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 "refs.h"
  30
  31 static char blame_usage[] = N_("git blame [<options>] [<rev-opts>] [<rev>] [--] <file>");
  32
  33 static const char *blame_opt_usage[] = {
  34         blame_usage,
  35         "",
  36         N_("<rev-opts> are documented in git-rev-list(1)"),
  37         NULL
  38 };
  39
  40 static int longest_file;
  41 static int longest_author;
  42 static int max_orig_digits;
  43 static int max_digits;
  44 static int max_score_digits;
  45 static int show_root;
  46 static int reverse;
  47 static int blank_boundary;
  48 static int incremental;
  49 static int xdl_opts;
  50 static int abbrev = -1;
  51 static int no_whole_file_rename;
  52 static int show_progress;
  53 static char repeated_meta_color[COLOR_MAXLEN];
  54 static int coloring_mode;
  55 static struct string_list ignore_revs_file_list = STRING_LIST_INIT_NODUP;
  56 static int mark_unblamable_lines;
  57 static int mark_ignored_lines;
  58
  59 static struct date_mode blame_date_mode = { DATE_ISO8601 };
  60 static size_t blame_date_width;
  61
  62 static struct string_list mailmap = STRING_LIST_INIT_NODUP;
  63
  64 #ifndef DEBUG_BLAME
  65 #define DEBUG_BLAME 0
  66 #endif
  67
  68 static unsigned blame_move_score;
  69 static unsigned blame_copy_score;
  70
  71 /* Remember to update object flag allocation in object.h */
  72 #define METAINFO_SHOWN          (1u<<12)
  73 #define MORE_THAN_ONE_PATH      (1u<<13)
  74
  75 struct progress_info {
  76         struct progress *progress;
  77         int blamed_lines;
  78 };
  79
  80 static const char *nth_line_cb(void *data, long lno)
  81 {
  82         return blame_nth_line((struct blame_scoreboard *)data, lno);
  83 }
  84
  85 /*
  86  * Information on commits, used for output.
  87  */
  88 struct commit_info {
  89         struct strbuf author;
  90         struct strbuf author_mail;
  91         timestamp_t author_time;
  92         struct strbuf author_tz;
  93
  94         /* filled only when asked for details */
  95         struct strbuf committer;
  96         struct strbuf committer_mail;
  97         timestamp_t committer_time;
  98         struct strbuf committer_tz;
  99
 100         struct strbuf summary;
 101 };
 102
 103 /*
 104  * Parse author/committer line in the commit object buffer
 105  */
 106 static void get_ac_line(const char *inbuf, const char *what,
 107         struct strbuf *name, struct strbuf *mail,
 108         timestamp_t *time, struct strbuf *tz)
 109 {
 110         struct ident_split ident;
 111         size_t len, maillen, namelen;
 112         char *tmp, *endp;
 113         const char *namebuf, *mailbuf;
 114
 115         tmp = strstr(inbuf, what);
 116         if (!tmp)
 117                 goto error_out;
 118         tmp += strlen(what);
 119         endp = strchr(tmp, '\n');
 120         if (!endp)
 121                 len = strlen(tmp);
 122         else
 123                 len = endp - tmp;
 124
 125         if (split_ident_line(&ident, tmp, len)) {
 126         error_out:
 127                 /* Ugh */
 128                 tmp = "(unknown)";
 129                 strbuf_addstr(name, tmp);
 130                 strbuf_addstr(mail, tmp);
 131                 strbuf_addstr(tz, tmp);
 132                 *time = 0;
 133                 return;
 134         }
 135
 136         namelen = ident.name_end - ident.name_begin;
 137         namebuf = ident.name_begin;
 138
 139         maillen = ident.mail_end - ident.mail_begin;
 140         mailbuf = ident.mail_begin;
 141
 142         if (ident.date_begin && ident.date_end)
 143                 *time = strtoul(ident.date_begin, NULL, 10);
 144         else
 145                 *time = 0;
 146
 147         if (ident.tz_begin && ident.tz_end)
 148                 strbuf_add(tz, ident.tz_begin, ident.tz_end - ident.tz_begin);
 149         else
 150                 strbuf_addstr(tz, "(unknown)");
 151
 152         /*
 153          * Now, convert both name and e-mail using mailmap
 154          */
 155         map_user(&mailmap, &mailbuf, &maillen,
 156                  &namebuf, &namelen);
 157
 158         strbuf_addf(mail, "<%.*s>", (int)maillen, mailbuf);
 159         strbuf_add(name, namebuf, namelen);
 160 }
 161
 162 static void commit_info_init(struct commit_info *ci)
 163 {
 164
 165         strbuf_init(&ci->author, 0);
 166         strbuf_init(&ci->author_mail, 0);
 167         strbuf_init(&ci->author_tz, 0);
 168         strbuf_init(&ci->committer, 0);
 169         strbuf_init(&ci->committer_mail, 0);
 170         strbuf_init(&ci->committer_tz, 0);
 171         strbuf_init(&ci->summary, 0);
 172 }
 173
 174 static void commit_info_destroy(struct commit_info *ci)
 175 {
 176
 177         strbuf_release(&ci->author);
 178         strbuf_release(&ci->author_mail);
 179         strbuf_release(&ci->author_tz);
 180         strbuf_release(&ci->committer);
 181         strbuf_release(&ci->committer_mail);
 182         strbuf_release(&ci->committer_tz);
 183         strbuf_release(&ci->summary);
 184 }
 185
 186 static void get_commit_info(struct commit *commit,
 187                             struct commit_info *ret,
 188                             int detailed)
 189 {
 190         int len;
 191         const char *subject, *encoding;
 192         const char *message;
 193
 194         commit_info_init(ret);
 195
 196         encoding = get_log_output_encoding();
 197         message = logmsg_reencode(commit, NULL, encoding);
 198         get_ac_line(message, "\nauthor ",
 199                     &ret->author, &ret->author_mail,
 200                     &ret->author_time, &ret->author_tz);
 201
 202         if (!detailed) {
 203                 unuse_commit_buffer(commit, message);
 204                 return;
 205         }
 206
 207         get_ac_line(message, "\ncommitter ",
 208                     &ret->committer, &ret->committer_mail,
 209                     &ret->committer_time, &ret->committer_tz);
 210
 211         len = find_commit_subject(message, &subject);
 212         if (len)
 213                 strbuf_add(&ret->summary, subject, len);
 214         else
 215                 strbuf_addf(&ret->summary, "(%s)", oid_to_hex(&commit->object.oid));
 216
 217         unuse_commit_buffer(commit, message);
 218 }
 219
 220 /*
 221  * Write out any suspect information which depends on the path. This must be
 222  * handled separately from emit_one_suspect_detail(), because a given commit
 223  * may have changes in multiple paths. So this needs to appear each time
 224  * we mention a new group.
