]>
Commit | Line | Data |
---|---|---|
cee7f245 JH |
1 | /* |
2 | * Pickaxe | |
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" |
f9567384 JH |
19 | #include "path-list.h" |
20 | #include "mailmap.h" | |
cee7f245 | 21 | |
acca687f | 22 | static char blame_usage[] = |
b82871b3 | 23 | "git-blame [-c] [-b] [-l] [--root] [-t] [-f] [-n] [-s] [-p] [-w] [-L n,m] [-S <revs-file>] [-M] [-C] [-C] [--contents <filename>] [--incremental] [commit] [--] file\n" |
635f4a30 | 24 | " -c Use the same output mode as git-annotate (Default: off)\n" |
4c10a5ca | 25 | " -b Show blank SHA-1 for boundary commits (Default: off)\n" |
635f4a30 | 26 | " -l Show long commit SHA1 (Default: off)\n" |
4c10a5ca | 27 | " --root Do not treat root commits as boundaries (Default: off)\n" |
635f4a30 | 28 | " -t Show raw timestamp (Default: off)\n" |
cee7f245 JH |
29 | " -f, --show-name Show original filename (Default: auto)\n" |
30 | " -n, --show-number Show original linenumber (Default: off)\n" | |
093dc5be | 31 | " -s Suppress author name and timestamp (Default: off)\n" |
cee7f245 | 32 | " -p, --porcelain Show in a format designed for machine consumption\n" |
b82871b3 | 33 | " -w Ignore whitespace differences\n" |
cee7f245 | 34 | " -L n,m Process only line range n,m, counting from 1\n" |
18abd745 | 35 | " -M, -C Find line movements within and across files\n" |
717d1462 | 36 | " --incremental Show blame entries as we find them, incrementally\n" |
1cfe7733 | 37 | " --contents file Use <file>'s contents as the final image\n" |
cee7f245 JH |
38 | " -S revs-file Use revisions from revs-file instead of calling git-rev-list\n"; |
39 | ||
40 | static int longest_file; | |
41 | static int longest_author; | |
42 | static int max_orig_digits; | |
43 | static int max_digits; | |
5ff62c30 | 44 | static int max_score_digits; |
4c10a5ca JH |
45 | static int show_root; |
46 | static int blank_boundary; | |
717d1462 | 47 | static int incremental; |
e68989a7 | 48 | static int cmd_is_annotate; |
b82871b3 | 49 | static int xdl_opts = XDF_NEED_MINIMAL; |
f9567384 | 50 | static struct path_list mailmap; |
cee7f245 | 51 | |
54a4c617 JH |
52 | #ifndef DEBUG |
53 | #define DEBUG 0 | |
54 | #endif | |
55 | ||
c2e525d9 JH |
56 | /* stats */ |
57 | static int num_read_blob; | |
58 | static int num_get_patch; | |
59 | static int num_commits; | |
60 | ||
d24bba80 | 61 | #define PICKAXE_BLAME_MOVE 01 |
18abd745 JH |
62 | #define PICKAXE_BLAME_COPY 02 |
63 | #define PICKAXE_BLAME_COPY_HARDER 04 | |
c63777c0 | 64 | #define PICKAXE_BLAME_COPY_HARDEST 010 |
d24bba80 | 65 | |
4a0fc95f JH |
66 | /* |
67 | * blame for a blame_entry with score lower than these thresholds | |
68 | * is not passed to the parent using move/copy logic. | |
69 | */ | |
70 | static unsigned blame_move_score; | |
71 | static unsigned blame_copy_score; | |
72 | #define BLAME_DEFAULT_MOVE_SCORE 20 | |
73 | #define BLAME_DEFAULT_COPY_SCORE 40 | |
74 | ||
cee7f245 JH |
75 | /* bits #0..7 in revision.h, #8..11 used for merge_bases() in commit.c */ |
76 | #define METAINFO_SHOWN (1u<<12) | |
77 | #define MORE_THAN_ONE_PATH (1u<<13) | |
78 | ||
79 | /* | |
54a4c617 | 80 | * One blob in a commit that is being suspected |
cee7f245 JH |
81 | */ |
82 | struct origin { | |
54a4c617 | 83 | int refcnt; |
cee7f245 | 84 | struct commit *commit; |
c2e525d9 | 85 | mmfile_t file; |
cee7f245 JH |
86 | unsigned char blob_sha1[20]; |
87 | char path[FLEX_ARRAY]; | |
88 | }; | |
89 | ||
1732a1fd JH |
90 | /* |
91 | * Given an origin, prepare mmfile_t structure to be used by the | |
92 | * diff machinery | |
93 | */ | |
f6c07d7d | 94 | static void fill_origin_blob(struct origin *o, mmfile_t *file) |
c2e525d9 JH |
95 | { |
96 | if (!o->file.ptr) { | |
21666f1a | 97 | enum object_type type; |
c2e525d9 | 98 | num_read_blob++; |
21666f1a | 99 | file->ptr = read_sha1_file(o->blob_sha1, &type, |
c2e525d9 | 100 | (unsigned long *)(&(file->size))); |
ab43e495 JH |
101 | if (!file->ptr) |
102 | die("Cannot read blob %s for path %s", | |
103 | sha1_to_hex(o->blob_sha1), | |
104 | o->path); | |
c2e525d9 JH |
105 | o->file = *file; |
106 | } | |
107 | else | |
108 | *file = o->file; | |
c2e525d9 JH |
109 | } |
110 | ||
1732a1fd JH |
111 | /* |
112 | * Origin is refcounted and usually we keep the blob contents to be | |
113 | * reused. | |
114 | */ | |
54a4c617 JH |
115 | static inline struct origin *origin_incref(struct origin *o) |
116 | { | |
117 | if (o) | |
118 | o->refcnt++; | |
119 | return o; | |
120 | } | |
121 | ||
122 | static void origin_decref(struct origin *o) | |
123 | { | |
124 | if (o && --o->refcnt <= 0) { | |
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 | ||
163 | /* the line number of the first line of this group in the | |
164 | * suspect's file; internally all line numbers are 0 based. | |
165 | */ | |
166 | int s_lno; | |
5ff62c30 JH |
167 | |
168 | /* how significant this entry is -- cached to avoid | |
1732a1fd | 169 | * scanning the lines over and over. |
5ff62c30 JH |
170 | */ |
171 | unsigned score; | |
cee7f245 JH |
172 | }; |
173 | ||
1732a1fd JH |
174 | /* |
175 | * The current state of the blame assignment. | |
176 | */ | |
cee7f245 JH |
177 | struct scoreboard { |
178 | /* the final commit (i.e. where we started digging from) */ | |
179 | struct commit *final; | |
180 | ||
181 | const char *path; | |
182 | ||
1732a1fd JH |
183 | /* |
184 | * The contents in the final image. | |
185 | * Used by many functions to obtain contents of the nth line, | |
186 | * indexed with scoreboard.lineno[blame_entry.lno]. | |
cee7f245 JH |
187 | */ |
188 | const char *final_buf; | |
189 | unsigned long final_buf_size; | |
190 | ||
191 | /* linked list of blames */ | |
192 | struct blame_entry *ent; | |
193 | ||
612702e8 | 194 | /* look-up a line in the final buffer */ |
cee7f245 JH |
195 | int num_lines; |
196 | int *lineno; | |
197 | }; | |
198 | ||
171dccd5 | 199 | static inline int same_suspect(struct origin *a, struct origin *b) |
46014766 | 200 | { |
171dccd5 | 201 | if (a == b) |
57584d9e | 202 | return 1; |
171dccd5 JH |
203 | if (a->commit != b->commit) |
204 | return 0; | |
205 | return !strcmp(a->path, b->path); | |
46014766 JH |
206 | } |
207 | ||
54a4c617 JH |
208 | static void sanity_check_refcnt(struct scoreboard *); |
209 | ||
1732a1fd JH |
210 | /* |
211 | * If two blame entries that are next to each other came from | |
212 | * contiguous lines in the same origin (i.e. <commit, path> pair), | |
213 | * merge them together. | |
214 | */ | |
cee7f245 JH |
215 | static void coalesce(struct scoreboard *sb) |
216 | { | |
217 | struct blame_entry *ent, *next; | |
218 | ||
219 | for (ent = sb->ent; ent && (next = ent->next); ent = next) { | |
171dccd5 | 220 | if (same_suspect(ent->suspect, next->suspect) && |
cee7f245 JH |
221 | ent->guilty == next->guilty && |
222 | ent->s_lno + ent->num_lines == next->s_lno) { | |
223 | ent->num_lines += next->num_lines; | |
224 | ent->next = next->next; | |
225 | if (ent->next) | |
226 | ent->next->prev = ent; | |
54a4c617 | 227 | origin_decref(next->suspect); |
cee7f245 | 228 | free(next); |
46014766 | 229 | ent->score = 0; |
cee7f245 JH |
230 | next = ent; /* again */ |
231 | } | |
232 | } | |
54a4c617 JH |
233 | |
234 | if (DEBUG) /* sanity */ | |
235 | sanity_check_refcnt(sb); | |
cee7f245 JH |
236 | } |
237 | ||
1732a1fd JH |
238 | /* |
239 | * Given a commit and a path in it, create a new origin structure. | |
240 | * The callers that add blame to the scoreboard should use | |
241 | * get_origin() to obtain shared, refcounted copy instead of calling | |
242 | * this function directly. | |
243 | */ | |
854b97f6 JH |
244 | static struct origin *make_origin(struct commit *commit, const char *path) |
245 | { | |
246 | struct origin *o; | |
247 | o = xcalloc(1, sizeof(*o) + strlen(path) + 1); | |
248 | o->commit = commit; | |
249 | o->refcnt = 1; | |
250 | strcpy(o->path, path); | |
251 | return o; | |
252 | } | |
253 | ||
1732a1fd JH |
254 | /* |
255 | * Locate an existing origin or create a new one. | |
256 | */ | |
f6c0e191 JH |
257 | static struct origin *get_origin(struct scoreboard *sb, |
258 | struct commit *commit, | |
259 | const char *path) | |
cee7f245 | 260 | { |
f6c0e191 | 261 | struct blame_entry *e; |
cee7f245 | 262 | |
f6c0e191 JH |
263 | for (e = sb->ent; e; e = e->next) { |
264 | if (e->suspect->commit == commit && | |
265 | !strcmp(e->suspect->path, path)) | |
54a4c617 | 266 | return origin_incref(e->suspect); |
cee7f245 | 267 | } |
854b97f6 | 268 | return make_origin(commit, path); |
cee7f245 JH |
269 | } |
270 | ||
1732a1fd JH |
271 | /* |
272 | * Fill the blob_sha1 field of an origin if it hasn't, so that later | |
273 | * call to fill_origin_blob() can use it to locate the data. blob_sha1 | |
274 | * for an origin is also used to pass the blame for the entire file to | |
275 | * the parent to detect the case where a child's blob is identical to | |
276 | * that of its parent's. | |
277 | */ | |
f6c0e191 JH |
278 | static int fill_blob_sha1(struct origin *origin) |
279 | { | |
280 | unsigned mode; | |
f6c0e191 JH |
281 | |
282 | if (!is_null_sha1(origin->blob_sha1)) | |
283 | return 0; | |
284 | if (get_tree_entry(origin->commit->object.sha1, | |
285 | origin->path, | |
286 | origin->blob_sha1, &mode)) | |
287 | goto error_out; | |
21666f1a | 288 | if (sha1_object_info(origin->blob_sha1, NULL) != OBJ_BLOB) |
f6c0e191 JH |
289 | goto error_out; |
290 | return 0; | |
291 | error_out: | |
292 | hashclr(origin->blob_sha1); | |
293 | return -1; | |
294 | } | |
295 | ||
1732a1fd JH |
296 | /* |
297 | * We have an origin -- check if the same path exists in the | |
298 | * parent and return an origin structure to represent it. | |
299 | */ | |
f6c0e191 | 300 | static struct origin *find_origin(struct scoreboard *sb, |
cee7f245 JH |
301 | struct commit *parent, |
302 | struct origin *origin) | |
303 | { | |
304 | struct origin *porigin = NULL; | |
305 | struct diff_options diff_opts; | |
f6c0e191 JH |
306 | const char *paths[2]; |
307 | ||
0d981c67 | 308 | if (parent->util) { |
1732a1fd JH |
309 | /* |
310 | * Each commit object can cache one origin in that | |
311 | * commit. This is a freestanding copy of origin and | |
312 | * not refcounted. | |
854b97f6 | 313 | */ |
0d981c67 | 314 | struct origin *cached = parent->util; |
854b97f6 | 315 | if (!strcmp(cached->path, origin->path)) { |
1732a1fd JH |
316 | /* |
317 | * The same path between origin and its parent | |
318 | * without renaming -- the most common case. | |
319 | */ | |
854b97f6 | 320 | porigin = get_origin(sb, parent, cached->path); |
1732a1fd JH |
321 | |
322 | /* | |
323 | * If the origin was newly created (i.e. get_origin | |
324 | * would call make_origin if none is found in the | |
325 | * scoreboard), it does not know the blob_sha1, | |
326 | * so copy it. Otherwise porigin was in the | |
327 | * scoreboard and already knows blob_sha1. | |
328 | */ | |
854b97f6 JH |
329 | if (porigin->refcnt == 1) |
330 | hashcpy(porigin->blob_sha1, cached->blob_sha1); | |
331 | return porigin; | |
332 | } | |
333 | /* otherwise it was not very useful; free it */ | |
334 | free(parent->util); | |
335 | parent->util = NULL; | |
0d981c67 JH |
336 | } |
337 | ||
f6c0e191 JH |
338 | /* See if the origin->path is different between parent |
339 | * and origin first. Most of the time they are the | |
