]>
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 | */ | |
c2e525d9 JH |
94 | static char *fill_origin_blob(struct origin *o, mmfile_t *file) |
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; | |
109 | return file->ptr; | |
110 | } | |
111 | ||
1732a1fd JH |
112 | /* |
113 | * Origin is refcounted and usually we keep the blob contents to be | |
114 | * reused. | |
115 | */ | |
54a4c617 JH |
116 | static inline struct origin *origin_incref(struct origin *o) |
117 | { | |
118 | if (o) | |
119 | o->refcnt++; | |
120 | return o; | |
121 | } | |
122 | ||
123 | static void origin_decref(struct origin *o) | |
124 | { | |
125 | if (o && --o->refcnt <= 0) { | |
c2e525d9 JH |
126 | if (o->file.ptr) |
127 | free(o->file.ptr); | |
54a4c617 JH |
128 | free(o); |
129 | } | |
130 | } | |
131 | ||
7c3c7962 JH |
132 | static void drop_origin_blob(struct origin *o) |
133 | { | |
134 | if (o->file.ptr) { | |
135 | free(o->file.ptr); | |
136 | o->file.ptr = NULL; | |
137 | } | |
138 | } | |
139 | ||
1732a1fd JH |
140 | /* |
141 | * Each group of lines is described by a blame_entry; it can be split | |
142 | * as we pass blame to the parents. They form a linked list in the | |
143 | * scoreboard structure, sorted by the target line number. | |
144 | */ | |
cee7f245 JH |
145 | struct blame_entry { |
146 | struct blame_entry *prev; | |
147 | struct blame_entry *next; | |
148 | ||
149 | /* the first line of this group in the final image; | |
150 | * internally all line numbers are 0 based. | |
151 | */ | |
152 | int lno; | |
153 | ||
154 | /* how many lines this group has */ | |
155 | int num_lines; | |
156 | ||
157 | /* the commit that introduced this group into the final image */ | |
158 | struct origin *suspect; | |
159 | ||
160 | /* true if the suspect is truly guilty; false while we have not | |
161 | * checked if the group came from one of its parents. | |
162 | */ | |
163 | char guilty; | |
164 | ||
165 | /* the line number of the first line of this group in the | |
166 | * suspect's file; internally all line numbers are 0 based. | |
167 | */ | |
168 | int s_lno; | |
5ff62c30 JH |
169 | |
170 | /* how significant this entry is -- cached to avoid | |
1732a1fd | 171 | * scanning the lines over and over. |
5ff62c30 JH |
172 | */ |
173 | unsigned score; | |
cee7f245 JH |
174 | }; |
175 | ||
1732a1fd JH |
176 | /* |
177 | * The current state of the blame assignment. | |
178 | */ | |
cee7f245 JH |
179 | struct scoreboard { |
180 | /* the final commit (i.e. where we started digging from) */ | |
181 | struct commit *final; | |
182 | ||
183 | const char *path; | |
184 | ||
1732a1fd JH |
185 | /* |
186 | * The contents in the final image. | |
187 | * Used by many functions to obtain contents of the nth line, | |
188 | * indexed with scoreboard.lineno[blame_entry.lno]. | |
cee7f245 JH |
189 | */ |
190 | const char *final_buf; | |
191 | unsigned long final_buf_size; | |
192 | ||
193 | /* linked list of blames */ | |
194 | struct blame_entry *ent; | |
195 | ||
612702e8 | 196 | /* look-up a line in the final buffer */ |
cee7f245 JH |
197 | int num_lines; |
198 | int *lineno; | |
199 | }; | |
200 | ||
171dccd5 | 201 | static inline int same_suspect(struct origin *a, struct origin *b) |
46014766 | 202 | { |
171dccd5 | 203 | if (a == b) |
57584d9e | 204 | return 1; |
171dccd5 JH |
205 | if (a->commit != b->commit) |
206 | return 0; | |
207 | return !strcmp(a->path, b->path); | |
46014766 JH |
208 | } |
209 | ||
54a4c617 JH |
210 | static void sanity_check_refcnt(struct scoreboard *); |
211 | ||
1732a1fd JH |
212 | /* |
213 | * If two blame entries that are next to each other came from | |
214 | * contiguous lines in the same origin (i.e. <commit, path> pair), | |
215 | * merge them together. | |
216 | */ | |
cee7f245 JH |
217 | static void coalesce(struct scoreboard *sb) |
218 | { | |
219 | struct blame_entry *ent, *next; | |
220 | ||
221 | for (ent = sb->ent; ent && (next = ent->next); ent = next) { | |
171dccd5 | 222 | if (same_suspect(ent->suspect, next->suspect) && |
cee7f245 JH |
223 | ent->guilty == next->guilty && |
224 | ent->s_lno + ent->num_lines == next->s_lno) { | |
225 | ent->num_lines += next->num_lines; | |
226 | ent->next = next->next; | |
227 | if (ent->next) | |
228 | ent->next->prev = ent; | |
54a4c617 | 229 | origin_decref(next->suspect); |
cee7f245 | 230 | free(next); |
46014766 | 231 | ent->score = 0; |
cee7f245 JH |
232 | next = ent; /* again */ |
233 | } | |
234 | } | |
54a4c617 JH |
235 | |
236 | if (DEBUG) /* sanity */ | |
237 | sanity_check_refcnt(sb); | |
cee7f245 JH |
238 | } |
239 | ||
1732a1fd JH |
240 | /* |
241 | * Given a commit and a path in it, create a new origin structure. | |
242 | * The callers that add blame to the scoreboard should use | |
243 | * get_origin() to obtain shared, refcounted copy instead of calling | |
244 | * this function directly. | |
245 | */ | |
854b97f6 JH |
246 | static struct origin *make_origin(struct commit *commit, const char *path) |
247 | { | |
248 | struct origin *o; | |
249 | o = xcalloc(1, sizeof(*o) + strlen(path) + 1); | |
250 | o->commit = commit; | |
251 | o->refcnt = 1; | |
252 | strcpy(o->path, path); | |
253 | return o; | |
254 | } | |
255 | ||
1732a1fd JH |
256 | /* |
257 | * Locate an existing origin or create a new one. | |
258 | */ | |
f6c0e191 JH |
259 | static struct origin *get_origin(struct scoreboard *sb, |
260 | struct commit *commit, | |
261 | const char *path) | |
cee7f245 | 262 | { |
f6c0e191 | 263 | struct blame_entry *e; |
cee7f245 | 264 | |
f6c0e191 JH |
265 | for (e = sb->ent; e; e = e->next) { |
266 | if (e->suspect->commit == commit && | |
267 | !strcmp(e->suspect->path, path)) | |
54a4c617 | 268 | return origin_incref(e->suspect); |
cee7f245 | 269 | } |
854b97f6 | 270 | return make_origin(commit, path); |
cee7f245 JH |
271 | } |
272 | ||
1732a1fd JH |
273 | /* |
274 | * Fill the blob_sha1 field of an origin if it hasn't, so that later | |
275 | * call to fill_origin_blob() can use it to locate the data. blob_sha1 | |
276 | * for an origin is also used to pass the blame for the entire file to | |
277 | * the parent to detect the case where a child's blob is identical to | |
278 | * that of its parent's. | |
279 | */ | |
f6c0e191 JH |
280 | static int fill_blob_sha1(struct origin *origin) |
281 | { | |
282 | unsigned mode; | |
f6c0e191 JH |
283 | |
284 | if (!is_null_sha1(origin->blob_sha1)) | |
285 | return 0; | |
286 | if (get_tree_entry(origin->commit->object.sha1, | |
287 | origin->path, | |
288 | origin->blob_sha1, &mode)) | |
289 | goto error_out; | |
21666f1a | 290 | if (sha1_object_info(origin->blob_sha1, NULL) != OBJ_BLOB) |
f6c0e191 JH |
291 | goto error_out; |
292 | return 0; | |
293 | error_out: | |
294 | hashclr(origin->blob_sha1); | |
295 | return -1; | |
296 | } | |
297 | ||
1732a1fd JH |
298 | /* |
299 | * We have an origin -- check if the same path exists in the | |
300 | * parent and return an origin structure to represent it. | |
301 | */ | |
f6c0e191 | 302 | static struct origin *find_origin(struct scoreboard *sb, |
cee7f245 JH |
303 | struct commit *parent, |
304 | struct origin *origin) | |
305 | { | |
306 | struct origin *porigin = NULL; | |
307 | struct diff_options diff_opts; | |
f6c0e191 JH |
308 | const char *paths[2]; |
309 | ||
0d981c67 | 310 | if (parent->util) { |
1732a1fd JH |
311 | /* |
312 | * Each commit object can cache one origin in that | |
313 | * commit. This is a freestanding copy of origin and | |
314 | * not refcounted. | |
854b97f6 | 315 | */ |
0d981c67 | 316 | struct origin *cached = parent->util; |
854b97f6 | 317 | if (!strcmp(cached->path, origin->path)) { |
1732a1fd JH |
318 | /* |
319 | * The same path between origin and its parent | |
320 | * without renaming -- the most common case. | |
321 | */ | |
854b97f6 | 322 | porigin = get_origin(sb, parent, cached->path); |
1732a1fd JH |
323 | |
324 | /* | |
325 | * If the origin was newly created (i.e. get_origin | |
326 | * would call make_origin if none is found in the | |
327 | * scoreboard), it does not know the blob_sha1, | |
328 | * so copy it. Otherwise porigin was in the | |
329 | * scoreboard and already knows blob_sha1. | |
330 | */ | |
854b97f6 JH |
331 | if (porigin->refcnt == 1) |
332 | hashcpy(porigin->blob_sha1, cached->blob_sha1); | |
333 | return porigin; | |
334 | } | |
335 | /* otherwise it was not very useful; free it */ | |
336 | free(parent->util); | |
337 | parent->util = NULL; | |
0d981c67 JH |
338 | } |
339 | ||
f6c0e191 JH |
340 | /* See if the origin->path is different between parent |
341 | * and origin first. Most of the time they are the | |
342 | * same and diff-tree is fairly efficient about this. | |
343 | */ | |
344 | diff_setup(&diff_opts); | |
8f67f8ae | 345 | DIFF_OPT_SET(&diff_opts, RECURSIVE); |
f6c0e191 JH |
346 | diff_opts.detect_rename = 0; |
347 | diff_opts.output_format = DIFF_FORMAT_NO_OUTPUT; | |
348 | paths[0] = origin->path; | |
349 | paths[1] = NULL; | |
350 | ||
351 | diff_tree_setup_paths(paths, &diff_opts); | |
352 | if (diff_setup_done(&diff_opts) < 0) | |
353 | die("diff-setup"); | |
1cfe7733 JH |
354 | |
355 | if (is_null_sha1(origin->commit->object.sha1)) | |
356 | do_diff_cache(parent->tree->object.sha1, &diff_opts); | |
357 | else | |
358 | diff_tree_sha1(parent->tree->object.sha1, | |
359 | origin->commit->tree->object.sha1, | |
360 | "", &diff_opts); | |
f6c0e191 JH |
361 | diffcore_std(&diff_opts); |
362 | ||
363 | /* It is either one entry that says "modified", or "created", | |
364 | * or nothing. | |
365 | */ | |
366 | if (!diff_queued_diff.nr) { | |
367 | /* The path is the same as parent */ | |
368 | porigin = get_origin(sb, parent, origin->path); | |
369 | hashcpy(porigin->blob_sha1, origin->blob_sha1); | |
370 | } | |
371 | else if (diff_queued_diff.nr != 1) | |
acca687f | 372 | die("internal error in blame::find_origin"); |
f6c0e191 JH |
373 | else { |
374 | struct diff_filepair *p = diff_queued_diff.queue[0]; | |
375 | switch (p->status) { | |
376 | default: | |
acca687f | 377 | die("internal error in blame::find_origin (%c)", |
f6c0e191 JH |
378 | p->status); |
379 | case 'M': | |
380 | porigin = get_origin(sb, parent, origin->path); | |
381 | hashcpy(porigin->blob_sha1, p->one->sha1); | |
382 | break; | |
383 | case 'A': | |
384 | case 'T': | |
385 | /* Did not exist in parent, or type changed */ | |
386 | break; | |
387 | } | |
388 | } | |
389 | diff_flush(&diff_opts); | |
03b69c76 | 390 | diff_tree_release_paths(&diff_opts); |
0d981c67 | 391 | if (porigin) { |
1732a1fd JH |
392 | /* |
393 | * Create a freestanding copy that is not part of | |
394 | * the refcounted origin found in the scoreboard, and | |
395 | * cache it in the commit. | |
396 | */ | |
854b97f6 | 397 | struct origin *cached; |