 225  *
 226  * To allow LF and other nonportable characters in pathnames,
 227  * they are c-style quoted as needed.
 228  */
 229 static void write_filename_info(struct blame_origin *suspect)
 230 {
 231         if (suspect->previous) {
 232                 struct blame_origin *prev = suspect->previous;
 233                 printf("previous %s ", oid_to_hex(&prev->commit->object.oid));
 234                 write_name_quoted(prev->path, stdout, '\n');
 235         }
 236         printf("filename ");
 237         write_name_quoted(suspect->path, stdout, '\n');
 238 }
 239
 240 /*
 241  * Porcelain/Incremental format wants to show a lot of details per
 242  * commit.  Instead of repeating this every line, emit it only once,
 243  * the first time each commit appears in the output (unless the
 244  * user has specifically asked for us to repeat).
 245  */
 246 static int emit_one_suspect_detail(struct blame_origin *suspect, int repeat)
 247 {
 248         struct commit_info ci;
 249
 250         if (!repeat && (suspect->commit->object.flags & METAINFO_SHOWN))
 251                 return 0;
 252
 253         suspect->commit->object.flags |= METAINFO_SHOWN;
 254         get_commit_info(suspect->commit, &ci, 1);
 255         printf("author %s\n", ci.author.buf);
 256         printf("author-mail %s\n", ci.author_mail.buf);
 257         printf("author-time %"PRItime"\n", ci.author_time);
 258         printf("author-tz %s\n", ci.author_tz.buf);
 259         printf("committer %s\n", ci.committer.buf);
 260         printf("committer-mail %s\n", ci.committer_mail.buf);
 261         printf("committer-time %"PRItime"\n", ci.committer_time);
 262         printf("committer-tz %s\n", ci.committer_tz.buf);
 263         printf("summary %s\n", ci.summary.buf);
 264         if (suspect->commit->object.flags & UNINTERESTING)
 265                 printf("boundary\n");
 266
 267         commit_info_destroy(&ci);
 268
 269         return 1;
 270 }
 271
 272 /*
 273  * The blame_entry is found to be guilty for the range.
 274  * Show it in incremental output.
 275  */
 276 static void found_guilty_entry(struct blame_entry *ent, void *data)
 277 {
 278         struct progress_info *pi = (struct progress_info *)data;
 279
 280         if (incremental) {
 281                 struct blame_origin *suspect = ent->suspect;
 282
 283                 printf("%s %d %d %d\n",
 284                        oid_to_hex(&suspect->commit->object.oid),
 285                        ent->s_lno + 1, ent->lno + 1, ent->num_lines);
 286                 emit_one_suspect_detail(suspect, 0);
 287                 write_filename_info(suspect);
 288                 maybe_flush_or_die(stdout, "stdout");
 289         }
 290         pi->blamed_lines += ent->num_lines;
 291         display_progress(pi->progress, pi->blamed_lines);
 292 }
 293
 294 static const char *format_time(timestamp_t time, const char *tz_str,
 295                                int show_raw_time)
 296 {
 297         static struct strbuf time_buf = STRBUF_INIT;
 298
 299         strbuf_reset(&time_buf);
 300         if (show_raw_time) {
 301                 strbuf_addf(&time_buf, "%"PRItime" %s", time, tz_str);
 302         }
 303         else {
 304                 const char *time_str;
 305                 size_t time_width;
 306                 int tz;
 307                 tz = atoi(tz_str);
 308                 time_str = show_date(time, tz, &blame_date_mode);
 309                 strbuf_addstr(&time_buf, time_str);
 310                 /*
 311                  * Add space paddings to time_buf to display a fixed width
 312                  * string, and use time_width for display width calibration.
