }
 566                 }
 567         }
 568
 569         for (ent = sb->ent; ent; ent = ent->next) {
 570                 if (option & OUTPUT_PORCELAIN)
 571                         emit_porcelain(sb, ent, option);
 572                 else {
 573                         emit_other(sb, ent, option);
 574                 }
 575         }
 576 }
 577
 578 /*
 579  * Add phony grafts for use with -S; this is primarily to
 580  * support git's cvsserver that wants to give a linear history
 581  * to its clients.
 582  */
 583 static int read_ancestry(const char *graft_file)
 584 {
 585         FILE *fp = fopen_or_warn(graft_file, "r");
 586         struct strbuf buf = STRBUF_INIT;
 587         if (!fp)
 588                 return -1;
 589         while (!strbuf_getwholeline(&buf, fp, '\n')) {
 590                 /* The format is just "Commit Parent1 Parent2 ...\n" */
 591                 struct commit_graft *graft = read_graft_line(&buf);
 592                 if (graft)
 593                         register_commit_graft(the_repository, graft, 0);
 594         }
 595         fclose(fp);
 596         strbuf_release(&buf);
 597         return 0;
 598 }
 599
 600 static int update_auto_abbrev(int auto_abbrev, struct blame_origin *suspect)
 601 {
 602         const char *uniq = find_unique_abbrev(&suspect->commit->object.oid,
 603                                               auto_abbrev);
 604         int len = strlen(uniq);
 605         if (auto_abbrev < len)
 606                 return len;
 607         return auto_abbrev;
 608 }
 609
 610 /*
 611  * How many columns do we need to show line numbers, authors,
 612  * and filenames?
 613  */
 614 static void find_alignment(struct blame_scoreboard *sb, int *option)
 615 {
 616         int longest_src_lines = 0;
 617         int longest_dst_lines = 0;
 618         unsigned largest_score = 0;
 619         struct blame_entry *e;
 620         int compute_auto_abbrev = (abbrev < 0);
 621         int auto_abbrev = DEFAULT_ABBREV;
 622
 623         for (e = sb->ent; e; e = e->next) {
 624                 struct blame_origin *suspect = e->suspect;
 625                 int num;
 626
 627                 if (compute_auto_abbrev)
 628                         auto_abbrev = update_auto_abbrev(auto_abbrev, suspect);
 629                 if (strcmp(suspect->path, sb->path))
 630                         *option |= OUTPUT_SHOW_NAME;
 631                 num = strlen(suspect->path);
 632                 if (longest_file < num)
 633                         longest_file = num;
 634                 if (!(suspect->commit->object.flags & METAINFO_SHOWN)) {
 635                         struct commit_info ci;
 636                         suspect->commit->object.flags |= METAINFO_SHOWN;
 637                         get_commit_info(suspect->commit, &ci, 1);
 638                         if (*option & OUTPUT_SHOW_EMAIL)
 639                                 num = utf8_strwidth(ci.author_mail.buf);
 640                         else
 641                                 num = utf8_strwidth(ci.author.buf);
 642                         if (longest_author < num)
 643                                 longest_author = num;
 644                         commit_info_destroy(&ci);
 645                 }
 646                 num = e->s_lno + e->num_lines;
 647                 if (longest_src_lines < num)
 648                         longest_src_lines = num;
 649                 num = e->lno + e->num_lines;
 650                 if (longest_dst_lines < num)
 651                         longest_dst_lines = num;
 652                 if (largest_score < blame_entry_score(sb, e))
 653                         largest_score = blame_entry_score(sb, e);
 654         }
 655         max_orig_digits = decimal_width(longest_src_lines);
 656         max_digits = decimal_width(longest_dst_lines);