340 | * same and diff-tree is fairly efficient about this. | |
341 | */ | |
342 | diff_setup(&diff_opts); | |
8f67f8ae | 343 | DIFF_OPT_SET(&diff_opts, RECURSIVE); |
f6c0e191 JH |
344 | diff_opts.detect_rename = 0; |
345 | diff_opts.output_format = DIFF_FORMAT_NO_OUTPUT; | |
346 | paths[0] = origin->path; | |
347 | paths[1] = NULL; | |
348 | ||
349 | diff_tree_setup_paths(paths, &diff_opts); | |
350 | if (diff_setup_done(&diff_opts) < 0) | |
351 | die("diff-setup"); | |
1cfe7733 JH |
352 | |
353 | if (is_null_sha1(origin->commit->object.sha1)) | |
354 | do_diff_cache(parent->tree->object.sha1, &diff_opts); | |
355 | else | |
356 | diff_tree_sha1(parent->tree->object.sha1, | |
357 | origin->commit->tree->object.sha1, | |
358 | "", &diff_opts); | |
f6c0e191 JH |
359 | diffcore_std(&diff_opts); |
360 | ||
361 | /* It is either one entry that says "modified", or "created", | |
362 | * or nothing. | |
363 | */ | |
364 | if (!diff_queued_diff.nr) { | |
365 | /* The path is the same as parent */ | |
366 | porigin = get_origin(sb, parent, origin->path); | |
367 | hashcpy(porigin->blob_sha1, origin->blob_sha1); | |
368 | } | |
369 | else if (diff_queued_diff.nr != 1) | |
acca687f | 370 | die("internal error in blame::find_origin"); |
f6c0e191 JH |
371 | else { |
372 | struct diff_filepair *p = diff_queued_diff.queue[0]; | |
373 | switch (p->status) { | |
374 | default: | |
acca687f | 375 | die("internal error in blame::find_origin (%c)", |
f6c0e191 JH |
376 | p->status); |
377 | case 'M': | |
378 | porigin = get_origin(sb, parent, origin->path); | |
379 | hashcpy(porigin->blob_sha1, p->one->sha1); | |
380 | break; | |
381 | case 'A': | |
382 | case 'T': | |
383 | /* Did not exist in parent, or type changed */ | |
384 | break; | |
385 | } | |
386 | } | |
387 | diff_flush(&diff_opts); | |
03b69c76 | 388 | diff_tree_release_paths(&diff_opts); |
0d981c67 | 389 | if (porigin) { |
1732a1fd JH |
390 | /* |
391 | * Create a freestanding copy that is not part of | |
392 | * the refcounted origin found in the scoreboard, and | |
393 | * cache it in the commit. | |
394 | */ | |
854b97f6 | 395 | struct origin *cached; |
1732a1fd | 396 | |
854b97f6 JH |
397 | cached = make_origin(porigin->commit, porigin->path); |
398 | hashcpy(cached->blob_sha1, porigin->blob_sha1); | |
399 | parent->util = cached; | |
0d981c67 | 400 | } |
f69e743d JH |
401 | return porigin; |
402 | } | |
f6c0e191 | 403 | |
1732a1fd JH |
404 | /* |
405 | * We have an origin -- find the path that corresponds to it in its | |
406 | * parent and return an origin structure to represent it. | |
407 | */ | |
f69e743d JH |
408 | static struct origin *find_rename(struct scoreboard *sb, |
409 | struct commit *parent, | |
410 | struct origin *origin) | |
411 | { | |
412 | struct origin *porigin = NULL; | |
413 | struct diff_options diff_opts; | |
414 | int i; | |
415 | const char *paths[2]; | |
cee7f245 JH |
416 | |
417 | diff_setup(&diff_opts); | |
8f67f8ae | 418 | DIFF_OPT_SET(&diff_opts, RECURSIVE); |
cee7f245 JH |
419 | diff_opts.detect_rename = DIFF_DETECT_RENAME; |
420 | diff_opts.output_format = DIFF_FORMAT_NO_OUTPUT; | |
2f3f8b21 | 421 | diff_opts.single_follow = origin->path; |
cee7f245 JH |
422 | paths[0] = NULL; |
423 | diff_tree_setup_paths(paths, &diff_opts); | |
424 | if (diff_setup_done(&diff_opts) < 0) | |
425 | die("diff-setup"); | |
1cfe7733 JH |
426 | |
427 | if (is_null_sha1(origin->commit->object.sha1)) | |
428 | do_diff_cache(parent->tree->object.sha1, &diff_opts); | |
429 | else | |
430 | diff_tree_sha1(parent->tree->object.sha1, | |
431 | origin->commit->tree->object.sha1, | |
432 | "", &diff_opts); | |
cee7f245 JH |
433 | diffcore_std(&diff_opts); |
434 | ||
435 | for (i = 0; i < diff_queued_diff.nr; i++) { | |
436 | struct diff_filepair *p = diff_queued_diff.queue[i]; | |
612702e8 | 437 | if ((p->status == 'R' || p->status == 'C') && |
f6c0e191 JH |
438 | !strcmp(p->two->path, origin->path)) { |
439 | porigin = get_origin(sb, parent, p->one->path); | |
440 | hashcpy(porigin->blob_sha1, p->one->sha1); | |
cee7f245 JH |
441 | break; |
442 | } | |
443 | } | |
444 | diff_flush(&diff_opts); | |
03b69c76 | 445 | diff_tree_release_paths(&diff_opts); |
cee7f245 JH |
446 | return porigin; |
447 | } | |
448 | ||
1732a1fd JH |
449 | /* |
450 | * Parsing of patch chunks... | |
451 | */ | |
cee7f245 JH |
452 | struct chunk { |
453 | /* line number in postimage; up to but not including this | |
454 | * line is the same as preimage | |
455 | */ | |
456 | int same; | |
457 | ||
458 | /* preimage line number after this chunk */ | |
459 | int p_next; | |
460 | ||
461 | /* postimage line number after this chunk */ | |
462 | int t_next; | |
463 | }; | |
464 | ||
465 | struct patch { | |
466 | struct chunk *chunks; | |
467 | int num; | |
468 | }; | |
469 | ||
470 | struct blame_diff_state { | |
471 | struct xdiff_emit_state xm; | |
472 | struct patch *ret; | |
473 | unsigned hunk_post_context; | |
474 | unsigned hunk_in_pre_context : 1; | |
475 | }; | |
476 | ||
477 | static void process_u_diff(void *state_, char *line, unsigned long len) | |
478 | { | |
479 | struct blame_diff_state *state = state_; | |
480 | struct chunk *chunk; | |
481 | int off1, off2, len1, len2, num; | |
482 | ||
cee7f245 JH |
483 | num = state->ret->num; |
484 | if (len < 4 || line[0] != '@' || line[1] != '@') { | |
485 | if (state->hunk_in_pre_context && line[0] == ' ') | |
486 | state->ret->chunks[num - 1].same++; | |
487 | else { | |
488 | state->hunk_in_pre_context = 0; | |
489 | if (line[0] == ' ') | |
490 | state->hunk_post_context++; | |
491 | else | |
492 | state->hunk_post_context = 0; | |
493 | } | |
494 | return; | |
495 | } | |
496 | ||
497 | if (num && state->hunk_post_context) { | |
498 | chunk = &state->ret->chunks[num - 1]; | |
499 | chunk->p_next -= state->hunk_post_context; | |
500 | chunk->t_next -= state->hunk_post_context; | |
501 | } | |
502 | state->ret->num = ++num; | |
503 | state->ret->chunks = xrealloc(state->ret->chunks, | |
504 | sizeof(struct chunk) * num); | |
505 | chunk = &state->ret->chunks[num - 1]; | |
506 | if (parse_hunk_header(line, len, &off1, &len1, &off2, &len2)) { | |
507 | state->ret->num--; | |
508 | return; | |
509 | } | |
510 | ||
511 | /* Line numbers in patch output are one based. */ | |
512 | off1--; | |
513 | off2--; | |
514 | ||
515 | chunk->same = len2 ? off2 : (off2 + 1); | |
516 | ||
517 | chunk->p_next = off1 + (len1 ? len1 : 1); | |
518 | chunk->t_next = chunk->same + len2; | |
519 | state->hunk_in_pre_context = 1; | |
520 | state->hunk_post_context = 0; | |
521 | } | |
522 | ||
523 | static struct patch *compare_buffer(mmfile_t *file_p, mmfile_t *file_o, | |
524 | int context) | |
525 | { | |
526 | struct blame_diff_state state; | |
527 | xpparam_t xpp; | |
528 | xdemitconf_t xecfg; | |
529 | xdemitcb_t ecb; | |
530 | ||
b82871b3 | 531 | xpp.flags = xdl_opts; |
30b25010 | 532 | memset(&xecfg, 0, sizeof(xecfg)); |
cee7f245 | 533 | xecfg.ctxlen = context; |
cee7f245 JH |
534 | ecb.outf = xdiff_outf; |
535 | ecb.priv = &state; | |
536 | memset(&state, 0, sizeof(state)); | |
537 | state.xm.consume = process_u_diff; | |
538 | state.ret = xmalloc(sizeof(struct patch)); | |
539 | state.ret->chunks = NULL; | |
540 | state.ret->num = 0; | |
541 | ||
c279d7e9 | 542 | xdi_diff(file_p, file_o, &xpp, &xecfg, &ecb); |
cee7f245 JH |
543 | |
544 | if (state.ret->num) { | |
545 | struct chunk *chunk; | |
546 | chunk = &state.ret->chunks[state.ret->num - 1]; | |
547 | chunk->p_next -= state.hunk_post_context; | |
548 | chunk->t_next -= state.hunk_post_context; | |
549 | } | |
550 | return state.ret; | |
551 | } | |
552 | ||
1732a1fd JH |
553 | /* |
554 | * Run diff between two origins and grab the patch output, so that | |
555 | * we can pass blame for lines origin is currently suspected for | |
556 | * to its parent. | |
557 | */ | |
cee7f245 JH |
558 | static struct patch *get_patch(struct origin *parent, struct origin *origin) |
559 | { | |
560 | mmfile_t file_p, file_o; | |
cee7f245 JH |
561 | struct patch *patch; |
562 | ||
c2e525d9 JH |
563 | fill_origin_blob(parent, &file_p); |
564 | fill_origin_blob(origin, &file_o); | |
565 | if (!file_p.ptr || !file_o.ptr) | |
cee7f245 | 566 | return NULL; |
cee7f245 | 567 | patch = compare_buffer(&file_p, &file_o, 0); |
c2e525d9 | 568 | num_get_patch++; |
cee7f245 JH |
569 | return patch; |
570 | } | |
571 | ||
572 | static void free_patch(struct patch *p) | |
573 | { | |
574 | free(p->chunks); | |
575 | free(p); | |
576 | } | |
577 | ||
1732a1fd | 578 | /* |
3dff5379 | 579 | * Link in a new blame entry to the scoreboard. Entries that cover the |
1732a1fd JH |
580 | * same line range have been removed from the scoreboard previously. |
581 | */ | |
cee7f245 JH |
582 | static void add_blame_entry(struct scoreboard *sb, struct blame_entry *e) |
583 | { | |
584 | struct blame_entry *ent, *prev = NULL; | |
585 | ||
54a4c617 JH |
586 | origin_incref(e->suspect); |
587 | ||
cee7f245 JH |
588 | for (ent = sb->ent; ent && ent->lno < e->lno; ent = ent->next) |
589 | prev = ent; | |
590 | ||
591 | /* prev, if not NULL, is the last one that is below e */ | |
592 | e->prev = prev; | |
593 | if (prev) { | |
594 | e->next = prev->next; | |
595 | prev->next = e; | |
596 | } | |
597 | else { | |
598 | e->next = sb->ent; | |
599 | sb->ent = e; | |
600 | } | |
601 | if (e->next) | |
602 | e->next->prev = e; | |
603 | } | |
604 | ||
1732a1fd JH |
605 | /* |
606 | * src typically is on-stack; we want to copy the information in it to | |
790296fd | 607 | * a malloced blame_entry that is already on the linked list of the |
1732a1fd JH |
608 | * scoreboard. The origin of dst loses a refcnt while the origin of src |
609 | * gains one. | |
610 | */ | |
cee7f245 JH |
611 | static void dup_entry(struct blame_entry *dst, struct blame_entry *src) |
612 | { | |
613 | struct blame_entry *p, *n; | |
54a4c617 | 614 | |
cee7f245 JH |
615 | p = dst->prev; |
616 | n = dst->next; | |
54a4c617 JH |
617 | origin_incref(src->suspect); |
618 | origin_decref(dst->suspect); | |
cee7f245 JH |
619 | memcpy(dst, src, sizeof(*src)); |
620 | dst->prev = p; | |
621 | dst->next = n; | |
5ff62c30 | 622 | dst->score = 0; |
cee7f245 JH |
623 | } |
624 | ||
625 | static const char *nth_line(struct scoreboard *sb, int lno) | |
626 | { | |
627 | return sb->final_buf + sb->lineno[lno]; | |
628 | } | |
629 | ||
1732a1fd JH |
630 | /* |
631 | * It is known that lines between tlno to same came from parent, and e | |
632 | * has an overlap with that range. it also is known that parent's | |
633 | * line plno corresponds to e's line tlno. | |
634 | * | |
635 | * <---- e -----> | |
636 | * <------> | |
637 | * <------------> | |
638 | * <------------> | |
639 | * <------------------> | |
640 | * | |
641 | * Split e into potentially three parts; before this chunk, the chunk | |
642 | * to be blamed for the parent, and after that portion. | |
643 | */ | |
54a4c617 | 644 | static void split_overlap(struct blame_entry *split, |
cee7f245 JH |
645 | struct blame_entry *e, |
646 | int tlno, int plno, int same, | |
647 | struct origin *parent) | |
648 | { | |
cee7f245 JH |
649 | int chunk_end_lno; |
650 | memset(split, 0, sizeof(struct blame_entry [3])); | |
651 | ||
652 | if (e->s_lno < tlno) { | |