1732a1fd | 398 | |
854b97f6 JH |
399 | cached = make_origin(porigin->commit, porigin->path); |
400 | hashcpy(cached->blob_sha1, porigin->blob_sha1); | |
401 | parent->util = cached; | |
0d981c67 | 402 | } |
f69e743d JH |
403 | return porigin; |
404 | } | |
f6c0e191 | 405 | |
1732a1fd JH |
406 | /* |
407 | * We have an origin -- find the path that corresponds to it in its | |
408 | * parent and return an origin structure to represent it. | |
409 | */ | |
f69e743d JH |
410 | static struct origin *find_rename(struct scoreboard *sb, |
411 | struct commit *parent, | |
412 | struct origin *origin) | |
413 | { | |
414 | struct origin *porigin = NULL; | |
415 | struct diff_options diff_opts; | |
416 | int i; | |
417 | const char *paths[2]; | |
cee7f245 JH |
418 | |
419 | diff_setup(&diff_opts); | |
8f67f8ae | 420 | DIFF_OPT_SET(&diff_opts, RECURSIVE); |
cee7f245 JH |
421 | diff_opts.detect_rename = DIFF_DETECT_RENAME; |
422 | diff_opts.output_format = DIFF_FORMAT_NO_OUTPUT; | |
2f3f8b21 | 423 | diff_opts.single_follow = origin->path; |
cee7f245 JH |
424 | paths[0] = NULL; |
425 | diff_tree_setup_paths(paths, &diff_opts); | |
426 | if (diff_setup_done(&diff_opts) < 0) | |
427 | die("diff-setup"); | |
1cfe7733 JH |
428 | |
429 | if (is_null_sha1(origin->commit->object.sha1)) | |
430 | do_diff_cache(parent->tree->object.sha1, &diff_opts); | |
431 | else | |
432 | diff_tree_sha1(parent->tree->object.sha1, | |
433 | origin->commit->tree->object.sha1, | |
434 | "", &diff_opts); | |
cee7f245 JH |
435 | diffcore_std(&diff_opts); |
436 | ||
437 | for (i = 0; i < diff_queued_diff.nr; i++) { | |
438 | struct diff_filepair *p = diff_queued_diff.queue[i]; | |
612702e8 | 439 | if ((p->status == 'R' || p->status == 'C') && |
f6c0e191 JH |
440 | !strcmp(p->two->path, origin->path)) { |
441 | porigin = get_origin(sb, parent, p->one->path); | |
442 | hashcpy(porigin->blob_sha1, p->one->sha1); | |
cee7f245 JH |
443 | break; |
444 | } | |
445 | } | |
446 | diff_flush(&diff_opts); | |
03b69c76 | 447 | diff_tree_release_paths(&diff_opts); |
cee7f245 JH |
448 | return porigin; |
449 | } | |
450 | ||
1732a1fd JH |
451 | /* |
452 | * Parsing of patch chunks... | |
453 | */ | |
cee7f245 JH |
454 | struct chunk { |
455 | /* line number in postimage; up to but not including this | |
456 | * line is the same as preimage | |
457 | */ | |
458 | int same; | |
459 | ||
460 | /* preimage line number after this chunk */ | |
461 | int p_next; | |
462 | ||
463 | /* postimage line number after this chunk */ | |
464 | int t_next; | |
465 | }; | |
466 | ||
467 | struct patch { | |
468 | struct chunk *chunks; | |
469 | int num; | |
470 | }; | |
471 | ||
472 | struct blame_diff_state { | |
473 | struct xdiff_emit_state xm; | |
474 | struct patch *ret; | |
475 | unsigned hunk_post_context; | |
476 | unsigned hunk_in_pre_context : 1; | |
477 | }; | |
478 | ||
479 | static void process_u_diff(void *state_, char *line, unsigned long len) | |
480 | { | |
481 | struct blame_diff_state *state = state_; | |
482 | struct chunk *chunk; | |
483 | int off1, off2, len1, len2, num; | |
484 | ||
cee7f245 JH |
485 | num = state->ret->num; |
486 | if (len < 4 || line[0] != '@' || line[1] != '@') { | |
487 | if (state->hunk_in_pre_context && line[0] == ' ') | |
488 | state->ret->chunks[num - 1].same++; | |
489 | else { | |
490 | state->hunk_in_pre_context = 0; | |
491 | if (line[0] == ' ') | |
492 | state->hunk_post_context++; | |
493 | else | |
494 | state->hunk_post_context = 0; | |
495 | } | |
496 | return; | |
497 | } | |
498 | ||
499 | if (num && state->hunk_post_context) { | |
500 | chunk = &state->ret->chunks[num - 1]; | |
501 | chunk->p_next -= state->hunk_post_context; | |
502 | chunk->t_next -= state->hunk_post_context; | |
503 | } | |
504 | state->ret->num = ++num; | |
505 | state->ret->chunks = xrealloc(state->ret->chunks, | |
506 | sizeof(struct chunk) * num); | |
507 | chunk = &state->ret->chunks[num - 1]; | |
508 | if (parse_hunk_header(line, len, &off1, &len1, &off2, &len2)) { | |
509 | state->ret->num--; | |
510 | return; | |
511 | } | |
512 | ||
513 | /* Line numbers in patch output are one based. */ | |
514 | off1--; | |
515 | off2--; | |
516 | ||
517 | chunk->same = len2 ? off2 : (off2 + 1); | |
518 | ||
519 | chunk->p_next = off1 + (len1 ? len1 : 1); | |
520 | chunk->t_next = chunk->same + len2; | |
521 | state->hunk_in_pre_context = 1; | |
522 | state->hunk_post_context = 0; | |
523 | } | |
524 | ||
525 | static struct patch *compare_buffer(mmfile_t *file_p, mmfile_t *file_o, | |
526 | int context) | |
527 | { | |
528 | struct blame_diff_state state; | |
529 | xpparam_t xpp; | |
530 | xdemitconf_t xecfg; | |
531 | xdemitcb_t ecb; | |
532 | ||
b82871b3 | 533 | xpp.flags = xdl_opts; |
30b25010 | 534 | memset(&xecfg, 0, sizeof(xecfg)); |
cee7f245 | 535 | xecfg.ctxlen = context; |
cee7f245 JH |
536 | ecb.outf = xdiff_outf; |
537 | ecb.priv = &state; | |
538 | memset(&state, 0, sizeof(state)); | |
539 | state.xm.consume = process_u_diff; | |
540 | state.ret = xmalloc(sizeof(struct patch)); | |
541 | state.ret->chunks = NULL; | |
542 | state.ret->num = 0; | |
543 | ||
c279d7e9 | 544 | xdi_diff(file_p, file_o, &xpp, &xecfg, &ecb); |
cee7f245 JH |
545 | |
546 | if (state.ret->num) { | |
547 | struct chunk *chunk; | |
548 | chunk = &state.ret->chunks[state.ret->num - 1]; | |
549 | chunk->p_next -= state.hunk_post_context; | |
550 | chunk->t_next -= state.hunk_post_context; | |
551 | } | |
552 | return state.ret; | |
553 | } | |
554 | ||
1732a1fd JH |
555 | /* |
556 | * Run diff between two origins and grab the patch output, so that | |
557 | * we can pass blame for lines origin is currently suspected for | |
558 | * to its parent. | |
559 | */ | |
cee7f245 JH |
560 | static struct patch *get_patch(struct origin *parent, struct origin *origin) |
561 | { | |
562 | mmfile_t file_p, file_o; | |
cee7f245 JH |
563 | struct patch *patch; |
564 | ||
c2e525d9 JH |
565 | fill_origin_blob(parent, &file_p); |
566 | fill_origin_blob(origin, &file_o); | |
567 | if (!file_p.ptr || !file_o.ptr) | |
cee7f245 | 568 | return NULL; |
cee7f245 | 569 | patch = compare_buffer(&file_p, &file_o, 0); |
c2e525d9 | 570 | num_get_patch++; |
cee7f245 JH |
571 | return patch; |
572 | } | |
573 | ||
574 | static void free_patch(struct patch *p) | |
575 | { | |
576 | free(p->chunks); | |
577 | free(p); | |
578 | } | |
579 | ||
1732a1fd | 580 | /* |
3dff5379 | 581 | * Link in a new blame entry to the scoreboard. Entries that cover the |
1732a1fd JH |
582 | * same line range have been removed from the scoreboard previously. |
583 | */ | |
cee7f245 JH |
584 | static void add_blame_entry(struct scoreboard *sb, struct blame_entry *e) |
585 | { | |
586 | struct blame_entry *ent, *prev = NULL; | |
587 | ||
54a4c617 JH |
588 | origin_incref(e->suspect); |
589 | ||
cee7f245 JH |
590 | for (ent = sb->ent; ent && ent->lno < e->lno; ent = ent->next) |
591 | prev = ent; | |
592 | ||
593 | /* prev, if not NULL, is the last one that is below e */ | |
594 | e->prev = prev; | |
595 | if (prev) { | |
596 | e->next = prev->next; | |
597 | prev->next = e; | |
598 | } | |
599 | else { | |
600 | e->next = sb->ent; | |
601 | sb->ent = e; | |
602 | } | |
603 | if (e->next) | |
604 | e->next->prev = e; | |
605 | } | |
606 | ||
1732a1fd JH |
607 | /* |
608 | * src typically is on-stack; we want to copy the information in it to | |
790296fd | 609 | * a malloced blame_entry that is already on the linked list of the |
1732a1fd JH |
610 | * scoreboard. The origin of dst loses a refcnt while the origin of src |
611 | * gains one. | |
612 | */ | |
cee7f245 JH |
613 | static void dup_entry(struct blame_entry *dst, struct blame_entry *src) |
614 | { | |
615 | struct blame_entry *p, *n; | |
54a4c617 | 616 | |
cee7f245 JH |
617 | p = dst->prev; |
618 | n = dst->next; | |
54a4c617 JH |
619 | origin_incref(src->suspect); |
620 | origin_decref(dst->suspect); | |
cee7f245 JH |
621 | memcpy(dst, src, sizeof(*src)); |
622 | dst->prev = p; | |
623 | dst->next = n; | |
5ff62c30 | 624 | dst->score = 0; |
cee7f245 JH |
625 | } |
626 | ||
627 | static const char *nth_line(struct scoreboard *sb, int lno) | |
628 | { | |
629 | return sb->final_buf + sb->lineno[lno]; | |
630 | } | |
631 | ||
1732a1fd JH |
632 | /* |
633 | * It is known that lines between tlno to same came from parent, and e | |
634 | * has an overlap with that range. it also is known that parent's | |
635 | * line plno corresponds to e's line tlno. | |
636 | * | |
637 | * <---- e -----> | |
638 | * <------> | |
639 | * <------------> | |
640 | * <------------> | |
641 | * <------------------> | |
642 | * | |
643 | * Split e into potentially three parts; before this chunk, the chunk | |
644 | * to be blamed for the parent, and after that portion. | |
645 | */ | |
54a4c617 | 646 | static void split_overlap(struct blame_entry *split, |
cee7f245 JH |
647 | struct blame_entry *e, |
648 | int tlno, int plno, int same, | |
649 | struct origin *parent) | |
650 | { | |
cee7f245 JH |
651 | int chunk_end_lno; |
652 | memset(split, 0, sizeof(struct blame_entry [3])); | |
653 | ||
654 | if (e->s_lno < tlno) { | |
655 | /* there is a pre-chunk part not blamed on parent */ | |
54a4c617 | 656 | split[0].suspect = origin_incref(e->suspect); |
cee7f245 JH |
657 | split[0].lno = e->lno; |
658 | split[0].s_lno = e->s_lno; | |
659 | split[0].num_lines = tlno - e->s_lno; | |
660 | split[1].lno = e->lno + tlno - e->s_lno; | |
661 | split[1].s_lno = plno; | |
662 | } | |
663 | else { | |
664 | split[1].lno = e->lno; | |
665 | split[1].s_lno = plno + (e->s_lno - tlno); | |
666 | } | |
667 | ||
668 | if (same < e->s_lno + e->num_lines) { | |
669 | /* there is a post-chunk part not blamed on parent */ | |
54a4c617 | 670 | split[2].suspect = origin_incref(e->suspect); |
cee7f245 JH |
671 | split[2].lno = e->lno + (same - e->s_lno); |
672 | split[2].s_lno = e->s_lno + (same - e->s_lno); | |
673 | split[2].num_lines = e->s_lno + e->num_lines - same; | |
674 | chunk_end_lno = split[2].lno; | |
675 | } | |
676 | else | |
677 | chunk_end_lno = e->lno + e->num_lines; | |
678 | split[1].num_lines = chunk_end_lno - split[1].lno; | |
679 | ||
1732a1fd JH |
680 | /* |
681 | * if it turns out there is nothing to blame the parent for, | |
682 | * forget about the splitting. !split[1].suspect signals this. | |
683 | */ | |
cee7f245 JH |
684 | if (split[1].num_lines < 1) |
685 | return; | |
54a4c617 | 686 | split[1].suspect = origin_incref(parent); |
cee7f245 JH |
687 | } |
688 | ||
1732a1fd JH |
689 | /* |
690 | * split_overlap() divided an existing blame e into up to three parts | |
691 | * in split. Adjust the linked list of blames in the scoreboard to | |
692 | * reflect the split. | |
693 | */ | |