 657         max_score_digits = decimal_width(largest_score);
 658
 659         if (compute_auto_abbrev)
 660                 /* one more abbrev length is needed for the boundary commit */
 661                 abbrev = auto_abbrev + 1;
 662 }
 663
 664 static void sanity_check_on_fail(struct blame_scoreboard *sb, int baa)
 665 {
 666         int opt = OUTPUT_SHOW_SCORE | OUTPUT_SHOW_NUMBER | OUTPUT_SHOW_NAME;
 667         find_alignment(sb, &opt);
 668         output(sb, opt);
 669         die("Baa %d!", baa);
 670 }
 671
 672 static unsigned parse_score(const char *arg)
 673 {
 674         char *end;
 675         unsigned long score = strtoul(arg, &end, 10);
 676         if (*end)
 677                 return 0;
 678         return score;
 679 }
 680
 681 static const char *add_prefix(const char *prefix, const char *path)
 682 {
 683         return prefix_path(prefix, prefix ? strlen(prefix) : 0, path);
 684 }
 685
 686 static int git_blame_config(const char *var, const char *value, void *cb)
 687 {
 688         if (!strcmp(var, "blame.showroot")) {
 689                 show_root = git_config_bool(var, value);
 690                 return 0;
 691         }
 692         if (!strcmp(var, "blame.blankboundary")) {
 693                 blank_boundary = git_config_bool(var, value);
 694                 return 0;
 695         }
 696         if (!strcmp(var, "blame.showemail")) {
 697                 int *output_option = cb;
 698                 if (git_config_bool(var, value))
 699                         *output_option |= OUTPUT_SHOW_EMAIL;
 700                 else
 701                         *output_option &= ~OUTPUT_SHOW_EMAIL;
 702                 return 0;
 703         }
 704         if (!strcmp(var, "blame.date")) {
 705                 if (!value)
 706                         return config_error_nonbool(var);
 707                 parse_date_format(value, &blame_date_mode);
 708                 return 0;
 709         }
 710         if (!strcmp(var, "blame.ignorerevsfile")) {
 711                 const char *str;
 712                 int ret;
 713
 714                 ret = git_config_pathname(&str, var, value);
 715                 if (ret)
 716                         return ret;
 717                 string_list_insert(&ignore_revs_file_list, str);
 718                 return 0;
 719         }
 720         if (!strcmp(var, "blame.markunblamablelines")) {
 721                 mark_unblamable_lines = git_config_bool(var, value);
 722                 return 0;
 723         }
 724         if (!strcmp(var, "blame.markignoredlines")) {
 725                 mark_ignored_lines = git_config_bool(var, value);
 726                 return 0;
 727         }
 728         if (!strcmp(var, "color.blame.repeatedlines")) {
 729                 if (color_parse_mem(value, strlen(value), repeated_meta_color))
 730                         warning(_("invalid color '%s' in color.blame.repeatedLines"),
 731                                 value);
 732                 return 0;
 733         }
 734         if (!strcmp(var, "color.blame.highlightrecent")) {
 735                 parse_color_fields(value);
 736                 return 0;
 737         }
 738
 739         if (!strcmp(var, "blame.coloring")) {
 740                 if (!strcmp(value, "repeatedLines")) {
 741                         coloring_mode |= OUTPUT_COLOR_LINE;
 742                 } else if (!strcmp(value, "highlightRecent")) {
 743                         coloring_mode |= OUTPUT_SHOW_AGE_WITH_COLOR;
 744                 } else if (!strcmp(value, "none")) {
 745                         coloring_mode &= ~(OUTPUT_COLOR_LINE |
 746                                             OUTPUT_SHOW_AGE_WITH_COLOR);
 747                 } else {
 748                         warning(_("invalid value for blame.coloring"));
 749                         return 0;