653 | /* there is a pre-chunk part not blamed on parent */ | |
54a4c617 | 654 | split[0].suspect = origin_incref(e->suspect); |
cee7f245 JH |
655 | split[0].lno = e->lno; |
656 | split[0].s_lno = e->s_lno; | |
657 | split[0].num_lines = tlno - e->s_lno; | |
658 | split[1].lno = e->lno + tlno - e->s_lno; | |
659 | split[1].s_lno = plno; | |
660 | } | |
661 | else { | |
662 | split[1].lno = e->lno; | |
663 | split[1].s_lno = plno + (e->s_lno - tlno); | |
664 | } | |
665 | ||
666 | if (same < e->s_lno + e->num_lines) { | |
667 | /* there is a post-chunk part not blamed on parent */ | |
54a4c617 | 668 | split[2].suspect = origin_incref(e->suspect); |
cee7f245 JH |
669 | split[2].lno = e->lno + (same - e->s_lno); |
670 | split[2].s_lno = e->s_lno + (same - e->s_lno); | |
671 | split[2].num_lines = e->s_lno + e->num_lines - same; | |
672 | chunk_end_lno = split[2].lno; | |
673 | } | |
674 | else | |
675 | chunk_end_lno = e->lno + e->num_lines; | |
676 | split[1].num_lines = chunk_end_lno - split[1].lno; | |
677 | ||
1732a1fd JH |
678 | /* |
679 | * if it turns out there is nothing to blame the parent for, | |
680 | * forget about the splitting. !split[1].suspect signals this. | |
681 | */ | |
cee7f245 JH |
682 | if (split[1].num_lines < 1) |
683 | return; | |
54a4c617 | 684 | split[1].suspect = origin_incref(parent); |
cee7f245 JH |
685 | } |
686 | ||
1732a1fd JH |
687 | /* |
688 | * split_overlap() divided an existing blame e into up to three parts | |
689 | * in split. Adjust the linked list of blames in the scoreboard to | |
690 | * reflect the split. | |
691 | */ | |
cee7f245 | 692 | static void split_blame(struct scoreboard *sb, |
54a4c617 | 693 | struct blame_entry *split, |
cee7f245 JH |
694 | struct blame_entry *e) |
695 | { | |
696 | struct blame_entry *new_entry; | |
697 | ||
698 | if (split[0].suspect && split[2].suspect) { | |
1732a1fd | 699 | /* The first part (reuse storage for the existing entry e) */ |
cee7f245 JH |
700 | dup_entry(e, &split[0]); |
701 | ||
1732a1fd | 702 | /* The last part -- me */ |
cee7f245 JH |
703 | new_entry = xmalloc(sizeof(*new_entry)); |
704 | memcpy(new_entry, &(split[2]), sizeof(struct blame_entry)); | |
705 | add_blame_entry(sb, new_entry); | |
706 | ||
1732a1fd | 707 | /* ... and the middle part -- parent */ |
cee7f245 JH |
708 | new_entry = xmalloc(sizeof(*new_entry)); |
709 | memcpy(new_entry, &(split[1]), sizeof(struct blame_entry)); | |
710 | add_blame_entry(sb, new_entry); | |
711 | } | |
712 | else if (!split[0].suspect && !split[2].suspect) | |
1732a1fd JH |
713 | /* |
714 | * The parent covers the entire area; reuse storage for | |
715 | * e and replace it with the parent. | |
716 | */ | |
cee7f245 JH |
717 | dup_entry(e, &split[1]); |
718 | else if (split[0].suspect) { | |
1732a1fd | 719 | /* me and then parent */ |
cee7f245 JH |
720 | dup_entry(e, &split[0]); |
721 | ||
722 | new_entry = xmalloc(sizeof(*new_entry)); | |
723 | memcpy(new_entry, &(split[1]), sizeof(struct blame_entry)); | |
724 | add_blame_entry(sb, new_entry); | |
725 | } | |
726 | else { | |
1732a1fd | 727 | /* parent and then me */ |
cee7f245 JH |
728 | dup_entry(e, &split[1]); |
729 | ||
730 | new_entry = xmalloc(sizeof(*new_entry)); | |
731 | memcpy(new_entry, &(split[2]), sizeof(struct blame_entry)); | |
732 | add_blame_entry(sb, new_entry); | |
733 | } | |
734 | ||
54a4c617 | 735 | if (DEBUG) { /* sanity */ |
cee7f245 | 736 | struct blame_entry *ent; |
612702e8 | 737 | int lno = sb->ent->lno, corrupt = 0; |
cee7f245 JH |
738 | |
739 | for (ent = sb->ent; ent; ent = ent->next) { | |
740 | if (lno != ent->lno) | |
741 | corrupt = 1; | |
742 | if (ent->s_lno < 0) | |
743 | corrupt = 1; | |
744 | lno += ent->num_lines; | |
745 | } | |
746 | if (corrupt) { | |
612702e8 | 747 | lno = sb->ent->lno; |
cee7f245 JH |
748 | for (ent = sb->ent; ent; ent = ent->next) { |
749 | printf("L %8d l %8d n %8d\n", | |
750 | lno, ent->lno, ent->num_lines); | |
751 | lno = ent->lno + ent->num_lines; | |
752 | } | |
753 | die("oops"); | |
754 | } | |
755 | } | |
756 | } | |
757 | ||
1732a1fd JH |
758 | /* |
759 | * After splitting the blame, the origins used by the | |
760 | * on-stack blame_entry should lose one refcnt each. | |
761 | */ | |
54a4c617 JH |
762 | static void decref_split(struct blame_entry *split) |
763 | { | |
764 | int i; | |
765 | ||
766 | for (i = 0; i < 3; i++) | |
767 | origin_decref(split[i].suspect); | |
768 | } | |
769 | ||
1732a1fd JH |
770 | /* |
771 | * Helper for blame_chunk(). blame_entry e is known to overlap with | |
772 | * the patch hunk; split it and pass blame to the parent. | |
773 | */ | |
cee7f245 JH |
774 | static void blame_overlap(struct scoreboard *sb, struct blame_entry *e, |
775 | int tlno, int plno, int same, | |
776 | struct origin *parent) | |
777 | { | |
778 | struct blame_entry split[3]; | |
779 | ||
780 | split_overlap(split, e, tlno, plno, same, parent); | |
54a4c617 JH |
781 | if (split[1].suspect) |
782 | split_blame(sb, split, e); | |
783 | decref_split(split); | |
cee7f245 JH |
784 | } |
785 | ||
1732a1fd JH |
786 | /* |
787 | * Find the line number of the last line the target is suspected for. | |
788 | */ | |
cee7f245 JH |
789 | static int find_last_in_target(struct scoreboard *sb, struct origin *target) |
790 | { | |
791 | struct blame_entry *e; | |
792 | int last_in_target = -1; | |
793 | ||
794 | for (e = sb->ent; e; e = e->next) { | |
171dccd5 | 795 | if (e->guilty || !same_suspect(e->suspect, target)) |
cee7f245 JH |
796 | continue; |
797 | if (last_in_target < e->s_lno + e->num_lines) | |
798 | last_in_target = e->s_lno + e->num_lines; | |
799 | } | |
800 | return last_in_target; | |
801 | } | |
802 | ||
1732a1fd JH |
803 | /* |
804 | * Process one hunk from the patch between the current suspect for | |
805 | * blame_entry e and its parent. Find and split the overlap, and | |
806 | * pass blame to the overlapping part to the parent. | |
807 | */ | |
cee7f245 JH |
808 | static void blame_chunk(struct scoreboard *sb, |
809 | int tlno, int plno, int same, | |
810 | struct origin *target, struct origin *parent) | |
811 | { | |
612702e8 | 812 | struct blame_entry *e; |
cee7f245 | 813 | |
612702e8 | 814 | for (e = sb->ent; e; e = e->next) { |
171dccd5 | 815 | if (e->guilty || !same_suspect(e->suspect, target)) |
cee7f245 JH |
816 | continue; |
817 | if (same <= e->s_lno) | |
818 | continue; | |
819 | if (tlno < e->s_lno + e->num_lines) | |
820 | blame_overlap(sb, e, tlno, plno, same, parent); | |
821 | } | |
822 | } | |
823 | ||
1732a1fd JH |
824 | /* |
825 | * We are looking at the origin 'target' and aiming to pass blame | |
826 | * for the lines it is suspected to its parent. Run diff to find | |
827 | * which lines came from parent and pass blame for them. | |
828 | */ | |
cee7f245 JH |
829 | static int pass_blame_to_parent(struct scoreboard *sb, |
830 | struct origin *target, | |
831 | struct origin *parent) | |
832 | { | |
833 | int i, last_in_target, plno, tlno; | |
834 | struct patch *patch; | |
835 | ||
836 | last_in_target = find_last_in_target(sb, target); | |
837 | if (last_in_target < 0) | |
838 | return 1; /* nothing remains for this target */ | |
839 | ||
840 | patch = get_patch(parent, target); | |
841 | plno = tlno = 0; | |
842 | for (i = 0; i < patch->num; i++) { | |
843 | struct chunk *chunk = &patch->chunks[i]; | |
844 | ||
cee7f245 JH |
845 | blame_chunk(sb, tlno, plno, chunk->same, target, parent); |
846 | plno = chunk->p_next; | |
847 | tlno = chunk->t_next; | |
848 | } | |
1732a1fd | 849 | /* The rest (i.e. anything after tlno) are the same as the parent */ |
cee7f245 JH |
850 | blame_chunk(sb, tlno, plno, last_in_target, target, parent); |
851 | ||
852 | free_patch(patch); | |
853 | return 0; | |
854 | } | |
855 | ||
1732a1fd JH |
856 | /* |
857 | * The lines in blame_entry after splitting blames many times can become | |
858 | * very small and trivial, and at some point it becomes pointless to | |
859 | * blame the parents. E.g. "\t\t}\n\t}\n\n" appears everywhere in any | |
860 | * ordinary C program, and it is not worth to say it was copied from | |
861 | * totally unrelated file in the parent. | |
862 | * | |
863 | * Compute how trivial the lines in the blame_entry are. | |
864 | */ | |
5ff62c30 JH |
865 | static unsigned ent_score(struct scoreboard *sb, struct blame_entry *e) |
866 | { | |
867 | unsigned score; | |
868 | const char *cp, *ep; | |
869 | ||
870 | if (e->score) | |
871 | return e->score; | |
872 | ||
612702e8 | 873 | score = 1; |
5ff62c30 JH |
874 | cp = nth_line(sb, e->lno); |
875 | ep = nth_line(sb, e->lno + e->num_lines); | |
876 | while (cp < ep) { | |
877 | unsigned ch = *((unsigned char *)cp); | |
878 | if (isalnum(ch)) | |
879 | score++; | |
880 | cp++; | |
881 | } | |
882 | e->score = score; | |
883 | return score; | |
884 | } | |
885 | ||
1732a1fd JH |
886 | /* |
887 | * best_so_far[] and this[] are both a split of an existing blame_entry | |
888 | * that passes blame to the parent. Maintain best_so_far the best split | |
889 | * so far, by comparing this and best_so_far and copying this into | |
890 | * bst_so_far as needed. | |
891 | */ | |
5ff62c30 | 892 | static void copy_split_if_better(struct scoreboard *sb, |
54a4c617 JH |
893 | struct blame_entry *best_so_far, |
894 | struct blame_entry *this) | |
d24bba80 | 895 | { |
54a4c617 JH |
896 | int i; |
897 | ||
d24bba80 JH |
898 | if (!this[1].suspect) |
899 | return; | |
5ff62c30 JH |
900 | if (best_so_far[1].suspect) { |
901 | if (ent_score(sb, &this[1]) < ent_score(sb, &best_so_far[1])) | |
902 | return; | |
903 | } | |
54a4c617 JH |
904 | |
905 | for (i = 0; i < 3; i++) | |
906 | origin_incref(this[i].suspect); | |
907 | decref_split(best_so_far); | |
d24bba80 JH |
908 | memcpy(best_so_far, this, sizeof(struct blame_entry [3])); |
909 | } | |
910 | ||
dd166aa8 JH |
911 | /* |
912 | * We are looking at a part of the final image represented by | |
913 | * ent (tlno and same are offset by ent->s_lno). | |
914 | * tlno is where we are looking at in the final image. | |
915 | * up to (but not including) same match preimage. | |
916 | * plno is where we are looking at in the preimage. | |
917 | * | |
918 | * <-------------- final image ----------------------> | |
919 | * <------ent------> | |
920 | * ^tlno ^same | |
921 | * <---------preimage-----> | |
922 | * ^plno | |
923 | * | |
924 | * All line numbers are 0-based. | |
925 | */ | |
926 | static void handle_split(struct scoreboard *sb, | |
927 | struct blame_entry *ent, | |
928 | int tlno, int plno, int same, | |
929 | struct origin *parent, | |
930 | struct blame_entry *split) | |
931 | { | |
932 | if (ent->num_lines <= tlno) | |
933 | return; | |
934 | if (tlno < same) { | |
935 | struct blame_entry this[3]; | |
936 | tlno += ent->s_lno; | |
937 | same += ent->s_lno; | |
938 | split_overlap(this, ent, tlno, plno, same, parent); | |
939 | copy_split_if_better(sb, split, this); | |
940 | decref_split(this); | |
941 | } | |
942 | } | |
943 | ||
1732a1fd JH |
944 | /* |
945 | * Find the lines from parent that are the same as ent so that | |
946 | * we can pass blames to it. file_p has the blob contents for | |
947 | * the parent. | |
948 | */ | |
d24bba80 JH |
949 | static void find_copy_in_blob(struct scoreboard *sb, |
950 | struct blame_entry *ent, | |
951 | struct origin *parent, | |