 313                  */
 314                 for (time_width = utf8_strwidth(time_str);
 315                      time_width < blame_date_width;
 316                      time_width++)
 317                         strbuf_addch(&time_buf, ' ');
 318         }
 319         return time_buf.buf;
 320 }
 321
 322 #define OUTPUT_ANNOTATE_COMPAT      (1U<<0)
 323 #define OUTPUT_LONG_OBJECT_NAME     (1U<<1)
 324 #define OUTPUT_RAW_TIMESTAMP        (1U<<2)
 325 #define OUTPUT_PORCELAIN            (1U<<3)
 326 #define OUTPUT_SHOW_NAME            (1U<<4)
 327 #define OUTPUT_SHOW_NUMBER          (1U<<5)
 328 #define OUTPUT_SHOW_SCORE           (1U<<6)
 329 #define OUTPUT_NO_AUTHOR            (1U<<7)
 330 #define OUTPUT_SHOW_EMAIL           (1U<<8)
 331 #define OUTPUT_LINE_PORCELAIN       (1U<<9)
 332 #define OUTPUT_COLOR_LINE           (1U<<10)
 333 #define OUTPUT_SHOW_AGE_WITH_COLOR  (1U<<11)
 334
 335 static void emit_porcelain_details(struct blame_origin *suspect, int repeat)
 336 {
 337         if (emit_one_suspect_detail(suspect, repeat) ||
 338             (suspect->commit->object.flags & MORE_THAN_ONE_PATH))
 339                 write_filename_info(suspect);
 340 }
 341
 342 static void emit_porcelain(struct blame_scoreboard *sb, struct blame_entry *ent,
 343                            int opt)
 344 {
 345         int repeat = opt & OUTPUT_LINE_PORCELAIN;
 346         int cnt;
 347         const char *cp;
 348         struct blame_origin *suspect = ent->suspect;
 349         char hex[GIT_MAX_HEXSZ + 1];
 350
 351         oid_to_hex_r(hex, &suspect->commit->object.oid);
 352         printf("%s %d %d %d\n",
 353                hex,
 354                ent->s_lno + 1,
 355                ent->lno + 1,
 356                ent->num_lines);
 357         emit_porcelain_details(suspect, repeat);
 358
 359         cp = blame_nth_line(sb, ent->lno);
 360         for (cnt = 0; cnt < ent->num_lines; cnt++) {
 361                 char ch;
 362                 if (cnt) {
 363                         printf("%s %d %d\n", hex,
 364                                ent->s_lno + 1 + cnt,
 365                                ent->lno + 1 + cnt);
 366                         if (repeat)
 367                                 emit_porcelain_details(suspect, 1);
 368                 }
 369                 putchar('\t');
 370                 do {
 371                         ch = *cp++;
 372                         putchar(ch);
 373                 } while (ch != '\n' &&
 374                          cp < sb->final_buf + sb->final_buf_size);
 375         }
 376
 377         if (sb->final_buf_size && cp[-1] != '\n')
 378                 putchar('\n');
 379 }
 380
 381 static struct color_field {
 382         timestamp_t hop;
 383         char col[COLOR_MAXLEN];
 384 } *colorfield;
 385 static int colorfield_nr, colorfield_alloc;
 386
 387 static void parse_color_fields(const char *s)
 388 {
 389         struct string_list l = STRING_LIST_INIT_DUP;
 390         struct string_list_item *item;
 391         enum { EXPECT_DATE, EXPECT_COLOR } next = EXPECT_COLOR;
 392
 393         colorfield_nr = 0;
 394
 395         /* Ideally this would be stripped and split at the same time? */
 396         string_list_split(&l, s, ',', -1);
 397         ALLOC_GROW(colorfield, colorfield_nr + 1, colorfield_alloc);
 398
 399         for_each_string_list_item(item, &l) {
 400                 switch (next) {
 401                 case EXPECT_DATE:
 402                         colorfield[colorfield_nr].hop = approxidate(item->string);
 403                         next = EXPECT_COLOR;
 404                         colorfield_nr++;
 405                         ALLOC_GROW(colorfield, colorfield_nr + 1, colorfield_alloc);
 406                         break;
 407                 case EXPECT_COLOR:
 408                         if (color_parse(item->string, colorfield[colorfield_nr].col))
 409                                 die(_("expecting a color: %s"), item->string);
 410                         next = EXPECT_DATE;
 411                         break;
 412                 }
 413         }
 414
 415         if (next == EXPECT_COLOR)
 416                 die(_("must end with a color"));
 417
 418         colorfield[colorfield_nr].hop = TIME_MAX;
 419         string_list_clear(&l, 0);
 420 }
 421
 422 static void setup_default_color_by_age(void)
 423 {
 424         parse_color_fields("blue,12 month ago,white,1 month ago,red");
 425 }
 426
 427 static void determine_line_heat(struct blame_entry *ent, const char **dest_color)
 428 {
 429         int i = 0;
 430         struct commit_info ci;
 431         get_commit_info(ent->suspect->commit, &ci, 1);
 432
 433         while (i < colorfield_nr && ci.author_time > colorfield[i].hop)
 434                 i++;
 435
 436         *dest_color = colorfield[i].col;
 437 }
 438
 439 static void emit_other(struct blame_scoreboard *sb, struct blame_entry *ent, int opt)
 440 {
 441         int cnt;
 442         const char *cp;
 443         struct blame_origin *suspect = ent->suspect;
 444         struct commit_info ci;
 445         char hex[GIT_MAX_HEXSZ + 1];
 446         int show_raw_time = !!(opt & OUTPUT_RAW_TIMESTAMP);
 447         const char *default_color = NULL, *color = NULL, *reset = NULL;
 448
 449         get_commit_info(suspect->commit, &ci, 1);
 450         oid_to_hex_r(hex, &suspect->commit->object.oid);
 451
 452         cp = blame_nth_line(sb, ent->lno);
 453
 454         if (opt & OUTPUT_SHOW_AGE_WITH_COLOR) {
 455                 determine_line_heat(ent, &default_color);
 456                 color = default_color;
 457                 reset = GIT_COLOR_RESET;
 458         }
 459
 460         for (cnt = 0; cnt < ent->num_lines; cnt++) {
 461                 char ch;
 462                 int length = (opt & OUTPUT_LONG_OBJECT_NAME) ? the_hash_algo->hexsz : abbrev;
 463
 464                 if (opt & OUTPUT_COLOR_LINE) {
 465                         if (cnt > 0) {
 466                                 color = repeated_meta_color;
 467                                 reset = GIT_COLOR_RESET;
 468                         } else  {
 469                                 color = default_color ? default_color : NULL;
 470                                 reset = default_color ? GIT_COLOR_RESET : NULL;