 750                 }
 751         }
 752
 753         if (git_diff_heuristic_config(var, value, cb) < 0)
 754                 return -1;
 755         if (userdiff_config(var, value) < 0)
 756                 return -1;
 757
 758         return git_default_config(var, value, cb);
 759 }
 760
 761 static int blame_copy_callback(const struct option *option, const char *arg, int unset)
 762 {
 763         int *opt = option->value;
 764
 765         BUG_ON_OPT_NEG(unset);
 766
 767         /*
 768          * -C enables copy from removed files;
 769          * -C -C enables copy from existing files, but only
 770          *       when blaming a new file;
 771          * -C -C -C enables copy from existing files for
 772          *          everybody
 773          */
 774         if (*opt & PICKAXE_BLAME_COPY_HARDER)
 775                 *opt |= PICKAXE_BLAME_COPY_HARDEST;
 776         if (*opt & PICKAXE_BLAME_COPY)
 777                 *opt |= PICKAXE_BLAME_COPY_HARDER;
 778         *opt |= PICKAXE_BLAME_COPY | PICKAXE_BLAME_MOVE;
 779
 780         if (arg)
 781                 blame_copy_score = parse_score(arg);
 782         return 0;
 783 }
 784
 785 static int blame_move_callback(const struct option *option, const char *arg, int unset)
 786 {
 787         int *opt = option->value;
 788
 789         BUG_ON_OPT_NEG(unset);
 790
 791         *opt |= PICKAXE_BLAME_MOVE;
 792
 793         if (arg)
 794                 blame_move_score = parse_score(arg);
 795         return 0;
 796 }
 797
 798 static int is_a_rev(const char *name)
 799 {
 800         struct object_id oid;
 801
 802         if (get_oid(name, &oid))
 803                 return 0;
 804         return OBJ_NONE < oid_object_info(the_repository, &oid, NULL);
 805 }
 806
 807 static void build_ignorelist(struct blame_scoreboard *sb,
 808                              struct string_list *ignore_revs_file_list,
 809                              struct string_list *ignore_rev_list)
 810 {
 811         struct string_list_item *i;
 812         struct object_id oid;
 813
 814         oidset_init(&sb->ignore_list, 0);
 815         for_each_string_list_item(i, ignore_revs_file_list) {
 816                 if (!strcmp(i->string, ""))
 817                         oidset_clear(&sb->ignore_list);
 818                 else
 819                         oidset_parse_file(&sb->ignore_list, i->string);
 820         }
 821         for_each_string_list_item(i, ignore_rev_list) {
 822                 if (get_oid_committish(i->string, &oid))
 823                         die(_("cannot find revision %s to ignore"), i->string);
 824                 oidset_insert(&sb->ignore_list, &oid);
 825         }
 826 }
 827
 828 int cmd_blame(int argc, const char **argv, const char *prefix)
 829 {
 830         struct rev_info revs;
 831         const char *path;
 832         struct blame_scoreboard sb;
 833         struct blame_origin *o;
 834         struct blame_entry *ent = NULL;
 835         long dashdash_pos, lno;
 836         struct progress_info pi = { NULL, 0 };
 837
 838         struct string_list range_list = STRING_LIST_INIT_NODUP;
 839         struct string_list ignore_rev_list = STRING_LIST_INIT_NODUP;
 840         int output_option = 0, opt = 0;
 841         int show_stats = 0;
 842         const char *revs_file = NULL;
 843         const char *contents_from = NULL;
 844         const struct option options[] = {
 845                 OPT_BOOL(0, "incremental", &incremental, N_("Show blame entries as we find them, incrementally")),
 846                 OPT_BOOL('b', NULL, &blank_boundary, N_("Show blank SHA-1 for boundary commits (Default: off)")),
 847                 OPT_BOOL(0, "root", &show_root, N_("Do not treat root commits as boundaries (Default: off)")),
 848                 OPT_BOOL(0, "show-stats", &show_stats, N_("Show work cost statistics")),
