]>
Commit | Line | Data |
---|---|---|
cee7f245 | 1 | /* |
31653c1a | 2 | * Blame |
cee7f245 JH |
3 | * |
4 | * Copyright (c) 2006, Junio C Hamano | |
5 | */ | |
6 | ||
7 | #include "cache.h" | |
8 | #include "builtin.h" | |
9 | #include "blob.h" | |
10 | #include "commit.h" | |
11 | #include "tag.h" | |
12 | #include "tree-walk.h" | |
13 | #include "diff.h" | |
14 | #include "diffcore.h" | |
15 | #include "revision.h" | |
717d1462 | 16 | #include "quote.h" |
cee7f245 | 17 | #include "xdiff-interface.h" |
1cfe7733 | 18 | #include "cache-tree.h" |
c455c87c | 19 | #include "string-list.h" |
f9567384 | 20 | #include "mailmap.h" |
5817da01 | 21 | #include "parse-options.h" |
ffaf9cc0 | 22 | #include "utf8.h" |
cee7f245 | 23 | |
588c038a | 24 | static char blame_usage[] = "git blame [options] [rev-opts] [rev] [--] file"; |
5817da01 PH |
25 | |
26 | static const char *blame_opt_usage[] = { | |
27 | blame_usage, | |
28 | "", | |
76447d23 | 29 | "[rev-opts] are documented in git-rev-list(1)", |
5817da01 PH |
30 | NULL |
31 | }; | |
cee7f245 JH |
32 | |
33 | static int longest_file; | |
34 | static int longest_author; | |
35 | static int max_orig_digits; | |
36 | static int max_digits; | |
5ff62c30 | 37 | static int max_score_digits; |
4c10a5ca | 38 | static int show_root; |
85af7929 | 39 | static int reverse; |
4c10a5ca | 40 | static int blank_boundary; |
717d1462 | 41 | static int incremental; |
b82871b3 | 42 | static int xdl_opts = XDF_NEED_MINIMAL; |
31653c1a EL |
43 | |
44 | static enum date_mode blame_date_mode = DATE_ISO8601; | |
45 | static size_t blame_date_width; | |
46 | ||
c455c87c | 47 | static struct string_list mailmap; |
cee7f245 | 48 | |
54a4c617 JH |
49 | #ifndef DEBUG |
50 | #define DEBUG 0 | |
51 | #endif | |
52 | ||
c2e525d9 JH |
53 | /* stats */ |
54 | static int num_read_blob; | |
55 | static int num_get_patch; | |
56 | static int num_commits; | |
57 | ||
d24bba80 | 58 | #define PICKAXE_BLAME_MOVE 01 |
18abd745 JH |
59 | #define PICKAXE_BLAME_COPY 02 |
60 | #define PICKAXE_BLAME_COPY_HARDER 04 | |
c63777c0 | 61 | #define PICKAXE_BLAME_COPY_HARDEST 010 |
d24bba80 | 62 | |
4a0fc95f JH |
63 | /* |
64 | * blame for a blame_entry with score lower than these thresholds | |
65 | * is not passed to the parent using move/copy logic. | |
66 | */ | |
67 | static unsigned blame_move_score; | |
68 | static unsigned blame_copy_score; | |
69 | #define BLAME_DEFAULT_MOVE_SCORE 20 | |
70 | #define BLAME_DEFAULT_COPY_SCORE 40 | |
71 | ||
cee7f245 JH |
72 | /* bits #0..7 in revision.h, #8..11 used for merge_bases() in commit.c */ |
73 | #define METAINFO_SHOWN (1u<<12) | |
74 | #define MORE_THAN_ONE_PATH (1u<<13) | |
75 | ||
76 | /* | |
54a4c617 | 77 | * One blob in a commit that is being suspected |
cee7f245 JH |
78 | */ |
79 | struct origin { | |
54a4c617 | 80 | int refcnt; |
96e11709 | 81 | struct origin *previous; |
cee7f245 | 82 | struct commit *commit; |
c2e525d9 | 83 | mmfile_t file; |
cee7f245 JH |
84 | unsigned char blob_sha1[20]; |
85 | char path[FLEX_ARRAY]; | |
86 | }; | |
87 | ||
1732a1fd JH |
88 | /* |
89 | * Given an origin, prepare mmfile_t structure to be used by the | |
90 | * diff machinery | |
91 | */ | |
f6c07d7d | 92 | static void fill_origin_blob(struct origin *o, mmfile_t *file) |
c2e525d9 JH |
93 | { |
94 | if (!o->file.ptr) { | |
21666f1a | 95 | enum object_type type; |
c2e525d9 | 96 | num_read_blob++; |
21666f1a | 97 | file->ptr = read_sha1_file(o->blob_sha1, &type, |
c2e525d9 | 98 | (unsigned long *)(&(file->size))); |
ab43e495 JH |
99 | if (!file->ptr) |
100 | die("Cannot read blob %s for path %s", | |
101 | sha1_to_hex(o->blob_sha1), | |
102 | o->path); | |
c2e525d9 JH |
103 | o->file = *file; |
104 | } | |
105 | else | |
106 | *file = o->file; | |
c2e525d9 JH |
107 | } |
108 | ||
1732a1fd JH |
109 | /* |
110 | * Origin is refcounted and usually we keep the blob contents to be | |
111 | * reused. | |
112 | */ | |
54a4c617 JH |
113 | static inline struct origin *origin_incref(struct origin *o) |
114 | { | |
115 | if (o) | |
116 | o->refcnt++; | |
117 | return o; | |
118 | } | |
119 | ||
120 | static void origin_decref(struct origin *o) | |
121 | { | |
122 | if (o && --o->refcnt <= 0) { | |
96e11709 JH |
123 | if (o->previous) |
124 | origin_decref(o->previous); | |
8e0f7003 | 125 | free(o->file.ptr); |
54a4c617 JH |
126 | free(o); |
127 | } | |
128 | } | |
129 | ||
7c3c7962 JH |
130 | static void drop_origin_blob(struct origin *o) |
131 | { | |
132 | if (o->file.ptr) { | |
133 | free(o->file.ptr); | |
134 | o->file.ptr = NULL; | |
135 | } | |
136 | } | |
137 | ||
1732a1fd JH |
138 | /* |
139 | * Each group of lines is described by a blame_entry; it can be split | |
140 | * as we pass blame to the parents. They form a linked list in the | |
141 | * scoreboard structure, sorted by the target line number. | |
142 | */ | |
cee7f245 JH |
143 | struct blame_entry { |
144 | struct blame_entry *prev; | |
145 | struct blame_entry *next; | |
146 | ||
147 | /* the first line of this group in the final image; | |
148 | * internally all line numbers are 0 based. | |
149 | */ | |
150 | int lno; | |
151 | ||
152 | /* how many lines this group has */ | |
153 | int num_lines; | |
154 | ||
155 | /* the commit that introduced this group into the final image */ | |
156 | struct origin *suspect; | |
157 | ||
158 | /* true if the suspect is truly guilty; false while we have not | |
159 | * checked if the group came from one of its parents. | |
160 | */ | |
161 | char guilty; | |
162 | ||
eb46328e AG |
163 | /* true if the entry has been scanned for copies in the current parent |
164 | */ | |
165 | char scanned; | |
166 | ||
cee7f245 JH |
167 | /* the line number of the first line of this group in the |
168 | * suspect's file; internally all line numbers are 0 based. | |
169 | */ | |
170 | int s_lno; | |
5ff62c30 JH |
171 | |
172 | /* how significant this entry is -- cached to avoid | |
1732a1fd | 173 | * scanning the lines over and over. |
5ff62c30 JH |
174 | */ |
175 | unsigned score; | |
cee7f245 JH |
176 | }; |
177 | ||
1732a1fd JH |
178 | /* |
179 | * The current state of the blame assignment. | |
180 | */ | |
cee7f245 JH |
181 | struct scoreboard { |
182 | /* the final commit (i.e. where we started digging from) */ | |
183 | struct commit *final; | |
85af7929 | 184 | struct rev_info *revs; |
cee7f245 JH |
185 | const char *path; |
186 | ||
1732a1fd JH |
187 | /* |
188 | * The contents in the final image. | |
189 | * Used by many functions to obtain contents of the nth line, | |
190 | * indexed with scoreboard.lineno[blame_entry.lno]. | |
cee7f245 JH |
191 | */ |
192 | const char *final_buf; | |
193 | unsigned long final_buf_size; | |
194 | ||
195 | /* linked list of blames */ | |
196 | struct blame_entry *ent; | |
197 | ||
612702e8 | 198 | /* look-up a line in the final buffer */ |
cee7f245 JH |
199 | int num_lines; |
200 | int *lineno; | |
201 | }; | |
202 | ||
171dccd5 | 203 | static inline int same_suspect(struct origin *a, struct origin *b) |
46014766 | 204 | { |
171dccd5 | 205 | if (a == b) |
57584d9e | 206 | return 1; |
171dccd5 JH |
207 | if (a->commit != b->commit) |
208 | return 0; | |
209 | return !strcmp(a->path, b->path); | |
46014766 JH |
210 | } |
211 | ||
54a4c617 JH |
212 | static void sanity_check_refcnt(struct scoreboard *); |
213 | ||
1732a1fd JH |
214 | /* |
215 | * If two blame entries that are next to each other came from | |
216 | * contiguous lines in the same origin (i.e. <commit, path> pair), | |
217 | * merge them together. | |
218 | */ | |
cee7f245 JH |
219 | static void coalesce(struct scoreboard *sb) |
220 | { | |
221 | struct blame_entry *ent, *next; | |
222 | ||
223 | for (ent = sb->ent; ent && (next = ent->next); ent = next) { | |
171dccd5 | 224 | if (same_suspect(ent->suspect, next->suspect) && |
cee7f245 JH |
225 | ent->guilty == next->guilty && |
226 | ent->s_lno + ent->num_lines == next->s_lno) { | |
227 | ent->num_lines += next->num_lines; | |
228 | ent->next = next->next; | |
229 | if (ent->next) | |
230 | ent->next->prev = ent; | |
54a4c617 | 231 | origin_decref(next->suspect); |
cee7f245 | 232 | free(next); |
46014766 | 233 | ent->score = 0; |
cee7f245 JH |
234 | next = ent; /* again */ |
235 | } | |
236 | } | |
54a4c617 JH |
237 | |
238 | if (DEBUG) /* sanity */ | |
239 | sanity_check_refcnt(sb); | |
cee7f245 JH |
240 | } |
241 | ||
1732a1fd JH |
242 | /* |
243 | * Given a commit and a path in it, create a new origin structure. | |
244 | * The callers that add blame to the scoreboard should use | |
245 | * get_origin() to obtain shared, refcounted copy instead of calling | |
246 | * this function directly. | |
247 | */ | |
854b97f6 JH |
248 | static struct origin *make_origin(struct commit *commit, const char *path) |
249 | { | |
250 | struct origin *o; | |
251 | o = xcalloc(1, sizeof(*o) + strlen(path) + 1); | |
252 | o->commit = commit; | |
253 | o->refcnt = 1; | |
254 | strcpy(o->path, path); | |
255 | return o; | |
256 | } | |
257 | ||
1732a1fd JH |
258 | /* |
259 | * Locate an existing origin or create a new one. | |
260 | */ | |
f6c0e191 JH |
261 | static struct origin *get_origin(struct scoreboard *sb, |
262 | struct commit *commit, | |
263 | const char *path) | |
cee7f245 | 264 | { |
f6c0e191 | 265 | struct blame_entry *e; |
cee7f245 | 266 | |
f6c0e191 JH |
267 | for (e = sb->ent; e; e = e->next) { |
268 | if (e->suspect->commit == commit && | |
269 | !strcmp(e->suspect->path, path)) | |
54a4c617 | 270 | return origin_incref(e->suspect); |
cee7f245 | 271 | } |
854b97f6 | 272 | return make_origin(commit, path); |
cee7f245 JH |
273 | } |
274 | ||
1732a1fd JH |
275 | /* |
276 | * Fill the blob_sha1 field of an origin if it hasn't, so that later | |
277 | * call to fill_origin_blob() can use it to locate the data. blob_sha1 | |
278 | * for an origin is also used to pass the blame for the entire file to | |
279 | * the parent to detect the case where a child's blob is identical to | |
280 | * that of its parent's. | |
281 | */ | |
f6c0e191 JH |
282 | static int fill_blob_sha1(struct origin *origin) |
283 | { | |
284 | unsigned mode; | |
f6c0e191 JH |
285 | |
286 | if (!is_null_sha1(origin->blob_sha1)) | |
287 | return 0; | |
288 | if (get_tree_entry(origin->commit->object.sha1, | |
289 | origin->path, | |
290 | origin->blob_sha1, &mode)) | |
291 | goto error_out; | |
21666f1a | 292 | if (sha1_object_info(origin->blob_sha1, NULL) != OBJ_BLOB) |
f6c0e191 JH |
293 | goto error_out; |
294 | return 0; | |
295 | error_out: | |
296 | hashclr(origin->blob_sha1); | |
297 | return -1; | |
298 | } | |
299 | ||
1732a1fd JH |
300 | /* |
301 | * We have an origin -- check if the same path exists in the | |
302 | * parent and return an origin structure to represent it. | |
303 | */ | |
f6c0e191 | 304 | static struct origin *find_origin(struct scoreboard *sb, |
cee7f245 JH |
305 | struct commit *parent, |
306 | struct origin *origin) | |
307 | { | |
308 | struct origin *porigin = NULL; | |
309 | struct diff_options diff_opts; | |
f6c0e191 JH |
310 | const char *paths[2]; |
311 | ||
0d981c67 | 312 | if (parent->util) { |
1732a1fd JH |
313 | /* |
314 | * Each commit object can cache one origin in that | |
315 | * commit. This is a freestanding copy of origin and | |
316 | * not refcounted. | |
854b97f6 | 317 | */ |
0d981c67 | 318 | struct origin *cached = parent->util; |
854b97f6 | 319 | if (!strcmp(cached->path, origin->path)) { |
1732a1fd JH |
320 | /* |
321 | * The same path between origin and its parent | |
322 | * without renaming -- the most common case. | |
323 | */ | |
854b97f6 | 324 | porigin = get_origin(sb, parent, cached->path); |