54a4c617 | 952 | struct blame_entry *split, |
d24bba80 JH |
953 | mmfile_t *file_p) |
954 | { | |
955 | const char *cp; | |
956 | int cnt; | |
957 | mmfile_t file_o; | |
958 | struct patch *patch; | |
959 | int i, plno, tlno; | |
960 | ||
1732a1fd JH |
961 | /* |
962 | * Prepare mmfile that contains only the lines in ent. | |
963 | */ | |
d24bba80 JH |
964 | cp = nth_line(sb, ent->lno); |
965 | file_o.ptr = (char*) cp; | |
966 | cnt = ent->num_lines; | |
967 | ||
968 | while (cnt && cp < sb->final_buf + sb->final_buf_size) { | |
969 | if (*cp++ == '\n') | |
970 | cnt--; | |
971 | } | |
972 | file_o.size = cp - file_o.ptr; | |
973 | ||
974 | patch = compare_buffer(file_p, &file_o, 1); | |
975 | ||
dd166aa8 JH |
976 | /* |
977 | * file_o is a part of final image we are annotating. | |
978 | * file_p partially may match that image. | |
979 | */ | |
d24bba80 JH |
980 | memset(split, 0, sizeof(struct blame_entry [3])); |
981 | plno = tlno = 0; | |
982 | for (i = 0; i < patch->num; i++) { | |
983 | struct chunk *chunk = &patch->chunks[i]; | |
984 | ||
dd166aa8 | 985 | handle_split(sb, ent, tlno, plno, chunk->same, parent, split); |
d24bba80 JH |
986 | plno = chunk->p_next; |
987 | tlno = chunk->t_next; | |
988 | } | |
dd166aa8 JH |
989 | /* remainder, if any, all match the preimage */ |
990 | handle_split(sb, ent, tlno, plno, ent->num_lines, parent, split); | |
d24bba80 JH |
991 | free_patch(patch); |
992 | } | |
993 | ||
1732a1fd JH |
994 | /* |
995 | * See if lines currently target is suspected for can be attributed to | |
996 | * parent. | |
997 | */ | |
d24bba80 JH |
998 | static int find_move_in_parent(struct scoreboard *sb, |
999 | struct origin *target, | |
1000 | struct origin *parent) | |
1001 | { | |
650e2f67 | 1002 | int last_in_target, made_progress; |
46014766 | 1003 | struct blame_entry *e, split[3]; |
d24bba80 | 1004 | mmfile_t file_p; |
d24bba80 JH |
1005 | |
1006 | last_in_target = find_last_in_target(sb, target); | |
1007 | if (last_in_target < 0) | |
1008 | return 1; /* nothing remains for this target */ | |
1009 | ||
c2e525d9 JH |
1010 | fill_origin_blob(parent, &file_p); |
1011 | if (!file_p.ptr) | |
d24bba80 | 1012 | return 0; |
d24bba80 | 1013 | |
650e2f67 JH |
1014 | made_progress = 1; |
1015 | while (made_progress) { | |
1016 | made_progress = 0; | |
1017 | for (e = sb->ent; e; e = e->next) { | |
171dccd5 | 1018 | if (e->guilty || !same_suspect(e->suspect, target)) |
650e2f67 JH |
1019 | continue; |
1020 | find_copy_in_blob(sb, e, parent, split, &file_p); | |
1021 | if (split[1].suspect && | |
1022 | blame_move_score < ent_score(sb, &split[1])) { | |
1023 | split_blame(sb, split, e); | |
1024 | made_progress = 1; | |
1025 | } | |
1026 | decref_split(split); | |
1027 | } | |
d24bba80 | 1028 | } |
d24bba80 JH |
1029 | return 0; |
1030 | } | |
1031 | ||
33494784 JH |
1032 | struct blame_list { |
1033 | struct blame_entry *ent; | |
1034 | struct blame_entry split[3]; | |
1035 | }; | |
1036 | ||
1732a1fd JH |
1037 | /* |
1038 | * Count the number of entries the target is suspected for, | |
1039 | * and prepare a list of entry and the best split. | |
1040 | */ | |
33494784 JH |
1041 | static struct blame_list *setup_blame_list(struct scoreboard *sb, |
1042 | struct origin *target, | |
1043 | int *num_ents_p) | |
1044 | { | |
1045 | struct blame_entry *e; | |
1046 | int num_ents, i; | |
1047 | struct blame_list *blame_list = NULL; | |
1048 | ||
33494784 | 1049 | for (e = sb->ent, num_ents = 0; e; e = e->next) |
171dccd5 | 1050 | if (!e->guilty && same_suspect(e->suspect, target)) |
33494784 JH |
1051 | num_ents++; |
1052 | if (num_ents) { | |
1053 | blame_list = xcalloc(num_ents, sizeof(struct blame_list)); | |
1054 | for (e = sb->ent, i = 0; e; e = e->next) | |
171dccd5 | 1055 | if (!e->guilty && same_suspect(e->suspect, target)) |
33494784 JH |
1056 | blame_list[i++].ent = e; |
1057 | } | |
1058 | *num_ents_p = num_ents; | |
1059 | return blame_list; | |
1060 | } | |
1061 | ||
1732a1fd JH |
1062 | /* |
1063 | * For lines target is suspected for, see if we can find code movement | |
1064 | * across file boundary from the parent commit. porigin is the path | |
1065 | * in the parent we already tried. | |
1066 | */ | |
18abd745 JH |
1067 | static int find_copy_in_parent(struct scoreboard *sb, |
1068 | struct origin *target, | |
1069 | struct commit *parent, | |
1070 | struct origin *porigin, | |
1071 | int opt) | |
1072 | { | |
1073 | struct diff_options diff_opts; | |
1074 | const char *paths[1]; | |
aec8fa1f | 1075 | int i, j; |
33494784 JH |
1076 | int retval; |
1077 | struct blame_list *blame_list; | |
aec8fa1f | 1078 | int num_ents; |
18abd745 | 1079 | |
33494784 JH |
1080 | blame_list = setup_blame_list(sb, target, &num_ents); |
1081 | if (!blame_list) | |
18abd745 JH |
1082 | return 1; /* nothing remains for this target */ |
1083 | ||
1084 | diff_setup(&diff_opts); | |
8f67f8ae | 1085 | DIFF_OPT_SET(&diff_opts, RECURSIVE); |
18abd745 JH |
1086 | diff_opts.output_format = DIFF_FORMAT_NO_OUTPUT; |
1087 | ||
18abd745 JH |
1088 | paths[0] = NULL; |
1089 | diff_tree_setup_paths(paths, &diff_opts); | |
1090 | if (diff_setup_done(&diff_opts) < 0) | |
1091 | die("diff-setup"); | |
33494784 JH |
1092 | |
1093 | /* Try "find copies harder" on new path if requested; | |
1094 | * we do not want to use diffcore_rename() actually to | |
1095 | * match things up; find_copies_harder is set only to | |
1096 | * force diff_tree_sha1() to feed all filepairs to diff_queue, | |
1097 | * and this code needs to be after diff_setup_done(), which | |
1098 | * usually makes find-copies-harder imply copy detection. | |
1099 | */ | |
c63777c0 JH |
1100 | if ((opt & PICKAXE_BLAME_COPY_HARDEST) |
1101 | || ((opt & PICKAXE_BLAME_COPY_HARDER) | |
1102 | && (!porigin || strcmp(target->path, porigin->path)))) | |
8f67f8ae | 1103 | DIFF_OPT_SET(&diff_opts, FIND_COPIES_HARDER); |
33494784 | 1104 | |
1cfe7733 JH |
1105 | if (is_null_sha1(target->commit->object.sha1)) |
1106 | do_diff_cache(parent->tree->object.sha1, &diff_opts); | |
1107 | else | |
1108 | diff_tree_sha1(parent->tree->object.sha1, | |
1109 | target->commit->tree->object.sha1, | |
1110 | "", &diff_opts); | |
18abd745 | 1111 | |
8f67f8ae | 1112 | if (!DIFF_OPT_TST(&diff_opts, FIND_COPIES_HARDER)) |
33494784 | 1113 | diffcore_std(&diff_opts); |
18abd745 | 1114 | |
33494784 JH |
1115 | retval = 0; |
1116 | while (1) { | |
1117 | int made_progress = 0; | |
1118 | ||
1119 | for (i = 0; i < diff_queued_diff.nr; i++) { | |
1120 | struct diff_filepair *p = diff_queued_diff.queue[i]; | |
1121 | struct origin *norigin; | |
1122 | mmfile_t file_p; | |
33494784 JH |
1123 | struct blame_entry this[3]; |
1124 | ||
1125 | if (!DIFF_FILE_VALID(p->one)) | |
1126 | continue; /* does not exist in parent */ | |
1127 | if (porigin && !strcmp(p->one->path, porigin->path)) | |
1128 | /* find_move already dealt with this path */ | |
1129 | continue; | |
1130 | ||
1131 | norigin = get_origin(sb, parent, p->one->path); | |
1132 | hashcpy(norigin->blob_sha1, p->one->sha1); | |
c2e525d9 | 1133 | fill_origin_blob(norigin, &file_p); |
33494784 JH |
1134 | if (!file_p.ptr) |
1135 | continue; | |
1136 | ||
1137 | for (j = 0; j < num_ents; j++) { | |
1138 | find_copy_in_blob(sb, blame_list[j].ent, | |
1139 | norigin, this, &file_p); | |
1140 | copy_split_if_better(sb, blame_list[j].split, | |
1141 | this); | |
1142 | decref_split(this); | |
1143 | } | |
33494784 JH |
1144 | origin_decref(norigin); |
1145 | } | |
18abd745 | 1146 | |
aec8fa1f | 1147 | for (j = 0; j < num_ents; j++) { |
33494784 JH |
1148 | struct blame_entry *split = blame_list[j].split; |
1149 | if (split[1].suspect && | |
1150 | blame_copy_score < ent_score(sb, &split[1])) { | |
1151 | split_blame(sb, split, blame_list[j].ent); | |
1152 | made_progress = 1; | |
1153 | } | |
1154 | decref_split(split); | |
18abd745 | 1155 | } |
33494784 | 1156 | free(blame_list); |
aec8fa1f | 1157 | |
33494784 JH |
1158 | if (!made_progress) |
1159 | break; | |
1160 | blame_list = setup_blame_list(sb, target, &num_ents); | |
1161 | if (!blame_list) { | |
1162 | retval = 1; | |
1163 | break; | |
1164 | } | |
18abd745 | 1165 | } |
33494784 | 1166 | diff_flush(&diff_opts); |
03b69c76 | 1167 | diff_tree_release_paths(&diff_opts); |
33494784 | 1168 | return retval; |
18abd745 JH |
1169 | } |
1170 | ||
1732a1fd JH |
1171 | /* |
1172 | * The blobs of origin and porigin exactly match, so everything | |
c2e525d9 JH |
1173 | * origin is suspected for can be blamed on the parent. |
1174 | */ | |
1175 | static void pass_whole_blame(struct scoreboard *sb, | |
1176 | struct origin *origin, struct origin *porigin) | |
1177 | { | |
1178 | struct blame_entry *e; | |
1179 | ||
1180 | if (!porigin->file.ptr && origin->file.ptr) { | |
1181 | /* Steal its file */ | |
1182 | porigin->file = origin->file; | |
1183 | origin->file.ptr = NULL; | |
1184 | } | |
1185 | for (e = sb->ent; e; e = e->next) { | |
171dccd5 | 1186 | if (!same_suspect(e->suspect, origin)) |
c2e525d9 JH |
1187 | continue; |
1188 | origin_incref(porigin); | |
1189 | origin_decref(e->suspect); | |
1190 | e->suspect = porigin; | |
1191 | } | |
1192 | } | |
1193 | ||
69264f46 JH |
1194 | /* |
1195 | * We pass blame from the current commit to its parents. We keep saying | |
1196 | * "parent" (and "porigin"), but what we mean is to find scapegoat to | |
1197 | * exonerate ourselves. | |
1198 | */ | |
1199 | static struct commit_list *first_scapegoat(struct commit *commit) | |
1200 | { | |
1201 | return commit->parents; | |
1202 | } | |
1203 | ||
1204 | static int num_scapegoats(struct commit *commit) | |
1205 | { | |
1206 | int cnt; | |
1207 | struct commit_list *l = first_scapegoat(commit); | |
1208 | for (cnt = 0; l; l = l->next) | |
1209 | cnt++; | |
1210 | return cnt; | |
1211 | } | |
1212 | ||
1213 | #define MAXSG 16 | |
cee7f245 | 1214 | |
d24bba80 | 1215 | static void pass_blame(struct scoreboard *sb, struct origin *origin, int opt) |
cee7f245 | 1216 | { |
69264f46 | 1217 | int i, pass, num_sg; |
cee7f245 | 1218 | struct commit *commit = origin->commit; |
69264f46 JH |
1219 | struct commit_list *sg; |
1220 | struct origin *sg_buf[MAXSG]; | |
1221 | struct origin *porigin, **sg_origin = sg_buf; | |
1222 | ||
1223 | num_sg = num_scapegoats(commit); | |
1224 | if (!num_sg) | |
1225 | goto finish; | |
1226 | else if (num_sg < ARRAY_SIZE(sg_buf)) | |
1227 | memset(sg_buf, 0, sizeof(sg_buf)); | |
1228 | else | |
1229 | sg_origin = xcalloc(num_sg, sizeof(*sg_origin)); | |
cee7f245 | 1230 | |
69264f46 JH |
1231 | /* |
1232 | * The first pass looks for unrenamed path to optimize for | |
f69e743d JH |
1233 | * common cases, then we look for renames in the second pass. |
1234 | */ | |
1235 | for (pass = 0; pass < 2; pass++) { | |
1236 | struct origin *(*find)(struct scoreboard *, | |
1237 | struct commit *, struct origin *); | |
1238 | find = pass ? find_rename : find_origin; | |
1239 | ||
69264f46 JH |
1240 | for (i = 0, sg = first_scapegoat(commit); |
1241 | i < num_sg && sg; | |
1242 | sg = sg->next, i++) { | |
1243 | struct commit *p = sg->item; | |
0421d9f8 | 1244 | int j, same; |
f69e743d | 1245 | |
69264f46 | 1246 | if (sg_origin[i]) |
f69e743d JH |
1247 | continue; |
1248 | if (parse_commit(p)) | |
1249 | continue; | |
0d981c67 | 1250 | porigin = find(sb, p, origin); |
f69e743d JH |