 849                 OPT_BOOL(0, "progress", &show_progress, N_("Force progress reporting")),
 850                 OPT_BIT(0, "score-debug", &output_option, N_("Show output score for blame entries"), OUTPUT_SHOW_SCORE),
 851                 OPT_BIT('f', "show-name", &output_option, N_("Show original filename (Default: auto)"), OUTPUT_SHOW_NAME),
 852                 OPT_BIT('n', "show-number", &output_option, N_("Show original linenumber (Default: off)"), OUTPUT_SHOW_NUMBER),
 853                 OPT_BIT('p', "porcelain", &output_option, N_("Show in a format designed for machine consumption"), OUTPUT_PORCELAIN),
 854                 OPT_BIT(0, "line-porcelain", &output_option, N_("Show porcelain format with per-line commit information"), OUTPUT_PORCELAIN|OUTPUT_LINE_PORCELAIN),
 855                 OPT_BIT('c', NULL, &output_option, N_("Use the same output mode as git-annotate (Default: off)"), OUTPUT_ANNOTATE_COMPAT),
 856                 OPT_BIT('t', NULL, &output_option, N_("Show raw timestamp (Default: off)"), OUTPUT_RAW_TIMESTAMP),
 857                 OPT_BIT('l', NULL, &output_option, N_("Show long commit SHA1 (Default: off)"), OUTPUT_LONG_OBJECT_NAME),
 858                 OPT_BIT('s', NULL, &output_option, N_("Suppress author name and timestamp (Default: off)"), OUTPUT_NO_AUTHOR),
 859                 OPT_BIT('e', "show-email", &output_option, N_("Show author email instead of name (Default: off)"), OUTPUT_SHOW_EMAIL),
 860                 OPT_BIT('w', NULL, &xdl_opts, N_("Ignore whitespace differences"), XDF_IGNORE_WHITESPACE),
 861                 OPT_STRING_LIST(0, "ignore-rev", &ignore_rev_list, N_("rev"), N_("Ignore <rev> when blaming")),
 862                 OPT_STRING_LIST(0, "ignore-revs-file", &ignore_revs_file_list, N_("file"), N_("Ignore revisions from <file>")),
 863                 OPT_BIT(0, "color-lines", &output_option, N_("color redundant metadata from previous line differently"), OUTPUT_COLOR_LINE),
 864                 OPT_BIT(0, "color-by-age", &output_option, N_("color lines by age"), OUTPUT_SHOW_AGE_WITH_COLOR),
 865
 866                 /*
 867                  * The following two options are parsed by parse_revision_opt()
 868                  * and are only included here to get included in the "-h"
 869                  * output:
 870                  */
 871                 { OPTION_LOWLEVEL_CALLBACK, 0, "indent-heuristic", NULL, NULL, N_("Use an experimental heuristic to improve diffs"), PARSE_OPT_NOARG, NULL, 0, parse_opt_unknown_cb },
 872
 873                 OPT_BIT(0, "minimal", &xdl_opts, N_("Spend extra cycles to find better match"), XDF_NEED_MINIMAL),
 874                 OPT_STRING('S', NULL, &revs_file, N_("file"), N_("Use revisions from <file> instead of calling git-rev-list")),
 875                 OPT_STRING(0, "contents", &contents_from, N_("file"), N_("Use <file>'s contents as the final image")),
 876                 { OPTION_CALLBACK, 'C', NULL, &opt, N_("score"), N_("Find line copies within and across files"), PARSE_OPT_OPTARG, blame_copy_callback },
 877                 { OPTION_CALLBACK, 'M', NULL, &opt, N_("score"), N_("Find line movements within and across files"), PARSE_OPT_OPTARG, blame_move_callback },
 878                 OPT_STRING_LIST('L', NULL, &range_list, N_("n,m"), N_("Process only line range n,m, counting from 1")),
 879                 OPT__ABBREV(&abbrev),
 880                 OPT_END()
 881         };
 882
 883         struct parse_opt_ctx_t ctx;
 884         int cmd_is_annotate = !strcmp(argv[0], "annotate");
 885         struct range_set ranges;
 886         unsigned int range_i;
 887         long anchor;
 888
 889         setup_default_color_by_age();
 890         git_config(git_blame_config, &output_option);
 891         repo_init_revisions(the_repository, &revs, NULL);
 892         revs.date_mode = blame_date_mode;