1732a1fd JH |
325 | |
326 | /* | |
327 | * If the origin was newly created (i.e. get_origin | |
328 | * would call make_origin if none is found in the | |
329 | * scoreboard), it does not know the blob_sha1, | |
330 | * so copy it. Otherwise porigin was in the | |
331 | * scoreboard and already knows blob_sha1. | |
332 | */ | |
854b97f6 JH |
333 | if (porigin->refcnt == 1) |
334 | hashcpy(porigin->blob_sha1, cached->blob_sha1); | |
335 | return porigin; | |
336 | } | |
337 | /* otherwise it was not very useful; free it */ | |
338 | free(parent->util); | |
339 | parent->util = NULL; | |
0d981c67 JH |
340 | } |
341 | ||
f6c0e191 JH |
342 | /* See if the origin->path is different between parent |
343 | * and origin first. Most of the time they are the | |
344 | * same and diff-tree is fairly efficient about this. | |
345 | */ | |
346 | diff_setup(&diff_opts); | |
8f67f8ae | 347 | DIFF_OPT_SET(&diff_opts, RECURSIVE); |
f6c0e191 JH |
348 | diff_opts.detect_rename = 0; |
349 | diff_opts.output_format = DIFF_FORMAT_NO_OUTPUT; | |
350 | paths[0] = origin->path; | |
351 | paths[1] = NULL; | |
352 | ||
353 | diff_tree_setup_paths(paths, &diff_opts); | |
354 | if (diff_setup_done(&diff_opts) < 0) | |
355 | die("diff-setup"); | |
1cfe7733 JH |
356 | |
357 | if (is_null_sha1(origin->commit->object.sha1)) | |
358 | do_diff_cache(parent->tree->object.sha1, &diff_opts); | |
359 | else | |
360 | diff_tree_sha1(parent->tree->object.sha1, | |
361 | origin->commit->tree->object.sha1, | |
362 | "", &diff_opts); | |
f6c0e191 JH |
363 | diffcore_std(&diff_opts); |
364 | ||
365 | /* It is either one entry that says "modified", or "created", | |
366 | * or nothing. | |
367 | */ | |
368 | if (!diff_queued_diff.nr) { | |
369 | /* The path is the same as parent */ | |
370 | porigin = get_origin(sb, parent, origin->path); | |
371 | hashcpy(porigin->blob_sha1, origin->blob_sha1); | |
372 | } | |
373 | else if (diff_queued_diff.nr != 1) | |
acca687f | 374 | die("internal error in blame::find_origin"); |
f6c0e191 JH |
375 | else { |
376 | struct diff_filepair *p = diff_queued_diff.queue[0]; | |
377 | switch (p->status) { | |
378 | default: | |
acca687f | 379 | die("internal error in blame::find_origin (%c)", |
f6c0e191 JH |
380 | p->status); |
381 | case 'M': | |
382 | porigin = get_origin(sb, parent, origin->path); | |
383 | hashcpy(porigin->blob_sha1, p->one->sha1); | |
384 | break; | |
385 | case 'A': | |
386 | case 'T': | |
387 | /* Did not exist in parent, or type changed */ | |
388 | break; | |
389 | } | |
390 | } | |
391 | diff_flush(&diff_opts); | |
03b69c76 | 392 | diff_tree_release_paths(&diff_opts); |
0d981c67 | 393 | if (porigin) { |
1732a1fd JH |
394 | /* |
395 | * Create a freestanding copy that is not part of | |
396 | * the refcounted origin found in the scoreboard, and | |
397 | * cache it in the commit. | |
398 | */ | |
854b97f6 | 399 | struct origin *cached; |
1732a1fd | 400 | |
854b97f6 JH |
401 | cached = make_origin(porigin->commit, porigin->path); |
402 | hashcpy(cached->blob_sha1, porigin->blob_sha1); | |
403 | parent->util = cached; | |
0d981c67 | 404 | } |
f69e743d JH |
405 | return porigin; |
406 | } | |
f6c0e191 | 407 | |
1732a1fd JH |
408 | /* |
409 | * We have an origin -- find the path that corresponds to it in its | |
410 | * parent and return an origin structure to represent it. | |
411 | */ | |
f69e743d JH |
412 | static struct origin *find_rename(struct scoreboard *sb, |
413 | struct commit *parent, | |
414 | struct origin *origin) | |
415 | { | |
416 | struct origin *porigin = NULL; | |
417 | struct diff_options diff_opts; | |
418 | int i; | |
419 | const char *paths[2]; | |
cee7f245 JH |
420 | |
421 | diff_setup(&diff_opts); | |
8f67f8ae | 422 | DIFF_OPT_SET(&diff_opts, RECURSIVE); |
cee7f245 JH |
423 | diff_opts.detect_rename = DIFF_DETECT_RENAME; |
424 | diff_opts.output_format = DIFF_FORMAT_NO_OUTPUT; | |
2f3f8b21 | 425 | diff_opts.single_follow = origin->path; |
cee7f245 JH |
426 | paths[0] = NULL; |
427 | diff_tree_setup_paths(paths, &diff_opts); | |
428 | if (diff_setup_done(&diff_opts) < 0) | |
429 | die("diff-setup"); | |
1cfe7733 JH |
430 | |
431 | if (is_null_sha1(origin->commit->object.sha1)) | |
432 | do_diff_cache(parent->tree->object.sha1, &diff_opts); | |
433 | else | |
434 | diff_tree_sha1(parent->tree->object.sha1, | |
435 | origin->commit->tree->object.sha1, | |
436 | "", &diff_opts); | |
cee7f245 JH |
437 | diffcore_std(&diff_opts); |
438 | ||
439 | for (i = 0; i < diff_queued_diff.nr; i++) { | |
440 | struct diff_filepair *p = diff_queued_diff.queue[i]; | |
612702e8 | 441 | if ((p->status == 'R' || p->status == 'C') && |
f6c0e191 JH |
442 | !strcmp(p->two->path, origin->path)) { |
443 | porigin = get_origin(sb, parent, p->one->path); | |
444 | hashcpy(porigin->blob_sha1, p->one->sha1); | |
cee7f245 JH |
445 | break; |
446 | } | |
447 | } | |
448 | diff_flush(&diff_opts); | |
03b69c76 | 449 | diff_tree_release_paths(&diff_opts); |
cee7f245 JH |
450 | return porigin; |
451 | } | |
452 | ||
1732a1fd | 453 | /* |
3dff5379 | 454 | * Link in a new blame entry to the scoreboard. Entries that cover the |
1732a1fd JH |
455 | * same line range have been removed from the scoreboard previously. |
456 | */ | |
cee7f245 JH |
457 | static void add_blame_entry(struct scoreboard *sb, struct blame_entry *e) |
458 | { | |
459 | struct blame_entry *ent, *prev = NULL; | |
460 | ||
54a4c617 JH |
461 | origin_incref(e->suspect); |
462 | ||
cee7f245 JH |
463 | for (ent = sb->ent; ent && ent->lno < e->lno; ent = ent->next) |
464 | prev = ent; | |
465 | ||
466 | /* prev, if not NULL, is the last one that is below e */ | |
467 | e->prev = prev; | |
468 | if (prev) { | |
469 | e->next = prev->next; | |
470 | prev->next = e; | |
471 | } | |
472 | else { | |
473 | e->next = sb->ent; | |
474 | sb->ent = e; | |
475 | } | |
476 | if (e->next) | |
477 | e->next->prev = e; | |
478 | } | |
479 | ||
1732a1fd JH |
480 | /* |
481 | * src typically is on-stack; we want to copy the information in it to | |
790296fd | 482 | * a malloced blame_entry that is already on the linked list of the |
1732a1fd JH |
483 | * scoreboard. The origin of dst loses a refcnt while the origin of src |
484 | * gains one. | |
485 | */ | |
cee7f245 JH |
486 | static void dup_entry(struct blame_entry *dst, struct blame_entry *src) |
487 | { | |
488 | struct blame_entry *p, *n; | |
54a4c617 | 489 | |
cee7f245 JH |
490 | p = dst->prev; |
491 | n = dst->next; | |
54a4c617 JH |
492 | origin_incref(src->suspect); |
493 | origin_decref(dst->suspect); | |
cee7f245 JH |
494 | memcpy(dst, src, sizeof(*src)); |
495 | dst->prev = p; | |
496 | dst->next = n; | |
5ff62c30 | 497 | dst->score = 0; |
cee7f245 JH |
498 | } |
499 | ||
500 | static const char *nth_line(struct scoreboard *sb, int lno) | |
501 | { | |
502 | return sb->final_buf + sb->lineno[lno]; | |
503 | } | |
504 | ||
1732a1fd JH |
505 | /* |
506 | * It is known that lines between tlno to same came from parent, and e | |
507 | * has an overlap with that range. it also is known that parent's | |
508 | * line plno corresponds to e's line tlno. | |
509 | * | |
510 | * <---- e -----> | |
511 | * <------> | |
512 | * <------------> | |
513 | * <------------> | |
514 | * <------------------> | |
515 | * | |
516 | * Split e into potentially three parts; before this chunk, the chunk | |
517 | * to be blamed for the parent, and after that portion. | |
518 | */ | |
54a4c617 | 519 | static void split_overlap(struct blame_entry *split, |
cee7f245 JH |
520 | struct blame_entry *e, |
521 | int tlno, int plno, int same, | |
522 | struct origin *parent) | |
523 | { | |
cee7f245 JH |
524 | int chunk_end_lno; |
525 | memset(split, 0, sizeof(struct blame_entry [3])); | |
526 | ||
527 | if (e->s_lno < tlno) { | |
528 | /* there is a pre-chunk part not blamed on parent */ | |
54a4c617 | 529 | split[0].suspect = origin_incref(e->suspect); |
cee7f245 JH |
530 | split[0].lno = e->lno; |
531 | split[0].s_lno = e->s_lno; | |
532 | split[0].num_lines = tlno - e->s_lno; | |
533 | split[1].lno = e->lno + tlno - e->s_lno; | |
534 | split[1].s_lno = plno; | |
535 | } | |
536 | else { | |
537 | split[1].lno = e->lno; | |
538 | split[1].s_lno = plno + (e->s_lno - tlno); | |
539 | } | |
540 | ||
541 | if (same < e->s_lno + e->num_lines) { | |
542 | /* there is a post-chunk part not blamed on parent */ | |
54a4c617 | 543 | split[2].suspect = origin_incref(e->suspect); |
cee7f245 JH |
544 | split[2].lno = e->lno + (same - e->s_lno); |
545 | split[2].s_lno = e->s_lno + (same - e->s_lno); | |
546 | split[2].num_lines = e->s_lno + e->num_lines - same; | |
547 | chunk_end_lno = split[2].lno; | |
548 | } | |
549 | else | |
550 | chunk_end_lno = e->lno + e->num_lines; | |
551 | split[1].num_lines = chunk_end_lno - split[1].lno; | |
552 | ||
1732a1fd JH |
553 | /* |
554 | * if it turns out there is nothing to blame the parent for, | |
555 | * forget about the splitting. !split[1].suspect signals this. | |
556 | */ | |
cee7f245 JH |
557 | if (split[1].num_lines < 1) |
558 | return; | |
54a4c617 | 559 | split[1].suspect = origin_incref(parent); |
cee7f245 JH |
560 | } |
561 | ||
1732a1fd JH |
562 | /* |
563 | * split_overlap() divided an existing blame e into up to three parts | |
564 | * in split. Adjust the linked list of blames in the scoreboard to | |
565 | * reflect the split. | |
566 | */ | |
cee7f245 | 567 | static void split_blame(struct scoreboard *sb, |
54a4c617 | 568 | struct blame_entry *split, |
cee7f245 JH |
569 | struct blame_entry *e) |
570 | { | |
571 | struct blame_entry *new_entry; | |
572 | ||
573 | if (split[0].suspect && split[2].suspect) { | |
1732a1fd | 574 | /* The first part (reuse storage for the existing entry e) */ |
cee7f245 JH |
575 | dup_entry(e, &split[0]); |
576 | ||
1732a1fd | 577 | /* The last part -- me */ |
cee7f245 JH |
578 | new_entry = xmalloc(sizeof(*new_entry)); |
579 | memcpy(new_entry, &(split[2]), sizeof(struct blame_entry)); | |
580 | add_blame_entry(sb, new_entry); | |
581 | ||
1732a1fd | 582 | /* ... and the middle part -- parent */ |
cee7f245 JH |
583 | new_entry = xmalloc(sizeof(*new_entry)); |
584 | memcpy(new_entry, &(split[1]), sizeof(struct blame_entry)); | |
585 | add_blame_entry(sb, new_entry); | |
586 | } | |
587 | else if (!split[0].suspect && !split[2].suspect) | |
1732a1fd JH |
588 | /* |
589 | * The parent covers the entire area; reuse storage for | |
590 | * e and replace it with the parent. | |
591 | */ | |
cee7f245 JH |
592 | dup_entry(e, &split[1]); |
593 | else if (split[0].suspect) { | |
1732a1fd | 594 | /* me and then parent */ |
cee7f245 JH |
595 | dup_entry(e, &split[0]); |
596 | ||
597 | new_entry = xmalloc(sizeof(*new_entry)); | |
598 | memcpy(new_entry, &(split[1]), sizeof(struct blame_entry)); | |
599 | add_blame_entry(sb, new_entry); | |
600 | } | |
601 | else { | |
1732a1fd | 602 | /* parent and then me */ |
cee7f245 JH |
603 | dup_entry(e, &split[1]); |
604 | ||
605 | new_entry = xmalloc(sizeof(*new_entry)); | |
606 | memcpy(new_entry, &(split[2]), sizeof(struct blame_entry)); | |
607 | add_blame_entry(sb, new_entry); | |
608 | } | |
609 | ||
54a4c617 | 610 | if (DEBUG) { /* sanity */ |
cee7f245 | 611 | struct blame_entry *ent; |
612702e8 | 612 | int lno = sb->ent->lno, corrupt = 0; |
cee7f245 JH |
613 | |
614 | for (ent = sb->ent; ent; ent = ent->next) { | |
615 | if (lno != ent->lno) | |
616 | corrupt = 1; | |
617 | if (ent->s_lno < 0) | |
618 | corrupt = 1; | |
619 | lno += ent->num_lines; | |
620 | } | |
621 | if (corrupt) { | |
612702e8 | 622 | lno = sb->ent->lno; |
cee7f245 JH |
623 | for (ent = sb->ent; ent; ent = ent->next) { |