1251 | if (!porigin) |
1252 | continue; | |
1253 | if (!hashcmp(porigin->blob_sha1, origin->blob_sha1)) { | |
c2e525d9 | 1254 | pass_whole_blame(sb, origin, porigin); |
f69e743d JH |
1255 | origin_decref(porigin); |
1256 | goto finish; | |
1257 | } | |
0421d9f8 | 1258 | for (j = same = 0; j < i; j++) |
69264f46 JH |
1259 | if (sg_origin[j] && |
1260 | !hashcmp(sg_origin[j]->blob_sha1, | |
0421d9f8 JH |
1261 | porigin->blob_sha1)) { |
1262 | same = 1; | |
1263 | break; | |
1264 | } | |
1265 | if (!same) | |
69264f46 | 1266 | sg_origin[i] = porigin; |
0421d9f8 JH |
1267 | else |
1268 | origin_decref(porigin); | |
cee7f245 | 1269 | } |
cee7f245 JH |
1270 | } |
1271 | ||
c2e525d9 | 1272 | num_commits++; |
69264f46 JH |
1273 | for (i = 0, sg = first_scapegoat(commit); |
1274 | i < num_sg && sg; | |
1275 | sg = sg->next, i++) { | |
1276 | struct origin *porigin = sg_origin[i]; | |
cee7f245 JH |
1277 | if (!porigin) |
1278 | continue; | |
1279 | if (pass_blame_to_parent(sb, origin, porigin)) | |
54a4c617 | 1280 | goto finish; |
cee7f245 | 1281 | } |
d24bba80 JH |
1282 | |
1283 | /* | |
1732a1fd | 1284 | * Optionally find moves in parents' files. |
d24bba80 JH |
1285 | */ |
1286 | if (opt & PICKAXE_BLAME_MOVE) | |
69264f46 JH |
1287 | for (i = 0, sg = first_scapegoat(commit); |
1288 | i < num_sg && sg; | |
1289 | sg = sg->next, i++) { | |
1290 | struct origin *porigin = sg_origin[i]; | |
d24bba80 JH |
1291 | if (!porigin) |
1292 | continue; | |
1293 | if (find_move_in_parent(sb, origin, porigin)) | |
54a4c617 | 1294 | goto finish; |
d24bba80 JH |
1295 | } |
1296 | ||
18abd745 | 1297 | /* |
1732a1fd | 1298 | * Optionally find copies from parents' files. |
18abd745 JH |
1299 | */ |
1300 | if (opt & PICKAXE_BLAME_COPY) | |
69264f46 JH |
1301 | for (i = 0, sg = first_scapegoat(commit); |
1302 | i < num_sg && sg; | |
1303 | sg = sg->next, i++) { | |
1304 | struct origin *porigin = sg_origin[i]; | |
1305 | if (find_copy_in_parent(sb, origin, sg->item, | |
18abd745 | 1306 | porigin, opt)) |
54a4c617 | 1307 | goto finish; |
18abd745 | 1308 | } |
54a4c617 JH |
1309 | |
1310 | finish: | |
69264f46 JH |
1311 | for (i = 0; i < num_sg; i++) { |
1312 | if (sg_origin[i]) { | |
1313 | drop_origin_blob(sg_origin[i]); | |
1314 | origin_decref(sg_origin[i]); | |
7c3c7962 JH |
1315 | } |
1316 | } | |
1317 | drop_origin_blob(origin); | |
69264f46 JH |
1318 | if (sg_buf != sg_origin) |
1319 | free(sg_origin); | |
cee7f245 JH |
1320 | } |
1321 | ||
1732a1fd JH |
1322 | /* |
1323 | * Information on commits, used for output. | |
1324 | */ | |
cee7f245 JH |
1325 | struct commit_info |
1326 | { | |
3a55602e SP |
1327 | const char *author; |
1328 | const char *author_mail; | |
cee7f245 | 1329 | unsigned long author_time; |
3a55602e | 1330 | const char *author_tz; |
cee7f245 JH |
1331 | |
1332 | /* filled only when asked for details */ | |
3a55602e SP |
1333 | const char *committer; |
1334 | const char *committer_mail; | |
cee7f245 | 1335 | unsigned long committer_time; |
3a55602e | 1336 | const char *committer_tz; |
cee7f245 | 1337 | |
3a55602e | 1338 | const char *summary; |
cee7f245 JH |
1339 | }; |
1340 | ||
1732a1fd JH |
1341 | /* |
1342 | * Parse author/committer line in the commit object buffer | |
1343 | */ | |
cee7f245 | 1344 | static void get_ac_line(const char *inbuf, const char *what, |
3a55602e SP |
1345 | int bufsz, char *person, const char **mail, |
1346 | unsigned long *time, const char **tz) | |
cee7f245 | 1347 | { |
f9567384 JH |
1348 | int len, tzlen, maillen; |
1349 | char *tmp, *endp, *timepos; | |
cee7f245 JH |
1350 | |
1351 | tmp = strstr(inbuf, what); | |
1352 | if (!tmp) | |
1353 | goto error_out; | |
1354 | tmp += strlen(what); | |
1355 | endp = strchr(tmp, '\n'); | |
1356 | if (!endp) | |
1357 | len = strlen(tmp); | |
1358 | else | |
1359 | len = endp - tmp; | |
1360 | if (bufsz <= len) { | |
1361 | error_out: | |
1362 | /* Ugh */ | |
3a55602e | 1363 | *mail = *tz = "(unknown)"; |
cee7f245 JH |
1364 | *time = 0; |
1365 | return; | |
1366 | } | |
1367 | memcpy(person, tmp, len); | |
1368 | ||
1369 | tmp = person; | |
1370 | tmp += len; | |
1371 | *tmp = 0; | |
1372 | while (*tmp != ' ') | |
1373 | tmp--; | |
1374 | *tz = tmp+1; | |
f9567384 | 1375 | tzlen = (person+len)-(tmp+1); |
cee7f245 JH |
1376 | |
1377 | *tmp = 0; | |
1378 | while (*tmp != ' ') | |
1379 | tmp--; | |
1380 | *time = strtoul(tmp, NULL, 10); | |
f9567384 | 1381 | timepos = tmp; |
cee7f245 JH |
1382 | |
1383 | *tmp = 0; | |
1384 | while (*tmp != ' ') | |
1385 | tmp--; | |
1386 | *mail = tmp + 1; | |
1387 | *tmp = 0; | |
f9567384 JH |
1388 | maillen = timepos - tmp; |
1389 | ||
1390 | if (!mailmap.nr) | |
1391 | return; | |
1392 | ||
1393 | /* | |
1394 | * mailmap expansion may make the name longer. | |
1395 | * make room by pushing stuff down. | |
1396 | */ | |
1397 | tmp = person + bufsz - (tzlen + 1); | |
1398 | memmove(tmp, *tz, tzlen); | |
1399 | tmp[tzlen] = 0; | |
1400 | *tz = tmp; | |
1401 | ||
1402 | tmp = tmp - (maillen + 1); | |
1403 | memmove(tmp, *mail, maillen); | |
1404 | tmp[maillen] = 0; | |
1405 | *mail = tmp; | |
1406 | ||
1407 | /* | |
1408 | * Now, convert e-mail using mailmap | |
1409 | */ | |
1410 | map_email(&mailmap, tmp + 1, person, tmp-person-1); | |
cee7f245 JH |
1411 | } |
1412 | ||
1413 | static void get_commit_info(struct commit *commit, | |
1414 | struct commit_info *ret, | |
1415 | int detailed) | |
1416 | { | |
1417 | int len; | |
1418 | char *tmp, *endp; | |
1419 | static char author_buf[1024]; | |
1420 | static char committer_buf[1024]; | |
1421 | static char summary_buf[1024]; | |
1422 | ||
1732a1fd JH |
1423 | /* |
1424 | * We've operated without save_commit_buffer, so | |
612702e8 JH |
1425 | * we now need to populate them for output. |
1426 | */ | |
1427 | if (!commit->buffer) { | |
21666f1a | 1428 | enum object_type type; |
612702e8 JH |
1429 | unsigned long size; |
1430 | commit->buffer = | |
21666f1a | 1431 | read_sha1_file(commit->object.sha1, &type, &size); |
ab43e495 JH |
1432 | if (!commit->buffer) |
1433 | die("Cannot read commit %s", | |
1434 | sha1_to_hex(commit->object.sha1)); | |
612702e8 | 1435 | } |
cee7f245 JH |
1436 | ret->author = author_buf; |
1437 | get_ac_line(commit->buffer, "\nauthor ", | |
1438 | sizeof(author_buf), author_buf, &ret->author_mail, | |
1439 | &ret->author_time, &ret->author_tz); | |
1440 | ||
1441 | if (!detailed) | |
1442 | return; | |
1443 | ||
1444 | ret->committer = committer_buf; | |
1445 | get_ac_line(commit->buffer, "\ncommitter ", | |
1446 | sizeof(committer_buf), committer_buf, &ret->committer_mail, | |
1447 | &ret->committer_time, &ret->committer_tz); | |
1448 | ||
1449 | ret->summary = summary_buf; | |
1450 | tmp = strstr(commit->buffer, "\n\n"); | |
1451 | if (!tmp) { | |
1452 | error_out: | |
1453 | sprintf(summary_buf, "(%s)", sha1_to_hex(commit->object.sha1)); | |
1454 | return; | |
1455 | } | |
1456 | tmp += 2; | |
1457 | endp = strchr(tmp, '\n'); | |
1458 | if (!endp) | |
1cfe7733 | 1459 | endp = tmp + strlen(tmp); |
cee7f245 | 1460 | len = endp - tmp; |
1cfe7733 | 1461 | if (len >= sizeof(summary_buf) || len == 0) |
cee7f245 JH |
1462 | goto error_out; |
1463 | memcpy(summary_buf, tmp, len); | |
1464 | summary_buf[len] = 0; | |
1465 | } | |
1466 | ||
1732a1fd JH |
1467 | /* |
1468 | * To allow LF and other nonportable characters in pathnames, | |
1469 | * they are c-style quoted as needed. | |
1470 | */ | |
46e5e69d JH |
1471 | static void write_filename_info(const char *path) |
1472 | { | |
1473 | printf("filename "); | |
663af342 | 1474 | write_name_quoted(path, stdout, '\n'); |
46e5e69d JH |
1475 | } |
1476 | ||
1732a1fd JH |
1477 | /* |
1478 | * The blame_entry is found to be guilty for the range. Mark it | |
1479 | * as such, and show it in incremental output. | |
1480 | */ | |
717d1462 LT |
1481 | static void found_guilty_entry(struct blame_entry *ent) |
1482 | { | |
1483 | if (ent->guilty) | |
1484 | return; | |
1485 | ent->guilty = 1; | |
1486 | if (incremental) { | |
1487 | struct origin *suspect = ent->suspect; | |
1488 | ||
1489 | printf("%s %d %d %d\n", | |
1490 | sha1_to_hex(suspect->commit->object.sha1), | |
1491 | ent->s_lno + 1, ent->lno + 1, ent->num_lines); | |
1492 | if (!(suspect->commit->object.flags & METAINFO_SHOWN)) { | |
1493 | struct commit_info ci; | |
1494 | suspect->commit->object.flags |= METAINFO_SHOWN; | |
1495 | get_commit_info(suspect->commit, &ci, 1); | |
1496 | printf("author %s\n", ci.author); | |
1497 | printf("author-mail %s\n", ci.author_mail); | |
1498 | printf("author-time %lu\n", ci.author_time); | |
1499 | printf("author-tz %s\n", ci.author_tz); | |
1500 | printf("committer %s\n", ci.committer); | |
1501 | printf("committer-mail %s\n", ci.committer_mail); | |
1502 | printf("committer-time %lu\n", ci.committer_time); | |
1503 | printf("committer-tz %s\n", ci.committer_tz); | |
1504 | printf("summary %s\n", ci.summary); | |
1505 | if (suspect->commit->object.flags & UNINTERESTING) | |
1506 | printf("boundary\n"); | |
1507 | } | |
46e5e69d | 1508 | write_filename_info(suspect->path); |
06f59e9f | 1509 | maybe_flush_or_die(stdout, "stdout"); |
717d1462 LT |
1510 | } |
1511 | } | |
1512 | ||
1732a1fd JH |
1513 | /* |
1514 | * The main loop -- while the scoreboard has lines whose true origin | |
3dff5379 | 1515 | * is still unknown, pick one blame_entry, and allow its current |
1732a1fd JH |
1516 | * suspect to pass blames to its parents. |
1517 | */ | |
717d1462 LT |
1518 | static void assign_blame(struct scoreboard *sb, struct rev_info *revs, int opt) |
1519 | { | |
1520 | while (1) { | |
1521 | struct blame_entry *ent; | |
1522 | struct commit *commit; | |
1523 | struct origin *suspect = NULL; | |
1524 | ||
1525 | /* find one suspect to break down */ | |
1526 | for (ent = sb->ent; !suspect && ent; ent = ent->next) | |
1527 | if (!ent->guilty) | |
1528 | suspect = ent->suspect; | |
1529 | if (!suspect) | |
1530 | return; /* all done */ | |
1531 | ||
1732a1fd JH |
1532 | /* |
1533 | * We will use this suspect later in the loop, | |
1534 | * so hold onto it in the meantime. | |
1535 | */ | |
717d1462 LT |
1536 | origin_incref(suspect); |
1537 | commit = suspect->commit; | |
1538 | if (!commit->object.parsed) | |
1539 | parse_commit(commit); | |
1540 | if (!(commit->object.flags & UNINTERESTING) && | |
1732a1fd | 1541 | !(revs->max_age != -1 && commit->date < revs->max_age)) |
717d1462 LT |
1542 | pass_blame(sb, suspect, opt); |
1543 | else { | |
1544 | commit->object.flags |= UNINTERESTING; | |
1545 | if (commit->object.parsed) | |
1546 | mark_parents_uninteresting(commit); | |
1547 | } | |
1548 | /* treat root commit as boundary */ | |
1549 | if (!commit->parents && !show_root) | |
1550 | commit->object.flags |= UNINTERESTING; | |
1551 | ||
1552 | /* Take responsibility for the remaining entries */ | |
1553 | for (ent = sb->ent; ent; ent = ent->next) | |
171dccd5 | 1554 | if (same_suspect(ent->suspect, suspect)) |
717d1462 LT |
1555 | found_guilty_entry(ent); |
1556 | origin_decref(suspect); | |
1557 | ||
1558 | if (DEBUG) /* sanity */ | |
1559 | sanity_check_refcnt(sb); | |
1560 | } | |
1561 | } | |
1562 | ||
1563 | static const char *format_time(unsigned long time, const char *tz_str, | |
1564 | int show_raw_time) | |
1565 | { | |
1566 | static char time_buf[128]; | |
1567 | time_t t = time; | |