 471                         }
 472                 }
 473                 if (color)
 474                         fputs(color, stdout);
 475
 476                 if (suspect->commit->object.flags & UNINTERESTING) {
 477                         if (blank_boundary)
 478                                 memset(hex, ' ', length);
 479                         else if (!(opt & OUTPUT_ANNOTATE_COMPAT)) {
 480                                 length--;
 481                                 putchar('^');
 482                         }
 483                 }
 484
 485                 if (mark_unblamable_lines && ent->unblamable) {
 486                         length--;
 487                         putchar('*');
 488                 }
 489                 if (mark_ignored_lines && ent->ignored) {
 490                         length--;
 491                         putchar('?');
 492                 }
 493                 printf("%.*s", length, hex);
 494                 if (opt & OUTPUT_ANNOTATE_COMPAT) {
 495                         const char *name;
 496                         if (opt & OUTPUT_SHOW_EMAIL)
 497                                 name = ci.author_mail.buf;
 498                         else
 499                                 name = ci.author.buf;
 500                         printf("\t(%10s\t%10s\t%d)", name,
 501                                format_time(ci.author_time, ci.author_tz.buf,
 502                                            show_raw_time),
 503                                ent->lno + 1 + cnt);
 504                 } else {
 505                         if (opt & OUTPUT_SHOW_SCORE)
 506                                 printf(" %*d %02d",
 507                                        max_score_digits, ent->score,
 508                                        ent->suspect->refcnt);
 509                         if (opt & OUTPUT_SHOW_NAME)
 510                                 printf(" %-*.*s", longest_file, longest_file,
 511                                        suspect->path);
 512                         if (opt & OUTPUT_SHOW_NUMBER)
 513                                 printf(" %*d", max_orig_digits,
 514                                        ent->s_lno + 1 + cnt);
 515
 516                         if (!(opt & OUTPUT_NO_AUTHOR)) {
 517                                 const char *name;
 518                                 int pad;
 519                                 if (opt & OUTPUT_SHOW_EMAIL)
 520                                         name = ci.author_mail.buf;
 521                                 else
 522                                         name = ci.author.buf;
 523                                 pad = longest_author - utf8_strwidth(name);
 524                                 printf(" (%s%*s %10s",
 525                                        name, pad, "",
 526                                        format_time(ci.author_time,
 527                                                    ci.author_tz.buf,
 528                                                    show_raw_time));
 529                         }
 530                         printf(" %*d) ",
 531                                max_digits, ent->lno + 1 + cnt);
 532                 }
 533                 if (reset)
 534                         fputs(reset, stdout);
 535                 do {
 536                         ch = *cp++;
 537                         putchar(ch);
 538                 } while (ch != '\n' &&
 539                          cp < sb->final_buf + sb->final_buf_size);
 540         }
 541
 542         if (sb->final_buf_size && cp[-1] != '\n')
 543                 putchar('\n');
 544
 545         commit_info_destroy(&ci);
 546 }
 547
 548 static void output(struct blame_scoreboard *sb, int option)
 549 {
 550         struct blame_entry *ent;
 551
 552         if (option & OUTPUT_PORCELAIN) {
 553                 for (ent = sb->ent; ent; ent = ent->next) {
 554                         int count = 0;
 555                         struct blame_origin *suspect;
 556                         struct commit *commit = ent->suspect->commit;
 557                         if (commit->object.flags & MORE_THAN_ONE_PATH)
 558                                 continue;
 559                         for (suspect = get_blame_suspects(commit); suspect; suspect = suspect->next) {
 560                                 if (suspect->guilty && count++) {
 561                                         commit->object.flags |= MORE_THAN_ONE_PATH;
 562                                         break;
 563                                 }
 564                         }
 565                 }
 566         }
 567
 568         for (ent = sb->ent; ent; ent = ent->next) {
 569                 if (option & OUTPUT_PORCELAIN)
 570                         emit_porcelain(sb, ent, option);
 571                 else {
 572                         emit_other(sb, ent, option);
 573                 }
 574         }
 575 }
 576
 577 /*
 578  * Add phony grafts for use with -S; this is primarily to
 579  * support git's cvsserver that wants to give a linear history
 580  * to its clients.
 581  */
 582 static int read_ancestry(const char *graft_file)
 583 {
 584         FILE *fp = fopen_or_warn(graft_file, "r");
 585         struct strbuf buf = STRBUF_INIT;
 586         if (!fp)
 587                 return -1;
 588         while (!strbuf_getwholeline(&buf, fp, '\n')) {
 589                 /* The format is just "Commit Parent1 Parent2 ...\n" */
 590                 struct commit_graft *graft = read_graft_line(&buf);
 591                 if (graft)
 592                         register_commit_graft(the_repository, graft, 0);
 593         }
 594         fclose(fp);
 595         strbuf_release(&buf);
 596         return 0;
 597 }
 598
 599 static int update_auto_abbrev(int auto_abbrev, struct blame_origin *suspect)
 600 {
 601         const char *uniq = find_unique_abbrev(&suspect->commit->object.oid,
 602                                               auto_abbrev);
 603         int len = strlen(uniq);
 604         if (auto_abbrev < len)
 605                 return len;
 606         return auto_abbrev;
 607 }
 608
 609 /*
 610  * How many columns do we need to show line numbers, authors,
 611  * and filenames?