624 | printf("L %8d l %8d n %8d\n", | |
625 | lno, ent->lno, ent->num_lines); | |
626 | lno = ent->lno + ent->num_lines; | |
627 | } | |
628 | die("oops"); | |
629 | } | |
630 | } | |
631 | } | |
632 | ||
1732a1fd JH |
633 | /* |
634 | * After splitting the blame, the origins used by the | |
635 | * on-stack blame_entry should lose one refcnt each. | |
636 | */ | |
54a4c617 JH |
637 | static void decref_split(struct blame_entry *split) |
638 | { | |
639 | int i; | |
640 | ||
641 | for (i = 0; i < 3; i++) | |
642 | origin_decref(split[i].suspect); | |
643 | } | |
644 | ||
1732a1fd JH |
645 | /* |
646 | * Helper for blame_chunk(). blame_entry e is known to overlap with | |
647 | * the patch hunk; split it and pass blame to the parent. | |
648 | */ | |
cee7f245 JH |
649 | static void blame_overlap(struct scoreboard *sb, struct blame_entry *e, |
650 | int tlno, int plno, int same, | |
651 | struct origin *parent) | |
652 | { | |
653 | struct blame_entry split[3]; | |
654 | ||
655 | split_overlap(split, e, tlno, plno, same, parent); | |
54a4c617 JH |
656 | if (split[1].suspect) |
657 | split_blame(sb, split, e); | |
658 | decref_split(split); | |
cee7f245 JH |
659 | } |
660 | ||
1732a1fd JH |
661 | /* |
662 | * Find the line number of the last line the target is suspected for. | |
663 | */ | |
cee7f245 JH |
664 | static int find_last_in_target(struct scoreboard *sb, struct origin *target) |
665 | { | |
666 | struct blame_entry *e; | |
667 | int last_in_target = -1; | |
668 | ||
669 | for (e = sb->ent; e; e = e->next) { | |
171dccd5 | 670 | if (e->guilty || !same_suspect(e->suspect, target)) |
cee7f245 JH |
671 | continue; |
672 | if (last_in_target < e->s_lno + e->num_lines) | |
673 | last_in_target = e->s_lno + e->num_lines; | |
674 | } | |
675 | return last_in_target; | |
676 | } | |
677 | ||
1732a1fd JH |
678 | /* |
679 | * Process one hunk from the patch between the current suspect for | |
680 | * blame_entry e and its parent. Find and split the overlap, and | |
681 | * pass blame to the overlapping part to the parent. | |
682 | */ | |
cee7f245 JH |
683 | static void blame_chunk(struct scoreboard *sb, |
684 | int tlno, int plno, int same, | |
685 | struct origin *target, struct origin *parent) | |
686 | { | |
612702e8 | 687 | struct blame_entry *e; |
cee7f245 | 688 | |
612702e8 | 689 | for (e = sb->ent; e; e = e->next) { |
171dccd5 | 690 | if (e->guilty || !same_suspect(e->suspect, target)) |
cee7f245 JH |
691 | continue; |
692 | if (same <= e->s_lno) | |
693 | continue; | |
694 | if (tlno < e->s_lno + e->num_lines) | |
695 | blame_overlap(sb, e, tlno, plno, same, parent); | |
696 | } | |
697 | } | |
698 | ||
cdcab133 RS |
699 | struct blame_chunk_cb_data { |
700 | struct scoreboard *sb; | |
701 | struct origin *target; | |
702 | struct origin *parent; | |
703 | long plno; | |
704 | long tlno; | |
705 | }; | |
706 | ||
707 | static void blame_chunk_cb(void *data, long same, long p_next, long t_next) | |
708 | { | |
709 | struct blame_chunk_cb_data *d = data; | |
710 | blame_chunk(d->sb, d->tlno, d->plno, same, d->target, d->parent); | |
711 | d->plno = p_next; | |
712 | d->tlno = t_next; | |
713 | } | |
714 | ||
1732a1fd JH |
715 | /* |
716 | * We are looking at the origin 'target' and aiming to pass blame | |
717 | * for the lines it is suspected to its parent. Run diff to find | |
718 | * which lines came from parent and pass blame for them. | |
719 | */ | |
cee7f245 JH |
720 | static int pass_blame_to_parent(struct scoreboard *sb, |
721 | struct origin *target, | |
722 | struct origin *parent) | |
723 | { | |
cdcab133 | 724 | int last_in_target; |
7c4ed8c8 | 725 | mmfile_t file_p, file_o; |
cdcab133 RS |
726 | struct blame_chunk_cb_data d = { sb, target, parent, 0, 0 }; |
727 | xpparam_t xpp; | |
728 | xdemitconf_t xecfg; | |
cee7f245 JH |
729 | |
730 | last_in_target = find_last_in_target(sb, target); | |
731 | if (last_in_target < 0) | |
732 | return 1; /* nothing remains for this target */ | |
733 | ||
7c4ed8c8 RS |
734 | fill_origin_blob(parent, &file_p); |
735 | fill_origin_blob(target, &file_o); | |
7c4ed8c8 | 736 | num_get_patch++; |
cee7f245 | 737 | |
cdcab133 RS |
738 | memset(&xpp, 0, sizeof(xpp)); |
739 | xpp.flags = xdl_opts; | |
740 | memset(&xecfg, 0, sizeof(xecfg)); | |
741 | xecfg.ctxlen = 0; | |
742 | xdi_diff_hunks(&file_p, &file_o, blame_chunk_cb, &d, &xpp, &xecfg); | |
1732a1fd | 743 | /* The rest (i.e. anything after tlno) are the same as the parent */ |
cdcab133 | 744 | blame_chunk(sb, d.tlno, d.plno, last_in_target, target, parent); |
cee7f245 | 745 | |
cee7f245 JH |
746 | return 0; |
747 | } | |
748 | ||
1732a1fd JH |
749 | /* |
750 | * The lines in blame_entry after splitting blames many times can become | |
751 | * very small and trivial, and at some point it becomes pointless to | |
752 | * blame the parents. E.g. "\t\t}\n\t}\n\n" appears everywhere in any | |
753 | * ordinary C program, and it is not worth to say it was copied from | |
754 | * totally unrelated file in the parent. | |
755 | * | |
756 | * Compute how trivial the lines in the blame_entry are. | |
757 | */ | |
5ff62c30 JH |
758 | static unsigned ent_score(struct scoreboard *sb, struct blame_entry *e) |
759 | { | |
760 | unsigned score; | |
761 | const char *cp, *ep; | |
762 | ||
763 | if (e->score) | |
764 | return e->score; | |
765 | ||
612702e8 | 766 | score = 1; |
5ff62c30 JH |
767 | cp = nth_line(sb, e->lno); |
768 | ep = nth_line(sb, e->lno + e->num_lines); | |
769 | while (cp < ep) { | |
770 | unsigned ch = *((unsigned char *)cp); | |
771 | if (isalnum(ch)) | |
772 | score++; | |
773 | cp++; | |
774 | } | |
775 | e->score = score; | |
776 | return score; | |
777 | } | |
778 | ||
1732a1fd JH |
779 | /* |
780 | * best_so_far[] and this[] are both a split of an existing blame_entry | |
781 | * that passes blame to the parent. Maintain best_so_far the best split | |
782 | * so far, by comparing this and best_so_far and copying this into | |
783 | * bst_so_far as needed. | |
784 | */ | |
5ff62c30 | 785 | static void copy_split_if_better(struct scoreboard *sb, |
54a4c617 JH |
786 | struct blame_entry *best_so_far, |
787 | struct blame_entry *this) | |
d24bba80 | 788 | { |
54a4c617 JH |
789 | int i; |
790 | ||
d24bba80 JH |
791 | if (!this[1].suspect) |
792 | return; | |
5ff62c30 JH |
793 | if (best_so_far[1].suspect) { |
794 | if (ent_score(sb, &this[1]) < ent_score(sb, &best_so_far[1])) | |
795 | return; | |
796 | } | |
54a4c617 JH |
797 | |
798 | for (i = 0; i < 3; i++) | |
799 | origin_incref(this[i].suspect); | |
800 | decref_split(best_so_far); | |
d24bba80 JH |
801 | memcpy(best_so_far, this, sizeof(struct blame_entry [3])); |
802 | } | |
803 | ||
dd166aa8 JH |
804 | /* |
805 | * We are looking at a part of the final image represented by | |
806 | * ent (tlno and same are offset by ent->s_lno). | |
807 | * tlno is where we are looking at in the final image. | |
808 | * up to (but not including) same match preimage. | |
809 | * plno is where we are looking at in the preimage. | |
810 | * | |
811 | * <-------------- final image ----------------------> | |
812 | * <------ent------> | |
813 | * ^tlno ^same | |
814 | * <---------preimage-----> | |
815 | * ^plno | |
816 | * | |
817 | * All line numbers are 0-based. | |
818 | */ | |
819 | static void handle_split(struct scoreboard *sb, | |
820 | struct blame_entry *ent, | |
821 | int tlno, int plno, int same, | |
822 | struct origin *parent, | |
823 | struct blame_entry *split) | |
824 | { | |
825 | if (ent->num_lines <= tlno) | |
826 | return; | |
827 | if (tlno < same) { | |
828 | struct blame_entry this[3]; | |
829 | tlno += ent->s_lno; | |
830 | same += ent->s_lno; | |
831 | split_overlap(this, ent, tlno, plno, same, parent); | |
832 | copy_split_if_better(sb, split, this); | |
833 | decref_split(this); | |
834 | } | |
835 | } | |
836 | ||
cdcab133 RS |
837 | struct handle_split_cb_data { |
838 | struct scoreboard *sb; | |
839 | struct blame_entry *ent; | |
840 | struct origin *parent; | |
841 | struct blame_entry *split; | |
842 | long plno; | |
843 | long tlno; | |
844 | }; | |
845 | ||
846 | static void handle_split_cb(void *data, long same, long p_next, long t_next) | |
847 | { | |
848 | struct handle_split_cb_data *d = data; | |
849 | handle_split(d->sb, d->ent, d->tlno, d->plno, same, d->parent, d->split); | |
850 | d->plno = p_next; | |
851 | d->tlno = t_next; | |
852 | } | |
853 | ||
1732a1fd JH |
854 | /* |
855 | * Find the lines from parent that are the same as ent so that | |
856 | * we can pass blames to it. file_p has the blob contents for | |
857 | * the parent. | |
858 | */ | |
d24bba80 JH |
859 | static void find_copy_in_blob(struct scoreboard *sb, |
860 | struct blame_entry *ent, | |
861 | struct origin *parent, | |
54a4c617 | 862 | struct blame_entry *split, |
d24bba80 JH |
863 | mmfile_t *file_p) |
864 | { | |
865 | const char *cp; | |
866 | int cnt; | |
867 | mmfile_t file_o; | |
cdcab133 RS |
868 | struct handle_split_cb_data d = { sb, ent, parent, split, 0, 0 }; |
869 | xpparam_t xpp; | |
870 | xdemitconf_t xecfg; | |
d24bba80 | 871 | |
1732a1fd JH |
872 | /* |
873 | * Prepare mmfile that contains only the lines in ent. | |
874 | */ | |
d24bba80 | 875 | cp = nth_line(sb, ent->lno); |
4b25d091 | 876 | file_o.ptr = (char *) cp; |
d24bba80 JH |
877 | cnt = ent->num_lines; |
878 | ||
879 | while (cnt && cp < sb->final_buf + sb->final_buf_size) { | |
880 | if (*cp++ == '\n') | |
881 | cnt--; | |
882 | } | |
883 | file_o.size = cp - file_o.ptr; | |
884 | ||
dd166aa8 JH |
885 | /* |
886 | * file_o is a part of final image we are annotating. | |
887 | * file_p partially may match that image. | |
888 | */ | |
cdcab133 RS |
889 | memset(&xpp, 0, sizeof(xpp)); |
890 | xpp.flags = xdl_opts; | |
891 | memset(&xecfg, 0, sizeof(xecfg)); | |
892 | xecfg.ctxlen = 1; | |
d24bba80 | 893 | memset(split, 0, sizeof(struct blame_entry [3])); |
cdcab133 | 894 | xdi_diff_hunks(file_p, &file_o, handle_split_cb, &d, &xpp, &xecfg); |
dd166aa8 | 895 | /* remainder, if any, all match the preimage */ |
cdcab133 | 896 | handle_split(sb, ent, d.tlno, d.plno, ent->num_lines, parent, split); |
d24bba80 JH |
897 | } |
898 | ||
1732a1fd JH |
899 | /* |
900 | * See if lines currently target is suspected for can be attributed to | |
901 | * parent. | |
902 | */ | |
d24bba80 JH |
903 | static int find_move_in_parent(struct scoreboard *sb, |
904 | struct origin *target, | |
905 | struct origin *parent) | |
906 | { | |
650e2f67 | 907 | int last_in_target, made_progress; |
46014766 | 908 | struct blame_entry *e, split[3]; |
d24bba80 | 909 | mmfile_t file_p; |
d24bba80 JH |
910 | |
911 | last_in_target = find_last_in_target(sb, target); | |
912 | if (last_in_target < 0) | |
913 | return 1; /* nothing remains for this target */ | |
914 | ||
c2e525d9 JH |
915 | fill_origin_blob(parent, &file_p); |
916 | if (!file_p.ptr) | |
d24bba80 | 917 | return 0; |
d24bba80 | 918 | |
650e2f67 JH |
919 | made_progress = 1; |
920 | while (made_progress) { | |
921 | made_progress = 0; | |
922 | for (e = sb->ent; e; e = e->next) { | |
1617fe98 AG |
923 | if (e->guilty || !same_suspect(e->suspect, target) || |
924 | ent_score(sb, e) < blame_move_score) | |
650e2f67 JH |
925 | continue; |
926 | find_copy_in_blob(sb, e, parent, split, &file_p); | |
927 | if (split[1].suspect && | |
928 | blame_move_score < ent_score(sb, &split[1])) { | |
929 | split_blame(sb, split, e); | |
930 | made_progress = 1; | |
931 | } | |
932 | decref_split(split); | |
933 | } | |
d24bba80 | 934 | } |
d24bba80 JH |
935 | return 0; |
936 | } | |
937 | ||
33494784 JH |
938 | struct blame_list { |