1568 | int minutes, tz; | |
1569 | struct tm *tm; | |
1570 | ||
1571 | if (show_raw_time) { | |
1572 | sprintf(time_buf, "%lu %s", time, tz_str); | |
1573 | return time_buf; | |
1574 | } | |
1575 | ||
1576 | tz = atoi(tz_str); | |
1577 | minutes = tz < 0 ? -tz : tz; | |
1578 | minutes = (minutes / 100)*60 + (minutes % 100); | |
1579 | minutes = tz < 0 ? -minutes : minutes; | |
1580 | t = time + minutes * 60; | |
1581 | tm = gmtime(&t); | |
1582 | ||
1583 | strftime(time_buf, sizeof(time_buf), "%Y-%m-%d %H:%M:%S ", tm); | |
1584 | strcat(time_buf, tz_str); | |
1585 | return time_buf; | |
1586 | } | |
1587 | ||
cee7f245 JH |
1588 | #define OUTPUT_ANNOTATE_COMPAT 001 |
1589 | #define OUTPUT_LONG_OBJECT_NAME 002 | |
1590 | #define OUTPUT_RAW_TIMESTAMP 004 | |
1591 | #define OUTPUT_PORCELAIN 010 | |
1592 | #define OUTPUT_SHOW_NAME 020 | |
1593 | #define OUTPUT_SHOW_NUMBER 040 | |
5ff62c30 | 1594 | #define OUTPUT_SHOW_SCORE 0100 |
093dc5be | 1595 | #define OUTPUT_NO_AUTHOR 0200 |
cee7f245 JH |
1596 | |
1597 | static void emit_porcelain(struct scoreboard *sb, struct blame_entry *ent) | |
1598 | { | |
1599 | int cnt; | |
1600 | const char *cp; | |
1601 | struct origin *suspect = ent->suspect; | |
1602 | char hex[41]; | |
1603 | ||
1604 | strcpy(hex, sha1_to_hex(suspect->commit->object.sha1)); | |
1605 | printf("%s%c%d %d %d\n", | |
1606 | hex, | |
1607 | ent->guilty ? ' ' : '*', // purely for debugging | |
1608 | ent->s_lno + 1, | |
1609 | ent->lno + 1, | |
1610 | ent->num_lines); | |
1611 | if (!(suspect->commit->object.flags & METAINFO_SHOWN)) { | |
1612 | struct commit_info ci; | |
1613 | suspect->commit->object.flags |= METAINFO_SHOWN; | |
1614 | get_commit_info(suspect->commit, &ci, 1); | |
1615 | printf("author %s\n", ci.author); | |
1616 | printf("author-mail %s\n", ci.author_mail); | |
1617 | printf("author-time %lu\n", ci.author_time); | |
1618 | printf("author-tz %s\n", ci.author_tz); | |
1619 | printf("committer %s\n", ci.committer); | |
1620 | printf("committer-mail %s\n", ci.committer_mail); | |
1621 | printf("committer-time %lu\n", ci.committer_time); | |
1622 | printf("committer-tz %s\n", ci.committer_tz); | |
46e5e69d | 1623 | write_filename_info(suspect->path); |
cee7f245 | 1624 | printf("summary %s\n", ci.summary); |
b11121d9 JH |
1625 | if (suspect->commit->object.flags & UNINTERESTING) |
1626 | printf("boundary\n"); | |
cee7f245 JH |
1627 | } |
1628 | else if (suspect->commit->object.flags & MORE_THAN_ONE_PATH) | |
46e5e69d | 1629 | write_filename_info(suspect->path); |
cee7f245 JH |
1630 | |
1631 | cp = nth_line(sb, ent->lno); | |
1632 | for (cnt = 0; cnt < ent->num_lines; cnt++) { | |
1633 | char ch; | |
1634 | if (cnt) | |
1635 | printf("%s %d %d\n", hex, | |
1636 | ent->s_lno + 1 + cnt, | |
1637 | ent->lno + 1 + cnt); | |
1638 | putchar('\t'); | |
1639 | do { | |
1640 | ch = *cp++; | |
1641 | putchar(ch); | |
1642 | } while (ch != '\n' && | |
1643 | cp < sb->final_buf + sb->final_buf_size); | |
1644 | } | |
1645 | } | |
1646 | ||
1647 | static void emit_other(struct scoreboard *sb, struct blame_entry *ent, int opt) | |
1648 | { | |
1649 | int cnt; | |
1650 | const char *cp; | |
1651 | struct origin *suspect = ent->suspect; | |
1652 | struct commit_info ci; | |
1653 | char hex[41]; | |
1654 | int show_raw_time = !!(opt & OUTPUT_RAW_TIMESTAMP); | |
1655 | ||
1656 | get_commit_info(suspect->commit, &ci, 1); | |
1657 | strcpy(hex, sha1_to_hex(suspect->commit->object.sha1)); | |
1658 | ||
1659 | cp = nth_line(sb, ent->lno); | |
1660 | for (cnt = 0; cnt < ent->num_lines; cnt++) { | |
1661 | char ch; | |
b11121d9 JH |
1662 | int length = (opt & OUTPUT_LONG_OBJECT_NAME) ? 40 : 8; |
1663 | ||
1664 | if (suspect->commit->object.flags & UNINTERESTING) { | |
e68989a7 JH |
1665 | if (blank_boundary) |
1666 | memset(hex, ' ', length); | |
1667 | else if (!cmd_is_annotate) { | |
4c10a5ca JH |
1668 | length--; |
1669 | putchar('^'); | |
1670 | } | |
b11121d9 | 1671 | } |
cee7f245 | 1672 | |
b11121d9 | 1673 | printf("%.*s", length, hex); |
cee7f245 JH |
1674 | if (opt & OUTPUT_ANNOTATE_COMPAT) |
1675 | printf("\t(%10s\t%10s\t%d)", ci.author, | |
1676 | format_time(ci.author_time, ci.author_tz, | |
1677 | show_raw_time), | |
1678 | ent->lno + 1 + cnt); | |
1679 | else { | |
5ff62c30 | 1680 | if (opt & OUTPUT_SHOW_SCORE) |
54a4c617 JH |
1681 | printf(" %*d %02d", |
1682 | max_score_digits, ent->score, | |
1683 | ent->suspect->refcnt); | |
cee7f245 JH |
1684 | if (opt & OUTPUT_SHOW_NAME) |
1685 | printf(" %-*.*s", longest_file, longest_file, | |
1686 | suspect->path); | |
1687 | if (opt & OUTPUT_SHOW_NUMBER) | |
1688 | printf(" %*d", max_orig_digits, | |
1689 | ent->s_lno + 1 + cnt); | |
093dc5be JH |
1690 | |
1691 | if (!(opt & OUTPUT_NO_AUTHOR)) | |
1692 | printf(" (%-*.*s %10s", | |
1693 | longest_author, longest_author, | |
1694 | ci.author, | |
1695 | format_time(ci.author_time, | |
1696 | ci.author_tz, | |
1697 | show_raw_time)); | |
1698 | printf(" %*d) ", | |
cee7f245 JH |
1699 | max_digits, ent->lno + 1 + cnt); |
1700 | } | |
1701 | do { | |
1702 | ch = *cp++; | |
1703 | putchar(ch); | |
1704 | } while (ch != '\n' && | |
1705 | cp < sb->final_buf + sb->final_buf_size); | |
1706 | } | |
1707 | } | |
1708 | ||
1709 | static void output(struct scoreboard *sb, int option) | |
1710 | { | |
1711 | struct blame_entry *ent; | |
1712 | ||
1713 | if (option & OUTPUT_PORCELAIN) { | |
1714 | for (ent = sb->ent; ent; ent = ent->next) { | |
1715 | struct blame_entry *oth; | |
1716 | struct origin *suspect = ent->suspect; | |
1717 | struct commit *commit = suspect->commit; | |
1718 | if (commit->object.flags & MORE_THAN_ONE_PATH) | |
1719 | continue; | |
1720 | for (oth = ent->next; oth; oth = oth->next) { | |
1721 | if ((oth->suspect->commit != commit) || | |
1722 | !strcmp(oth->suspect->path, suspect->path)) | |
1723 | continue; | |
1724 | commit->object.flags |= MORE_THAN_ONE_PATH; | |
1725 | break; | |
1726 | } | |
1727 | } | |
1728 | } | |
1729 | ||
1730 | for (ent = sb->ent; ent; ent = ent->next) { | |
1731 | if (option & OUTPUT_PORCELAIN) | |
1732 | emit_porcelain(sb, ent); | |
5ff62c30 | 1733 | else { |
cee7f245 | 1734 | emit_other(sb, ent, option); |
5ff62c30 | 1735 | } |
cee7f245 JH |
1736 | } |
1737 | } | |
1738 | ||
1732a1fd JH |
1739 | /* |
1740 | * To allow quick access to the contents of nth line in the | |
1741 | * final image, prepare an index in the scoreboard. | |
1742 | */ | |
cee7f245 JH |
1743 | static int prepare_lines(struct scoreboard *sb) |
1744 | { | |
1745 | const char *buf = sb->final_buf; | |
1746 | unsigned long len = sb->final_buf_size; | |
1747 | int num = 0, incomplete = 0, bol = 1; | |
1748 | ||
1749 | if (len && buf[len-1] != '\n') | |
1750 | incomplete++; /* incomplete line at the end */ | |
1751 | while (len--) { | |
1752 | if (bol) { | |
1753 | sb->lineno = xrealloc(sb->lineno, | |
1754 | sizeof(int* ) * (num + 1)); | |
1755 | sb->lineno[num] = buf - sb->final_buf; | |
1756 | bol = 0; | |
1757 | } | |
1758 | if (*buf++ == '\n') { | |
1759 | num++; | |
1760 | bol = 1; | |
1761 | } | |
1762 | } | |
1ca6ca87 JH |
1763 | sb->lineno = xrealloc(sb->lineno, |
1764 | sizeof(int* ) * (num + incomplete + 1)); | |
1765 | sb->lineno[num + incomplete] = buf - sb->final_buf; | |
cee7f245 JH |
1766 | sb->num_lines = num + incomplete; |
1767 | return sb->num_lines; | |
1768 | } | |
1769 | ||
1732a1fd JH |
1770 | /* |
1771 | * Add phony grafts for use with -S; this is primarily to | |
1772 | * support git-cvsserver that wants to give a linear history | |
1773 | * to its clients. | |
1774 | */ | |
cee7f245 JH |
1775 | static int read_ancestry(const char *graft_file) |
1776 | { | |
1777 | FILE *fp = fopen(graft_file, "r"); | |
1778 | char buf[1024]; | |
1779 | if (!fp) | |
1780 | return -1; | |
1781 | while (fgets(buf, sizeof(buf), fp)) { | |
1782 | /* The format is just "Commit Parent1 Parent2 ...\n" */ | |
1783 | int len = strlen(buf); | |
1784 | struct commit_graft *graft = read_graft_line(buf, len); | |
8eaf7986 JH |
1785 | if (graft) |
1786 | register_commit_graft(graft, 0); | |
cee7f245 JH |
1787 | } |
1788 | fclose(fp); | |
1789 | return 0; | |
1790 | } | |
1791 | ||
1732a1fd JH |
1792 | /* |
1793 | * How many columns do we need to show line numbers in decimal? | |
1794 | */ | |
cee7f245 JH |
1795 | static int lineno_width(int lines) |
1796 | { | |
f2915045 | 1797 | int i, width; |
cee7f245 | 1798 | |
f2915045 JH |
1799 | for (width = 1, i = 10; i <= lines + 1; width++) |
1800 | i *= 10; | |
1801 | return width; | |
cee7f245 JH |
1802 | } |
1803 | ||
1732a1fd JH |
1804 | /* |
1805 | * How many columns do we need to show line numbers, authors, | |
1806 | * and filenames? | |
1807 | */ | |
cee7f245 JH |
1808 | static void find_alignment(struct scoreboard *sb, int *option) |
1809 | { | |
1810 | int longest_src_lines = 0; | |
1811 | int longest_dst_lines = 0; | |
5ff62c30 | 1812 | unsigned largest_score = 0; |
cee7f245 JH |
1813 | struct blame_entry *e; |
1814 | ||
1815 | for (e = sb->ent; e; e = e->next) { | |
1816 | struct origin *suspect = e->suspect; | |
1817 | struct commit_info ci; | |
1818 | int num; | |
1819 | ||
ab3bb800 JH |
1820 | if (strcmp(suspect->path, sb->path)) |
1821 | *option |= OUTPUT_SHOW_NAME; | |
1822 | num = strlen(suspect->path); | |
1823 | if (longest_file < num) | |
1824 | longest_file = num; | |
cee7f245 JH |
1825 | if (!(suspect->commit->object.flags & METAINFO_SHOWN)) { |
1826 | suspect->commit->object.flags |= METAINFO_SHOWN; | |
1827 | get_commit_info(suspect->commit, &ci, 1); | |
cee7f245 JH |
1828 | num = strlen(ci.author); |
1829 | if (longest_author < num) | |
1830 | longest_author = num; | |
1831 | } | |
1832 | num = e->s_lno + e->num_lines; | |
1833 | if (longest_src_lines < num) | |
1834 | longest_src_lines = num; | |
1835 | num = e->lno + e->num_lines; | |
1836 | if (longest_dst_lines < num) | |
1837 | longest_dst_lines = num; | |
5ff62c30 JH |
1838 | if (largest_score < ent_score(sb, e)) |
1839 | largest_score = ent_score(sb, e); | |
cee7f245 JH |
1840 | } |
1841 | max_orig_digits = lineno_width(longest_src_lines); | |
1842 | max_digits = lineno_width(longest_dst_lines); | |
5ff62c30 | 1843 | max_score_digits = lineno_width(largest_score); |
cee7f245 JH |
1844 | } |
1845 | ||
1732a1fd JH |
1846 | /* |
1847 | * For debugging -- origin is refcounted, and this asserts that | |
1848 | * we do not underflow. | |
1849 | */ | |
54a4c617 JH |
1850 | static void sanity_check_refcnt(struct scoreboard *sb) |
1851 | { | |
1852 | int baa = 0; | |
1853 | struct blame_entry *ent; | |
1854 | ||
1855 | for (ent = sb->ent; ent; ent = ent->next) { | |
ae86ad65 | 1856 | /* Nobody should have zero or negative refcnt */ |
854b97f6 JH |
1857 | if (ent->suspect->refcnt <= 0) { |
1858 | fprintf(stderr, "%s in %s has negative refcnt %d\n", | |
1859 | ent->suspect->path, | |
1860 | sha1_to_hex(ent->suspect->commit->object.sha1), | |
1861 | ent->suspect->refcnt); | |
ae86ad65 | 1862 | baa = 1; |
854b97f6 | 1863 | } |
ae86ad65 JH |
1864 | } |
1865 | for (ent = sb->ent; ent; ent = ent->next) { | |
1866 | /* Mark the ones that haven't been checked */ | |
54a4c617 JH |
1867 | if (0 < ent->suspect->refcnt) |
1868 | ent->suspect->refcnt = -ent->suspect->refcnt; | |
54a4c617 JH |
1869 | } |
1870 | for (ent = sb->ent; ent; ent = ent->next) { | |