cee7f245 | 694 | static void split_blame(struct scoreboard *sb, |
54a4c617 | 695 | struct blame_entry *split, |
cee7f245 JH |
696 | struct blame_entry *e) |
697 | { | |
698 | struct blame_entry *new_entry; | |
699 | ||
700 | if (split[0].suspect && split[2].suspect) { | |
1732a1fd | 701 | /* The first part (reuse storage for the existing entry e) */ |
cee7f245 JH |
702 | dup_entry(e, &split[0]); |
703 | ||
1732a1fd | 704 | /* The last part -- me */ |
cee7f245 JH |
705 | new_entry = xmalloc(sizeof(*new_entry)); |
706 | memcpy(new_entry, &(split[2]), sizeof(struct blame_entry)); | |
707 | add_blame_entry(sb, new_entry); | |
708 | ||
1732a1fd | 709 | /* ... and the middle part -- parent */ |
cee7f245 JH |
710 | new_entry = xmalloc(sizeof(*new_entry)); |
711 | memcpy(new_entry, &(split[1]), sizeof(struct blame_entry)); | |
712 | add_blame_entry(sb, new_entry); | |
713 | } | |
714 | else if (!split[0].suspect && !split[2].suspect) | |
1732a1fd JH |
715 | /* |
716 | * The parent covers the entire area; reuse storage for | |
717 | * e and replace it with the parent. | |
718 | */ | |
cee7f245 JH |
719 | dup_entry(e, &split[1]); |
720 | else if (split[0].suspect) { | |
1732a1fd | 721 | /* me and then parent */ |
cee7f245 JH |
722 | dup_entry(e, &split[0]); |
723 | ||
724 | new_entry = xmalloc(sizeof(*new_entry)); | |
725 | memcpy(new_entry, &(split[1]), sizeof(struct blame_entry)); | |
726 | add_blame_entry(sb, new_entry); | |
727 | } | |
728 | else { | |
1732a1fd | 729 | /* parent and then me */ |
cee7f245 JH |
730 | dup_entry(e, &split[1]); |
731 | ||
732 | new_entry = xmalloc(sizeof(*new_entry)); | |
733 | memcpy(new_entry, &(split[2]), sizeof(struct blame_entry)); | |
734 | add_blame_entry(sb, new_entry); | |
735 | } | |
736 | ||
54a4c617 | 737 | if (DEBUG) { /* sanity */ |
cee7f245 | 738 | struct blame_entry *ent; |
612702e8 | 739 | int lno = sb->ent->lno, corrupt = 0; |
cee7f245 JH |
740 | |
741 | for (ent = sb->ent; ent; ent = ent->next) { | |
742 | if (lno != ent->lno) | |
743 | corrupt = 1; | |
744 | if (ent->s_lno < 0) | |
745 | corrupt = 1; | |
746 | lno += ent->num_lines; | |
747 | } | |
748 | if (corrupt) { | |
612702e8 | 749 | lno = sb->ent->lno; |
cee7f245 JH |
750 | for (ent = sb->ent; ent; ent = ent->next) { |
751 | printf("L %8d l %8d n %8d\n", | |
752 | lno, ent->lno, ent->num_lines); | |
753 | lno = ent->lno + ent->num_lines; | |
754 | } | |
755 | die("oops"); | |
756 | } | |
757 | } | |
758 | } | |
759 | ||
1732a1fd JH |
760 | /* |
761 | * After splitting the blame, the origins used by the | |
762 | * on-stack blame_entry should lose one refcnt each. | |
763 | */ | |
54a4c617 JH |
764 | static void decref_split(struct blame_entry *split) |
765 | { | |
766 | int i; | |
767 | ||
768 | for (i = 0; i < 3; i++) | |
769 | origin_decref(split[i].suspect); | |
770 | } | |
771 | ||
1732a1fd JH |
772 | /* |
773 | * Helper for blame_chunk(). blame_entry e is known to overlap with | |
774 | * the patch hunk; split it and pass blame to the parent. | |
775 | */ | |
cee7f245 JH |
776 | static void blame_overlap(struct scoreboard *sb, struct blame_entry *e, |
777 | int tlno, int plno, int same, | |
778 | struct origin *parent) | |
779 | { | |
780 | struct blame_entry split[3]; | |
781 | ||
782 | split_overlap(split, e, tlno, plno, same, parent); | |
54a4c617 JH |
783 | if (split[1].suspect) |
784 | split_blame(sb, split, e); | |
785 | decref_split(split); | |
cee7f245 JH |
786 | } |
787 | ||
1732a1fd JH |
788 | /* |
789 | * Find the line number of the last line the target is suspected for. | |
790 | */ | |
cee7f245 JH |
791 | static int find_last_in_target(struct scoreboard *sb, struct origin *target) |
792 | { | |
793 | struct blame_entry *e; | |
794 | int last_in_target = -1; | |
795 | ||
796 | for (e = sb->ent; e; e = e->next) { | |
171dccd5 | 797 | if (e->guilty || !same_suspect(e->suspect, target)) |
cee7f245 JH |
798 | continue; |
799 | if (last_in_target < e->s_lno + e->num_lines) | |
800 | last_in_target = e->s_lno + e->num_lines; | |
801 | } | |
802 | return last_in_target; | |
803 | } | |
804 | ||
1732a1fd JH |
805 | /* |
806 | * Process one hunk from the patch between the current suspect for | |
807 | * blame_entry e and its parent. Find and split the overlap, and | |
808 | * pass blame to the overlapping part to the parent. | |
809 | */ | |
cee7f245 JH |
810 | static void blame_chunk(struct scoreboard *sb, |
811 | int tlno, int plno, int same, | |
812 | struct origin *target, struct origin *parent) | |
813 | { | |
612702e8 | 814 | struct blame_entry *e; |
cee7f245 | 815 | |
612702e8 | 816 | for (e = sb->ent; e; e = e->next) { |
171dccd5 | 817 | if (e->guilty || !same_suspect(e->suspect, target)) |
cee7f245 JH |
818 | continue; |
819 | if (same <= e->s_lno) | |
820 | continue; | |
821 | if (tlno < e->s_lno + e->num_lines) | |
822 | blame_overlap(sb, e, tlno, plno, same, parent); | |
823 | } | |
824 | } | |
825 | ||
1732a1fd JH |
826 | /* |
827 | * We are looking at the origin 'target' and aiming to pass blame | |
828 | * for the lines it is suspected to its parent. Run diff to find | |
829 | * which lines came from parent and pass blame for them. | |
830 | */ | |
cee7f245 JH |
831 | static int pass_blame_to_parent(struct scoreboard *sb, |
832 | struct origin *target, | |
833 | struct origin *parent) | |
834 | { | |
835 | int i, last_in_target, plno, tlno; | |
836 | struct patch *patch; | |
837 | ||
838 | last_in_target = find_last_in_target(sb, target); | |
839 | if (last_in_target < 0) | |
840 | return 1; /* nothing remains for this target */ | |
841 | ||
842 | patch = get_patch(parent, target); | |
843 | plno = tlno = 0; | |
844 | for (i = 0; i < patch->num; i++) { | |
845 | struct chunk *chunk = &patch->chunks[i]; | |
846 | ||
cee7f245 JH |
847 | blame_chunk(sb, tlno, plno, chunk->same, target, parent); |
848 | plno = chunk->p_next; | |
849 | tlno = chunk->t_next; | |
850 | } | |
1732a1fd | 851 | /* The rest (i.e. anything after tlno) are the same as the parent */ |
cee7f245 JH |
852 | blame_chunk(sb, tlno, plno, last_in_target, target, parent); |
853 | ||
854 | free_patch(patch); | |
855 | return 0; | |
856 | } | |
857 | ||
1732a1fd JH |
858 | /* |
859 | * The lines in blame_entry after splitting blames many times can become | |
860 | * very small and trivial, and at some point it becomes pointless to | |
861 | * blame the parents. E.g. "\t\t}\n\t}\n\n" appears everywhere in any | |
862 | * ordinary C program, and it is not worth to say it was copied from | |
863 | * totally unrelated file in the parent. | |
864 | * | |
865 | * Compute how trivial the lines in the blame_entry are. | |
866 | */ | |
5ff62c30 JH |
867 | static unsigned ent_score(struct scoreboard *sb, struct blame_entry *e) |
868 | { | |
869 | unsigned score; | |
870 | const char *cp, *ep; | |
871 | ||
872 | if (e->score) | |
873 | return e->score; | |
874 | ||
612702e8 | 875 | score = 1; |
5ff62c30 JH |
876 | cp = nth_line(sb, e->lno); |
877 | ep = nth_line(sb, e->lno + e->num_lines); | |
878 | while (cp < ep) { | |
879 | unsigned ch = *((unsigned char *)cp); | |
880 | if (isalnum(ch)) | |
881 | score++; | |
882 | cp++; | |
883 | } | |
884 | e->score = score; | |
885 | return score; | |
886 | } | |
887 | ||
1732a1fd JH |
888 | /* |
889 | * best_so_far[] and this[] are both a split of an existing blame_entry | |
890 | * that passes blame to the parent. Maintain best_so_far the best split | |
891 | * so far, by comparing this and best_so_far and copying this into | |
892 | * bst_so_far as needed. | |
893 | */ | |
5ff62c30 | 894 | static void copy_split_if_better(struct scoreboard *sb, |
54a4c617 JH |
895 | struct blame_entry *best_so_far, |
896 | struct blame_entry *this) | |
d24bba80 | 897 | { |
54a4c617 JH |
898 | int i; |
899 | ||
d24bba80 JH |
900 | if (!this[1].suspect) |
901 | return; | |
5ff62c30 JH |
902 | if (best_so_far[1].suspect) { |
903 | if (ent_score(sb, &this[1]) < ent_score(sb, &best_so_far[1])) | |
904 | return; | |
905 | } | |
54a4c617 JH |
906 | |
907 | for (i = 0; i < 3; i++) | |
908 | origin_incref(this[i].suspect); | |
909 | decref_split(best_so_far); | |
d24bba80 JH |
910 | memcpy(best_so_far, this, sizeof(struct blame_entry [3])); |
911 | } | |
912 | ||
dd166aa8 JH |
913 | /* |
914 | * We are looking at a part of the final image represented by | |
915 | * ent (tlno and same are offset by ent->s_lno). | |
916 | * tlno is where we are looking at in the final image. | |
917 | * up to (but not including) same match preimage. | |
918 | * plno is where we are looking at in the preimage. | |
919 | * | |
920 | * <-------------- final image ----------------------> | |
921 | * <------ent------> | |
922 | * ^tlno ^same | |
923 | * <---------preimage-----> | |
924 | * ^plno | |
925 | * | |
926 | * All line numbers are 0-based. | |
927 | */ | |
928 | static void handle_split(struct scoreboard *sb, | |
929 | struct blame_entry *ent, | |
930 | int tlno, int plno, int same, | |
931 | struct origin *parent, | |
932 | struct blame_entry *split) | |
933 | { | |
934 | if (ent->num_lines <= tlno) | |
935 | return; | |
936 | if (tlno < same) { | |
937 | struct blame_entry this[3]; | |
938 | tlno += ent->s_lno; | |
939 | same += ent->s_lno; | |
940 | split_overlap(this, ent, tlno, plno, same, parent); | |
941 | copy_split_if_better(sb, split, this); | |
942 | decref_split(this); | |
943 | } | |
944 | } | |
945 | ||
1732a1fd JH |
946 | /* |
947 | * Find the lines from parent that are the same as ent so that | |
948 | * we can pass blames to it. file_p has the blob contents for | |
949 | * the parent. | |
950 | */ | |
d24bba80 JH |
951 | static void find_copy_in_blob(struct scoreboard *sb, |
952 | struct blame_entry *ent, | |
953 | struct origin *parent, | |
54a4c617 | 954 | struct blame_entry *split, |
d24bba80 JH |
955 | mmfile_t *file_p) |
956 | { | |
957 | const char *cp; | |
958 | int cnt; | |
959 | mmfile_t file_o; | |
960 | struct patch *patch; | |
961 | int i, plno, tlno; | |
962 | ||
1732a1fd JH |
963 | /* |
964 | * Prepare mmfile that contains only the lines in ent. | |
965 | */ | |
d24bba80 JH |
966 | cp = nth_line(sb, ent->lno); |
967 | file_o.ptr = (char*) cp; | |
968 | cnt = ent->num_lines; | |
969 | ||
970 | while (cnt && cp < sb->final_buf + sb->final_buf_size) { | |
971 | if (*cp++ == '\n') | |
972 | cnt--; | |
973 | } | |
974 | file_o.size = cp - file_o.ptr; | |
975 | ||
976 | patch = compare_buffer(file_p, &file_o, 1); | |
977 | ||
dd166aa8 JH |
978 | /* |
979 | * file_o is a part of final image we are annotating. | |
980 | * file_p partially may match that image. | |
981 | */ | |
d24bba80 JH |
982 | memset(split, 0, sizeof(struct blame_entry [3])); |
983 | plno = tlno = 0; | |
984 | for (i = 0; i < patch->num; i++) { | |