939 | struct blame_entry *ent; | |
940 | struct blame_entry split[3]; | |
941 | }; | |
942 | ||
1732a1fd JH |
943 | /* |
944 | * Count the number of entries the target is suspected for, | |
945 | * and prepare a list of entry and the best split. | |
946 | */ | |
33494784 JH |
947 | static struct blame_list *setup_blame_list(struct scoreboard *sb, |
948 | struct origin *target, | |
1617fe98 | 949 | int min_score, |
33494784 JH |
950 | int *num_ents_p) |
951 | { | |
952 | struct blame_entry *e; | |
953 | int num_ents, i; | |
954 | struct blame_list *blame_list = NULL; | |
955 | ||
33494784 | 956 | for (e = sb->ent, num_ents = 0; e; e = e->next) |
1617fe98 AG |
957 | if (!e->scanned && !e->guilty && |
958 | same_suspect(e->suspect, target) && | |
959 | min_score < ent_score(sb, e)) | |
33494784 JH |
960 | num_ents++; |
961 | if (num_ents) { | |
962 | blame_list = xcalloc(num_ents, sizeof(struct blame_list)); | |
963 | for (e = sb->ent, i = 0; e; e = e->next) | |
1617fe98 AG |
964 | if (!e->scanned && !e->guilty && |
965 | same_suspect(e->suspect, target) && | |
966 | min_score < ent_score(sb, e)) | |
33494784 JH |
967 | blame_list[i++].ent = e; |
968 | } | |
969 | *num_ents_p = num_ents; | |
970 | return blame_list; | |
971 | } | |
972 | ||
eb46328e AG |
973 | /* |
974 | * Reset the scanned status on all entries. | |
975 | */ | |
976 | static void reset_scanned_flag(struct scoreboard *sb) | |
977 | { | |
978 | struct blame_entry *e; | |
979 | for (e = sb->ent; e; e = e->next) | |
980 | e->scanned = 0; | |
981 | } | |
982 | ||
1732a1fd JH |
983 | /* |
984 | * For lines target is suspected for, see if we can find code movement | |
985 | * across file boundary from the parent commit. porigin is the path | |
986 | * in the parent we already tried. | |
987 | */ | |
18abd745 JH |
988 | static int find_copy_in_parent(struct scoreboard *sb, |
989 | struct origin *target, | |
990 | struct commit *parent, | |
991 | struct origin *porigin, | |
992 | int opt) | |
993 | { | |
994 | struct diff_options diff_opts; | |
995 | const char *paths[1]; | |
aec8fa1f | 996 | int i, j; |
33494784 JH |
997 | int retval; |
998 | struct blame_list *blame_list; | |
aec8fa1f | 999 | int num_ents; |
18abd745 | 1000 | |
1617fe98 | 1001 | blame_list = setup_blame_list(sb, target, blame_copy_score, &num_ents); |
33494784 | 1002 | if (!blame_list) |
18abd745 JH |
1003 | return 1; /* nothing remains for this target */ |
1004 | ||
1005 | diff_setup(&diff_opts); | |
8f67f8ae | 1006 | DIFF_OPT_SET(&diff_opts, RECURSIVE); |
18abd745 JH |
1007 | diff_opts.output_format = DIFF_FORMAT_NO_OUTPUT; |
1008 | ||
18abd745 JH |
1009 | paths[0] = NULL; |
1010 | diff_tree_setup_paths(paths, &diff_opts); | |
1011 | if (diff_setup_done(&diff_opts) < 0) | |
1012 | die("diff-setup"); | |
33494784 JH |
1013 | |
1014 | /* Try "find copies harder" on new path if requested; | |
1015 | * we do not want to use diffcore_rename() actually to | |
1016 | * match things up; find_copies_harder is set only to | |
1017 | * force diff_tree_sha1() to feed all filepairs to diff_queue, | |
1018 | * and this code needs to be after diff_setup_done(), which | |
1019 | * usually makes find-copies-harder imply copy detection. | |
1020 | */ | |
c63777c0 JH |
1021 | if ((opt & PICKAXE_BLAME_COPY_HARDEST) |
1022 | || ((opt & PICKAXE_BLAME_COPY_HARDER) | |
1023 | && (!porigin || strcmp(target->path, porigin->path)))) | |
8f67f8ae | 1024 | DIFF_OPT_SET(&diff_opts, FIND_COPIES_HARDER); |
33494784 | 1025 | |
1cfe7733 JH |
1026 | if (is_null_sha1(target->commit->object.sha1)) |
1027 | do_diff_cache(parent->tree->object.sha1, &diff_opts); | |
1028 | else | |
1029 | diff_tree_sha1(parent->tree->object.sha1, | |
1030 | target->commit->tree->object.sha1, | |
1031 | "", &diff_opts); | |
18abd745 | 1032 | |
8f67f8ae | 1033 | if (!DIFF_OPT_TST(&diff_opts, FIND_COPIES_HARDER)) |
33494784 | 1034 | diffcore_std(&diff_opts); |
18abd745 | 1035 | |
33494784 JH |
1036 | retval = 0; |
1037 | while (1) { | |
1038 | int made_progress = 0; | |
1039 | ||
1040 | for (i = 0; i < diff_queued_diff.nr; i++) { | |
1041 | struct diff_filepair *p = diff_queued_diff.queue[i]; | |
1042 | struct origin *norigin; | |
1043 | mmfile_t file_p; | |
33494784 JH |
1044 | struct blame_entry this[3]; |
1045 | ||
1046 | if (!DIFF_FILE_VALID(p->one)) | |
1047 | continue; /* does not exist in parent */ | |
5209ac4d AG |
1048 | if (S_ISGITLINK(p->one->mode)) |
1049 | continue; /* ignore git links */ | |
33494784 JH |
1050 | if (porigin && !strcmp(p->one->path, porigin->path)) |
1051 | /* find_move already dealt with this path */ | |
1052 | continue; | |
1053 | ||
1054 | norigin = get_origin(sb, parent, p->one->path); | |
1055 | hashcpy(norigin->blob_sha1, p->one->sha1); | |
c2e525d9 | 1056 | fill_origin_blob(norigin, &file_p); |
33494784 JH |
1057 | if (!file_p.ptr) |
1058 | continue; | |
1059 | ||
1060 | for (j = 0; j < num_ents; j++) { | |
1061 | find_copy_in_blob(sb, blame_list[j].ent, | |
1062 | norigin, this, &file_p); | |
1063 | copy_split_if_better(sb, blame_list[j].split, | |
1064 | this); | |
1065 | decref_split(this); | |
1066 | } | |
33494784 JH |
1067 | origin_decref(norigin); |
1068 | } | |
18abd745 | 1069 | |
aec8fa1f | 1070 | for (j = 0; j < num_ents; j++) { |
33494784 JH |
1071 | struct blame_entry *split = blame_list[j].split; |
1072 | if (split[1].suspect && | |
1073 | blame_copy_score < ent_score(sb, &split[1])) { | |
1074 | split_blame(sb, split, blame_list[j].ent); | |
1075 | made_progress = 1; | |
1076 | } | |
eb46328e AG |
1077 | else |
1078 | blame_list[j].ent->scanned = 1; | |
33494784 | 1079 | decref_split(split); |
18abd745 | 1080 | } |
33494784 | 1081 | free(blame_list); |
aec8fa1f | 1082 | |
33494784 JH |
1083 | if (!made_progress) |
1084 | break; | |
1617fe98 | 1085 | blame_list = setup_blame_list(sb, target, blame_copy_score, &num_ents); |
33494784 JH |
1086 | if (!blame_list) { |
1087 | retval = 1; | |
1088 | break; | |
1089 | } | |
18abd745 | 1090 | } |
eb46328e | 1091 | reset_scanned_flag(sb); |
33494784 | 1092 | diff_flush(&diff_opts); |
03b69c76 | 1093 | diff_tree_release_paths(&diff_opts); |
33494784 | 1094 | return retval; |
18abd745 JH |
1095 | } |
1096 | ||
1732a1fd JH |
1097 | /* |
1098 | * The blobs of origin and porigin exactly match, so everything | |
c2e525d9 JH |
1099 | * origin is suspected for can be blamed on the parent. |
1100 | */ | |
1101 | static void pass_whole_blame(struct scoreboard *sb, | |
1102 | struct origin *origin, struct origin *porigin) | |
1103 | { | |
1104 | struct blame_entry *e; | |
1105 | ||
1106 | if (!porigin->file.ptr && origin->file.ptr) { | |
1107 | /* Steal its file */ | |
1108 | porigin->file = origin->file; | |
1109 | origin->file.ptr = NULL; | |
1110 | } | |
1111 | for (e = sb->ent; e; e = e->next) { | |
171dccd5 | 1112 | if (!same_suspect(e->suspect, origin)) |
c2e525d9 JH |
1113 | continue; |
1114 | origin_incref(porigin); | |
1115 | origin_decref(e->suspect); | |
1116 | e->suspect = porigin; | |
1117 | } | |
1118 | } | |
1119 | ||
69264f46 JH |
1120 | /* |
1121 | * We pass blame from the current commit to its parents. We keep saying | |
1122 | * "parent" (and "porigin"), but what we mean is to find scapegoat to | |
1123 | * exonerate ourselves. | |
1124 | */ | |
85af7929 | 1125 | static struct commit_list *first_scapegoat(struct rev_info *revs, struct commit *commit) |
69264f46 | 1126 | { |
85af7929 JH |
1127 | if (!reverse) |
1128 | return commit->parents; | |
1129 | return lookup_decoration(&revs->children, &commit->object); | |
69264f46 JH |
1130 | } |
1131 | ||
85af7929 | 1132 | static int num_scapegoats(struct rev_info *revs, struct commit *commit) |
69264f46 JH |
1133 | { |
1134 | int cnt; | |
85af7929 | 1135 | struct commit_list *l = first_scapegoat(revs, commit); |
69264f46 JH |
1136 | for (cnt = 0; l; l = l->next) |
1137 | cnt++; | |
1138 | return cnt; | |
1139 | } | |
1140 | ||
1141 | #define MAXSG 16 | |
cee7f245 | 1142 | |
d24bba80 | 1143 | static void pass_blame(struct scoreboard *sb, struct origin *origin, int opt) |
cee7f245 | 1144 | { |
85af7929 | 1145 | struct rev_info *revs = sb->revs; |
69264f46 | 1146 | int i, pass, num_sg; |
cee7f245 | 1147 | struct commit *commit = origin->commit; |
69264f46 JH |
1148 | struct commit_list *sg; |
1149 | struct origin *sg_buf[MAXSG]; | |
1150 | struct origin *porigin, **sg_origin = sg_buf; | |
1151 | ||
85af7929 | 1152 | num_sg = num_scapegoats(revs, commit); |
69264f46 JH |
1153 | if (!num_sg) |
1154 | goto finish; | |
1155 | else if (num_sg < ARRAY_SIZE(sg_buf)) | |
1156 | memset(sg_buf, 0, sizeof(sg_buf)); | |
1157 | else | |
1158 | sg_origin = xcalloc(num_sg, sizeof(*sg_origin)); | |
cee7f245 | 1159 | |
69264f46 JH |
1160 | /* |
1161 | * The first pass looks for unrenamed path to optimize for | |
f69e743d JH |
1162 | * common cases, then we look for renames in the second pass. |
1163 | */ | |
1164 | for (pass = 0; pass < 2; pass++) { | |
1165 | struct origin *(*find)(struct scoreboard *, | |
1166 | struct commit *, struct origin *); | |
1167 | find = pass ? find_rename : find_origin; | |
1168 | ||
85af7929 | 1169 | for (i = 0, sg = first_scapegoat(revs, commit); |
69264f46 JH |
1170 | i < num_sg && sg; |
1171 | sg = sg->next, i++) { | |
1172 | struct commit *p = sg->item; | |
0421d9f8 | 1173 | int j, same; |
f69e743d | 1174 | |
69264f46 | 1175 | if (sg_origin[i]) |
f69e743d JH |
1176 | continue; |
1177 | if (parse_commit(p)) | |
1178 | continue; | |
0d981c67 | 1179 | porigin = find(sb, p, origin); |
f69e743d JH |
1180 | if (!porigin) |
1181 | continue; | |
1182 | if (!hashcmp(porigin->blob_sha1, origin->blob_sha1)) { | |
c2e525d9 | 1183 | pass_whole_blame(sb, origin, porigin); |
f69e743d JH |
1184 | origin_decref(porigin); |
1185 | goto finish; | |
1186 | } | |
0421d9f8 | 1187 | for (j = same = 0; j < i; j++) |
69264f46 JH |
1188 | if (sg_origin[j] && |
1189 | !hashcmp(sg_origin[j]->blob_sha1, | |
0421d9f8 JH |
1190 | porigin->blob_sha1)) { |
1191 | same = 1; | |
1192 | break; | |
1193 | } | |
1194 | if (!same) | |
69264f46 | 1195 | sg_origin[i] = porigin; |
0421d9f8 JH |
1196 | else |
1197 | origin_decref(porigin); | |
cee7f245 | 1198 | } |
cee7f245 JH |
1199 | } |
1200 | ||
c2e525d9 | 1201 | num_commits++; |
85af7929 | 1202 | for (i = 0, sg = first_scapegoat(revs, commit); |
69264f46 JH |
1203 | i < num_sg && sg; |
1204 | sg = sg->next, i++) { | |
1205 | struct origin *porigin = sg_origin[i]; | |
cee7f245 JH |
1206 | if (!porigin) |
1207 | continue; | |
96e11709 JH |
1208 | if (!origin->previous) { |
1209 | origin_incref(porigin); | |
1210 | origin->previous = porigin; | |
1211 | } | |
cee7f245 | 1212 | if (pass_blame_to_parent(sb, origin, porigin)) |
54a4c617 | 1213 | goto finish; |
cee7f245 | 1214 | } |
d24bba80 JH |
1215 | |
1216 | /* | |
1732a1fd | 1217 | * Optionally find moves in parents' files. |
d24bba80 JH |
1218 | */ |
1219 | if (opt & PICKAXE_BLAME_MOVE) | |
85af7929 | 1220 | for (i = 0, sg = first_scapegoat(revs, commit); |
69264f46 JH |
1221 | i < num_sg && sg; |
1222 | sg = sg->next, i++) { | |
1223 | struct origin *porigin = sg_origin[i]; | |
d24bba80 JH |
1224 | if (!porigin) |
1225 | continue; | |
1226 | if (find_move_in_parent(sb, origin, porigin)) | |
54a4c617 | 1227 | goto finish; |
d24bba80 JH |
1228 | } |
1229 | ||
18abd745 | 1230 | /* |
1732a1fd | 1231 | * Optionally find copies from parents' files. |
18abd745 JH |
1232 | */ |
1233 | if (opt & PICKAXE_BLAME_COPY) | |
85af7929 | 1234 | for (i = 0, sg = first_scapegoat(revs, commit); |
69264f46 JH |
1235 | i < num_sg && sg; |
1236 | sg = sg->next, i++) { | |
1237 | struct origin *porigin = sg_origin[i]; | |