 612  */
 613 static void find_alignment(struct blame_scoreboard *sb, int *option)
 614 {
 615         int longest_src_lines = 0;
 616         int longest_dst_lines = 0;
 617         unsigned largest_score = 0;
 618         struct blame_entry *e;
 619         int compute_auto_abbrev = (abbrev < 0);
 620         int auto_abbrev = DEFAULT_ABBREV;
 621
 622         for (e = sb->ent; e; e = e->next) {
 623                 struct blame_origin *suspect = e->suspect;
 624                 int num;
 625
 626                 if (compute_auto_abbrev)
 627                         auto_abbrev = update_auto_abbrev(auto_abbrev, suspect);
 628                 if (strcmp(suspect->path, sb->path))
 629                         *option |= OUTPUT_SHOW_NAME;
 630                 num = strlen(suspect->path);
 631                 if (longest_file < num)
 632                         longest_file = num;
 633                 if (!(suspect->commit->object.flags & METAINFO_SHOWN)) {
 634                         struct commit_info ci;
 635                         suspect->commit->object.flags |= METAINFO_SHOWN;
 636                         get_commit_info(suspect->commit, &ci, 1);
 637                         if (*option & OUTPUT_SHOW_EMAIL)
 638                                 num = utf8_strwidth(ci.author_mail.buf);
 639                         else
 640                                 num = utf8_strwidth(ci.author.buf);
 641                         if (longest_author < num)
 642                                 longest_author = num;
 643                         commit_info_destroy(&ci);
 644                 }
 645                 num = e->s_lno + e->num_lines;
 646                 if (longest_src_lines < num)
 647                         longest_src_lines = num;
 648                 num = e->lno + e->num_lines;
 649                 if (longest_dst_lines < num)
 650                         longest_dst_lines = num;
 651                 if (largest_score < blame_entry_score(sb, e))
 652                         largest_score = blame_entry_score(sb, e);
 653         }
 654         max_orig_digits = decimal_width(longest_src_lines);
 655         max_digits = decimal_width(longest_dst_lines);
 656         max_score_digits = decimal_width(largest_score);
 657
 658         if (compute_auto_abbrev)
 659                 /* one more abbrev length is needed for the boundary commit */
 660                 abbrev = auto_abbrev + 1;
 661 }
 662
 663 static void sanity_check_on_fail(struct blame_scoreboard *sb, int baa)
 664 {
 665         int opt = OUTPUT_SHOW_SCORE | OUTPUT_SHOW_NUMBER | OUTPUT_SHOW_NAME;
 666         find_alignment(sb, &opt);
 667         output(sb, opt);
 668         die("Baa %d!", baa);
 669 }
 670
 671 static unsigned parse_score(const char *arg)
 672 {
 673         char *end;
 674         unsigned long score = strtoul(arg, &end, 10);
 675         if (*end)
 676                 return 0;
 677         return score;
 678 }
 679
 680 static const char *add_prefix(const char *prefix, const char *path)
 681 {
 682         return prefix_path(prefix, prefix ? strlen(prefix) : 0, path);
 683 }
 684
 685 static int git_blame_config(const char *var, const char *value, void *cb)
 686 {
 687         if (!strcmp(var, "blame.showroot")) {
 688                 show_root = git_config_bool(var, value);
 689                 return 0;
 690         }
 691         if (!strcmp(var, "blame.blankboundary")) {
 692                 blank_boundary = git_config_bool(var, value);
 693                 return 0;
 694         }
 695         if (!strcmp(var, "blame.showemail")) {
 696                 int *output_option = cb;
 697                 if (git_config_bool(var, value))
 698                         *output_option |= OUTPUT_SHOW_EMAIL;
 699                 else
 700                         *output_option &= ~OUTPUT_SHOW_EMAIL;
 701                 return 0;
 702         }
 703         if (!strcmp(var, "blame.date")) {
 704                 if (!value)
 705                         return config_error_nonbool(var);
 706                 parse_date_format(value, &blame_date_mode);
 707                 return 0;
 708         }
 709         if (!strcmp(var, "blame.ignorerevsfile")) {
 710                 const char *str;
 711                 int ret;
 712
 713                 ret = git_config_pathname(&str, var, value);
 714                 if (ret)
 715                         return ret;
 716                 string_list_insert(&ignore_revs_file_list, str);
 717                 return 0;
 718         }
 719         if (!strcmp(var, "blame.markunblamablelines")) {
 720                 mark_unblamable_lines = git_config_bool(var, value);
 721                 return 0;
 722         }
 723         if (!strcmp(var, "blame.markignoredlines")) {
 724                 mark_ignored_lines = git_config_bool(var, value);
 725                 return 0;
 726         }
 727         if (!strcmp(var, "color.blame.repeatedlines")) {
 728                 if (color_parse_mem(value, strlen(value), repeated_meta_color))
 729                         warning(_("invalid color '%s' in color.blame.repeatedLines"),
 730                                 value);
 731                 return 0;
 732         }
 733         if (!strcmp(var, "color.blame.highlightrecent")) {
 734                 parse_color_fields(value);
 735                 return 0;
 736         }
 737
 738         if (!strcmp(var, "blame.coloring")) {
 739                 if (!strcmp(value, "repeatedLines")) {
 740                         coloring_mode |= OUTPUT_COLOR_LINE;
 741                 } else if (!strcmp(value, "highlightRecent")) {
 742                         coloring_mode |= OUTPUT_SHOW_AGE_WITH_COLOR;
 743                 } else if (!strcmp(value, "none")) {