985 | struct chunk *chunk = &patch->chunks[i]; | |
986 | ||
dd166aa8 | 987 | handle_split(sb, ent, tlno, plno, chunk->same, parent, split); |
d24bba80 JH |
988 | plno = chunk->p_next; |
989 | tlno = chunk->t_next; | |
990 | } | |
dd166aa8 JH |
991 | /* remainder, if any, all match the preimage */ |
992 | handle_split(sb, ent, tlno, plno, ent->num_lines, parent, split); | |
d24bba80 JH |
993 | free_patch(patch); |
994 | } | |
995 | ||
1732a1fd JH |
996 | /* |
997 | * See if lines currently target is suspected for can be attributed to | |
998 | * parent. | |
999 | */ | |
d24bba80 JH |
1000 | static int find_move_in_parent(struct scoreboard *sb, |
1001 | struct origin *target, | |
1002 | struct origin *parent) | |
1003 | { | |
650e2f67 | 1004 | int last_in_target, made_progress; |
46014766 | 1005 | struct blame_entry *e, split[3]; |
d24bba80 | 1006 | mmfile_t file_p; |
d24bba80 JH |
1007 | |
1008 | last_in_target = find_last_in_target(sb, target); | |
1009 | if (last_in_target < 0) | |
1010 | return 1; /* nothing remains for this target */ | |
1011 | ||
c2e525d9 JH |
1012 | fill_origin_blob(parent, &file_p); |
1013 | if (!file_p.ptr) | |
d24bba80 | 1014 | return 0; |
d24bba80 | 1015 | |
650e2f67 JH |
1016 | made_progress = 1; |
1017 | while (made_progress) { | |
1018 | made_progress = 0; | |
1019 | for (e = sb->ent; e; e = e->next) { | |
171dccd5 | 1020 | if (e->guilty || !same_suspect(e->suspect, target)) |
650e2f67 JH |
1021 | continue; |
1022 | find_copy_in_blob(sb, e, parent, split, &file_p); | |
1023 | if (split[1].suspect && | |
1024 | blame_move_score < ent_score(sb, &split[1])) { | |
1025 | split_blame(sb, split, e); | |
1026 | made_progress = 1; | |
1027 | } | |
1028 | decref_split(split); | |
1029 | } | |
d24bba80 | 1030 | } |
d24bba80 JH |
1031 | return 0; |
1032 | } | |
1033 | ||
33494784 JH |
1034 | struct blame_list { |
1035 | struct blame_entry *ent; | |
1036 | struct blame_entry split[3]; | |
1037 | }; | |
1038 | ||
1732a1fd JH |
1039 | /* |
1040 | * Count the number of entries the target is suspected for, | |
1041 | * and prepare a list of entry and the best split. | |
1042 | */ | |
33494784 JH |
1043 | static struct blame_list *setup_blame_list(struct scoreboard *sb, |
1044 | struct origin *target, | |
1045 | int *num_ents_p) | |
1046 | { | |
1047 | struct blame_entry *e; | |
1048 | int num_ents, i; | |
1049 | struct blame_list *blame_list = NULL; | |
1050 | ||
33494784 | 1051 | for (e = sb->ent, num_ents = 0; e; e = e->next) |
171dccd5 | 1052 | if (!e->guilty && same_suspect(e->suspect, target)) |
33494784 JH |
1053 | num_ents++; |
1054 | if (num_ents) { | |
1055 | blame_list = xcalloc(num_ents, sizeof(struct blame_list)); | |
1056 | for (e = sb->ent, i = 0; e; e = e->next) | |
171dccd5 | 1057 | if (!e->guilty && same_suspect(e->suspect, target)) |
33494784 JH |
1058 | blame_list[i++].ent = e; |
1059 | } | |
1060 | *num_ents_p = num_ents; | |
1061 | return blame_list; | |
1062 | } | |
1063 | ||
1732a1fd JH |
1064 | /* |
1065 | * For lines target is suspected for, see if we can find code movement | |
1066 | * across file boundary from the parent commit. porigin is the path | |
1067 | * in the parent we already tried. | |
1068 | */ | |
18abd745 JH |
1069 | static int find_copy_in_parent(struct scoreboard *sb, |
1070 | struct origin *target, | |
1071 | struct commit *parent, | |
1072 | struct origin *porigin, | |
1073 | int opt) | |
1074 | { | |
1075 | struct diff_options diff_opts; | |
1076 | const char *paths[1]; | |
aec8fa1f | 1077 | int i, j; |
33494784 JH |
1078 | int retval; |
1079 | struct blame_list *blame_list; | |
aec8fa1f | 1080 | int num_ents; |
18abd745 | 1081 | |
33494784 JH |
1082 | blame_list = setup_blame_list(sb, target, &num_ents); |
1083 | if (!blame_list) | |
18abd745 JH |
1084 | return 1; /* nothing remains for this target */ |
1085 | ||
1086 | diff_setup(&diff_opts); | |
8f67f8ae | 1087 | DIFF_OPT_SET(&diff_opts, RECURSIVE); |
18abd745 JH |
1088 | diff_opts.output_format = DIFF_FORMAT_NO_OUTPUT; |
1089 | ||
18abd745 JH |
1090 | paths[0] = NULL; |
1091 | diff_tree_setup_paths(paths, &diff_opts); | |
1092 | if (diff_setup_done(&diff_opts) < 0) | |
1093 | die("diff-setup"); | |
33494784 JH |
1094 | |
1095 | /* Try "find copies harder" on new path if requested; | |
1096 | * we do not want to use diffcore_rename() actually to | |
1097 | * match things up; find_copies_harder is set only to | |
1098 | * force diff_tree_sha1() to feed all filepairs to diff_queue, | |
1099 | * and this code needs to be after diff_setup_done(), which | |
1100 | * usually makes find-copies-harder imply copy detection. | |
1101 | */ | |
c63777c0 JH |
1102 | if ((opt & PICKAXE_BLAME_COPY_HARDEST) |
1103 | || ((opt & PICKAXE_BLAME_COPY_HARDER) | |
1104 | && (!porigin || strcmp(target->path, porigin->path)))) | |
8f67f8ae | 1105 | DIFF_OPT_SET(&diff_opts, FIND_COPIES_HARDER); |
33494784 | 1106 | |
1cfe7733 JH |
1107 | if (is_null_sha1(target->commit->object.sha1)) |
1108 | do_diff_cache(parent->tree->object.sha1, &diff_opts); | |
1109 | else | |
1110 | diff_tree_sha1(parent->tree->object.sha1, | |
1111 | target->commit->tree->object.sha1, | |
1112 | "", &diff_opts); | |
18abd745 | 1113 | |
8f67f8ae | 1114 | if (!DIFF_OPT_TST(&diff_opts, FIND_COPIES_HARDER)) |
33494784 | 1115 | diffcore_std(&diff_opts); |
18abd745 | 1116 | |
33494784 JH |
1117 | retval = 0; |
1118 | while (1) { | |
1119 | int made_progress = 0; | |
1120 | ||
1121 | for (i = 0; i < diff_queued_diff.nr; i++) { | |
1122 | struct diff_filepair *p = diff_queued_diff.queue[i]; | |
1123 | struct origin *norigin; | |
1124 | mmfile_t file_p; | |
33494784 JH |
1125 | struct blame_entry this[3]; |
1126 | ||
1127 | if (!DIFF_FILE_VALID(p->one)) | |
1128 | continue; /* does not exist in parent */ | |
1129 | if (porigin && !strcmp(p->one->path, porigin->path)) | |
1130 | /* find_move already dealt with this path */ | |
1131 | continue; | |
1132 | ||
1133 | norigin = get_origin(sb, parent, p->one->path); | |
1134 | hashcpy(norigin->blob_sha1, p->one->sha1); | |
c2e525d9 | 1135 | fill_origin_blob(norigin, &file_p); |
33494784 JH |
1136 | if (!file_p.ptr) |
1137 | continue; | |
1138 | ||
1139 | for (j = 0; j < num_ents; j++) { | |
1140 | find_copy_in_blob(sb, blame_list[j].ent, | |
1141 | norigin, this, &file_p); | |
1142 | copy_split_if_better(sb, blame_list[j].split, | |
1143 | this); | |
1144 | decref_split(this); | |
1145 | } | |
33494784 JH |
1146 | origin_decref(norigin); |
1147 | } | |
18abd745 | 1148 | |
aec8fa1f | 1149 | for (j = 0; j < num_ents; j++) { |
33494784 JH |
1150 | struct blame_entry *split = blame_list[j].split; |
1151 | if (split[1].suspect && | |
1152 | blame_copy_score < ent_score(sb, &split[1])) { | |
1153 | split_blame(sb, split, blame_list[j].ent); | |
1154 | made_progress = 1; | |
1155 | } | |
1156 | decref_split(split); | |
18abd745 | 1157 | } |
33494784 | 1158 | free(blame_list); |
aec8fa1f | 1159 | |
33494784 JH |
1160 | if (!made_progress) |
1161 | break; | |
1162 | blame_list = setup_blame_list(sb, target, &num_ents); | |
1163 | if (!blame_list) { | |
1164 | retval = 1; | |
1165 | break; | |
1166 | } | |
18abd745 | 1167 | } |
33494784 | 1168 | diff_flush(&diff_opts); |
03b69c76 | 1169 | diff_tree_release_paths(&diff_opts); |
33494784 | 1170 | return retval; |
18abd745 JH |
1171 | } |
1172 | ||
1732a1fd JH |
1173 | /* |
1174 | * The blobs of origin and porigin exactly match, so everything | |
c2e525d9 JH |
1175 | * origin is suspected for can be blamed on the parent. |
1176 | */ | |
1177 | static void pass_whole_blame(struct scoreboard *sb, | |
1178 | struct origin *origin, struct origin *porigin) | |
1179 | { | |
1180 | struct blame_entry *e; | |
1181 | ||
1182 | if (!porigin->file.ptr && origin->file.ptr) { | |
1183 | /* Steal its file */ | |
1184 | porigin->file = origin->file; | |
1185 | origin->file.ptr = NULL; | |
1186 | } | |
1187 | for (e = sb->ent; e; e = e->next) { | |
171dccd5 | 1188 | if (!same_suspect(e->suspect, origin)) |
c2e525d9 JH |
1189 | continue; |
1190 | origin_incref(porigin); | |
1191 | origin_decref(e->suspect); | |
1192 | e->suspect = porigin; | |
1193 | } | |
1194 | } | |
1195 | ||
cee7f245 JH |
1196 | #define MAXPARENT 16 |
1197 | ||
d24bba80 | 1198 | static void pass_blame(struct scoreboard *sb, struct origin *origin, int opt) |
cee7f245 | 1199 | { |
f69e743d | 1200 | int i, pass; |
cee7f245 JH |
1201 | struct commit *commit = origin->commit; |
1202 | struct commit_list *parent; | |
1203 | struct origin *parent_origin[MAXPARENT], *porigin; | |
1204 | ||
1205 | memset(parent_origin, 0, sizeof(parent_origin)); | |
cee7f245 | 1206 | |
f69e743d JH |
1207 | /* The first pass looks for unrenamed path to optimize for |
1208 | * common cases, then we look for renames in the second pass. | |
1209 | */ | |
1210 | for (pass = 0; pass < 2; pass++) { | |
1211 | struct origin *(*find)(struct scoreboard *, | |
1212 | struct commit *, struct origin *); | |
1213 | find = pass ? find_rename : find_origin; | |
1214 | ||
1215 | for (i = 0, parent = commit->parents; | |
1216 | i < MAXPARENT && parent; | |
1217 | parent = parent->next, i++) { | |
1218 | struct commit *p = parent->item; | |
0421d9f8 | 1219 | int j, same; |
f69e743d JH |
1220 | |
1221 | if (parent_origin[i]) | |
1222 | continue; | |
1223 | if (parse_commit(p)) | |
1224 | continue; | |
0d981c67 | 1225 | porigin = find(sb, p, origin); |
f69e743d JH |
1226 | if (!porigin) |
1227 | continue; | |
1228 | if (!hashcmp(porigin->blob_sha1, origin->blob_sha1)) { | |
c2e525d9 | 1229 | pass_whole_blame(sb, origin, porigin); |
f69e743d JH |
1230 | origin_decref(porigin); |
1231 | goto finish; | |
1232 | } | |
0421d9f8 | 1233 | for (j = same = 0; j < i; j++) |
33494784 JH |
1234 | if (parent_origin[j] && |
1235 | !hashcmp(parent_origin[j]->blob_sha1, | |
0421d9f8 JH |
1236 | porigin->blob_sha1)) { |
1237 | same = 1; | |
1238 | break; | |
1239 | } | |
1240 | if (!same) | |
1241 | parent_origin[i] = porigin; | |
1242 | else | |
1243 | origin_decref(porigin); | |
cee7f245 | 1244 | } |
cee7f245 JH |
1245 | } |
1246 | ||
c2e525d9 | 1247 | num_commits++; |
cee7f245 JH |
1248 | for (i = 0, parent = commit->parents; |
1249 | i < MAXPARENT && parent; | |
1250 | parent = parent->next, i++) { | |
1251 | struct origin *porigin = parent_origin[i]; | |
1252 | if (!porigin) | |
1253 | continue; | |
1254 | if (pass_blame_to_parent(sb, origin, porigin)) | |
54a4c617 | 1255 | goto finish; |
cee7f245 | 1256 | } |
d24bba80 JH |
1257 | |
1258 | /* | |
1732a1fd | 1259 | * Optionally find moves in parents' files. |
d24bba80 JH |
1260 | */ |
1261 | if (opt & PICKAXE_BLAME_MOVE) | |
1262 | for (i = 0, parent = commit->parents; | |
1263 | i < MAXPARENT && parent; | |
1264 | parent = parent->next, i++) { | |