1238 | if (find_copy_in_parent(sb, origin, sg->item, | |
18abd745 | 1239 | porigin, opt)) |
54a4c617 | 1240 | goto finish; |
18abd745 | 1241 | } |
54a4c617 JH |
1242 | |
1243 | finish: | |
69264f46 JH |
1244 | for (i = 0; i < num_sg; i++) { |
1245 | if (sg_origin[i]) { | |
1246 | drop_origin_blob(sg_origin[i]); | |
1247 | origin_decref(sg_origin[i]); | |
7c3c7962 JH |
1248 | } |
1249 | } | |
1250 | drop_origin_blob(origin); | |
69264f46 JH |
1251 | if (sg_buf != sg_origin) |
1252 | free(sg_origin); | |
cee7f245 JH |
1253 | } |
1254 | ||
1732a1fd JH |
1255 | /* |
1256 | * Information on commits, used for output. | |
1257 | */ | |
cee7f245 JH |
1258 | struct commit_info |
1259 | { | |
3a55602e SP |
1260 | const char *author; |
1261 | const char *author_mail; | |
cee7f245 | 1262 | unsigned long author_time; |
3a55602e | 1263 | const char *author_tz; |
cee7f245 JH |
1264 | |
1265 | /* filled only when asked for details */ | |
3a55602e SP |
1266 | const char *committer; |
1267 | const char *committer_mail; | |
cee7f245 | 1268 | unsigned long committer_time; |
3a55602e | 1269 | const char *committer_tz; |
cee7f245 | 1270 | |
3a55602e | 1271 | const char *summary; |
cee7f245 JH |
1272 | }; |
1273 | ||
1732a1fd JH |
1274 | /* |
1275 | * Parse author/committer line in the commit object buffer | |
1276 | */ | |
cee7f245 | 1277 | static void get_ac_line(const char *inbuf, const char *what, |
d20d654f MSO |
1278 | int person_len, char *person, |
1279 | int mail_len, char *mail, | |
3a55602e | 1280 | unsigned long *time, const char **tz) |
cee7f245 | 1281 | { |
f9567384 | 1282 | int len, tzlen, maillen; |
d20d654f | 1283 | char *tmp, *endp, *timepos, *mailpos; |
cee7f245 JH |
1284 | |
1285 | tmp = strstr(inbuf, what); | |
1286 | if (!tmp) | |
1287 | goto error_out; | |
1288 | tmp += strlen(what); | |
1289 | endp = strchr(tmp, '\n'); | |
1290 | if (!endp) | |
1291 | len = strlen(tmp); | |
1292 | else | |
1293 | len = endp - tmp; | |
d20d654f | 1294 | if (person_len <= len) { |
cee7f245 JH |
1295 | error_out: |
1296 | /* Ugh */ | |
d20d654f MSO |
1297 | *tz = "(unknown)"; |
1298 | strcpy(mail, *tz); | |
cee7f245 JH |
1299 | *time = 0; |
1300 | return; | |
1301 | } | |
1302 | memcpy(person, tmp, len); | |
1303 | ||
1304 | tmp = person; | |
1305 | tmp += len; | |
1306 | *tmp = 0; | |
1307 | while (*tmp != ' ') | |
1308 | tmp--; | |
1309 | *tz = tmp+1; | |
f9567384 | 1310 | tzlen = (person+len)-(tmp+1); |
cee7f245 JH |
1311 | |
1312 | *tmp = 0; | |
1313 | while (*tmp != ' ') | |
1314 | tmp--; | |
1315 | *time = strtoul(tmp, NULL, 10); | |
f9567384 | 1316 | timepos = tmp; |
cee7f245 JH |
1317 | |
1318 | *tmp = 0; | |
1319 | while (*tmp != ' ') | |
1320 | tmp--; | |
d20d654f | 1321 | mailpos = tmp + 1; |
cee7f245 | 1322 | *tmp = 0; |
f9567384 | 1323 | maillen = timepos - tmp; |
d20d654f | 1324 | memcpy(mail, mailpos, maillen); |
f9567384 JH |
1325 | |
1326 | if (!mailmap.nr) | |
1327 | return; | |
1328 | ||
1329 | /* | |
1330 | * mailmap expansion may make the name longer. | |
1331 | * make room by pushing stuff down. | |
1332 | */ | |
d20d654f | 1333 | tmp = person + person_len - (tzlen + 1); |
f9567384 JH |
1334 | memmove(tmp, *tz, tzlen); |
1335 | tmp[tzlen] = 0; | |
1336 | *tz = tmp; | |
1337 | ||
f9567384 | 1338 | /* |
d20d654f | 1339 | * Now, convert both name and e-mail using mailmap |
f9567384 | 1340 | */ |
d20d654f MSO |
1341 | if(map_user(&mailmap, mail+1, mail_len-1, person, tmp-person-1)) { |
1342 | /* Add a trailing '>' to email, since map_user returns plain emails | |
1343 | Note: It already has '<', since we replace from mail+1 */ | |
1344 | mailpos = memchr(mail, '\0', mail_len); | |
1345 | if (mailpos && mailpos-mail < mail_len - 1) { | |
1346 | *mailpos = '>'; | |
1347 | *(mailpos+1) = '\0'; | |
1348 | } | |
1349 | } | |
cee7f245 JH |
1350 | } |
1351 | ||
1352 | static void get_commit_info(struct commit *commit, | |
1353 | struct commit_info *ret, | |
1354 | int detailed) | |
1355 | { | |
1356 | int len; | |
69cd8f63 | 1357 | char *tmp, *endp, *reencoded, *message; |
d20d654f MSO |
1358 | static char author_name[1024]; |
1359 | static char author_mail[1024]; | |
1360 | static char committer_name[1024]; | |
1361 | static char committer_mail[1024]; | |
cee7f245 JH |
1362 | static char summary_buf[1024]; |
1363 | ||
1732a1fd JH |
1364 | /* |
1365 | * We've operated without save_commit_buffer, so | |
612702e8 JH |
1366 | * we now need to populate them for output. |
1367 | */ | |
1368 | if (!commit->buffer) { | |
21666f1a | 1369 | enum object_type type; |
612702e8 JH |
1370 | unsigned long size; |
1371 | commit->buffer = | |
21666f1a | 1372 | read_sha1_file(commit->object.sha1, &type, &size); |
ab43e495 JH |
1373 | if (!commit->buffer) |
1374 | die("Cannot read commit %s", | |
1375 | sha1_to_hex(commit->object.sha1)); | |
612702e8 | 1376 | } |
69cd8f63 AG |
1377 | reencoded = reencode_commit_message(commit, NULL); |
1378 | message = reencoded ? reencoded : commit->buffer; | |
d20d654f MSO |
1379 | ret->author = author_name; |
1380 | ret->author_mail = author_mail; | |
69cd8f63 | 1381 | get_ac_line(message, "\nauthor ", |
d20d654f MSO |
1382 | sizeof(author_name), author_name, |
1383 | sizeof(author_mail), author_mail, | |
cee7f245 JH |
1384 | &ret->author_time, &ret->author_tz); |
1385 | ||
69cd8f63 AG |
1386 | if (!detailed) { |
1387 | free(reencoded); | |
cee7f245 | 1388 | return; |
69cd8f63 | 1389 | } |
cee7f245 | 1390 | |
d20d654f MSO |
1391 | ret->committer = committer_name; |
1392 | ret->committer_mail = committer_mail; | |
69cd8f63 | 1393 | get_ac_line(message, "\ncommitter ", |
d20d654f MSO |
1394 | sizeof(committer_name), committer_name, |
1395 | sizeof(committer_mail), committer_mail, | |
cee7f245 JH |
1396 | &ret->committer_time, &ret->committer_tz); |
1397 | ||
1398 | ret->summary = summary_buf; | |
69cd8f63 | 1399 | tmp = strstr(message, "\n\n"); |
cee7f245 JH |
1400 | if (!tmp) { |
1401 | error_out: | |
1402 | sprintf(summary_buf, "(%s)", sha1_to_hex(commit->object.sha1)); | |
69cd8f63 | 1403 | free(reencoded); |
cee7f245 JH |
1404 | return; |
1405 | } | |
1406 | tmp += 2; | |
1407 | endp = strchr(tmp, '\n'); | |
1408 | if (!endp) | |
1cfe7733 | 1409 | endp = tmp + strlen(tmp); |
cee7f245 | 1410 | len = endp - tmp; |
1cfe7733 | 1411 | if (len >= sizeof(summary_buf) || len == 0) |
cee7f245 JH |
1412 | goto error_out; |
1413 | memcpy(summary_buf, tmp, len); | |
1414 | summary_buf[len] = 0; | |
69cd8f63 | 1415 | free(reencoded); |
cee7f245 JH |
1416 | } |
1417 | ||
1732a1fd JH |
1418 | /* |
1419 | * To allow LF and other nonportable characters in pathnames, | |
1420 | * they are c-style quoted as needed. | |
1421 | */ | |
46e5e69d JH |
1422 | static void write_filename_info(const char *path) |
1423 | { | |
1424 | printf("filename "); | |
663af342 | 1425 | write_name_quoted(path, stdout, '\n'); |
46e5e69d JH |
1426 | } |
1427 | ||
9991030c JH |
1428 | /* |
1429 | * Porcelain/Incremental format wants to show a lot of details per | |
1430 | * commit. Instead of repeating this every line, emit it only once, | |
1431 | * the first time each commit appears in the output. | |
1432 | */ | |
1433 | static int emit_one_suspect_detail(struct origin *suspect) | |
1434 | { | |
1435 | struct commit_info ci; | |
1436 | ||
1437 | if (suspect->commit->object.flags & METAINFO_SHOWN) | |
1438 | return 0; | |
1439 | ||
1440 | suspect->commit->object.flags |= METAINFO_SHOWN; | |
1441 | get_commit_info(suspect->commit, &ci, 1); | |
1442 | printf("author %s\n", ci.author); | |
1443 | printf("author-mail %s\n", ci.author_mail); | |
1444 | printf("author-time %lu\n", ci.author_time); | |
1445 | printf("author-tz %s\n", ci.author_tz); | |
1446 | printf("committer %s\n", ci.committer); | |
1447 | printf("committer-mail %s\n", ci.committer_mail); | |
1448 | printf("committer-time %lu\n", ci.committer_time); | |
1449 | printf("committer-tz %s\n", ci.committer_tz); | |
1450 | printf("summary %s\n", ci.summary); | |
1451 | if (suspect->commit->object.flags & UNINTERESTING) | |
1452 | printf("boundary\n"); | |
96e11709 JH |
1453 | if (suspect->previous) { |
1454 | struct origin *prev = suspect->previous; | |
1455 | printf("previous %s ", sha1_to_hex(prev->commit->object.sha1)); | |
1456 | write_name_quoted(prev->path, stdout, '\n'); | |
1457 | } | |
9991030c JH |
1458 | return 1; |
1459 | } | |
1460 | ||
1732a1fd JH |
1461 | /* |
1462 | * The blame_entry is found to be guilty for the range. Mark it | |
1463 | * as such, and show it in incremental output. | |
1464 | */ | |
717d1462 LT |
1465 | static void found_guilty_entry(struct blame_entry *ent) |
1466 | { | |
1467 | if (ent->guilty) | |
1468 | return; | |
1469 | ent->guilty = 1; | |
1470 | if (incremental) { | |
1471 | struct origin *suspect = ent->suspect; | |
1472 | ||
1473 | printf("%s %d %d %d\n", | |
1474 | sha1_to_hex(suspect->commit->object.sha1), | |
1475 | ent->s_lno + 1, ent->lno + 1, ent->num_lines); | |
9991030c | 1476 | emit_one_suspect_detail(suspect); |
46e5e69d | 1477 | write_filename_info(suspect->path); |
06f59e9f | 1478 | maybe_flush_or_die(stdout, "stdout"); |
717d1462 LT |
1479 | } |
1480 | } | |
1481 | ||
1732a1fd JH |
1482 | /* |
1483 | * The main loop -- while the scoreboard has lines whose true origin | |
3dff5379 | 1484 | * is still unknown, pick one blame_entry, and allow its current |
1732a1fd JH |
1485 | * suspect to pass blames to its parents. |
1486 | */ | |
85af7929 | 1487 | static void assign_blame(struct scoreboard *sb, int opt) |
717d1462 | 1488 | { |
85af7929 JH |
1489 | struct rev_info *revs = sb->revs; |
1490 | ||
717d1462 LT |
1491 | while (1) { |
1492 | struct blame_entry *ent; | |
1493 | struct commit *commit; | |
1494 | struct origin *suspect = NULL; | |
1495 | ||
1496 | /* find one suspect to break down */ | |
1497 | for (ent = sb->ent; !suspect && ent; ent = ent->next) | |
1498 | if (!ent->guilty) | |
1499 | suspect = ent->suspect; | |
1500 | if (!suspect) | |
1501 | return; /* all done */ | |
1502 | ||
1732a1fd JH |
1503 | /* |
1504 | * We will use this suspect later in the loop, | |
1505 | * so hold onto it in the meantime. | |
1506 | */ | |
717d1462 LT |
1507 | origin_incref(suspect); |
1508 | commit = suspect->commit; | |
1509 | if (!commit->object.parsed) | |
1510 | parse_commit(commit); | |
85af7929 JH |
1511 | if (reverse || |
1512 | (!(commit->object.flags & UNINTERESTING) && | |
1513 | !(revs->max_age != -1 && commit->date < revs->max_age))) | |
717d1462 LT |
1514 | pass_blame(sb, suspect, opt); |
1515 | else { | |
1516 | commit->object.flags |= UNINTERESTING; | |
1517 | if (commit->object.parsed) | |
1518 | mark_parents_uninteresting(commit); | |
1519 | } | |
1520 | /* treat root commit as boundary */ | |
1521 | if (!commit->parents && !show_root) | |
1522 | commit->object.flags |= UNINTERESTING; | |
1523 | ||
1524 | /* Take responsibility for the remaining entries */ | |
1525 | for (ent = sb->ent; ent; ent = ent->next) | |
171dccd5 | 1526 | if (same_suspect(ent->suspect, suspect)) |
717d1462 LT |
1527 | found_guilty_entry(ent); |
1528 | origin_decref(suspect); | |
1529 | ||
1530 | if (DEBUG) /* sanity */ | |
1531 | sanity_check_refcnt(sb); | |
1532 | } | |
1533 | } | |
1534 | ||
1535 | static const char *format_time(unsigned long time, const char *tz_str, | |
1536 | int show_raw_time) | |
1537 | { | |
1538 | static char time_buf[128]; | |
31653c1a EL |
1539 | const char *time_str; |
1540 | int time_len; | |
1541 | int tz; | |
717d1462 LT |
1542 | |
1543 | if (show_raw_time) { | |
1544 | sprintf(time_buf, "%lu %s", time, tz_str); | |
717d1462 | 1545 | } |
31653c1a EL |