1732a1fd JH |
1871 | /* |
1872 | * ... then pick each and see if they have the the | |
1873 | * correct refcnt. | |
54a4c617 JH |
1874 | */ |
1875 | int found; | |
1876 | struct blame_entry *e; | |
1877 | struct origin *suspect = ent->suspect; | |
1878 | ||
1879 | if (0 < suspect->refcnt) | |
1880 | continue; | |
ae86ad65 | 1881 | suspect->refcnt = -suspect->refcnt; /* Unmark */ |
54a4c617 JH |
1882 | for (found = 0, e = sb->ent; e; e = e->next) { |
1883 | if (e->suspect != suspect) | |
1884 | continue; | |
1885 | found++; | |
1886 | } | |
854b97f6 JH |
1887 | if (suspect->refcnt != found) { |
1888 | fprintf(stderr, "%s in %s has refcnt %d, not %d\n", | |
1889 | ent->suspect->path, | |
1890 | sha1_to_hex(ent->suspect->commit->object.sha1), | |
1891 | ent->suspect->refcnt, found); | |
1892 | baa = 2; | |
1893 | } | |
54a4c617 JH |
1894 | } |
1895 | if (baa) { | |
1896 | int opt = 0160; | |
1897 | find_alignment(sb, &opt); | |
1898 | output(sb, opt); | |
854b97f6 | 1899 | die("Baa %d!", baa); |
54a4c617 JH |
1900 | } |
1901 | } | |
1902 | ||
1732a1fd JH |
1903 | /* |
1904 | * Used for the command line parsing; check if the path exists | |
1905 | * in the working tree. | |
1906 | */ | |
cee7f245 JH |
1907 | static int has_path_in_work_tree(const char *path) |
1908 | { | |
1909 | struct stat st; | |
1910 | return !lstat(path, &st); | |
1911 | } | |
1912 | ||
4a0fc95f JH |
1913 | static unsigned parse_score(const char *arg) |
1914 | { | |
1915 | char *end; | |
1916 | unsigned long score = strtoul(arg, &end, 10); | |
1917 | if (*end) | |
1918 | return 0; | |
1919 | return score; | |
1920 | } | |
1921 | ||
20239bae JK |
1922 | static const char *add_prefix(const char *prefix, const char *path) |
1923 | { | |
097971f5 | 1924 | return prefix_path(prefix, prefix ? strlen(prefix) : 0, path); |
20239bae JK |
1925 | } |
1926 | ||
1732a1fd JH |
1927 | /* |
1928 | * Parsing of (comma separated) one item in the -L option | |
1929 | */ | |
931233bc JH |
1930 | static const char *parse_loc(const char *spec, |
1931 | struct scoreboard *sb, long lno, | |
1932 | long begin, long *ret) | |
1933 | { | |
1934 | char *term; | |
1935 | const char *line; | |
1936 | long num; | |
1937 | int reg_error; | |
1938 | regex_t regexp; | |
1939 | regmatch_t match[1]; | |
1940 | ||
7bd9641d JH |
1941 | /* Allow "-L <something>,+20" to mean starting at <something> |
1942 | * for 20 lines, or "-L <something>,-5" for 5 lines ending at | |
1943 | * <something>. | |
1944 | */ | |
1945 | if (1 < begin && (spec[0] == '+' || spec[0] == '-')) { | |
1946 | num = strtol(spec + 1, &term, 10); | |
1947 | if (term != spec + 1) { | |
1948 | if (spec[0] == '-') | |
1949 | num = 0 - num; | |
1950 | if (0 < num) | |
1951 | *ret = begin + num - 2; | |
1952 | else if (!num) | |
1953 | *ret = begin; | |
1954 | else | |
1955 | *ret = begin + num; | |
1956 | return term; | |
1957 | } | |
1958 | return spec; | |
1959 | } | |
931233bc JH |
1960 | num = strtol(spec, &term, 10); |
1961 | if (term != spec) { | |
1962 | *ret = num; | |
1963 | return term; | |
1964 | } | |
1965 | if (spec[0] != '/') | |
1966 | return spec; | |
1967 | ||
1968 | /* it could be a regexp of form /.../ */ | |
1969 | for (term = (char*) spec + 1; *term && *term != '/'; term++) { | |
1970 | if (*term == '\\') | |
1971 | term++; | |
1972 | } | |
1973 | if (*term != '/') | |
1974 | return spec; | |
1975 | ||
1976 | /* try [spec+1 .. term-1] as regexp */ | |
1977 | *term = 0; | |
1978 | begin--; /* input is in human terms */ | |
1979 | line = nth_line(sb, begin); | |
1980 | ||
1981 | if (!(reg_error = regcomp(®exp, spec + 1, REG_NEWLINE)) && | |
1982 | !(reg_error = regexec(®exp, line, 1, match, 0))) { | |
1983 | const char *cp = line + match[0].rm_so; | |
1984 | const char *nline; | |
1985 | ||
1986 | while (begin++ < lno) { | |
1987 | nline = nth_line(sb, begin); | |
1988 | if (line <= cp && cp < nline) | |
1989 | break; | |
1990 | line = nline; | |
1991 | } | |
1992 | *ret = begin; | |
1993 | regfree(®exp); | |
1994 | *term++ = '/'; | |
1995 | return term; | |
1996 | } | |
1997 | else { | |
1998 | char errbuf[1024]; | |
1999 | regerror(reg_error, ®exp, errbuf, 1024); | |
2000 | die("-L parameter '%s': %s", spec + 1, errbuf); | |
2001 | } | |
2002 | } | |
2003 | ||
1732a1fd JH |
2004 | /* |
2005 | * Parsing of -L option | |
2006 | */ | |
931233bc JH |
2007 | static void prepare_blame_range(struct scoreboard *sb, |
2008 | const char *bottomtop, | |
2009 | long lno, | |
2010 | long *bottom, long *top) | |
2011 | { | |
2012 | const char *term; | |
2013 | ||
2014 | term = parse_loc(bottomtop, sb, lno, 1, bottom); | |
2015 | if (*term == ',') { | |
2016 | term = parse_loc(term + 1, sb, lno, *bottom + 1, top); | |
2017 | if (*term) | |
acca687f | 2018 | usage(blame_usage); |
931233bc JH |
2019 | } |
2020 | if (*term) | |
acca687f | 2021 | usage(blame_usage); |
931233bc JH |
2022 | } |
2023 | ||
4c10a5ca JH |
2024 | static int git_blame_config(const char *var, const char *value) |
2025 | { | |
2026 | if (!strcmp(var, "blame.showroot")) { | |
2027 | show_root = git_config_bool(var, value); | |
2028 | return 0; | |
2029 | } | |
2030 | if (!strcmp(var, "blame.blankboundary")) { | |
2031 | blank_boundary = git_config_bool(var, value); | |
2032 | return 0; | |
2033 | } | |
2034 | return git_default_config(var, value); | |
2035 | } | |
2036 | ||
f6c07d7d JH |
2037 | /* |
2038 | * Prepare a dummy commit that represents the work tree (or staged) item. | |
2039 | * Note that annotating work tree item never works in the reverse. | |
2040 | */ | |
1cfe7733 JH |
2041 | static struct commit *fake_working_tree_commit(const char *path, const char *contents_from) |
2042 | { | |
2043 | struct commit *commit; | |
2044 | struct origin *origin; | |
2045 | unsigned char head_sha1[20]; | |
af6eb822 | 2046 | struct strbuf buf; |
1cfe7733 | 2047 | const char *ident; |
1cfe7733 | 2048 | time_t now; |
1cfe7733 JH |
2049 | int size, len; |
2050 | struct cache_entry *ce; | |
2051 | unsigned mode; | |
2052 | ||
2053 | if (get_sha1("HEAD", head_sha1)) | |
2054 | die("No such ref: HEAD"); | |
2055 | ||
2056 | time(&now); | |
2057 | commit = xcalloc(1, sizeof(*commit)); | |
2058 | commit->parents = xcalloc(1, sizeof(*commit->parents)); | |
2059 | commit->parents->item = lookup_commit_reference(head_sha1); | |
2060 | commit->object.parsed = 1; | |
2061 | commit->date = now; | |
2062 | commit->object.type = OBJ_COMMIT; | |
2063 | ||
2064 | origin = make_origin(commit, path); | |
2065 | ||
f1696ee3 | 2066 | strbuf_init(&buf, 0); |
1cfe7733 JH |
2067 | if (!contents_from || strcmp("-", contents_from)) { |
2068 | struct stat st; | |
2069 | const char *read_from; | |
af6eb822 | 2070 | unsigned long fin_size; |
1cfe7733 JH |
2071 | |
2072 | if (contents_from) { | |
2073 | if (stat(contents_from, &st) < 0) | |
2074 | die("Cannot stat %s", contents_from); | |
2075 | read_from = contents_from; | |
2076 | } | |
2077 | else { | |
2078 | if (lstat(path, &st) < 0) | |
2079 | die("Cannot lstat %s", path); | |
2080 | read_from = path; | |
2081 | } | |
dc49cd76 | 2082 | fin_size = xsize_t(st.st_size); |
1cfe7733 JH |
2083 | mode = canon_mode(st.st_mode); |
2084 | switch (st.st_mode & S_IFMT) { | |
2085 | case S_IFREG: | |
387e7e19 PH |
2086 | if (strbuf_read_file(&buf, read_from, st.st_size) != st.st_size) |
2087 | die("cannot open or read %s", read_from); | |
1cfe7733 JH |
2088 | break; |
2089 | case S_IFLNK: | |
af6eb822 | 2090 | if (readlink(read_from, buf.buf, buf.alloc) != fin_size) |
1cfe7733 | 2091 | die("cannot readlink %s", read_from); |
af6eb822 | 2092 | buf.len = fin_size; |
1cfe7733 JH |
2093 | break; |
2094 | default: | |
2095 | die("unsupported file type %s", read_from); | |
2096 | } | |
2097 | } | |
2098 | else { | |
2099 | /* Reading from stdin */ | |
2100 | contents_from = "standard input"; | |
1cfe7733 | 2101 | mode = 0; |
f1696ee3 | 2102 | if (strbuf_read(&buf, 0, 0) < 0) |
af6eb822 | 2103 | die("read error %s from stdin", strerror(errno)); |
1cfe7733 | 2104 | } |
21e5ad50 | 2105 | convert_to_git(path, buf.buf, buf.len, &buf, 0); |
af6eb822 PH |
2106 | origin->file.ptr = buf.buf; |
2107 | origin->file.size = buf.len; | |
2108 | pretend_sha1_file(buf.buf, buf.len, OBJ_BLOB, origin->blob_sha1); | |
1cfe7733 JH |
2109 | commit->util = origin; |
2110 | ||
2111 | /* | |
2112 | * Read the current index, replace the path entry with | |
2113 | * origin->blob_sha1 without mucking with its mode or type | |
2114 | * bits; we are not going to write this index out -- we just | |
2115 | * want to run "diff-index --cached". | |
2116 | */ | |
2117 | discard_cache(); | |
2118 | read_cache(); | |
2119 | ||
2120 | len = strlen(path); | |
2121 | if (!mode) { | |
2122 | int pos = cache_name_pos(path, len); | |
2123 | if (0 <= pos) | |
7a51ed66 | 2124 | mode = active_cache[pos]->ce_mode; |
1cfe7733 JH |
2125 | else |
2126 | /* Let's not bother reading from HEAD tree */ | |
2127 | mode = S_IFREG | 0644; | |
2128 | } | |
2129 | size = cache_entry_size(len); | |
2130 | ce = xcalloc(1, size); | |
2131 | hashcpy(ce->sha1, origin->blob_sha1); | |
2132 | memcpy(ce->name, path, len); | |
2133 | ce->ce_flags = create_ce_flags(len, 0); | |
2134 | ce->ce_mode = create_ce_mode(mode); | |
2135 | add_cache_entry(ce, ADD_CACHE_OK_TO_ADD|ADD_CACHE_OK_TO_REPLACE); | |
2136 | ||
2137 | /* | |
2138 | * We are not going to write this out, so this does not matter | |
2139 | * right now, but someday we might optimize diff-index --cached | |
2140 | * with cache-tree information. | |
2141 | */ | |
2142 | cache_tree_invalidate_path(active_cache_tree, path); | |
2143 | ||
2144 | commit->buffer = xmalloc(400); | |
2145 | ident = fmt_ident("Not Committed Yet", "not.committed.yet", NULL, 0); | |
1bb88be9 | 2146 | snprintf(commit->buffer, 400, |
1cfe7733 JH |
2147 | "tree 0000000000000000000000000000000000000000\n" |
2148 | "parent %s\n" | |
2149 | "author %s\n" | |
2150 | "committer %s\n\n" | |
2151 | "Version of %s from %s\n", | |
2152 | sha1_to_hex(head_sha1), | |
2153 | ident, ident, path, contents_from ? contents_from : path); | |
2154 | return commit; | |
2155 | } | |
2156 | ||
f6c07d7d JH |
2157 | static const char *prepare_final(struct scoreboard *sb, struct rev_info *revs) |
2158 | { | |
2159 | int i; | |
2160 | const char *final_commit_name = NULL; | |
2161 | ||
2162 | /* | |
2163 | * There must be one and only one positive commit in the | |
2164 | * revs->pending array. | |
2165 | */ | |
2166 | for (i = 0; i < revs->pending.nr; i++) { | |
2167 | struct object *obj = revs->pending.objects[i].item; | |
2168 | if (obj->flags & UNINTERESTING) | |
2169 | continue; | |
2170 | while (obj->type == OBJ_TAG) | |
2171 | obj = deref_tag(obj, NULL, 0); | |
2172 | if (obj->type != OBJ_COMMIT) | |
2173 | die("Non commit %s?", revs->pending.objects[i].name); | |
2174 | if (sb->final) | |
2175 | die("More than one commit to dig from %s and %s?", | |
2176 | revs->pending.objects[i].name, | |
2177 | final_commit_name); | |
2178 | sb->final = (struct commit *) obj; | |
2179 | final_commit_name = revs->pending.objects[i].name; | |
2180 | } | |
2181 | return final_commit_name; | |
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; | |
d24bba80 | 2191 | int i, seen_dashdash, unk, opt; |
cee7f245 JH |
2192 | long bottom, top, lno; |
2193 | int output_option = 0; | |