1265 | struct origin *porigin = parent_origin[i]; | |
1266 | if (!porigin) | |
1267 | continue; | |
1268 | if (find_move_in_parent(sb, origin, porigin)) | |
54a4c617 | 1269 | goto finish; |
d24bba80 JH |
1270 | } |
1271 | ||
18abd745 | 1272 | /* |
1732a1fd | 1273 | * Optionally find copies from parents' files. |
18abd745 JH |
1274 | */ |
1275 | if (opt & PICKAXE_BLAME_COPY) | |
1276 | for (i = 0, parent = commit->parents; | |
1277 | i < MAXPARENT && parent; | |
1278 | parent = parent->next, i++) { | |
1279 | struct origin *porigin = parent_origin[i]; | |
1280 | if (find_copy_in_parent(sb, origin, parent->item, | |
1281 | porigin, opt)) | |
54a4c617 | 1282 | goto finish; |
18abd745 | 1283 | } |
54a4c617 JH |
1284 | |
1285 | finish: | |
7c3c7962 JH |
1286 | for (i = 0; i < MAXPARENT; i++) { |
1287 | if (parent_origin[i]) { | |
1288 | drop_origin_blob(parent_origin[i]); | |
1289 | origin_decref(parent_origin[i]); | |
1290 | } | |
1291 | } | |
1292 | drop_origin_blob(origin); | |
cee7f245 JH |
1293 | } |
1294 | ||
1732a1fd JH |
1295 | /* |
1296 | * Information on commits, used for output. | |
1297 | */ | |
cee7f245 JH |
1298 | struct commit_info |
1299 | { | |
3a55602e SP |
1300 | const char *author; |
1301 | const char *author_mail; | |
cee7f245 | 1302 | unsigned long author_time; |
3a55602e | 1303 | const char *author_tz; |
cee7f245 JH |
1304 | |
1305 | /* filled only when asked for details */ | |
3a55602e SP |
1306 | const char *committer; |
1307 | const char *committer_mail; | |
cee7f245 | 1308 | unsigned long committer_time; |
3a55602e | 1309 | const char *committer_tz; |
cee7f245 | 1310 | |
3a55602e | 1311 | const char *summary; |
cee7f245 JH |
1312 | }; |
1313 | ||
1732a1fd JH |
1314 | /* |
1315 | * Parse author/committer line in the commit object buffer | |
1316 | */ | |
cee7f245 | 1317 | static void get_ac_line(const char *inbuf, const char *what, |
3a55602e SP |
1318 | int bufsz, char *person, const char **mail, |
1319 | unsigned long *time, const char **tz) | |
cee7f245 | 1320 | { |
f9567384 JH |
1321 | int len, tzlen, maillen; |
1322 | char *tmp, *endp, *timepos; | |
cee7f245 JH |
1323 | |
1324 | tmp = strstr(inbuf, what); | |
1325 | if (!tmp) | |
1326 | goto error_out; | |
1327 | tmp += strlen(what); | |
1328 | endp = strchr(tmp, '\n'); | |
1329 | if (!endp) | |
1330 | len = strlen(tmp); | |
1331 | else | |
1332 | len = endp - tmp; | |
1333 | if (bufsz <= len) { | |
1334 | error_out: | |
1335 | /* Ugh */ | |
3a55602e | 1336 | *mail = *tz = "(unknown)"; |
cee7f245 JH |
1337 | *time = 0; |
1338 | return; | |
1339 | } | |
1340 | memcpy(person, tmp, len); | |
1341 | ||
1342 | tmp = person; | |
1343 | tmp += len; | |
1344 | *tmp = 0; | |
1345 | while (*tmp != ' ') | |
1346 | tmp--; | |
1347 | *tz = tmp+1; | |
f9567384 | 1348 | tzlen = (person+len)-(tmp+1); |
cee7f245 JH |
1349 | |
1350 | *tmp = 0; | |
1351 | while (*tmp != ' ') | |
1352 | tmp--; | |
1353 | *time = strtoul(tmp, NULL, 10); | |
f9567384 | 1354 | timepos = tmp; |
cee7f245 JH |
1355 | |
1356 | *tmp = 0; | |
1357 | while (*tmp != ' ') | |
1358 | tmp--; | |
1359 | *mail = tmp + 1; | |
1360 | *tmp = 0; | |
f9567384 JH |
1361 | maillen = timepos - tmp; |
1362 | ||
1363 | if (!mailmap.nr) | |
1364 | return; | |
1365 | ||
1366 | /* | |
1367 | * mailmap expansion may make the name longer. | |
1368 | * make room by pushing stuff down. | |
1369 | */ | |
1370 | tmp = person + bufsz - (tzlen + 1); | |
1371 | memmove(tmp, *tz, tzlen); | |
1372 | tmp[tzlen] = 0; | |
1373 | *tz = tmp; | |
1374 | ||
1375 | tmp = tmp - (maillen + 1); | |
1376 | memmove(tmp, *mail, maillen); | |
1377 | tmp[maillen] = 0; | |
1378 | *mail = tmp; | |
1379 | ||
1380 | /* | |
1381 | * Now, convert e-mail using mailmap | |
1382 | */ | |
1383 | map_email(&mailmap, tmp + 1, person, tmp-person-1); | |
cee7f245 JH |
1384 | } |
1385 | ||
1386 | static void get_commit_info(struct commit *commit, | |
1387 | struct commit_info *ret, | |
1388 | int detailed) | |
1389 | { | |
1390 | int len; | |
1391 | char *tmp, *endp; | |
1392 | static char author_buf[1024]; | |
1393 | static char committer_buf[1024]; | |
1394 | static char summary_buf[1024]; | |
1395 | ||
1732a1fd JH |
1396 | /* |
1397 | * We've operated without save_commit_buffer, so | |
612702e8 JH |
1398 | * we now need to populate them for output. |
1399 | */ | |
1400 | if (!commit->buffer) { | |
21666f1a | 1401 | enum object_type type; |
612702e8 JH |
1402 | unsigned long size; |
1403 | commit->buffer = | |
21666f1a | 1404 | read_sha1_file(commit->object.sha1, &type, &size); |
ab43e495 JH |
1405 | if (!commit->buffer) |
1406 | die("Cannot read commit %s", | |
1407 | sha1_to_hex(commit->object.sha1)); | |
612702e8 | 1408 | } |
cee7f245 JH |
1409 | ret->author = author_buf; |
1410 | get_ac_line(commit->buffer, "\nauthor ", | |
1411 | sizeof(author_buf), author_buf, &ret->author_mail, | |
1412 | &ret->author_time, &ret->author_tz); | |
1413 | ||
1414 | if (!detailed) | |
1415 | return; | |
1416 | ||
1417 | ret->committer = committer_buf; | |
1418 | get_ac_line(commit->buffer, "\ncommitter ", | |
1419 | sizeof(committer_buf), committer_buf, &ret->committer_mail, | |
1420 | &ret->committer_time, &ret->committer_tz); | |
1421 | ||
1422 | ret->summary = summary_buf; | |
1423 | tmp = strstr(commit->buffer, "\n\n"); | |
1424 | if (!tmp) { | |
1425 | error_out: | |
1426 | sprintf(summary_buf, "(%s)", sha1_to_hex(commit->object.sha1)); | |
1427 | return; | |
1428 | } | |
1429 | tmp += 2; | |
1430 | endp = strchr(tmp, '\n'); | |
1431 | if (!endp) | |
1cfe7733 | 1432 | endp = tmp + strlen(tmp); |
cee7f245 | 1433 | len = endp - tmp; |
1cfe7733 | 1434 | if (len >= sizeof(summary_buf) || len == 0) |
cee7f245 JH |
1435 | goto error_out; |
1436 | memcpy(summary_buf, tmp, len); | |
1437 | summary_buf[len] = 0; | |
1438 | } | |
1439 | ||
1732a1fd JH |
1440 | /* |
1441 | * To allow LF and other nonportable characters in pathnames, | |
1442 | * they are c-style quoted as needed. | |
1443 | */ | |
46e5e69d JH |
1444 | static void write_filename_info(const char *path) |
1445 | { | |
1446 | printf("filename "); | |
663af342 | 1447 | write_name_quoted(path, stdout, '\n'); |
46e5e69d JH |
1448 | } |
1449 | ||
1732a1fd JH |
1450 | /* |
1451 | * The blame_entry is found to be guilty for the range. Mark it | |
1452 | * as such, and show it in incremental output. | |
1453 | */ | |
717d1462 LT |
1454 | static void found_guilty_entry(struct blame_entry *ent) |
1455 | { | |
1456 | if (ent->guilty) | |
1457 | return; | |
1458 | ent->guilty = 1; | |
1459 | if (incremental) { | |
1460 | struct origin *suspect = ent->suspect; | |
1461 | ||
1462 | printf("%s %d %d %d\n", | |
1463 | sha1_to_hex(suspect->commit->object.sha1), | |
1464 | ent->s_lno + 1, ent->lno + 1, ent->num_lines); | |
1465 | if (!(suspect->commit->object.flags & METAINFO_SHOWN)) { | |
1466 | struct commit_info ci; | |
1467 | suspect->commit->object.flags |= METAINFO_SHOWN; | |
1468 | get_commit_info(suspect->commit, &ci, 1); | |
1469 | printf("author %s\n", ci.author); | |
1470 | printf("author-mail %s\n", ci.author_mail); | |
1471 | printf("author-time %lu\n", ci.author_time); | |
1472 | printf("author-tz %s\n", ci.author_tz); | |
1473 | printf("committer %s\n", ci.committer); | |
1474 | printf("committer-mail %s\n", ci.committer_mail); | |
1475 | printf("committer-time %lu\n", ci.committer_time); | |
1476 | printf("committer-tz %s\n", ci.committer_tz); | |
1477 | printf("summary %s\n", ci.summary); | |
1478 | if (suspect->commit->object.flags & UNINTERESTING) | |
1479 | printf("boundary\n"); | |
1480 | } | |
46e5e69d | 1481 | write_filename_info(suspect->path); |
06f59e9f | 1482 | maybe_flush_or_die(stdout, "stdout"); |
717d1462 LT |
1483 | } |
1484 | } | |
1485 | ||
1732a1fd JH |
1486 | /* |
1487 | * The main loop -- while the scoreboard has lines whose true origin | |
3dff5379 | 1488 | * is still unknown, pick one blame_entry, and allow its current |
1732a1fd JH |
1489 | * suspect to pass blames to its parents. |
1490 | */ | |
717d1462 LT |
1491 | static void assign_blame(struct scoreboard *sb, struct rev_info *revs, int opt) |
1492 | { | |
1493 | while (1) { | |
1494 | struct blame_entry *ent; | |
1495 | struct commit *commit; | |
1496 | struct origin *suspect = NULL; | |
1497 | ||
1498 | /* find one suspect to break down */ | |
1499 | for (ent = sb->ent; !suspect && ent; ent = ent->next) | |
1500 | if (!ent->guilty) | |
1501 | suspect = ent->suspect; | |
1502 | if (!suspect) | |
1503 | return; /* all done */ | |
1504 | ||
1732a1fd JH |
1505 | /* |
1506 | * We will use this suspect later in the loop, | |
1507 | * so hold onto it in the meantime. | |
1508 | */ | |
717d1462 LT |
1509 | origin_incref(suspect); |
1510 | commit = suspect->commit; | |
1511 | if (!commit->object.parsed) | |
1512 | parse_commit(commit); | |
1513 | if (!(commit->object.flags & UNINTERESTING) && | |
1732a1fd | 1514 | !(revs->max_age != -1 && commit->date < revs->max_age)) |
717d1462 LT |
1515 | pass_blame(sb, suspect, opt); |
1516 | else { | |
1517 | commit->object.flags |= UNINTERESTING; | |
1518 | if (commit->object.parsed) | |
1519 | mark_parents_uninteresting(commit); | |
1520 | } | |
1521 | /* treat root commit as boundary */ | |
1522 | if (!commit->parents && !show_root) | |
1523 | commit->object.flags |= UNINTERESTING; | |
1524 | ||
1525 | /* Take responsibility for the remaining entries */ | |
1526 | for (ent = sb->ent; ent; ent = ent->next) | |
171dccd5 | 1527 | if (same_suspect(ent->suspect, suspect)) |
717d1462 LT |
1528 | found_guilty_entry(ent); |
1529 | origin_decref(suspect); | |
1530 | ||
1531 | if (DEBUG) /* sanity */ | |
1532 | sanity_check_refcnt(sb); | |
1533 | } | |
1534 | } | |
1535 | ||
1536 | static const char *format_time(unsigned long time, const char *tz_str, | |
1537 | int show_raw_time) | |
1538 | { | |
1539 | static char time_buf[128]; | |
1540 | time_t t = time; | |
1541 | int minutes, tz; | |
1542 | struct tm *tm; | |
1543 | ||
1544 | if (show_raw_time) { | |
1545 | sprintf(time_buf, "%lu %s", time, tz_str); | |
1546 | return time_buf; | |
1547 | } | |
1548 | ||
1549 | tz = atoi(tz_str); | |
1550 | minutes = tz < 0 ? -tz : tz; | |
1551 | minutes = (minutes / 100)*60 + (minutes % 100); | |
1552 | minutes = tz < 0 ? -minutes : minutes; | |
1553 | t = time + minutes * 60; | |
1554 | tm = gmtime(&t); | |
1555 | ||
1556 | strftime(time_buf, sizeof(time_buf), "%Y-%m-%d %H:%M:%S ", tm); | |
1557 | strcat(time_buf, tz_str); | |
1558 | return time_buf; | |
1559 | } | |
1560 | ||
cee7f245 JH |
1561 | #define OUTPUT_ANNOTATE_COMPAT 001 |
1562 | #define OUTPUT_LONG_OBJECT_NAME 002 | |
1563 | #define OUTPUT_RAW_TIMESTAMP 004 | |
1564 | #define OUTPUT_PORCELAIN 010 | |
1565 | #define OUTPUT_SHOW_NAME 020 | |
1566 | #define OUTPUT_SHOW_NUMBER 040 | |