 893         revs.diffopt.flags.allow_textconv = 1;
 894         revs.diffopt.flags.follow_renames = 1;
 895
 896         save_commit_buffer = 0;
 897         dashdash_pos = 0;
 898         show_progress = -1;
 899
 900         parse_options_start(&ctx, argc, argv, prefix, options,
 901                             PARSE_OPT_KEEP_DASHDASH | PARSE_OPT_KEEP_ARGV0);
 902         for (;;) {
 903                 switch (parse_options_step(&ctx, options, blame_opt_usage)) {
 904                 case PARSE_OPT_HELP:
 905                 case PARSE_OPT_ERROR:
 906                         exit(129);
 907                 case PARSE_OPT_COMPLETE:
 908                         exit(0);
 909                 case PARSE_OPT_DONE:
 910                         if (ctx.argv[0])
 911                                 dashdash_pos = ctx.cpidx;
 912                         goto parse_done;
 913                 }
 914
 915                 if (!strcmp(ctx.argv[0], "--reverse")) {
 916                         ctx.argv[0] = "--children";
 917                         reverse = 1;
 918                 }
 919                 parse_revision_opt(&revs, &ctx, options, blame_opt_usage);
 920         }
 921 parse_done:
 922         no_whole_file_rename = !revs.diffopt.flags.follow_renames;
 923         xdl_opts |= revs.diffopt.xdl_opts & XDF_INDENT_HEURISTIC;
 924         revs.diffopt.flags.follow_renames = 0;
 925         argc = parse_options_end(&ctx);
 926
 927         if (incremental || (output_option & OUTPUT_PORCELAIN)) {
 928                 if (show_progress > 0)
 929                         die(_("--progress can't be used with --incremental or porcelain formats"));
 930                 show_progress = 0;
 931         } else if (show_progress < 0)
 932                 show_progress = isatty(2);
 933
 934         if (0 < abbrev && abbrev < GIT_SHA1_HEXSZ)
 935                 /* one more abbrev length is needed for the boundary commit */
 936                 abbrev++;
 937         else if (!abbrev)
 938                 abbrev = GIT_SHA1_HEXSZ;
 939
 940         if (revs_file && read_ancestry(revs_file))
 941                 die_errno("reading graft file '%s' failed", revs_file);
 942
 943         if (cmd_is_annotate) {
 944                 output_option |= OUTPUT_ANNOTATE_COMPAT;
 945                 blame_date_mode.type = DATE_ISO8601;
 946         } else {
 947                 blame_date_mode = revs.date_mode;
 948         }
 949
 950         /* The maximum width used to show the dates */
 951         switch (blame_date_mode.type) {
 952         case DATE_RFC2822:
 953                 blame_date_width = sizeof("Thu, 19 Oct 2006 16:00:04 -0700");
 954                 break;
 955         case DATE_ISO8601_STRICT:
 956                 blame_date_width = sizeof("2006-10-19T16:00:04-07:00");
 957                 break;
 958         case DATE_ISO8601:
 959                 blame_date_width = sizeof("2006-10-19 16:00:04 -0700");
 960                 break;
 961         case DATE_RAW:
 962                 blame_date_width = sizeof("1161298804 -0700");
 963                 break;
 964         case DATE_UNIX:
 965                 blame_date_width = sizeof("1161298804");
 966                 break;
 967         case DATE_SHORT:
 968                 blame_date_width = sizeof("2006-10-19");
 969                 break;
 970         case DATE_RELATIVE:
 971                 /*
 972                  * TRANSLATORS: This string is used to tell us the
 973                  * maximum display width for a relative timestamp in
 974                  * "git blame" output.  For C locale, "4 years, 11
 975                  * months ago", which takes 22 places, is the longest
 976                  * among various forms of relative timestamps, but
 977                  * your language may need more or fewer display
 978                  * columns.