 744                         coloring_mode &= ~(OUTPUT_COLOR_LINE |
 745                                             OUTPUT_SHOW_AGE_WITH_COLOR);
 746                 } else {
 747                         warning(_("invalid value for blame.coloring"));
 748                         return 0;
 749                 }
 750         }
 751
 752         if (git_diff_heuristic_config(var, value, cb) < 0)
 753                 return -1;
 754         if (userdiff_config(var, value) < 0)
 755                 return -1;
 756
 757         return git_default_config(var, value, cb);
 758 }
 759
 760 static int blame_copy_callback(const struct option *option, const char *arg, int unset)
 761 {
 762         int *opt = option->value;
 763
 764         BUG_ON_OPT_NEG(unset);
 765
 766         /*
 767          * -C enables copy from removed files;
 768          * -C -C enables copy from existing files, but only
 769          *       when blaming a new file;
 770          * -C -C -C enables copy from existing files for
 771          *          everybody
 772          */
 773         if (*opt & PICKAXE_BLAME_COPY_HARDER)
 774                 *opt |= PICKAXE_BLAME_COPY_HARDEST;
 775         if (*opt & PICKAXE_BLAME_COPY)
 776                 *opt |= PICKAXE_BLAME_COPY_HARDER;
 777         *opt |= PICKAXE_BLAME_COPY | PICKAXE_BLAME_MOVE;
 778
 779         if (arg)
 780                 blame_copy_score = parse_score(arg);
 781         return 0;
 782 }
 783
 784 static int blame_move_callback(const struct option *option, const char *arg, int unset)
 785 {
 786         int *opt = option->value;
 787
 788         BUG_ON_OPT_NEG(unset);
 789
 790         *opt |= PICKAXE_BLAME_MOVE;
 791
 792         if (arg)
 793                 blame_move_score = parse_score(arg);
 794         return 0;
 795 }
 796
 797 static int is_a_rev(const char *name)
 798 {
 799         struct object_id oid;
 800
 801         if (get_oid(name, &oid))
 802                 return 0;
 803         return OBJ_NONE < oid_object_info(the_repository, &oid, NULL);
 804 }
 805
 806 static void build_ignorelist(struct blame_scoreboard *sb,
 807                              struct string_list *ignore_revs_file_list,
 808                              struct string_list *ignore_rev_list)
 809 {
 810         struct string_list_item *i;
 811         struct object_id oid;
 812
 813         oidset_init(&sb->ignore_list, 0);
 814         for_each_string_list_item(i, ignore_revs_file_list) {
 815                 if (!strcmp(i->string, ""))
 816                         oidset_clear(&sb->ignore_list);
 817                 else
 818                         oidset_parse_file(&sb->ignore_list, i->string);
 819         }
 820         for_each_string_list_item(i, ignore_rev_list) {
 821                 if (get_oid_committish(i->string, &oid))
 822                         die(_("cannot find revision %s to ignore"), i->string);
 823                 oidset_insert(&sb->ignore_list, &oid);
 824         }
 825 }
 826
 827 int cmd_blame(int argc, const char **argv, const char *prefix)
 828 {
 829         struct rev_info revs;
 830         const char *path;
 831         struct blame_scoreboard sb;
 832         struct blame_origin *o;
 833         struct blame_entry *ent = NULL;
 834         long dashdash_pos, lno;
 835         struct progress_info pi = { NULL, 0 };
 836
 837         struct string_list range_list = STRING_LIST_INIT_NODUP;
 838         struct string_list ignore_rev_list = STRING_LIST_INIT_NODUP;
 839         int output_option = 0, opt = 0;
 840         int show_stats = 0;
 841         const char *revs_file = NULL;
 842         const char *contents_from = NULL;
 843         const struct option options[] = {
 844                 OPT_BOOL(0, "incremental", &incremental, N_("Show blame entries as we find them, incrementally")),
 845                 OPT_BOOL('b', NULL, &blank_boundary, N_("Show blank SHA-1 for boundary commits (Default: off)")),
 846                 OPT_BOOL(0, "root", &show_root, N_("Do not treat root commits as boundaries (Default: off)")),
 847                 OPT_BOOL(0, "show-stats", &show_stats, N_("Show work cost statistics")),
 848                 OPT_BOOL(0, "progress", &show_progress, N_("Force progress reporting")),
 849                 OPT_BIT(0, "score-debug", &output_option, N_("Show output score for blame entries"), OUTPUT_SHOW_SCORE),
 850                 OPT_BIT('f', "show-name", &output_option, N_("Show original filename (Default: auto)"), OUTPUT_SHOW_NAME),
 851                 OPT_BIT('n', "show-number", &output_option, N_("Show original linenumber (Default: off)"), OUTPUT_SHOW_NUMBER),
 852                 OPT_BIT('p', "porcelain", &output_option, N_("Show in a format designed for machine consumption"), OUTPUT_PORCELAIN),
 853                 OPT_BIT(0, "line-porcelain", &output_option, N_("Show porcelain format with per-line commit information"), OUTPUT_PORCELAIN|OUTPUT_LINE_PORCELAIN),
 854                 OPT_BIT('c', NULL, &output_option, N_("Use the same output mode as git-annotate (Default: off)"), OUTPUT_ANNOTATE_COMPAT),
 855                 OPT_BIT('t', NULL, &output_option, N_("Show raw timestamp (Default: off)"), OUTPUT_RAW_TIMESTAMP),
 856                 OPT_BIT('l', NULL, &output_option, N_("Show long commit SHA1 (Default: off)"), OUTPUT_LONG_OBJECT_NAME),
 857                 OPT_BIT('s', NULL, &output_option, N_("Suppress author name and timestamp (Default: off)"), OUTPUT_NO_AUTHOR),
 858                 OPT_BIT('e', "show-email", &output_option, N_("Show author email instead of name (Default: off)"), OUTPUT_SHOW_EMAIL),
 859                 OPT_BIT('w', NULL, &xdl_opts, N_("Ignore whitespace differences"), XDF_IGNORE_WHITESPACE),