5ff62c30 | 1567 | #define OUTPUT_SHOW_SCORE 0100 |
093dc5be | 1568 | #define OUTPUT_NO_AUTHOR 0200 |
cee7f245 JH |
1569 | |
1570 | static void emit_porcelain(struct scoreboard *sb, struct blame_entry *ent) | |
1571 | { | |
1572 | int cnt; | |
1573 | const char *cp; | |
1574 | struct origin *suspect = ent->suspect; | |
1575 | char hex[41]; | |
1576 | ||
1577 | strcpy(hex, sha1_to_hex(suspect->commit->object.sha1)); | |
1578 | printf("%s%c%d %d %d\n", | |
1579 | hex, | |
1580 | ent->guilty ? ' ' : '*', // purely for debugging | |
1581 | ent->s_lno + 1, | |
1582 | ent->lno + 1, | |
1583 | ent->num_lines); | |
1584 | if (!(suspect->commit->object.flags & METAINFO_SHOWN)) { | |
1585 | struct commit_info ci; | |
1586 | suspect->commit->object.flags |= METAINFO_SHOWN; | |
1587 | get_commit_info(suspect->commit, &ci, 1); | |
1588 | printf("author %s\n", ci.author); | |
1589 | printf("author-mail %s\n", ci.author_mail); | |
1590 | printf("author-time %lu\n", ci.author_time); | |
1591 | printf("author-tz %s\n", ci.author_tz); | |
1592 | printf("committer %s\n", ci.committer); | |
1593 | printf("committer-mail %s\n", ci.committer_mail); | |
1594 | printf("committer-time %lu\n", ci.committer_time); | |
1595 | printf("committer-tz %s\n", ci.committer_tz); | |
46e5e69d | 1596 | write_filename_info(suspect->path); |
cee7f245 | 1597 | printf("summary %s\n", ci.summary); |
b11121d9 JH |
1598 | if (suspect->commit->object.flags & UNINTERESTING) |
1599 | printf("boundary\n"); | |
cee7f245 JH |
1600 | } |
1601 | else if (suspect->commit->object.flags & MORE_THAN_ONE_PATH) | |
46e5e69d | 1602 | write_filename_info(suspect->path); |
cee7f245 JH |
1603 | |
1604 | cp = nth_line(sb, ent->lno); | |
1605 | for (cnt = 0; cnt < ent->num_lines; cnt++) { | |
1606 | char ch; | |
1607 | if (cnt) | |
1608 | printf("%s %d %d\n", hex, | |
1609 | ent->s_lno + 1 + cnt, | |
1610 | ent->lno + 1 + cnt); | |
1611 | putchar('\t'); | |
1612 | do { | |
1613 | ch = *cp++; | |
1614 | putchar(ch); | |
1615 | } while (ch != '\n' && | |
1616 | cp < sb->final_buf + sb->final_buf_size); | |
1617 | } | |
1618 | } | |
1619 | ||
1620 | static void emit_other(struct scoreboard *sb, struct blame_entry *ent, int opt) | |
1621 | { | |
1622 | int cnt; | |
1623 | const char *cp; | |
1624 | struct origin *suspect = ent->suspect; | |
1625 | struct commit_info ci; | |
1626 | char hex[41]; | |
1627 | int show_raw_time = !!(opt & OUTPUT_RAW_TIMESTAMP); | |
1628 | ||
1629 | get_commit_info(suspect->commit, &ci, 1); | |
1630 | strcpy(hex, sha1_to_hex(suspect->commit->object.sha1)); | |
1631 | ||
1632 | cp = nth_line(sb, ent->lno); | |
1633 | for (cnt = 0; cnt < ent->num_lines; cnt++) { | |
1634 | char ch; | |
b11121d9 JH |
1635 | int length = (opt & OUTPUT_LONG_OBJECT_NAME) ? 40 : 8; |
1636 | ||
1637 | if (suspect->commit->object.flags & UNINTERESTING) { | |
e68989a7 JH |
1638 | if (blank_boundary) |
1639 | memset(hex, ' ', length); | |
1640 | else if (!cmd_is_annotate) { | |
4c10a5ca JH |
1641 | length--; |
1642 | putchar('^'); | |
1643 | } | |
b11121d9 | 1644 | } |
cee7f245 | 1645 | |
b11121d9 | 1646 | printf("%.*s", length, hex); |
cee7f245 JH |
1647 | if (opt & OUTPUT_ANNOTATE_COMPAT) |
1648 | printf("\t(%10s\t%10s\t%d)", ci.author, | |
1649 | format_time(ci.author_time, ci.author_tz, | |
1650 | show_raw_time), | |
1651 | ent->lno + 1 + cnt); | |
1652 | else { | |
5ff62c30 | 1653 | if (opt & OUTPUT_SHOW_SCORE) |
54a4c617 JH |
1654 | printf(" %*d %02d", |
1655 | max_score_digits, ent->score, | |
1656 | ent->suspect->refcnt); | |
cee7f245 JH |
1657 | if (opt & OUTPUT_SHOW_NAME) |
1658 | printf(" %-*.*s", longest_file, longest_file, | |
1659 | suspect->path); | |
1660 | if (opt & OUTPUT_SHOW_NUMBER) | |
1661 | printf(" %*d", max_orig_digits, | |
1662 | ent->s_lno + 1 + cnt); | |
093dc5be JH |
1663 | |
1664 | if (!(opt & OUTPUT_NO_AUTHOR)) | |
1665 | printf(" (%-*.*s %10s", | |
1666 | longest_author, longest_author, | |
1667 | ci.author, | |
1668 | format_time(ci.author_time, | |
1669 | ci.author_tz, | |
1670 | show_raw_time)); | |
1671 | printf(" %*d) ", | |
cee7f245 JH |
1672 | max_digits, ent->lno + 1 + cnt); |
1673 | } | |
1674 | do { | |
1675 | ch = *cp++; | |
1676 | putchar(ch); | |
1677 | } while (ch != '\n' && | |
1678 | cp < sb->final_buf + sb->final_buf_size); | |
1679 | } | |
1680 | } | |
1681 | ||
1682 | static void output(struct scoreboard *sb, int option) | |
1683 | { | |
1684 | struct blame_entry *ent; | |
1685 | ||
1686 | if (option & OUTPUT_PORCELAIN) { | |
1687 | for (ent = sb->ent; ent; ent = ent->next) { | |
1688 | struct blame_entry *oth; | |
1689 | struct origin *suspect = ent->suspect; | |
1690 | struct commit *commit = suspect->commit; | |
1691 | if (commit->object.flags & MORE_THAN_ONE_PATH) | |
1692 | continue; | |
1693 | for (oth = ent->next; oth; oth = oth->next) { | |
1694 | if ((oth->suspect->commit != commit) || | |
1695 | !strcmp(oth->suspect->path, suspect->path)) | |
1696 | continue; | |
1697 | commit->object.flags |= MORE_THAN_ONE_PATH; | |
1698 | break; | |
1699 | } | |
1700 | } | |
1701 | } | |
1702 | ||
1703 | for (ent = sb->ent; ent; ent = ent->next) { | |
1704 | if (option & OUTPUT_PORCELAIN) | |
1705 | emit_porcelain(sb, ent); | |
5ff62c30 | 1706 | else { |
cee7f245 | 1707 | emit_other(sb, ent, option); |
5ff62c30 | 1708 | } |
cee7f245 JH |
1709 | } |
1710 | } | |
1711 | ||
1732a1fd JH |
1712 | /* |
1713 | * To allow quick access to the contents of nth line in the | |
1714 | * final image, prepare an index in the scoreboard. | |
1715 | */ | |
cee7f245 JH |
1716 | static int prepare_lines(struct scoreboard *sb) |
1717 | { | |
1718 | const char *buf = sb->final_buf; | |
1719 | unsigned long len = sb->final_buf_size; | |
1720 | int num = 0, incomplete = 0, bol = 1; | |
1721 | ||
1722 | if (len && buf[len-1] != '\n') | |
1723 | incomplete++; /* incomplete line at the end */ | |
1724 | while (len--) { | |
1725 | if (bol) { | |
1726 | sb->lineno = xrealloc(sb->lineno, | |
1727 | sizeof(int* ) * (num + 1)); | |
1728 | sb->lineno[num] = buf - sb->final_buf; | |
1729 | bol = 0; | |
1730 | } | |
1731 | if (*buf++ == '\n') { | |
1732 | num++; | |
1733 | bol = 1; | |
1734 | } | |
1735 | } | |
1ca6ca87 JH |
1736 | sb->lineno = xrealloc(sb->lineno, |
1737 | sizeof(int* ) * (num + incomplete + 1)); | |
1738 | sb->lineno[num + incomplete] = buf - sb->final_buf; | |
cee7f245 JH |
1739 | sb->num_lines = num + incomplete; |
1740 | return sb->num_lines; | |
1741 | } | |
1742 | ||
1732a1fd JH |
1743 | /* |
1744 | * Add phony grafts for use with -S; this is primarily to | |
1745 | * support git-cvsserver that wants to give a linear history | |
1746 | * to its clients. | |
1747 | */ | |
cee7f245 JH |
1748 | static int read_ancestry(const char *graft_file) |
1749 | { | |
1750 | FILE *fp = fopen(graft_file, "r"); | |
1751 | char buf[1024]; | |
1752 | if (!fp) | |
1753 | return -1; | |
1754 | while (fgets(buf, sizeof(buf), fp)) { | |
1755 | /* The format is just "Commit Parent1 Parent2 ...\n" */ | |
1756 | int len = strlen(buf); | |
1757 | struct commit_graft *graft = read_graft_line(buf, len); | |
8eaf7986 JH |
1758 | if (graft) |
1759 | register_commit_graft(graft, 0); | |
cee7f245 JH |
1760 | } |
1761 | fclose(fp); | |
1762 | return 0; | |
1763 | } | |
1764 | ||
1732a1fd JH |
1765 | /* |
1766 | * How many columns do we need to show line numbers in decimal? | |
1767 | */ | |
cee7f245 JH |
1768 | static int lineno_width(int lines) |
1769 | { | |
f2915045 | 1770 | int i, width; |
cee7f245 | 1771 | |
f2915045 JH |
1772 | for (width = 1, i = 10; i <= lines + 1; width++) |
1773 | i *= 10; | |
1774 | return width; | |
cee7f245 JH |
1775 | } |
1776 | ||
1732a1fd JH |
1777 | /* |
1778 | * How many columns do we need to show line numbers, authors, | |
1779 | * and filenames? | |
1780 | */ | |
cee7f245 JH |
1781 | static void find_alignment(struct scoreboard *sb, int *option) |
1782 | { | |
1783 | int longest_src_lines = 0; | |
1784 | int longest_dst_lines = 0; | |
5ff62c30 | 1785 | unsigned largest_score = 0; |
cee7f245 JH |
1786 | struct blame_entry *e; |
1787 | ||
1788 | for (e = sb->ent; e; e = e->next) { | |
1789 | struct origin *suspect = e->suspect; | |
1790 | struct commit_info ci; | |
1791 | int num; | |
1792 | ||
ab3bb800 JH |
1793 | if (strcmp(suspect->path, sb->path)) |
1794 | *option |= OUTPUT_SHOW_NAME; | |
1795 | num = strlen(suspect->path); | |
1796 | if (longest_file < num) | |
1797 | longest_file = num; | |
cee7f245 JH |
1798 | if (!(suspect->commit->object.flags & METAINFO_SHOWN)) { |
1799 | suspect->commit->object.flags |= METAINFO_SHOWN; | |
1800 | get_commit_info(suspect->commit, &ci, 1); | |
cee7f245 JH |
1801 | num = strlen(ci.author); |
1802 | if (longest_author < num) | |
1803 | longest_author = num; | |
1804 | } | |
1805 | num = e->s_lno + e->num_lines; | |
1806 | if (longest_src_lines < num) | |
1807 | longest_src_lines = num; | |
1808 | num = e->lno + e->num_lines; | |
1809 | if (longest_dst_lines < num) | |
1810 | longest_dst_lines = num; | |
5ff62c30 JH |
1811 | if (largest_score < ent_score(sb, e)) |
1812 | largest_score = ent_score(sb, e); | |
cee7f245 JH |
1813 | } |
1814 | max_orig_digits = lineno_width(longest_src_lines); | |
1815 | max_digits = lineno_width(longest_dst_lines); | |
5ff62c30 | 1816 | max_score_digits = lineno_width(largest_score); |
cee7f245 JH |
1817 | } |
1818 | ||
1732a1fd JH |
1819 | /* |
1820 | * For debugging -- origin is refcounted, and this asserts that | |
1821 | * we do not underflow. | |
1822 | */ | |
54a4c617 JH |
1823 | static void sanity_check_refcnt(struct scoreboard *sb) |
1824 | { | |
1825 | int baa = 0; | |
1826 | struct blame_entry *ent; | |
1827 | ||
1828 | for (ent = sb->ent; ent; ent = ent->next) { | |
ae86ad65 | 1829 | /* Nobody should have zero or negative refcnt */ |
854b97f6 JH |
1830 | if (ent->suspect->refcnt <= 0) { |
1831 | fprintf(stderr, "%s in %s has negative refcnt %d\n", | |
1832 | ent->suspect->path, | |
1833 | sha1_to_hex(ent->suspect->commit->object.sha1), | |
1834 | ent->suspect->refcnt); | |
ae86ad65 | 1835 | baa = 1; |
854b97f6 | 1836 | } |
ae86ad65 JH |
1837 | } |
1838 | for (ent = sb->ent; ent; ent = ent->next) { | |
1839 | /* Mark the ones that haven't been checked */ | |
54a4c617 JH |
1840 | if (0 < ent->suspect->refcnt) |
1841 | ent->suspect->refcnt = -ent->suspect->refcnt; | |
54a4c617 JH |
1842 | } |
1843 | for (ent = sb->ent; ent; ent = ent->next) { | |
1732a1fd JH |
1844 | /* |
1845 | * ... then pick each and see if they have the the | |
1846 | * correct refcnt. | |
54a4c617 JH |
1847 | */ |
1848 | int found; | |
1849 | struct blame_entry *e; | |
1850 | struct origin *suspect = ent->suspect; | |
1851 | ||
1852 | if (0 < suspect->refcnt) | |
1853 | continue; | |