1546 | else { |
1547 | tz = atoi(tz_str); | |
1548 | time_str = show_date(time, tz, blame_date_mode); | |
1549 | time_len = strlen(time_str); | |
1550 | memcpy(time_buf, time_str, time_len); | |
1551 | memset(time_buf + time_len, ' ', blame_date_width - time_len); | |
1552 | } | |
717d1462 LT |
1553 | return time_buf; |
1554 | } | |
1555 | ||
cee7f245 JH |
1556 | #define OUTPUT_ANNOTATE_COMPAT 001 |
1557 | #define OUTPUT_LONG_OBJECT_NAME 002 | |
1558 | #define OUTPUT_RAW_TIMESTAMP 004 | |
1559 | #define OUTPUT_PORCELAIN 010 | |
1560 | #define OUTPUT_SHOW_NAME 020 | |
1561 | #define OUTPUT_SHOW_NUMBER 040 | |
5ff62c30 | 1562 | #define OUTPUT_SHOW_SCORE 0100 |
093dc5be | 1563 | #define OUTPUT_NO_AUTHOR 0200 |
cee7f245 JH |
1564 | |
1565 | static void emit_porcelain(struct scoreboard *sb, struct blame_entry *ent) | |
1566 | { | |
1567 | int cnt; | |
1568 | const char *cp; | |
1569 | struct origin *suspect = ent->suspect; | |
1570 | char hex[41]; | |
1571 | ||
1572 | strcpy(hex, sha1_to_hex(suspect->commit->object.sha1)); | |
1573 | printf("%s%c%d %d %d\n", | |
1574 | hex, | |
1575 | ent->guilty ? ' ' : '*', // purely for debugging | |
1576 | ent->s_lno + 1, | |
1577 | ent->lno + 1, | |
1578 | ent->num_lines); | |
9991030c JH |
1579 | if (emit_one_suspect_detail(suspect) || |
1580 | (suspect->commit->object.flags & MORE_THAN_ONE_PATH)) | |
46e5e69d | 1581 | write_filename_info(suspect->path); |
cee7f245 JH |
1582 | |
1583 | cp = nth_line(sb, ent->lno); | |
1584 | for (cnt = 0; cnt < ent->num_lines; cnt++) { | |
1585 | char ch; | |
1586 | if (cnt) | |
1587 | printf("%s %d %d\n", hex, | |
1588 | ent->s_lno + 1 + cnt, | |
1589 | ent->lno + 1 + cnt); | |
1590 | putchar('\t'); | |
1591 | do { | |
1592 | ch = *cp++; | |
1593 | putchar(ch); | |
1594 | } while (ch != '\n' && | |
1595 | cp < sb->final_buf + sb->final_buf_size); | |
1596 | } | |
1597 | } | |
1598 | ||
1599 | static void emit_other(struct scoreboard *sb, struct blame_entry *ent, int opt) | |
1600 | { | |
1601 | int cnt; | |
1602 | const char *cp; | |
1603 | struct origin *suspect = ent->suspect; | |
1604 | struct commit_info ci; | |
1605 | char hex[41]; | |
1606 | int show_raw_time = !!(opt & OUTPUT_RAW_TIMESTAMP); | |
1607 | ||
1608 | get_commit_info(suspect->commit, &ci, 1); | |
1609 | strcpy(hex, sha1_to_hex(suspect->commit->object.sha1)); | |
1610 | ||
1611 | cp = nth_line(sb, ent->lno); | |
1612 | for (cnt = 0; cnt < ent->num_lines; cnt++) { | |
1613 | char ch; | |
b11121d9 JH |
1614 | int length = (opt & OUTPUT_LONG_OBJECT_NAME) ? 40 : 8; |
1615 | ||
1616 | if (suspect->commit->object.flags & UNINTERESTING) { | |
e68989a7 JH |
1617 | if (blank_boundary) |
1618 | memset(hex, ' ', length); | |
7ceacdff | 1619 | else if (!(opt & OUTPUT_ANNOTATE_COMPAT)) { |
4c10a5ca JH |
1620 | length--; |
1621 | putchar('^'); | |
1622 | } | |
b11121d9 | 1623 | } |
cee7f245 | 1624 | |
b11121d9 | 1625 | printf("%.*s", length, hex); |
cee7f245 JH |
1626 | if (opt & OUTPUT_ANNOTATE_COMPAT) |
1627 | printf("\t(%10s\t%10s\t%d)", ci.author, | |
1628 | format_time(ci.author_time, ci.author_tz, | |
1629 | show_raw_time), | |
1630 | ent->lno + 1 + cnt); | |
1631 | else { | |
5ff62c30 | 1632 | if (opt & OUTPUT_SHOW_SCORE) |
54a4c617 JH |
1633 | printf(" %*d %02d", |
1634 | max_score_digits, ent->score, | |
1635 | ent->suspect->refcnt); | |
cee7f245 JH |
1636 | if (opt & OUTPUT_SHOW_NAME) |
1637 | printf(" %-*.*s", longest_file, longest_file, | |
1638 | suspect->path); | |
1639 | if (opt & OUTPUT_SHOW_NUMBER) | |
1640 | printf(" %*d", max_orig_digits, | |
1641 | ent->s_lno + 1 + cnt); | |
093dc5be | 1642 | |
ffaf9cc0 GT |
1643 | if (!(opt & OUTPUT_NO_AUTHOR)) { |
1644 | int pad = longest_author - utf8_strwidth(ci.author); | |
1645 | printf(" (%s%*s %10s", | |
1646 | ci.author, pad, "", | |
093dc5be JH |
1647 | format_time(ci.author_time, |
1648 | ci.author_tz, | |
1649 | show_raw_time)); | |
ffaf9cc0 | 1650 | } |
093dc5be | 1651 | printf(" %*d) ", |
cee7f245 JH |
1652 | max_digits, ent->lno + 1 + cnt); |
1653 | } | |
1654 | do { | |
1655 | ch = *cp++; | |
1656 | putchar(ch); | |
1657 | } while (ch != '\n' && | |
1658 | cp < sb->final_buf + sb->final_buf_size); | |
1659 | } | |
1660 | } | |
1661 | ||
1662 | static void output(struct scoreboard *sb, int option) | |
1663 | { | |
1664 | struct blame_entry *ent; | |
1665 | ||
1666 | if (option & OUTPUT_PORCELAIN) { | |
1667 | for (ent = sb->ent; ent; ent = ent->next) { | |
1668 | struct blame_entry *oth; | |
1669 | struct origin *suspect = ent->suspect; | |
1670 | struct commit *commit = suspect->commit; | |
1671 | if (commit->object.flags & MORE_THAN_ONE_PATH) | |
1672 | continue; | |
1673 | for (oth = ent->next; oth; oth = oth->next) { | |
1674 | if ((oth->suspect->commit != commit) || | |
1675 | !strcmp(oth->suspect->path, suspect->path)) | |
1676 | continue; | |
1677 | commit->object.flags |= MORE_THAN_ONE_PATH; | |
1678 | break; | |
1679 | } | |
1680 | } | |
1681 | } | |
1682 | ||
1683 | for (ent = sb->ent; ent; ent = ent->next) { | |
1684 | if (option & OUTPUT_PORCELAIN) | |
1685 | emit_porcelain(sb, ent); | |
5ff62c30 | 1686 | else { |
cee7f245 | 1687 | emit_other(sb, ent, option); |
5ff62c30 | 1688 | } |
cee7f245 JH |
1689 | } |
1690 | } | |
1691 | ||
1732a1fd JH |
1692 | /* |
1693 | * To allow quick access to the contents of nth line in the | |
1694 | * final image, prepare an index in the scoreboard. | |
1695 | */ | |
cee7f245 JH |
1696 | static int prepare_lines(struct scoreboard *sb) |
1697 | { | |
1698 | const char *buf = sb->final_buf; | |
1699 | unsigned long len = sb->final_buf_size; | |
1700 | int num = 0, incomplete = 0, bol = 1; | |
1701 | ||
1702 | if (len && buf[len-1] != '\n') | |
1703 | incomplete++; /* incomplete line at the end */ | |
1704 | while (len--) { | |
1705 | if (bol) { | |
1706 | sb->lineno = xrealloc(sb->lineno, | |
4b25d091 | 1707 | sizeof(int *) * (num + 1)); |
cee7f245 JH |
1708 | sb->lineno[num] = buf - sb->final_buf; |
1709 | bol = 0; | |
1710 | } | |
1711 | if (*buf++ == '\n') { | |
1712 | num++; | |
1713 | bol = 1; | |
1714 | } | |
1715 | } | |
1ca6ca87 | 1716 | sb->lineno = xrealloc(sb->lineno, |
4b25d091 | 1717 | sizeof(int *) * (num + incomplete + 1)); |
1ca6ca87 | 1718 | sb->lineno[num + incomplete] = buf - sb->final_buf; |
cee7f245 JH |
1719 | sb->num_lines = num + incomplete; |
1720 | return sb->num_lines; | |
1721 | } | |
1722 | ||
1732a1fd JH |
1723 | /* |
1724 | * Add phony grafts for use with -S; this is primarily to | |
34baebce | 1725 | * support git's cvsserver that wants to give a linear history |
1732a1fd JH |
1726 | * to its clients. |
1727 | */ | |
cee7f245 JH |
1728 | static int read_ancestry(const char *graft_file) |
1729 | { | |
1730 | FILE *fp = fopen(graft_file, "r"); | |
1731 | char buf[1024]; | |
1732 | if (!fp) | |
1733 | return -1; | |
1734 | while (fgets(buf, sizeof(buf), fp)) { | |
1735 | /* The format is just "Commit Parent1 Parent2 ...\n" */ | |
1736 | int len = strlen(buf); | |
1737 | struct commit_graft *graft = read_graft_line(buf, len); | |
8eaf7986 JH |
1738 | if (graft) |
1739 | register_commit_graft(graft, 0); | |
cee7f245 JH |
1740 | } |
1741 | fclose(fp); | |
1742 | return 0; | |
1743 | } | |
1744 | ||
1732a1fd JH |
1745 | /* |
1746 | * How many columns do we need to show line numbers in decimal? | |
1747 | */ | |
cee7f245 JH |
1748 | static int lineno_width(int lines) |
1749 | { | |
f2915045 | 1750 | int i, width; |
cee7f245 | 1751 | |
f2915045 JH |
1752 | for (width = 1, i = 10; i <= lines + 1; width++) |
1753 | i *= 10; | |
1754 | return width; | |
cee7f245 JH |
1755 | } |
1756 | ||
1732a1fd JH |
1757 | /* |
1758 | * How many columns do we need to show line numbers, authors, | |
1759 | * and filenames? | |
1760 | */ | |
cee7f245 JH |
1761 | static void find_alignment(struct scoreboard *sb, int *option) |
1762 | { | |
1763 | int longest_src_lines = 0; | |
1764 | int longest_dst_lines = 0; | |
5ff62c30 | 1765 | unsigned largest_score = 0; |
cee7f245 JH |
1766 | struct blame_entry *e; |
1767 | ||
1768 | for (e = sb->ent; e; e = e->next) { | |
1769 | struct origin *suspect = e->suspect; | |
1770 | struct commit_info ci; | |
1771 | int num; | |
1772 | ||
ab3bb800 JH |
1773 | if (strcmp(suspect->path, sb->path)) |
1774 | *option |= OUTPUT_SHOW_NAME; | |
1775 | num = strlen(suspect->path); | |
1776 | if (longest_file < num) | |
1777 | longest_file = num; | |
cee7f245 JH |
1778 | if (!(suspect->commit->object.flags & METAINFO_SHOWN)) { |
1779 | suspect->commit->object.flags |= METAINFO_SHOWN; | |
1780 | get_commit_info(suspect->commit, &ci, 1); | |
ffaf9cc0 | 1781 | num = utf8_strwidth(ci.author); |
cee7f245 JH |
1782 | if (longest_author < num) |
1783 | longest_author = num; | |
1784 | } | |
1785 | num = e->s_lno + e->num_lines; | |
1786 | if (longest_src_lines < num) | |
1787 | longest_src_lines = num; | |
1788 | num = e->lno + e->num_lines; | |
1789 | if (longest_dst_lines < num) | |
1790 | longest_dst_lines = num; | |
5ff62c30 JH |
1791 | if (largest_score < ent_score(sb, e)) |
1792 | largest_score = ent_score(sb, e); | |
cee7f245 JH |
1793 | } |
1794 | max_orig_digits = lineno_width(longest_src_lines); | |
1795 | max_digits = lineno_width(longest_dst_lines); | |
5ff62c30 | 1796 | max_score_digits = lineno_width(largest_score); |
cee7f245 JH |
1797 | } |
1798 | ||
1732a1fd JH |
1799 | /* |
1800 | * For debugging -- origin is refcounted, and this asserts that | |
1801 | * we do not underflow. | |
1802 | */ | |
54a4c617 JH |
1803 | static void sanity_check_refcnt(struct scoreboard *sb) |
1804 | { | |
1805 | int baa = 0; | |
1806 | struct blame_entry *ent; | |
1807 | ||
1808 | for (ent = sb->ent; ent; ent = ent->next) { | |
ae86ad65 | 1809 | /* Nobody should have zero or negative refcnt */ |
854b97f6 JH |
1810 | if (ent->suspect->refcnt <= 0) { |
1811 | fprintf(stderr, "%s in %s has negative refcnt %d\n", | |
1812 | ent->suspect->path, | |
1813 | sha1_to_hex(ent->suspect->commit->object.sha1), | |
1814 | ent->suspect->refcnt); | |
ae86ad65 | 1815 | baa = 1; |
854b97f6 | 1816 | } |
ae86ad65 | 1817 | } |
54a4c617 JH |
1818 | if (baa) { |
1819 | int opt = 0160; | |
1820 | find_alignment(sb, &opt); | |
1821 | output(sb, opt); | |
854b97f6 | 1822 | die("Baa %d!", baa); |
54a4c617 JH |
1823 | } |
1824 | } | |
1825 | ||
1732a1fd JH |
1826 | /* |
1827 | * Used for the command line parsing; check if the path exists | |
1828 | * in the working tree. | |
1829 | */ | |
c455c87c | 1830 | static int has_string_in_work_tree(const char *path) |
cee7f245 JH |
1831 | { |
1832 | struct stat st; | |
1833 | return !lstat(path, &st); | |
1834 | } | |
1835 | ||
4a0fc95f JH |
1836 | static unsigned parse_score(const char *arg) |
1837 | { | |
1838 | char *end; | |
1839 | unsigned long score = strtoul(arg, &end, 10); | |
1840 | if (*end) | |
1841 | return 0; | |
1842 | return score; | |
1843 | } | |
1844 | ||
20239bae JK |
1845 | static const char *add_prefix(const char *prefix, const char *path) |
1846 | { | |
097971f5 | 1847 | return prefix_path(prefix, prefix ? strlen(prefix) : 0, path); |
20239bae JK |
1848 | } |
1849 | ||
1732a1fd JH |
1850 | /* |
1851 | * Parsing of (comma separated) one item in the -L option | |
1852 | */ | |
931233bc JH |
1853 | static const char *parse_loc(const char *spec, |
1854 | struct scoreboard *sb, long lno, | |
1855 | long begin, long *ret) | |
1856 | { | |
1857 | char *term; | |
1858 | const char *line; | |
1859 | long num; | |
1860 | int reg_error; | |
1861 | regex_t regexp; | |
1862 | regmatch_t match[1]; | |
1863 | ||
7bd9641d JH |
1864 | /* Allow "-L <something>,+20" to mean starting at <something> |