870b39c1 | 2194 | int show_stats = 0; |
cee7f245 JH |
2195 | const char *revs_file = NULL; |
2196 | const char *final_commit_name = NULL; | |
21666f1a | 2197 | enum object_type type; |
931233bc | 2198 | const char *bottomtop = NULL; |
1cfe7733 | 2199 | const char *contents_from = NULL; |
cee7f245 | 2200 | |
e68989a7 JH |
2201 | cmd_is_annotate = !strcmp(argv[0], "annotate"); |
2202 | ||
4c10a5ca | 2203 | git_config(git_blame_config); |
612702e8 JH |
2204 | save_commit_buffer = 0; |
2205 | ||
d24bba80 | 2206 | opt = 0; |
cee7f245 JH |
2207 | seen_dashdash = 0; |
2208 | for (unk = i = 1; i < argc; i++) { | |
2209 | const char *arg = argv[i]; | |
2210 | if (*arg != '-') | |
2211 | break; | |
4c10a5ca JH |
2212 | else if (!strcmp("-b", arg)) |
2213 | blank_boundary = 1; | |
2214 | else if (!strcmp("--root", arg)) | |
2215 | show_root = 1; | |
870b39c1 JH |
2216 | else if (!strcmp(arg, "--show-stats")) |
2217 | show_stats = 1; | |
cee7f245 JH |
2218 | else if (!strcmp("-c", arg)) |
2219 | output_option |= OUTPUT_ANNOTATE_COMPAT; | |
2220 | else if (!strcmp("-t", arg)) | |
2221 | output_option |= OUTPUT_RAW_TIMESTAMP; | |
2222 | else if (!strcmp("-l", arg)) | |
2223 | output_option |= OUTPUT_LONG_OBJECT_NAME; | |
093dc5be JH |
2224 | else if (!strcmp("-s", arg)) |
2225 | output_option |= OUTPUT_NO_AUTHOR; | |
b82871b3 JH |
2226 | else if (!strcmp("-w", arg)) |
2227 | xdl_opts |= XDF_IGNORE_WHITESPACE; | |
cee7f245 JH |
2228 | else if (!strcmp("-S", arg) && ++i < argc) |
2229 | revs_file = argv[i]; | |
599065a3 | 2230 | else if (!prefixcmp(arg, "-M")) { |
d24bba80 | 2231 | opt |= PICKAXE_BLAME_MOVE; |
4a0fc95f JH |
2232 | blame_move_score = parse_score(arg+2); |
2233 | } | |
599065a3 | 2234 | else if (!prefixcmp(arg, "-C")) { |
c63777c0 JH |
2235 | /* |
2236 | * -C enables copy from removed files; | |
2237 | * -C -C enables copy from existing files, but only | |
2238 | * when blaming a new file; | |
2239 | * -C -C -C enables copy from existing files for | |
2240 | * everybody | |
2241 | */ | |
2242 | if (opt & PICKAXE_BLAME_COPY_HARDER) | |
2243 | opt |= PICKAXE_BLAME_COPY_HARDEST; | |
18abd745 JH |
2244 | if (opt & PICKAXE_BLAME_COPY) |
2245 | opt |= PICKAXE_BLAME_COPY_HARDER; | |
2246 | opt |= PICKAXE_BLAME_COPY | PICKAXE_BLAME_MOVE; | |
4a0fc95f | 2247 | blame_copy_score = parse_score(arg+2); |
18abd745 | 2248 | } |
599065a3 | 2249 | else if (!prefixcmp(arg, "-L")) { |
2c40f984 JH |
2250 | if (!arg[2]) { |
2251 | if (++i >= argc) | |
acca687f | 2252 | usage(blame_usage); |
2c40f984 JH |
2253 | arg = argv[i]; |
2254 | } | |
2255 | else | |
2256 | arg += 2; | |
931233bc | 2257 | if (bottomtop) |
cee7f245 | 2258 | die("More than one '-L n,m' option given"); |
931233bc | 2259 | bottomtop = arg; |
cee7f245 | 2260 | } |
1cfe7733 JH |
2261 | else if (!strcmp("--contents", arg)) { |
2262 | if (++i >= argc) | |
2263 | usage(blame_usage); | |
2264 | contents_from = argv[i]; | |
2265 | } | |
717d1462 LT |
2266 | else if (!strcmp("--incremental", arg)) |
2267 | incremental = 1; | |
5ff62c30 JH |
2268 | else if (!strcmp("--score-debug", arg)) |
2269 | output_option |= OUTPUT_SHOW_SCORE; | |
cee7f245 JH |
2270 | else if (!strcmp("-f", arg) || |
2271 | !strcmp("--show-name", arg)) | |
2272 | output_option |= OUTPUT_SHOW_NAME; | |
2273 | else if (!strcmp("-n", arg) || | |
2274 | !strcmp("--show-number", arg)) | |
2275 | output_option |= OUTPUT_SHOW_NUMBER; | |
2276 | else if (!strcmp("-p", arg) || | |
2277 | !strcmp("--porcelain", arg)) | |
2278 | output_option |= OUTPUT_PORCELAIN; | |
2279 | else if (!strcmp("--", arg)) { | |
2280 | seen_dashdash = 1; | |
2281 | i++; | |
2282 | break; | |
2283 | } | |
2284 | else | |
2285 | argv[unk++] = arg; | |
2286 | } | |
2287 | ||
4a0fc95f JH |
2288 | if (!blame_move_score) |
2289 | blame_move_score = BLAME_DEFAULT_MOVE_SCORE; | |
2290 | if (!blame_copy_score) | |
2291 | blame_copy_score = BLAME_DEFAULT_COPY_SCORE; | |
2292 | ||
1732a1fd JH |
2293 | /* |
2294 | * We have collected options unknown to us in argv[1..unk] | |
cee7f245 | 2295 | * which are to be passed to revision machinery if we are |
3dff5379 | 2296 | * going to do the "bottom" processing. |
cee7f245 JH |
2297 | * |
2298 | * The remaining are: | |
2299 | * | |
2300 | * (1) if seen_dashdash, its either | |
2301 | * "-options -- <path>" or | |
2302 | * "-options -- <path> <rev>". | |
2303 | * but the latter is allowed only if there is no | |
2304 | * options that we passed to revision machinery. | |
2305 | * | |
2306 | * (2) otherwise, we may have "--" somewhere later and | |
2307 | * might be looking at the first one of multiple 'rev' | |
2308 | * parameters (e.g. " master ^next ^maint -- path"). | |
2309 | * See if there is a dashdash first, and give the | |
2310 | * arguments before that to revision machinery. | |
2311 | * After that there must be one 'path'. | |
2312 | * | |
2313 | * (3) otherwise, its one of the three: | |
2314 | * "-options <path> <rev>" | |
2315 | * "-options <rev> <path>" | |
2316 | * "-options <path>" | |
2317 | * but again the first one is allowed only if | |
2318 | * there is no options that we passed to revision | |
2319 | * machinery. | |
2320 | */ | |
2321 | ||
2322 | if (seen_dashdash) { | |
2323 | /* (1) */ | |
2324 | if (argc <= i) | |
acca687f | 2325 | usage(blame_usage); |
20239bae | 2326 | path = add_prefix(prefix, argv[i]); |
cee7f245 JH |
2327 | if (i + 1 == argc - 1) { |
2328 | if (unk != 1) | |
acca687f | 2329 | usage(blame_usage); |
cee7f245 JH |
2330 | argv[unk++] = argv[i + 1]; |
2331 | } | |
2332 | else if (i + 1 != argc) | |
2333 | /* garbage at end */ | |
acca687f | 2334 | usage(blame_usage); |
cee7f245 JH |
2335 | } |
2336 | else { | |
2337 | int j; | |
2338 | for (j = i; !seen_dashdash && j < argc; j++) | |
2339 | if (!strcmp(argv[j], "--")) | |
2340 | seen_dashdash = j; | |
2341 | if (seen_dashdash) { | |
f4421325 | 2342 | /* (2) */ |
cee7f245 | 2343 | if (seen_dashdash + 1 != argc - 1) |
acca687f | 2344 | usage(blame_usage); |
20239bae | 2345 | path = add_prefix(prefix, argv[seen_dashdash + 1]); |
cee7f245 JH |
2346 | for (j = i; j < seen_dashdash; j++) |
2347 | argv[unk++] = argv[j]; | |
2348 | } | |
2349 | else { | |
2350 | /* (3) */ | |
f4421325 TK |
2351 | if (argc <= i) |
2352 | usage(blame_usage); | |
20239bae | 2353 | path = add_prefix(prefix, argv[i]); |
cee7f245 JH |
2354 | if (i + 1 == argc - 1) { |
2355 | final_commit_name = argv[i + 1]; | |
2356 | ||
2357 | /* if (unk == 1) we could be getting | |
2358 | * old-style | |
2359 | */ | |
2360 | if (unk == 1 && !has_path_in_work_tree(path)) { | |
20239bae | 2361 | path = add_prefix(prefix, argv[i + 1]); |
cee7f245 JH |
2362 | final_commit_name = argv[i]; |
2363 | } | |
2364 | } | |
2365 | else if (i != argc - 1) | |
acca687f | 2366 | usage(blame_usage); /* garbage at end */ |
cee7f245 | 2367 | |
354e6534 | 2368 | setup_work_tree(); |
cee7f245 JH |
2369 | if (!has_path_in_work_tree(path)) |
2370 | die("cannot stat path %s: %s", | |
2371 | path, strerror(errno)); | |
2372 | } | |
2373 | } | |
2374 | ||
2375 | if (final_commit_name) | |
2376 | argv[unk++] = final_commit_name; | |
2377 | ||
1732a1fd JH |
2378 | /* |
2379 | * Now we got rev and path. We do not want the path pruning | |
cee7f245 JH |
2380 | * but we may want "bottom" processing. |
2381 | */ | |
6bee4e40 | 2382 | argv[unk++] = "--"; /* terminate the rev name */ |
cee7f245 JH |
2383 | argv[unk] = NULL; |
2384 | ||
2385 | init_revisions(&revs, NULL); | |
1cfe7733 | 2386 | setup_revisions(unk, argv, &revs, NULL); |
cee7f245 JH |
2387 | memset(&sb, 0, sizeof(sb)); |
2388 | ||
f6c07d7d | 2389 | final_commit_name = prepare_final(&sb, &revs); |
cee7f245 | 2390 | if (!sb.final) { |
1732a1fd JH |
2391 | /* |
2392 | * "--not A B -- path" without anything positive; | |
1cfe7733 JH |
2393 | * do not default to HEAD, but use the working tree |
2394 | * or "--contents". | |
1732a1fd | 2395 | */ |
1981820b | 2396 | setup_work_tree(); |
1cfe7733 JH |
2397 | sb.final = fake_working_tree_commit(path, contents_from); |
2398 | add_pending_object(&revs, &(sb.final->object), ":"); | |
cee7f245 | 2399 | } |
1cfe7733 JH |
2400 | else if (contents_from) |
2401 | die("Cannot use --contents with final commit object name"); | |
cee7f245 | 2402 | |
1732a1fd JH |
2403 | /* |
2404 | * If we have bottom, this will mark the ancestors of the | |
cee7f245 JH |
2405 | * bottom commits we would reach while traversing as |
2406 | * uninteresting. | |
2407 | */ | |
3d51e1b5 MK |
2408 | if (prepare_revision_walk(&revs)) |
2409 | die("revision walk setup failed"); | |
cee7f245 | 2410 | |
1cfe7733 JH |
2411 | if (is_null_sha1(sb.final->object.sha1)) { |
2412 | char *buf; | |
2413 | o = sb.final->util; | |
2414 | buf = xmalloc(o->file.size + 1); | |
2415 | memcpy(buf, o->file.ptr, o->file.size + 1); | |
2416 | sb.final_buf = buf; | |
2417 | sb.final_buf_size = o->file.size; | |
2418 | } | |
2419 | else { | |
2420 | o = get_origin(&sb, sb.final, path); | |
2421 | if (fill_blob_sha1(o)) | |
2422 | die("no such path %s in %s", path, final_commit_name); | |
cee7f245 | 2423 | |
21666f1a | 2424 | sb.final_buf = read_sha1_file(o->blob_sha1, &type, |
1cfe7733 | 2425 | &sb.final_buf_size); |
ab43e495 JH |
2426 | if (!sb.final_buf) |
2427 | die("Cannot read blob %s for path %s", | |
2428 | sha1_to_hex(o->blob_sha1), | |
2429 | path); | |
1cfe7733 | 2430 | } |
c2e525d9 | 2431 | num_read_blob++; |
cee7f245 JH |
2432 | lno = prepare_lines(&sb); |
2433 | ||
931233bc JH |
2434 | bottom = top = 0; |
2435 | if (bottomtop) | |
2436 | prepare_blame_range(&sb, bottomtop, lno, &bottom, &top); | |
2437 | if (bottom && top && top < bottom) { | |
2438 | long tmp; | |
2439 | tmp = top; top = bottom; bottom = tmp; | |
2440 | } | |
cee7f245 JH |
2441 | if (bottom < 1) |
2442 | bottom = 1; | |
2443 | if (top < 1) | |
2444 | top = lno; | |
2445 | bottom--; | |
2446 | if (lno < top) | |
2447 | die("file %s has only %lu lines", path, lno); | |
2448 | ||
2449 | ent = xcalloc(1, sizeof(*ent)); | |
2450 | ent->lno = bottom; | |
2451 | ent->num_lines = top - bottom; | |
2452 | ent->suspect = o; | |
2453 | ent->s_lno = bottom; | |
2454 | ||
2455 | sb.ent = ent; | |
2456 | sb.path = path; | |
2457 | ||
2458 | if (revs_file && read_ancestry(revs_file)) | |
2459 | die("reading graft file %s failed: %s", | |
2460 | revs_file, strerror(errno)); | |
2461 | ||
50acc589 | 2462 | read_mailmap(&mailmap, ".mailmap", NULL); |
f9567384 | 2463 | |
b92565dc MH |
2464 | if (!incremental) |
2465 | setup_pager(); | |
2466 | ||
d24bba80 | 2467 | assign_blame(&sb, &revs, opt); |
cee7f245 | 2468 | |
717d1462 LT |
2469 | if (incremental) |
2470 | return 0; | |
2471 | ||
cee7f245 JH |
2472 | coalesce(&sb); |
2473 | ||
2474 | if (!(output_option & OUTPUT_PORCELAIN)) | |
2475 | find_alignment(&sb, &output_option); | |
2476 | ||
2477 | output(&sb, output_option); | |
2478 | free((void *)sb.final_buf); | |
2479 | for (ent = sb.ent; ent; ) { | |
2480 | struct blame_entry *e = ent->next; | |
2481 | free(ent); | |
2482 | ent = e; | |
2483 | } | |
c2e525d9 | 2484 | |
870b39c1 | 2485 | if (show_stats) { |
c2e525d9 JH |
2486 | printf("num read blob: %d\n", num_read_blob); |
2487 | printf("num get patch: %d\n", num_get_patch); | |
2488 | printf("num commits: %d\n", num_commits); | |
2489 | } | |
cee7f245 JH |
2490 | return 0; |
2491 | } |