 979                  */
 980                 blame_date_width = utf8_strwidth(_("4 years, 11 months ago")) + 1; /* add the null */
 981                 break;
 982         case DATE_HUMAN:
 983                 /* If the year is shown, no time is shown */
 984                 blame_date_width = sizeof("Thu Oct 19 16:00");
 985                 break;
 986         case DATE_NORMAL:
 987                 blame_date_width = sizeof("Thu Oct 19 16:00:04 2006 -0700");
 988                 break;
 989         case DATE_STRFTIME:
 990                 blame_date_width = strlen(show_date(0, 0, &blame_date_mode)) + 1; /* add the null */
 991                 break;
 992         }
 993         blame_date_width -= 1; /* strip the null */
 994
 995         if (revs.diffopt.flags.find_copies_harder)
 996                 opt |= (PICKAXE_BLAME_COPY | PICKAXE_BLAME_MOVE |
 997                         PICKAXE_BLAME_COPY_HARDER);
 998
 999         /*
1000          * We have collected options unknown to us in argv[1..unk]
1001          * which are to be passed to revision machinery if we are
1002          * going to do the "bottom" processing.
1003          *
1004          * The remaining are:
1005          *
1006          * (1) if dashdash_pos != 0, it is either
1007          *     "blame [revisions] -- <path>" or
1008          *     "blame -- <path> <rev>"
1009          *
1010          * (2) otherwise, it is one of the two:
1011          *     "blame [revisions] <path>"
1012          *     "blame <path> <rev>"
1013          *
1014          * Note that we must strip out <path> from the arguments: we do not
1015          * want the path pruning but we may want "bottom" processing.
1016          */
1017         if (dashdash_pos) {
1018                 switch (argc - dashdash_pos - 1) {
1019                 case 2: /* (1b) */
1020                         if (argc != 4)
1021                                 usage_with_options(blame_opt_usage, options);
1022                         /* reorder for the new way: <rev> -- <path> */
1023                         argv[1] = argv[3];
1024                         argv[3] = argv[2];
1025                         argv[2] = "--";
1026                         /* FALLTHROUGH */
1027                 case 1: /* (1a) */
1028                         path = add_prefix(prefix, argv[--argc]);
1029                         argv[argc] = NULL;
1030                         break;
1031                 default:
1032                         usage_with_options(blame_opt_usage, options);
1033                 }
1034         } else {
1035                 if (argc < 2)
1036                         usage_with_options(blame_opt_usage, options);
1037                 if (argc == 3 && is_a_rev(argv[argc - 1])) { /* (2b) */
1038                         path = add_prefix(prefix, argv[1]);
1039                         argv[1] = argv[2];
1040                 } else {        /* (2a) */
1041                         if (argc == 2 && is_a_rev(argv[1]) && !get_git_work_tree())
1042                                 die("missing <path> to blame");
1043                         path = add_prefix(prefix, argv[argc - 1]);
1044                 }
1045                 argv[argc - 1] = "--";
1046         }
1047
1048         revs.disable_stdin = 1;
1049         setup_revisions(argc, argv, &revs, NULL);
1050         if (!revs.pending.nr && is_bare_repository()) {
1051                 struct commit *head_commit;
1052                 struct object_id head_oid;
1053
1054                 if (!resolve_ref_unsafe("HEAD", RESOLVE_REF_READING,
1055                                         &head_oid, NULL) ||
1056                     !(head_commit = lookup_commit_reference_gently(revs.repo,
1057                                                              &head_oid, 1)))
1058                         die("no such ref: HEAD");
1059
1060                 add_pending_object(&revs, &head_commit->object, "HEAD");
1061         }
1062
1063         init_scoreboard(&sb);
1064         sb.revs = &revs;
1065         sb.contents_from = contents_from;
1066         sb.reverse = reverse;
1067         sb.repo = the_repository;
1068         build_ignorelist(&sb, &ignore_revs_file_list, &ignore_rev_list);