ae86ad65 | 1854 | suspect->refcnt = -suspect->refcnt; /* Unmark */ |
54a4c617 JH |
1855 | for (found = 0, e = sb->ent; e; e = e->next) { |
1856 | if (e->suspect != suspect) | |
1857 | continue; | |
1858 | found++; | |
1859 | } | |
854b97f6 JH |
1860 | if (suspect->refcnt != found) { |
1861 | fprintf(stderr, "%s in %s has refcnt %d, not %d\n", | |
1862 | ent->suspect->path, | |
1863 | sha1_to_hex(ent->suspect->commit->object.sha1), | |
1864 | ent->suspect->refcnt, found); | |
1865 | baa = 2; | |
1866 | } | |
54a4c617 JH |
1867 | } |
1868 | if (baa) { | |
1869 | int opt = 0160; | |
1870 | find_alignment(sb, &opt); | |
1871 | output(sb, opt); | |
854b97f6 | 1872 | die("Baa %d!", baa); |
54a4c617 JH |
1873 | } |
1874 | } | |
1875 | ||
1732a1fd JH |
1876 | /* |
1877 | * Used for the command line parsing; check if the path exists | |
1878 | * in the working tree. | |
1879 | */ | |
cee7f245 JH |
1880 | static int has_path_in_work_tree(const char *path) |
1881 | { | |
1882 | struct stat st; | |
1883 | return !lstat(path, &st); | |
1884 | } | |
1885 | ||
4a0fc95f JH |
1886 | static unsigned parse_score(const char *arg) |
1887 | { | |
1888 | char *end; | |
1889 | unsigned long score = strtoul(arg, &end, 10); | |
1890 | if (*end) | |
1891 | return 0; | |
1892 | return score; | |
1893 | } | |
1894 | ||
20239bae JK |
1895 | static const char *add_prefix(const char *prefix, const char *path) |
1896 | { | |
097971f5 | 1897 | return prefix_path(prefix, prefix ? strlen(prefix) : 0, path); |
20239bae JK |
1898 | } |
1899 | ||
1732a1fd JH |
1900 | /* |
1901 | * Parsing of (comma separated) one item in the -L option | |
1902 | */ | |
931233bc JH |
1903 | static const char *parse_loc(const char *spec, |
1904 | struct scoreboard *sb, long lno, | |
1905 | long begin, long *ret) | |
1906 | { | |
1907 | char *term; | |
1908 | const char *line; | |
1909 | long num; | |
1910 | int reg_error; | |
1911 | regex_t regexp; | |
1912 | regmatch_t match[1]; | |
1913 | ||
7bd9641d JH |
1914 | /* Allow "-L <something>,+20" to mean starting at <something> |
1915 | * for 20 lines, or "-L <something>,-5" for 5 lines ending at | |
1916 | * <something>. | |
1917 | */ | |
1918 | if (1 < begin && (spec[0] == '+' || spec[0] == '-')) { | |
1919 | num = strtol(spec + 1, &term, 10); | |
1920 | if (term != spec + 1) { | |
1921 | if (spec[0] == '-') | |
1922 | num = 0 - num; | |
1923 | if (0 < num) | |
1924 | *ret = begin + num - 2; | |
1925 | else if (!num) | |
1926 | *ret = begin; | |
1927 | else | |
1928 | *ret = begin + num; | |
1929 | return term; | |
1930 | } | |
1931 | return spec; | |
1932 | } | |
931233bc JH |
1933 | num = strtol(spec, &term, 10); |
1934 | if (term != spec) { | |
1935 | *ret = num; | |
1936 | return term; | |
1937 | } | |
1938 | if (spec[0] != '/') | |
1939 | return spec; | |
1940 | ||
1941 | /* it could be a regexp of form /.../ */ | |
1942 | for (term = (char*) spec + 1; *term && *term != '/'; term++) { | |
1943 | if (*term == '\\') | |
1944 | term++; | |
1945 | } | |
1946 | if (*term != '/') | |
1947 | return spec; | |
1948 | ||
1949 | /* try [spec+1 .. term-1] as regexp */ | |
1950 | *term = 0; | |
1951 | begin--; /* input is in human terms */ | |
1952 | line = nth_line(sb, begin); | |
1953 | ||
1954 | if (!(reg_error = regcomp(®exp, spec + 1, REG_NEWLINE)) && | |
1955 | !(reg_error = regexec(®exp, line, 1, match, 0))) { | |
1956 | const char *cp = line + match[0].rm_so; | |
1957 | const char *nline; | |
1958 | ||
1959 | while (begin++ < lno) { | |
1960 | nline = nth_line(sb, begin); | |
1961 | if (line <= cp && cp < nline) | |
1962 | break; | |
1963 | line = nline; | |
1964 | } | |
1965 | *ret = begin; | |
1966 | regfree(®exp); | |
1967 | *term++ = '/'; | |
1968 | return term; | |
1969 | } | |
1970 | else { | |
1971 | char errbuf[1024]; | |
1972 | regerror(reg_error, ®exp, errbuf, 1024); | |
1973 | die("-L parameter '%s': %s", spec + 1, errbuf); | |
1974 | } | |
1975 | } | |
1976 | ||
1732a1fd JH |
1977 | /* |
1978 | * Parsing of -L option | |
1979 | */ | |
931233bc JH |
1980 | static void prepare_blame_range(struct scoreboard *sb, |
1981 | const char *bottomtop, | |
1982 | long lno, | |
1983 | long *bottom, long *top) | |
1984 | { | |
1985 | const char *term; | |
1986 | ||
1987 | term = parse_loc(bottomtop, sb, lno, 1, bottom); | |
1988 | if (*term == ',') { | |
1989 | term = parse_loc(term + 1, sb, lno, *bottom + 1, top); | |
1990 | if (*term) | |
acca687f | 1991 | usage(blame_usage); |
931233bc JH |
1992 | } |
1993 | if (*term) | |
acca687f | 1994 | usage(blame_usage); |
931233bc JH |
1995 | } |
1996 | ||
4c10a5ca JH |
1997 | static int git_blame_config(const char *var, const char *value) |
1998 | { | |
1999 | if (!strcmp(var, "blame.showroot")) { | |
2000 | show_root = git_config_bool(var, value); | |
2001 | return 0; | |
2002 | } | |
2003 | if (!strcmp(var, "blame.blankboundary")) { | |
2004 | blank_boundary = git_config_bool(var, value); | |
2005 | return 0; | |
2006 | } | |
2007 | return git_default_config(var, value); | |
2008 | } | |
2009 | ||
1cfe7733 JH |
2010 | static struct commit *fake_working_tree_commit(const char *path, const char *contents_from) |
2011 | { | |
2012 | struct commit *commit; | |
2013 | struct origin *origin; | |
2014 | unsigned char head_sha1[20]; | |
af6eb822 | 2015 | struct strbuf buf; |
1cfe7733 | 2016 | const char *ident; |
1cfe7733 | 2017 | time_t now; |
1cfe7733 JH |
2018 | int size, len; |
2019 | struct cache_entry *ce; | |
2020 | unsigned mode; | |
2021 | ||
2022 | if (get_sha1("HEAD", head_sha1)) | |
2023 | die("No such ref: HEAD"); | |
2024 | ||
2025 | time(&now); | |
2026 | commit = xcalloc(1, sizeof(*commit)); | |
2027 | commit->parents = xcalloc(1, sizeof(*commit->parents)); | |
2028 | commit->parents->item = lookup_commit_reference(head_sha1); | |
2029 | commit->object.parsed = 1; | |
2030 | commit->date = now; | |
2031 | commit->object.type = OBJ_COMMIT; | |
2032 | ||
2033 | origin = make_origin(commit, path); | |
2034 | ||
f1696ee3 | 2035 | strbuf_init(&buf, 0); |
1cfe7733 JH |
2036 | if (!contents_from || strcmp("-", contents_from)) { |
2037 | struct stat st; | |
2038 | const char *read_from; | |
af6eb822 | 2039 | unsigned long fin_size; |
1cfe7733 JH |
2040 | |
2041 | if (contents_from) { | |
2042 | if (stat(contents_from, &st) < 0) | |
2043 | die("Cannot stat %s", contents_from); | |
2044 | read_from = contents_from; | |
2045 | } | |
2046 | else { | |
2047 | if (lstat(path, &st) < 0) | |
2048 | die("Cannot lstat %s", path); | |
2049 | read_from = path; | |
2050 | } | |
dc49cd76 | 2051 | fin_size = xsize_t(st.st_size); |
1cfe7733 JH |
2052 | mode = canon_mode(st.st_mode); |
2053 | switch (st.st_mode & S_IFMT) { | |
2054 | case S_IFREG: | |
387e7e19 PH |
2055 | if (strbuf_read_file(&buf, read_from, st.st_size) != st.st_size) |
2056 | die("cannot open or read %s", read_from); | |
1cfe7733 JH |
2057 | break; |
2058 | case S_IFLNK: | |
af6eb822 | 2059 | if (readlink(read_from, buf.buf, buf.alloc) != fin_size) |
1cfe7733 | 2060 | die("cannot readlink %s", read_from); |
af6eb822 | 2061 | buf.len = fin_size; |
1cfe7733 JH |
2062 | break; |
2063 | default: | |
2064 | die("unsupported file type %s", read_from); | |
2065 | } | |
2066 | } | |
2067 | else { | |
2068 | /* Reading from stdin */ | |
2069 | contents_from = "standard input"; | |
1cfe7733 | 2070 | mode = 0; |
f1696ee3 | 2071 | if (strbuf_read(&buf, 0, 0) < 0) |
af6eb822 | 2072 | die("read error %s from stdin", strerror(errno)); |
1cfe7733 | 2073 | } |
21e5ad50 | 2074 | convert_to_git(path, buf.buf, buf.len, &buf, 0); |
af6eb822 PH |
2075 | origin->file.ptr = buf.buf; |
2076 | origin->file.size = buf.len; | |
2077 | pretend_sha1_file(buf.buf, buf.len, OBJ_BLOB, origin->blob_sha1); | |
1cfe7733 JH |
2078 | commit->util = origin; |
2079 | ||
2080 | /* | |
2081 | * Read the current index, replace the path entry with | |
2082 | * origin->blob_sha1 without mucking with its mode or type | |
2083 | * bits; we are not going to write this index out -- we just | |
2084 | * want to run "diff-index --cached". | |
2085 | */ | |
2086 | discard_cache(); | |
2087 | read_cache(); | |
2088 | ||
2089 | len = strlen(path); | |
2090 | if (!mode) { | |
2091 | int pos = cache_name_pos(path, len); | |
2092 | if (0 <= pos) | |
7a51ed66 | 2093 | mode = active_cache[pos]->ce_mode; |
1cfe7733 JH |
2094 | else |
2095 | /* Let's not bother reading from HEAD tree */ | |
2096 | mode = S_IFREG | 0644; | |
2097 | } | |
2098 | size = cache_entry_size(len); | |
2099 | ce = xcalloc(1, size); | |
2100 | hashcpy(ce->sha1, origin->blob_sha1); | |
2101 | memcpy(ce->name, path, len); | |
2102 | ce->ce_flags = create_ce_flags(len, 0); | |
2103 | ce->ce_mode = create_ce_mode(mode); | |
2104 | add_cache_entry(ce, ADD_CACHE_OK_TO_ADD|ADD_CACHE_OK_TO_REPLACE); | |
2105 | ||
2106 | /* | |
2107 | * We are not going to write this out, so this does not matter | |
2108 | * right now, but someday we might optimize diff-index --cached | |
2109 | * with cache-tree information. | |
2110 | */ | |
2111 | cache_tree_invalidate_path(active_cache_tree, path); | |
2112 | ||
2113 | commit->buffer = xmalloc(400); | |
2114 | ident = fmt_ident("Not Committed Yet", "not.committed.yet", NULL, 0); | |
1bb88be9 | 2115 | snprintf(commit->buffer, 400, |
1cfe7733 JH |
2116 | "tree 0000000000000000000000000000000000000000\n" |
2117 | "parent %s\n" | |
2118 | "author %s\n" | |
2119 | "committer %s\n\n" | |
2120 | "Version of %s from %s\n", | |
2121 | sha1_to_hex(head_sha1), | |
2122 | ident, ident, path, contents_from ? contents_from : path); | |
2123 | return commit; | |
2124 | } | |
2125 | ||
acca687f | 2126 | int cmd_blame(int argc, const char **argv, const char *prefix) |
cee7f245 JH |
2127 | { |
2128 | struct rev_info revs; | |
2129 | const char *path; | |
2130 | struct scoreboard sb; | |
2131 | struct origin *o; | |
2132 | struct blame_entry *ent; | |
d24bba80 | 2133 | int i, seen_dashdash, unk, opt; |
cee7f245 JH |
2134 | long bottom, top, lno; |
2135 | int output_option = 0; | |
870b39c1 | 2136 | int show_stats = 0; |
cee7f245 JH |
2137 | const char *revs_file = NULL; |
2138 | const char *final_commit_name = NULL; | |
21666f1a | 2139 | enum object_type type; |
931233bc | 2140 | const char *bottomtop = NULL; |
1cfe7733 | 2141 | const char *contents_from = NULL; |
cee7f245 | 2142 | |
e68989a7 JH |
2143 | cmd_is_annotate = !strcmp(argv[0], "annotate"); |
2144 | ||
4c10a5ca | 2145 | git_config(git_blame_config); |
612702e8 JH |
2146 | save_commit_buffer = 0; |
2147 | ||
d24bba80 | 2148 | opt = 0; |
cee7f245 JH |
2149 | seen_dashdash = 0; |
2150 | for (unk = i = 1; i < argc; i++) { | |
2151 | const char *arg = argv[i]; | |
2152 | if (*arg != '-') | |
2153 | break; | |
4c10a5ca JH |
2154 | else if (!strcmp("-b", arg)) |