 860                 OPT_STRING_LIST(0, "ignore-rev", &ignore_rev_list, N_("rev"), N_("Ignore <rev> when blaming")),
 861                 OPT_STRING_LIST(0, "ignore-revs-file", &ignore_revs_file_list, N_("file"), N_("Ignore revisions from <file>")),
 862                 OPT_BIT(0, "color-lines", &output_option, N_("color redundant metadata from previous line differently"), OUTPUT_COLOR_LINE),
 863                 OPT_BIT(0, "color-by-age", &output_option, N_("color lines by age"), OUTPUT_SHOW_AGE_WITH_COLOR),
 864                 OPT_BIT(0, "minimal", &xdl_opts, N_("Spend extra cycles to find better match"), XDF_NEED_MINIMAL),
 865                 OPT_STRING('S', NULL, &revs_file, N_("file"), N_("Use revisions from <file> instead of calling git-rev-list")),
 866                 OPT_STRING(0, "contents", &contents_from, N_("file"), N_("Use <file>'s contents as the final image")),
 867                 OPT_CALLBACK_F('C', NULL, &opt, N_("score"), N_("Find line copies within and across files"), PARSE_OPT_OPTARG, blame_copy_callback),
 868                 OPT_CALLBACK_F('M', NULL, &opt, N_("score"), N_("Find line movements within and across files"), PARSE_OPT_OPTARG, blame_move_callback),
 869                 OPT_STRING_LIST('L', NULL, &range_list, N_("n,m"), N_("Process only line range n,m, counting from 1")),
 870                 OPT__ABBREV(&abbrev),
 871                 OPT_END()
 872         };
 873
 874         struct parse_opt_ctx_t ctx;
 875         int cmd_is_annotate = !strcmp(argv[0], "annotate");
 876         struct range_set ranges;
 877         unsigned int range_i;
 878         long anchor;
 879         const int hexsz = the_hash_algo->hexsz;
 880
 881         setup_default_color_by_age();
 882         git_config(git_blame_config, &output_option);
 883         repo_init_revisions(the_repository, &revs, NULL);
 884         revs.date_mode = blame_date_mode;
 885         revs.diffopt.flags.allow_textconv = 1;
 886         revs.diffopt.flags.follow_renames = 1;
 887
 888         save_commit_buffer = 0;
 889         dashdash_pos = 0;
 890         show_progress = -1;
 891
 892         parse_options_start(&ctx, argc, argv, prefix, options,
 893                             PARSE_OPT_KEEP_DASHDASH | PARSE_OPT_KEEP_ARGV0);
 894         for (;;) {
 895                 switch (parse_options_step(&ctx, options, blame_opt_usage)) {
 896                 case PARSE_OPT_HELP:
 897                 case PARSE_OPT_ERROR:
 898                         exit(129);
 899                 case PARSE_OPT_COMPLETE:
 900                         exit(0);
 901                 case PARSE_OPT_DONE:
 902                         if (ctx.argv[0])
 903                                 dashdash_pos = ctx.cpidx;
 904                         goto parse_done;
 905                 }
 906
 907                 if (!strcmp(ctx.argv[0], "--reverse")) {
 908                         ctx.argv[0] = "--children";
 909                         reverse = 1;
 910                 }
 911                 parse_revision_opt(&revs, &ctx, options, blame_opt_usage);
 912         }
 913 parse_done:
 914         no_whole_file_rename = !revs.diffopt.flags.follow_renames;
 915         xdl_opts |= revs.diffopt.xdl_opts & XDF_INDENT_HEURISTIC;
 916         revs.diffopt.flags.follow_renames = 0;
 917         argc = parse_options_end(&ctx);
 918
 919         if (incremental || (output_option & OUTPUT_PORCELAIN)) {
 920                 if (show_progress > 0)
 921                         die(_("--progress can't be used with --incremental or porcelain formats"));
 922                 show_progress = 0;
 923         } else if (show_progress < 0)
 924                 show_progress = isatty(2);
 925
 926         if (0 < abbrev && abbrev < hexsz)
 927                 /* one more abbrev length is needed for the boundary commit */
 928                 abbrev++;
 929         else if (!abbrev)
 930                 abbrev = hexsz;
 931
 932         if (revs_file && read_ancestry(revs_file))
 933                 die_errno("reading graft file '%s' failed", revs_file);
 934
 935         if (cmd_is_annotate) {
 936                 output_option |= OUTPUT_ANNOTATE_COMPAT;
 937                 blame_date_mode.type = DATE_ISO8601;
 938         } else {
 939                 blame_date_mode = revs.date_mode;
 940         }
 941
 942         /* The maximum width used to show the dates */
 943         switch (blame_date_mode.type) {
 944         case DATE_RFC2822:
 945                 blame_date_width = sizeof("Thu, 19 Oct 2006 16:00:04 -0700");
 946                 break;
 947         case DATE_ISO8601_STRICT:
 948                 blame_date_width = sizeof("2006-10-19T16:00:04-07:00");
 949                 break;
 950         case DATE_ISO8601:
 951                 blame_date_width = sizeof("2006-10-19 16:00:04 -0700");
 952                 break;
 953         case DATE_RAW:
 954                 blame_date_width = sizeof("1161298804 -0700");
 955                 break;
 956         case DATE_UNIX:
 957                 blame_date_width = sizeof("1161298804");
 958                 break;
 959         case DATE_SHORT:
 960                 blame_date_width = sizeof("2006-10-19");
 961                 break;
 962         case DATE_RELATIVE:
 963                 /*
 964                  * TRANSLATORS: This string is used to tell us the
 965                  * maximum display width for a relative timestamp in
 966                  * "git blame" output.  For C locale, "4 years, 11
 967                  * months ago", which takes 22 places, is the longest
 968                  * among various forms of relative timestamps, but
 969                  * your language may need more or fewer display
 970                  * columns.