1865 | * for 20 lines, or "-L <something>,-5" for 5 lines ending at | |
1866 | * <something>. | |
1867 | */ | |
1868 | if (1 < begin && (spec[0] == '+' || spec[0] == '-')) { | |
1869 | num = strtol(spec + 1, &term, 10); | |
1870 | if (term != spec + 1) { | |
1871 | if (spec[0] == '-') | |
1872 | num = 0 - num; | |
1873 | if (0 < num) | |
1874 | *ret = begin + num - 2; | |
1875 | else if (!num) | |
1876 | *ret = begin; | |
1877 | else | |
1878 | *ret = begin + num; | |
1879 | return term; | |
1880 | } | |
1881 | return spec; | |
1882 | } | |
931233bc JH |
1883 | num = strtol(spec, &term, 10); |
1884 | if (term != spec) { | |
1885 | *ret = num; | |
1886 | return term; | |
1887 | } | |
1888 | if (spec[0] != '/') | |
1889 | return spec; | |
1890 | ||
1891 | /* it could be a regexp of form /.../ */ | |
4b25d091 | 1892 | for (term = (char *) spec + 1; *term && *term != '/'; term++) { |
931233bc JH |
1893 | if (*term == '\\') |
1894 | term++; | |
1895 | } | |
1896 | if (*term != '/') | |
1897 | return spec; | |
1898 | ||
1899 | /* try [spec+1 .. term-1] as regexp */ | |
1900 | *term = 0; | |
1901 | begin--; /* input is in human terms */ | |
1902 | line = nth_line(sb, begin); | |
1903 | ||
1904 | if (!(reg_error = regcomp(®exp, spec + 1, REG_NEWLINE)) && | |
1905 | !(reg_error = regexec(®exp, line, 1, match, 0))) { | |
1906 | const char *cp = line + match[0].rm_so; | |
1907 | const char *nline; | |
1908 | ||
1909 | while (begin++ < lno) { | |
1910 | nline = nth_line(sb, begin); | |
1911 | if (line <= cp && cp < nline) | |
1912 | break; | |
1913 | line = nline; | |
1914 | } | |
1915 | *ret = begin; | |
1916 | regfree(®exp); | |
1917 | *term++ = '/'; | |
1918 | return term; | |
1919 | } | |
1920 | else { | |
1921 | char errbuf[1024]; | |
1922 | regerror(reg_error, ®exp, errbuf, 1024); | |
1923 | die("-L parameter '%s': %s", spec + 1, errbuf); | |
1924 | } | |
1925 | } | |
1926 | ||
1732a1fd JH |
1927 | /* |
1928 | * Parsing of -L option | |
1929 | */ | |
931233bc JH |
1930 | static void prepare_blame_range(struct scoreboard *sb, |
1931 | const char *bottomtop, | |
1932 | long lno, | |
1933 | long *bottom, long *top) | |
1934 | { | |
1935 | const char *term; | |
1936 | ||
1937 | term = parse_loc(bottomtop, sb, lno, 1, bottom); | |
1938 | if (*term == ',') { | |
1939 | term = parse_loc(term + 1, sb, lno, *bottom + 1, top); | |
1940 | if (*term) | |
acca687f | 1941 | usage(blame_usage); |
931233bc JH |
1942 | } |
1943 | if (*term) | |
acca687f | 1944 | usage(blame_usage); |
931233bc JH |
1945 | } |
1946 | ||
ef90d6d4 | 1947 | static int git_blame_config(const char *var, const char *value, void *cb) |
4c10a5ca JH |
1948 | { |
1949 | if (!strcmp(var, "blame.showroot")) { | |
1950 | show_root = git_config_bool(var, value); | |
1951 | return 0; | |
1952 | } | |
1953 | if (!strcmp(var, "blame.blankboundary")) { | |
1954 | blank_boundary = git_config_bool(var, value); | |
1955 | return 0; | |
1956 | } | |
31653c1a EL |
1957 | if (!strcmp(var, "blame.date")) { |
1958 | if (!value) | |
1959 | return config_error_nonbool(var); | |
1960 | blame_date_mode = parse_date_format(value); | |
1961 | return 0; | |
1962 | } | |
ef90d6d4 | 1963 | return git_default_config(var, value, cb); |
4c10a5ca JH |
1964 | } |
1965 | ||
f6c07d7d JH |
1966 | /* |
1967 | * Prepare a dummy commit that represents the work tree (or staged) item. | |
1968 | * Note that annotating work tree item never works in the reverse. | |
1969 | */ | |
1cfe7733 JH |
1970 | static struct commit *fake_working_tree_commit(const char *path, const char *contents_from) |
1971 | { | |
1972 | struct commit *commit; | |
1973 | struct origin *origin; | |
1974 | unsigned char head_sha1[20]; | |
f285a2d7 | 1975 | struct strbuf buf = STRBUF_INIT; |
1cfe7733 | 1976 | const char *ident; |
1cfe7733 | 1977 | time_t now; |
1cfe7733 JH |
1978 | int size, len; |
1979 | struct cache_entry *ce; | |
1980 | unsigned mode; | |
1981 | ||
1982 | if (get_sha1("HEAD", head_sha1)) | |
1983 | die("No such ref: HEAD"); | |
1984 | ||
1985 | time(&now); | |
1986 | commit = xcalloc(1, sizeof(*commit)); | |
1987 | commit->parents = xcalloc(1, sizeof(*commit->parents)); | |
1988 | commit->parents->item = lookup_commit_reference(head_sha1); | |
1989 | commit->object.parsed = 1; | |
1990 | commit->date = now; | |
1991 | commit->object.type = OBJ_COMMIT; | |
1992 | ||
1993 | origin = make_origin(commit, path); | |
1994 | ||
1995 | if (!contents_from || strcmp("-", contents_from)) { | |
1996 | struct stat st; | |
1997 | const char *read_from; | |
1998 | ||
1999 | if (contents_from) { | |
2000 | if (stat(contents_from, &st) < 0) | |
2001 | die("Cannot stat %s", contents_from); | |
2002 | read_from = contents_from; | |
2003 | } | |
2004 | else { | |
2005 | if (lstat(path, &st) < 0) | |
2006 | die("Cannot lstat %s", path); | |
2007 | read_from = path; | |
2008 | } | |
1cfe7733 JH |
2009 | mode = canon_mode(st.st_mode); |
2010 | switch (st.st_mode & S_IFMT) { | |
2011 | case S_IFREG: | |
387e7e19 PH |
2012 | if (strbuf_read_file(&buf, read_from, st.st_size) != st.st_size) |
2013 | die("cannot open or read %s", read_from); | |
1cfe7733 JH |
2014 | break; |
2015 | case S_IFLNK: | |
edfd45d2 | 2016 | if (strbuf_readlink(&buf, read_from, st.st_size) < 0) |
1cfe7733 JH |
2017 | die("cannot readlink %s", read_from); |
2018 | break; | |
2019 | default: | |
2020 | die("unsupported file type %s", read_from); | |
2021 | } | |
2022 | } | |
2023 | else { | |
2024 | /* Reading from stdin */ | |
2025 | contents_from = "standard input"; | |
1cfe7733 | 2026 | mode = 0; |
f1696ee3 | 2027 | if (strbuf_read(&buf, 0, 0) < 0) |
af6eb822 | 2028 | die("read error %s from stdin", strerror(errno)); |
1cfe7733 | 2029 | } |
21e5ad50 | 2030 | convert_to_git(path, buf.buf, buf.len, &buf, 0); |
af6eb822 PH |
2031 | origin->file.ptr = buf.buf; |
2032 | origin->file.size = buf.len; | |
2033 | pretend_sha1_file(buf.buf, buf.len, OBJ_BLOB, origin->blob_sha1); | |
1cfe7733 JH |
2034 | commit->util = origin; |
2035 | ||
2036 | /* | |
2037 | * Read the current index, replace the path entry with | |
2038 | * origin->blob_sha1 without mucking with its mode or type | |
2039 | * bits; we are not going to write this index out -- we just | |
2040 | * want to run "diff-index --cached". | |
2041 | */ | |
2042 | discard_cache(); | |
2043 | read_cache(); | |
2044 | ||
2045 | len = strlen(path); | |
2046 | if (!mode) { | |
2047 | int pos = cache_name_pos(path, len); | |
2048 | if (0 <= pos) | |
7a51ed66 | 2049 | mode = active_cache[pos]->ce_mode; |
1cfe7733 JH |
2050 | else |
2051 | /* Let's not bother reading from HEAD tree */ | |
2052 | mode = S_IFREG | 0644; | |
2053 | } | |
2054 | size = cache_entry_size(len); | |
2055 | ce = xcalloc(1, size); | |
2056 | hashcpy(ce->sha1, origin->blob_sha1); | |
2057 | memcpy(ce->name, path, len); | |
2058 | ce->ce_flags = create_ce_flags(len, 0); | |
2059 | ce->ce_mode = create_ce_mode(mode); | |
2060 | add_cache_entry(ce, ADD_CACHE_OK_TO_ADD|ADD_CACHE_OK_TO_REPLACE); | |
2061 | ||
2062 | /* | |
2063 | * We are not going to write this out, so this does not matter | |
2064 | * right now, but someday we might optimize diff-index --cached | |
2065 | * with cache-tree information. | |
2066 | */ | |
2067 | cache_tree_invalidate_path(active_cache_tree, path); | |
2068 | ||
2069 | commit->buffer = xmalloc(400); | |
2070 | ident = fmt_ident("Not Committed Yet", "not.committed.yet", NULL, 0); | |
1bb88be9 | 2071 | snprintf(commit->buffer, 400, |
1cfe7733 JH |
2072 | "tree 0000000000000000000000000000000000000000\n" |
2073 | "parent %s\n" | |
2074 | "author %s\n" | |
2075 | "committer %s\n\n" | |
2076 | "Version of %s from %s\n", | |
2077 | sha1_to_hex(head_sha1), | |
2078 | ident, ident, path, contents_from ? contents_from : path); | |
2079 | return commit; | |
2080 | } | |
2081 | ||
85af7929 | 2082 | static const char *prepare_final(struct scoreboard *sb) |
f6c07d7d JH |
2083 | { |
2084 | int i; | |
2085 | const char *final_commit_name = NULL; | |
85af7929 | 2086 | struct rev_info *revs = sb->revs; |
f6c07d7d JH |
2087 | |
2088 | /* | |
2089 | * There must be one and only one positive commit in the | |
2090 | * revs->pending array. | |
2091 | */ | |
2092 | for (i = 0; i < revs->pending.nr; i++) { | |
2093 | struct object *obj = revs->pending.objects[i].item; | |
2094 | if (obj->flags & UNINTERESTING) | |
2095 | continue; | |
2096 | while (obj->type == OBJ_TAG) | |
2097 | obj = deref_tag(obj, NULL, 0); | |
2098 | if (obj->type != OBJ_COMMIT) | |
2099 | die("Non commit %s?", revs->pending.objects[i].name); | |
2100 | if (sb->final) | |
2101 | die("More than one commit to dig from %s and %s?", | |
2102 | revs->pending.objects[i].name, | |
2103 | final_commit_name); | |
2104 | sb->final = (struct commit *) obj; | |
2105 | final_commit_name = revs->pending.objects[i].name; | |
2106 | } | |
2107 | return final_commit_name; | |
2108 | } | |
2109 | ||
85af7929 JH |
2110 | static const char *prepare_initial(struct scoreboard *sb) |
2111 | { | |
2112 | int i; | |
2113 | const char *final_commit_name = NULL; | |
2114 | struct rev_info *revs = sb->revs; | |
2115 | ||
2116 | /* | |
2117 | * There must be one and only one negative commit, and it must be | |
2118 | * the boundary. | |
2119 | */ | |
2120 | for (i = 0; i < revs->pending.nr; i++) { | |
2121 | struct object *obj = revs->pending.objects[i].item; | |
2122 | if (!(obj->flags & UNINTERESTING)) | |
2123 | continue; | |
2124 | while (obj->type == OBJ_TAG) | |
2125 | obj = deref_tag(obj, NULL, 0); | |
2126 | if (obj->type != OBJ_COMMIT) | |
2127 | die("Non commit %s?", revs->pending.objects[i].name); | |
2128 | if (sb->final) | |
2129 | die("More than one commit to dig down to %s and %s?", | |
2130 | revs->pending.objects[i].name, | |
2131 | final_commit_name); | |
2132 | sb->final = (struct commit *) obj; | |
2133 | final_commit_name = revs->pending.objects[i].name; | |
2134 | } | |
2135 | if (!final_commit_name) | |
2136 | die("No commit to dig down to?"); | |
2137 | return final_commit_name; | |
2138 | } | |
2139 | ||
5817da01 PH |
2140 | static int blame_copy_callback(const struct option *option, const char *arg, int unset) |
2141 | { | |
2142 | int *opt = option->value; | |
2143 | ||
2144 | /* | |
2145 | * -C enables copy from removed files; | |
2146 | * -C -C enables copy from existing files, but only | |
2147 | * when blaming a new file; | |
2148 | * -C -C -C enables copy from existing files for | |
2149 | * everybody | |
2150 | */ | |
2151 | if (*opt & PICKAXE_BLAME_COPY_HARDER) | |
2152 | *opt |= PICKAXE_BLAME_COPY_HARDEST; | |
2153 | if (*opt & PICKAXE_BLAME_COPY) | |
2154 | *opt |= PICKAXE_BLAME_COPY_HARDER; | |
2155 | *opt |= PICKAXE_BLAME_COPY | PICKAXE_BLAME_MOVE; | |
2156 | ||
2157 | if (arg) | |
2158 | blame_copy_score = parse_score(arg); | |
2159 | return 0; | |
2160 | } | |
2161 | ||
2162 | static int blame_move_callback(const struct option *option, const char *arg, int unset) | |
2163 | { | |
2164 | int *opt = option->value; | |
2165 | ||
2166 | *opt |= PICKAXE_BLAME_MOVE; | |
2167 | ||
2168 | if (arg) | |
2169 | blame_move_score = parse_score(arg); | |
2170 | return 0; | |
2171 | } | |
2172 | ||
2173 | static int blame_bottomtop_callback(const struct option *option, const char *arg, int unset) | |
2174 | { | |
2175 | const char **bottomtop = option->value; | |
2176 | if (!arg) | |
2177 | return -1; | |
2178 | if (*bottomtop) | |
2179 | die("More than one '-L n,m' option given"); | |
2180 | *bottomtop = arg; | |
2181 | return 0; | |
2182 | } | |
2183 | ||
acca687f | 2184 | int cmd_blame(int argc, const char **argv, const char *prefix) |
cee7f245 JH |