1069         string_list_clear(&ignore_revs_file_list, 0);
1070         string_list_clear(&ignore_rev_list, 0);
1071         setup_scoreboard(&sb, path, &o);
1072         lno = sb.num_lines;
1073
1074         if (lno && !range_list.nr)
1075                 string_list_append(&range_list, "1");
1076
1077         anchor = 1;
1078         range_set_init(&ranges, range_list.nr);
1079         for (range_i = 0; range_i < range_list.nr; ++range_i) {
1080                 long bottom, top;
1081                 if (parse_range_arg(range_list.items[range_i].string,
1082                                     nth_line_cb, &sb, lno, anchor,
1083                                     &bottom, &top, sb.path,
1084                                     the_repository->index))
1085                         usage(blame_usage);
1086                 if ((!lno && (top || bottom)) || lno < bottom)
1087                         die(Q_("file %s has only %lu line",
1088                                "file %s has only %lu lines",
1089                                lno), path, lno);
1090                 if (bottom < 1)
1091                         bottom = 1;
1092                 if (top < 1 || lno < top)
1093                         top = lno;
1094                 bottom--;
1095                 range_set_append_unsafe(&ranges, bottom, top);
1096                 anchor = top + 1;
1097         }
1098         sort_and_merge_range_set(&ranges);
1099
1100         for (range_i = ranges.nr; range_i > 0; --range_i) {
1101                 const struct range *r = &ranges.ranges[range_i - 1];
1102                 ent = blame_entry_prepend(ent, r->start, r->end, o);
1103         }
1104
1105         o->suspects = ent;
1106         prio_queue_put(&sb.commits, o->commit);
1107
1108         blame_origin_decref(o);
1109
1110         range_set_release(&ranges);
1111         string_list_clear(&range_list, 0);
1112
1113         sb.ent = NULL;
1114         sb.path = path;
1115
1116         if (blame_move_score)
1117                 sb.move_score = blame_move_score;
1118         if (blame_copy_score)
1119                 sb.copy_score = blame_copy_score;
1120
1121         sb.debug = DEBUG_BLAME;
1122         sb.on_sanity_fail = &sanity_check_on_fail;
1123
1124         sb.show_root = show_root;
1125         sb.xdl_opts = xdl_opts;
1126         sb.no_whole_file_rename = no_whole_file_rename;
1127
1128         read_mailmap(&mailmap, NULL);
1129
1130         sb.found_guilty_entry = &found_guilty_entry;
1131         sb.found_guilty_entry_data = &pi;
1132         if (show_progress)
1133                 pi.progress = start_delayed_progress(_("Blaming lines"), sb.num_lines);
1134
1135         assign_blame(&sb, opt);
1136
1137         stop_progress(&pi.progress);
1138
1139         if (!incremental)
1140                 setup_pager();
1141         else
1142                 return 0;
1143
1144         blame_sort_final(&sb);
1145
1146         blame_coalesce(&sb);
1147
1148         if (!(output_option & (OUTPUT_COLOR_LINE | OUTPUT_SHOW_AGE_WITH_COLOR)))
1149                 output_option |= coloring_mode;
1150
1151         if (!(output_option & OUTPUT_PORCELAIN)) {
1152                 find_alignment(&sb, &output_option);
1153                 if (!*repeated_meta_color &&
1154                     (output_option & OUTPUT_COLOR_LINE))
1155                         xsnprintf(repeated_meta_color,
1156                                   sizeof(repeated_meta_color),
1157                                   "%s", GIT_COLOR_CYAN);
1158         }
1159         if (output_option & OUTPUT_ANNOTATE_COMPAT)
1160                 output_option &= ~(OUTPUT_COLOR_LINE | OUTPUT_SHOW_AGE_WITH_COLOR);
1161
1162         output(&sb, output_option);
1163         free((void *)sb.final_buf);
1164         for (ent = sb.ent; ent; ) {
1165                 struct blame_entry *e = ent->next;
1166                 free(ent);
1167                 ent = e;
1168         }
1169
1170         if (show_stats) {
1171                 printf("num read blob: %d\n", sb.num_read_blob);
1172                 printf("num get patch: %d\n", sb.num_get_patch);
1173                 printf("num commits: %d\n", sb.num_commits);
1174         }
1175         return 0;
1176 }