 971                  */
 972                 blame_date_width = utf8_strwidth(_("4 years, 11 months ago")) + 1; /* add the null */
 973                 break;
 974         case DATE_HUMAN:
 975                 /* If the year is shown, no time is shown */
 976                 blame_date_width = sizeof("Thu Oct 19 16:00");
 977                 break;
 978         case DATE_NORMAL:
 979                 blame_date_width = sizeof("Thu Oct 19 16:00:04 2006 -0700");
 980                 break;
 981         case DATE_STRFTIME:
 982                 blame_date_width = strlen(show_date(0, 0, &blame_date_mode)) + 1; /* add the null */
 983                 break;
 984         }
 985         blame_date_width -= 1; /* strip the null */
 986
 987         if (revs.diffopt.flags.find_copies_harder)
 988                 opt |= (PICKAXE_BLAME_COPY | PICKAXE_BLAME_MOVE |
 989                         PICKAXE_BLAME_COPY_HARDER);
 990
 991         /*
 992          * We have collected options unknown to us in argv[1..unk]
 993          * which are to be passed to revision machinery if we are
 994          * going to do the "bottom" processing.
 995          *
 996          * The remaining are:
 997          *
 998          * (1) if dashdash_pos != 0, it is either
 999          *     "blame [revisions] -- <path>" or
1000          *     "blame -- <path> <rev>"
1001          *
1002          * (2) otherwise, it is one of the two:
1003          *     "blame [revisions] <path>"
1004          *     "blame <path> <rev>"
1005          *
1006          * Note that we must strip out <path> from the arguments: we do not
1007          * want the path pruning but we may want "bottom" processing.
1008          */
1009         if (dashdash_pos) {
1010                 switch (argc - dashdash_pos - 1) {
1011                 case 2: /* (1b) */
1012                         if (argc != 4)
1013                                 usage_with_options(blame_opt_usage, options);
1014                         /* reorder for the new way: <rev> -- <path> */
1015                         argv[1] = argv[3];
1016                         argv[3] = argv[2];
1017                         argv[2] = "--";
1018                         /* FALLTHROUGH */
1019                 case 1: /* (1a) */
1020                         path = add_prefix(prefix, argv[--argc]);
1021                         argv[argc] = NULL;
1022                         break;
1023                 default:
1024                         usage_with_options(blame_opt_usage, options);
1025                 }
1026         } else {
1027                 if (argc < 2)
1028                         usage_with_options(blame_opt_usage, options);
1029                 if (argc == 3 && is_a_rev(argv[argc - 1])) { /* (2b) */
1030                         path = add_prefix(prefix, argv[1]);
1031                         argv[1] = argv[2];
1032                 } else {        /* (2a) */
1033                         if (argc == 2 && is_a_rev(argv[1]) && !get_git_work_tree())
1034                                 die("missing <path> to blame");
1035                         path = add_prefix(prefix, argv[argc - 1]);
1036                 }
1037                 argv[argc - 1] = "--";
1038         }
1039
1040         revs.disable_stdin = 1;
1041         setup_revisions(argc, argv, &revs, NULL);
1042         if (!revs.pending.nr && is_bare_repository()) {
1043                 struct commit *head_commit;
1044                 struct object_id head_oid;
1045
1046                 if (!resolve_ref_unsafe("HEAD", RESOLVE_REF_READING,
1047                                         &head_oid, NULL) ||
1048                     !(head_commit = lookup_commit_reference_gently(revs.repo,
1049                                                              &head_oid, 1)))
1050                         die("no such ref: HEAD");
1051
1052                 add_pending_object(&revs, &head_commit->object, "HEAD");
1053         }
1054
1055         init_scoreboard(&sb);
1056         sb.revs = &revs;
1057         sb.contents_from = contents_from;
1058         sb.reverse = reverse;
1059         sb.repo = the_repository;
1060         build_ignorelist(&sb, &ignore_revs_file_list, &ignore_rev_list);
1061         string_list_clear(&ignore_revs_file_list, 0);
1062         string_list_clear(&ignore_rev_list, 0);
1063         setup_scoreboard(&sb, path, &o);
1064
1065         /*
1066          * Changed-path Bloom filters are disabled when looking
1067          * for copies.
1068          */
1069         if (!(opt & PICKAXE_BLAME_COPY))
1070                 setup_blame_bloom_data(&sb, path);
1071
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
1176         cleanup_scoreboard(&sb);
1177         return 0;
1178 }