2185 | { |
2186 | struct rev_info revs; | |
2187 | const char *path; | |
2188 | struct scoreboard sb; | |
2189 | struct origin *o; | |
2190 | struct blame_entry *ent; | |
3f8d5204 | 2191 | long dashdash_pos, bottom, top, lno; |
cee7f245 | 2192 | const char *final_commit_name = NULL; |
21666f1a | 2193 | enum object_type type; |
5817da01 PH |
2194 | |
2195 | static const char *bottomtop = NULL; | |
2196 | static int output_option = 0, opt = 0; | |
2197 | static int show_stats = 0; | |
2198 | static const char *revs_file = NULL; | |
2199 | static const char *contents_from = NULL; | |
2200 | static const struct option options[] = { | |
2201 | OPT_BOOLEAN(0, "incremental", &incremental, "Show blame entries as we find them, incrementally"), | |
2202 | OPT_BOOLEAN('b', NULL, &blank_boundary, "Show blank SHA-1 for boundary commits (Default: off)"), | |
2203 | OPT_BOOLEAN(0, "root", &show_root, "Do not treat root commits as boundaries (Default: off)"), | |
2204 | OPT_BOOLEAN(0, "show-stats", &show_stats, "Show work cost statistics"), | |
2205 | OPT_BIT(0, "score-debug", &output_option, "Show output score for blame entries", OUTPUT_SHOW_SCORE), | |
2206 | OPT_BIT('f', "show-name", &output_option, "Show original filename (Default: auto)", OUTPUT_SHOW_NAME), | |
2207 | OPT_BIT('n', "show-number", &output_option, "Show original linenumber (Default: off)", OUTPUT_SHOW_NUMBER), | |
2208 | OPT_BIT('p', "porcelain", &output_option, "Show in a format designed for machine consumption", OUTPUT_PORCELAIN), | |
2209 | OPT_BIT('c', NULL, &output_option, "Use the same output mode as git-annotate (Default: off)", OUTPUT_ANNOTATE_COMPAT), | |
2210 | OPT_BIT('t', NULL, &output_option, "Show raw timestamp (Default: off)", OUTPUT_RAW_TIMESTAMP), | |
2211 | OPT_BIT('l', NULL, &output_option, "Show long commit SHA1 (Default: off)", OUTPUT_LONG_OBJECT_NAME), | |
2212 | OPT_BIT('s', NULL, &output_option, "Suppress author name and timestamp (Default: off)", OUTPUT_NO_AUTHOR), | |
2213 | OPT_BIT('w', NULL, &xdl_opts, "Ignore whitespace differences", XDF_IGNORE_WHITESPACE), | |
2214 | OPT_STRING('S', NULL, &revs_file, "file", "Use revisions from <file> instead of calling git-rev-list"), | |
2215 | OPT_STRING(0, "contents", &contents_from, "file", "Use <file>'s contents as the final image"), | |
2216 | { OPTION_CALLBACK, 'C', NULL, &opt, "score", "Find line copies within and across files", PARSE_OPT_OPTARG, blame_copy_callback }, | |
2217 | { OPTION_CALLBACK, 'M', NULL, &opt, "score", "Find line movements within and across files", PARSE_OPT_OPTARG, blame_move_callback }, | |
2218 | OPT_CALLBACK('L', NULL, &bottomtop, "n,m", "Process only line range n,m, counting from 1", blame_bottomtop_callback), | |
2219 | OPT_END() | |
2220 | }; | |
2221 | ||
2222 | struct parse_opt_ctx_t ctx; | |
7ceacdff | 2223 | int cmd_is_annotate = !strcmp(argv[0], "annotate"); |
e68989a7 | 2224 | |
ef90d6d4 | 2225 | git_config(git_blame_config, NULL); |
5817da01 | 2226 | init_revisions(&revs, NULL); |
31653c1a EL |
2227 | revs.date_mode = blame_date_mode; |
2228 | ||
612702e8 | 2229 | save_commit_buffer = 0; |
3f8d5204 | 2230 | dashdash_pos = 0; |
612702e8 | 2231 | |
37782920 | 2232 | parse_options_start(&ctx, argc, argv, prefix, PARSE_OPT_KEEP_DASHDASH | |
5817da01 PH |
2233 | PARSE_OPT_KEEP_ARGV0); |
2234 | for (;;) { | |
5817da01 PH |
2235 | switch (parse_options_step(&ctx, options, blame_opt_usage)) { |
2236 | case PARSE_OPT_HELP: | |
2237 | exit(129); | |
2238 | case PARSE_OPT_DONE: | |
3f8d5204 PH |
2239 | if (ctx.argv[0]) |
2240 | dashdash_pos = ctx.cpidx; | |
5817da01 PH |
2241 | goto parse_done; |
2242 | } | |
2243 | ||
2244 | if (!strcmp(ctx.argv[0], "--reverse")) { | |
2245 | ctx.argv[0] = "--children"; | |
2246 | reverse = 1; | |
2247 | } | |
6b61ec05 | 2248 | parse_revision_opt(&revs, &ctx, options, blame_opt_usage); |
5817da01 PH |
2249 | } |
2250 | parse_done: | |
2251 | argc = parse_options_end(&ctx); | |
2252 | ||
aa9ea77d JH |
2253 | if (revs_file && read_ancestry(revs_file)) |
2254 | die("reading graft file %s failed: %s", | |
2255 | revs_file, strerror(errno)); | |
2256 | ||
31653c1a | 2257 | if (cmd_is_annotate) { |
7ceacdff | 2258 | output_option |= OUTPUT_ANNOTATE_COMPAT; |
31653c1a EL |
2259 | blame_date_mode = DATE_ISO8601; |
2260 | } else { | |
2261 | blame_date_mode = revs.date_mode; | |
2262 | } | |
2263 | ||
2264 | /* The maximum width used to show the dates */ | |
2265 | switch (blame_date_mode) { | |
2266 | case DATE_RFC2822: | |
2267 | blame_date_width = sizeof("Thu, 19 Oct 2006 16:00:04 -0700"); | |
2268 | break; | |
2269 | case DATE_ISO8601: | |
2270 | blame_date_width = sizeof("2006-10-19 16:00:04 -0700"); | |
2271 | break; | |
2272 | case DATE_RAW: | |
2273 | blame_date_width = sizeof("1161298804 -0700"); | |
2274 | break; | |
2275 | case DATE_SHORT: | |
2276 | blame_date_width = sizeof("2006-10-19"); | |
2277 | break; | |
2278 | case DATE_RELATIVE: | |
2279 | /* "normal" is used as the fallback for "relative" */ | |
2280 | case DATE_LOCAL: | |
2281 | case DATE_NORMAL: | |
2282 | blame_date_width = sizeof("Thu Oct 19 16:00:04 2006 -0700"); | |
2283 | break; | |
2284 | } | |
2285 | blame_date_width -= 1; /* strip the null */ | |
7ceacdff | 2286 | |
b3123f98 JH |
2287 | if (DIFF_OPT_TST(&revs.diffopt, FIND_COPIES_HARDER)) |
2288 | opt |= (PICKAXE_BLAME_COPY | PICKAXE_BLAME_MOVE | | |
2289 | PICKAXE_BLAME_COPY_HARDER); | |
2290 | ||
4a0fc95f JH |
2291 | if (!blame_move_score) |
2292 | blame_move_score = BLAME_DEFAULT_MOVE_SCORE; | |
2293 | if (!blame_copy_score) | |
2294 | blame_copy_score = BLAME_DEFAULT_COPY_SCORE; | |
2295 | ||
1732a1fd JH |
2296 | /* |
2297 | * We have collected options unknown to us in argv[1..unk] | |
cee7f245 | 2298 | * which are to be passed to revision machinery if we are |
3dff5379 | 2299 | * going to do the "bottom" processing. |
cee7f245 JH |
2300 | * |
2301 | * The remaining are: | |
2302 | * | |
3f8d5204 PH |
2303 | * (1) if dashdash_pos != 0, its either |
2304 | * "blame [revisions] -- <path>" or | |
2305 | * "blame -- <path> <rev>" | |
cee7f245 | 2306 | * |
3f8d5204 PH |
2307 | * (2) otherwise, its one of the two: |
2308 | * "blame [revisions] <path>" | |
2309 | * "blame <path> <rev>" | |
cee7f245 | 2310 | * |
3f8d5204 PH |
2311 | * Note that we must strip out <path> from the arguments: we do not |
2312 | * want the path pruning but we may want "bottom" processing. | |
cee7f245 | 2313 | */ |
3f8d5204 PH |
2314 | if (dashdash_pos) { |
2315 | switch (argc - dashdash_pos - 1) { | |
2316 | case 2: /* (1b) */ | |
2317 | if (argc != 4) | |
5817da01 | 2318 | usage_with_options(blame_opt_usage, options); |
3f8d5204 PH |
2319 | /* reorder for the new way: <rev> -- <path> */ |
2320 | argv[1] = argv[3]; | |
2321 | argv[3] = argv[2]; | |
2322 | argv[2] = "--"; | |
2323 | /* FALLTHROUGH */ | |
2324 | case 1: /* (1a) */ | |
2325 | path = add_prefix(prefix, argv[--argc]); | |
2326 | argv[argc] = NULL; | |
2327 | break; | |
2328 | default: | |
2329 | usage_with_options(blame_opt_usage, options); | |
cee7f245 | 2330 | } |
3f8d5204 PH |
2331 | } else { |
2332 | if (argc < 2) | |
5817da01 | 2333 | usage_with_options(blame_opt_usage, options); |
3f8d5204 | 2334 | path = add_prefix(prefix, argv[argc - 1]); |
c455c87c | 2335 | if (argc == 3 && !has_string_in_work_tree(path)) { /* (2b) */ |
3f8d5204 PH |
2336 | path = add_prefix(prefix, argv[1]); |
2337 | argv[1] = argv[2]; | |
cee7f245 | 2338 | } |
3f8d5204 | 2339 | argv[argc - 1] = "--"; |
cee7f245 | 2340 | |
3f8d5204 | 2341 | setup_work_tree(); |
c455c87c | 2342 | if (!has_string_in_work_tree(path)) |
3f8d5204 | 2343 | die("cannot stat path %s: %s", path, strerror(errno)); |
cee7f245 JH |
2344 | } |
2345 | ||
3f8d5204 | 2346 | setup_revisions(argc, argv, &revs, NULL); |
cee7f245 JH |
2347 | memset(&sb, 0, sizeof(sb)); |
2348 | ||
85af7929 JH |
2349 | sb.revs = &revs; |
2350 | if (!reverse) | |
2351 | final_commit_name = prepare_final(&sb); | |
2352 | else if (contents_from) | |
2353 | die("--contents and --children do not blend well."); | |
2354 | else | |
2355 | final_commit_name = prepare_initial(&sb); | |
cee7f245 JH |
2356 | |
2357 | if (!sb.final) { | |
1732a1fd JH |
2358 | /* |
2359 | * "--not A B -- path" without anything positive; | |
1cfe7733 JH |
2360 | * do not default to HEAD, but use the working tree |
2361 | * or "--contents". | |
1732a1fd | 2362 | */ |
1981820b | 2363 | setup_work_tree(); |
1cfe7733 JH |
2364 | sb.final = fake_working_tree_commit(path, contents_from); |
2365 | add_pending_object(&revs, &(sb.final->object), ":"); | |
cee7f245 | 2366 | } |
1cfe7733 JH |
2367 | else if (contents_from) |
2368 | die("Cannot use --contents with final commit object name"); | |
cee7f245 | 2369 | |
1732a1fd JH |
2370 | /* |
2371 | * If we have bottom, this will mark the ancestors of the | |
cee7f245 JH |
2372 | * bottom commits we would reach while traversing as |
2373 | * uninteresting. | |
2374 | */ | |
3d51e1b5 MK |
2375 | if (prepare_revision_walk(&revs)) |
2376 | die("revision walk setup failed"); | |
cee7f245 | 2377 | |
1cfe7733 JH |
2378 | if (is_null_sha1(sb.final->object.sha1)) { |
2379 | char *buf; | |
2380 | o = sb.final->util; | |
2381 | buf = xmalloc(o->file.size + 1); | |
2382 | memcpy(buf, o->file.ptr, o->file.size + 1); | |
2383 | sb.final_buf = buf; | |
2384 | sb.final_buf_size = o->file.size; | |
2385 | } | |
2386 | else { | |
2387 | o = get_origin(&sb, sb.final, path); | |
2388 | if (fill_blob_sha1(o)) | |
2389 | die("no such path %s in %s", path, final_commit_name); | |
cee7f245 | 2390 | |
21666f1a | 2391 | sb.final_buf = read_sha1_file(o->blob_sha1, &type, |
1cfe7733 | 2392 | &sb.final_buf_size); |
ab43e495 JH |
2393 | if (!sb.final_buf) |
2394 | die("Cannot read blob %s for path %s", | |
2395 | sha1_to_hex(o->blob_sha1), | |
2396 | path); | |
1cfe7733 | 2397 | } |
c2e525d9 | 2398 | num_read_blob++; |
cee7f245 JH |
2399 | lno = prepare_lines(&sb); |
2400 | ||
931233bc JH |
2401 | bottom = top = 0; |
2402 | if (bottomtop) | |
2403 | prepare_blame_range(&sb, bottomtop, lno, &bottom, &top); | |
2404 | if (bottom && top && top < bottom) { | |
2405 | long tmp; | |
2406 | tmp = top; top = bottom; bottom = tmp; | |
2407 | } | |
cee7f245 JH |
2408 | if (bottom < 1) |
2409 | bottom = 1; | |
2410 | if (top < 1) | |
2411 | top = lno; | |
2412 | bottom--; | |
2413 | if (lno < top) | |
2414 | die("file %s has only %lu lines", path, lno); | |
2415 | ||
2416 | ent = xcalloc(1, sizeof(*ent)); | |
2417 | ent->lno = bottom; | |
2418 | ent->num_lines = top - bottom; | |
2419 | ent->suspect = o; | |
2420 | ent->s_lno = bottom; | |
2421 | ||
2422 | sb.ent = ent; | |
2423 | sb.path = path; | |
2424 | ||
d551a488 | 2425 | read_mailmap(&mailmap, NULL); |
f9567384 | 2426 | |
b92565dc MH |
2427 | if (!incremental) |
2428 | setup_pager(); | |
2429 | ||
85af7929 | 2430 | assign_blame(&sb, opt); |
cee7f245 | 2431 | |
717d1462 LT |
2432 | if (incremental) |
2433 | return 0; | |
2434 | ||
cee7f245 JH |
2435 | coalesce(&sb); |
2436 | ||
2437 | if (!(output_option & OUTPUT_PORCELAIN)) | |
2438 | find_alignment(&sb, &output_option); | |
2439 | ||
2440 | output(&sb, output_option); | |
2441 | free((void *)sb.final_buf); | |
2442 | for (ent = sb.ent; ent; ) { | |
2443 | struct blame_entry *e = ent->next; | |
2444 | free(ent); | |
2445 | ent = e; | |
2446 | } | |
c2e525d9 | 2447 | |
870b39c1 | 2448 | if (show_stats) { |
c2e525d9 JH |
2449 | printf("num read blob: %d\n", num_read_blob); |
2450 | printf("num get patch: %d\n", num_get_patch); | |
2451 | printf("num commits: %d\n", num_commits); | |
2452 | } | |
cee7f245 JH |
2453 | return 0; |
2454 | } |