2155 | blank_boundary = 1; | |
2156 | else if (!strcmp("--root", arg)) | |
2157 | show_root = 1; | |
870b39c1 JH |
2158 | else if (!strcmp(arg, "--show-stats")) |
2159 | show_stats = 1; | |
cee7f245 JH |
2160 | else if (!strcmp("-c", arg)) |
2161 | output_option |= OUTPUT_ANNOTATE_COMPAT; | |
2162 | else if (!strcmp("-t", arg)) | |
2163 | output_option |= OUTPUT_RAW_TIMESTAMP; | |
2164 | else if (!strcmp("-l", arg)) | |
2165 | output_option |= OUTPUT_LONG_OBJECT_NAME; | |
093dc5be JH |
2166 | else if (!strcmp("-s", arg)) |
2167 | output_option |= OUTPUT_NO_AUTHOR; | |
b82871b3 JH |
2168 | else if (!strcmp("-w", arg)) |
2169 | xdl_opts |= XDF_IGNORE_WHITESPACE; | |
cee7f245 JH |
2170 | else if (!strcmp("-S", arg) && ++i < argc) |
2171 | revs_file = argv[i]; | |
599065a3 | 2172 | else if (!prefixcmp(arg, "-M")) { |
d24bba80 | 2173 | opt |= PICKAXE_BLAME_MOVE; |
4a0fc95f JH |
2174 | blame_move_score = parse_score(arg+2); |
2175 | } | |
599065a3 | 2176 | else if (!prefixcmp(arg, "-C")) { |
c63777c0 JH |
2177 | /* |
2178 | * -C enables copy from removed files; | |
2179 | * -C -C enables copy from existing files, but only | |
2180 | * when blaming a new file; | |
2181 | * -C -C -C enables copy from existing files for | |
2182 | * everybody | |
2183 | */ | |
2184 | if (opt & PICKAXE_BLAME_COPY_HARDER) | |
2185 | opt |= PICKAXE_BLAME_COPY_HARDEST; | |
18abd745 JH |
2186 | if (opt & PICKAXE_BLAME_COPY) |
2187 | opt |= PICKAXE_BLAME_COPY_HARDER; | |
2188 | opt |= PICKAXE_BLAME_COPY | PICKAXE_BLAME_MOVE; | |
4a0fc95f | 2189 | blame_copy_score = parse_score(arg+2); |
18abd745 | 2190 | } |
599065a3 | 2191 | else if (!prefixcmp(arg, "-L")) { |
2c40f984 JH |
2192 | if (!arg[2]) { |
2193 | if (++i >= argc) | |
acca687f | 2194 | usage(blame_usage); |
2c40f984 JH |
2195 | arg = argv[i]; |
2196 | } | |
2197 | else | |
2198 | arg += 2; | |
931233bc | 2199 | if (bottomtop) |
cee7f245 | 2200 | die("More than one '-L n,m' option given"); |
931233bc | 2201 | bottomtop = arg; |
cee7f245 | 2202 | } |
1cfe7733 JH |
2203 | else if (!strcmp("--contents", arg)) { |
2204 | if (++i >= argc) | |
2205 | usage(blame_usage); | |
2206 | contents_from = argv[i]; | |
2207 | } | |
717d1462 LT |
2208 | else if (!strcmp("--incremental", arg)) |
2209 | incremental = 1; | |
5ff62c30 JH |
2210 | else if (!strcmp("--score-debug", arg)) |
2211 | output_option |= OUTPUT_SHOW_SCORE; | |
cee7f245 JH |
2212 | else if (!strcmp("-f", arg) || |
2213 | !strcmp("--show-name", arg)) | |
2214 | output_option |= OUTPUT_SHOW_NAME; | |
2215 | else if (!strcmp("-n", arg) || | |
2216 | !strcmp("--show-number", arg)) | |
2217 | output_option |= OUTPUT_SHOW_NUMBER; | |
2218 | else if (!strcmp("-p", arg) || | |
2219 | !strcmp("--porcelain", arg)) | |
2220 | output_option |= OUTPUT_PORCELAIN; | |
2221 | else if (!strcmp("--", arg)) { | |
2222 | seen_dashdash = 1; | |
2223 | i++; | |
2224 | break; | |
2225 | } | |
2226 | else | |
2227 | argv[unk++] = arg; | |
2228 | } | |
2229 | ||
4a0fc95f JH |
2230 | if (!blame_move_score) |
2231 | blame_move_score = BLAME_DEFAULT_MOVE_SCORE; | |
2232 | if (!blame_copy_score) | |
2233 | blame_copy_score = BLAME_DEFAULT_COPY_SCORE; | |
2234 | ||
1732a1fd JH |
2235 | /* |
2236 | * We have collected options unknown to us in argv[1..unk] | |
cee7f245 | 2237 | * which are to be passed to revision machinery if we are |
3dff5379 | 2238 | * going to do the "bottom" processing. |
cee7f245 JH |
2239 | * |
2240 | * The remaining are: | |
2241 | * | |
2242 | * (1) if seen_dashdash, its either | |
2243 | * "-options -- <path>" or | |
2244 | * "-options -- <path> <rev>". | |
2245 | * but the latter is allowed only if there is no | |
2246 | * options that we passed to revision machinery. | |
2247 | * | |
2248 | * (2) otherwise, we may have "--" somewhere later and | |
2249 | * might be looking at the first one of multiple 'rev' | |
2250 | * parameters (e.g. " master ^next ^maint -- path"). | |
2251 | * See if there is a dashdash first, and give the | |
2252 | * arguments before that to revision machinery. | |
2253 | * After that there must be one 'path'. | |
2254 | * | |
2255 | * (3) otherwise, its one of the three: | |
2256 | * "-options <path> <rev>" | |
2257 | * "-options <rev> <path>" | |
2258 | * "-options <path>" | |
2259 | * but again the first one is allowed only if | |
2260 | * there is no options that we passed to revision | |
2261 | * machinery. | |
2262 | */ | |
2263 | ||
2264 | if (seen_dashdash) { | |
2265 | /* (1) */ | |
2266 | if (argc <= i) | |
acca687f | 2267 | usage(blame_usage); |
20239bae | 2268 | path = add_prefix(prefix, argv[i]); |
cee7f245 JH |
2269 | if (i + 1 == argc - 1) { |
2270 | if (unk != 1) | |
acca687f | 2271 | usage(blame_usage); |
cee7f245 JH |
2272 | argv[unk++] = argv[i + 1]; |
2273 | } | |
2274 | else if (i + 1 != argc) | |
2275 | /* garbage at end */ | |
acca687f | 2276 | usage(blame_usage); |
cee7f245 JH |
2277 | } |
2278 | else { | |
2279 | int j; | |
2280 | for (j = i; !seen_dashdash && j < argc; j++) | |
2281 | if (!strcmp(argv[j], "--")) | |
2282 | seen_dashdash = j; | |
2283 | if (seen_dashdash) { | |
f4421325 | 2284 | /* (2) */ |
cee7f245 | 2285 | if (seen_dashdash + 1 != argc - 1) |
acca687f | 2286 | usage(blame_usage); |
20239bae | 2287 | path = add_prefix(prefix, argv[seen_dashdash + 1]); |
cee7f245 JH |
2288 | for (j = i; j < seen_dashdash; j++) |
2289 | argv[unk++] = argv[j]; | |
2290 | } | |
2291 | else { | |
2292 | /* (3) */ | |
f4421325 TK |
2293 | if (argc <= i) |
2294 | usage(blame_usage); | |
20239bae | 2295 | path = add_prefix(prefix, argv[i]); |
cee7f245 JH |
2296 | if (i + 1 == argc - 1) { |
2297 | final_commit_name = argv[i + 1]; | |
2298 | ||
2299 | /* if (unk == 1) we could be getting | |
2300 | * old-style | |
2301 | */ | |
2302 | if (unk == 1 && !has_path_in_work_tree(path)) { | |
20239bae | 2303 | path = add_prefix(prefix, argv[i + 1]); |
cee7f245 JH |
2304 | final_commit_name = argv[i]; |
2305 | } | |
2306 | } | |
2307 | else if (i != argc - 1) | |
acca687f | 2308 | usage(blame_usage); /* garbage at end */ |
cee7f245 | 2309 | |
354e6534 | 2310 | setup_work_tree(); |
cee7f245 JH |
2311 | if (!has_path_in_work_tree(path)) |
2312 | die("cannot stat path %s: %s", | |
2313 | path, strerror(errno)); | |
2314 | } | |
2315 | } | |
2316 | ||
2317 | if (final_commit_name) | |
2318 | argv[unk++] = final_commit_name; | |
2319 | ||
1732a1fd JH |
2320 | /* |
2321 | * Now we got rev and path. We do not want the path pruning | |
cee7f245 JH |
2322 | * but we may want "bottom" processing. |
2323 | */ | |
6bee4e40 | 2324 | argv[unk++] = "--"; /* terminate the rev name */ |
cee7f245 JH |
2325 | argv[unk] = NULL; |
2326 | ||
2327 | init_revisions(&revs, NULL); | |
1cfe7733 | 2328 | setup_revisions(unk, argv, &revs, NULL); |
cee7f245 JH |
2329 | memset(&sb, 0, sizeof(sb)); |
2330 | ||
1732a1fd JH |
2331 | /* |
2332 | * There must be one and only one positive commit in the | |
cee7f245 JH |
2333 | * revs->pending array. |
2334 | */ | |
2335 | for (i = 0; i < revs.pending.nr; i++) { | |
2336 | struct object *obj = revs.pending.objects[i].item; | |
2337 | if (obj->flags & UNINTERESTING) | |
2338 | continue; | |
2339 | while (obj->type == OBJ_TAG) | |
2340 | obj = deref_tag(obj, NULL, 0); | |
2341 | if (obj->type != OBJ_COMMIT) | |
2342 | die("Non commit %s?", | |
2343 | revs.pending.objects[i].name); | |
2344 | if (sb.final) | |
2345 | die("More than one commit to dig from %s and %s?", | |
2346 | revs.pending.objects[i].name, | |
2347 | final_commit_name); | |
2348 | sb.final = (struct commit *) obj; | |
2349 | final_commit_name = revs.pending.objects[i].name; | |
2350 | } | |
2351 | ||
2352 | if (!sb.final) { | |
1732a1fd JH |
2353 | /* |
2354 | * "--not A B -- path" without anything positive; | |
1cfe7733 JH |
2355 | * do not default to HEAD, but use the working tree |
2356 | * or "--contents". | |
1732a1fd | 2357 | */ |
1981820b | 2358 | setup_work_tree(); |
1cfe7733 JH |
2359 | sb.final = fake_working_tree_commit(path, contents_from); |
2360 | add_pending_object(&revs, &(sb.final->object), ":"); | |
cee7f245 | 2361 | } |
1cfe7733 JH |
2362 | else if (contents_from) |
2363 | die("Cannot use --contents with final commit object name"); | |
cee7f245 | 2364 | |
1732a1fd JH |
2365 | /* |
2366 | * If we have bottom, this will mark the ancestors of the | |
cee7f245 JH |
2367 | * bottom commits we would reach while traversing as |
2368 | * uninteresting. | |
2369 | */ | |
3d51e1b5 MK |
2370 | if (prepare_revision_walk(&revs)) |
2371 | die("revision walk setup failed"); | |
cee7f245 | 2372 | |
1cfe7733 JH |
2373 | if (is_null_sha1(sb.final->object.sha1)) { |
2374 | char *buf; | |
2375 | o = sb.final->util; | |
2376 | buf = xmalloc(o->file.size + 1); | |
2377 | memcpy(buf, o->file.ptr, o->file.size + 1); | |
2378 | sb.final_buf = buf; | |
2379 | sb.final_buf_size = o->file.size; | |
2380 | } | |
2381 | else { | |
2382 | o = get_origin(&sb, sb.final, path); | |
2383 | if (fill_blob_sha1(o)) | |
2384 | die("no such path %s in %s", path, final_commit_name); | |
cee7f245 | 2385 | |
21666f1a | 2386 | sb.final_buf = read_sha1_file(o->blob_sha1, &type, |
1cfe7733 | 2387 | &sb.final_buf_size); |
ab43e495 JH |
2388 | if (!sb.final_buf) |
2389 | die("Cannot read blob %s for path %s", | |
2390 | sha1_to_hex(o->blob_sha1), | |
2391 | path); | |
1cfe7733 | 2392 | } |
c2e525d9 | 2393 | num_read_blob++; |
cee7f245 JH |
2394 | lno = prepare_lines(&sb); |
2395 | ||
931233bc JH |
2396 | bottom = top = 0; |
2397 | if (bottomtop) | |
2398 | prepare_blame_range(&sb, bottomtop, lno, &bottom, &top); | |
2399 | if (bottom && top && top < bottom) { | |
2400 | long tmp; | |
2401 | tmp = top; top = bottom; bottom = tmp; | |
2402 | } | |
cee7f245 JH |
2403 | if (bottom < 1) |
2404 | bottom = 1; | |
2405 | if (top < 1) | |
2406 | top = lno; | |
2407 | bottom--; | |
2408 | if (lno < top) | |
2409 | die("file %s has only %lu lines", path, lno); | |
2410 | ||
2411 | ent = xcalloc(1, sizeof(*ent)); | |
2412 | ent->lno = bottom; | |
2413 | ent->num_lines = top - bottom; | |
2414 | ent->suspect = o; | |
2415 | ent->s_lno = bottom; | |
2416 | ||
2417 | sb.ent = ent; | |
2418 | sb.path = path; | |
2419 | ||
2420 | if (revs_file && read_ancestry(revs_file)) | |
2421 | die("reading graft file %s failed: %s", | |
2422 | revs_file, strerror(errno)); | |
2423 | ||
50acc589 | 2424 | read_mailmap(&mailmap, ".mailmap", NULL); |
f9567384 | 2425 | |
b92565dc MH |
2426 | if (!incremental) |
2427 | setup_pager(); | |
2428 | ||
d24bba80 | 2429 | assign_blame(&sb, &revs, opt); |
cee7f245 | 2430 | |
717d1462 LT |
2431 | if (incremental) |
2432 | return 0; | |
2433 | ||
cee7f245 JH |
2434 | coalesce(&sb); |
2435 | ||
2436 | if (!(output_option & OUTPUT_PORCELAIN)) | |
2437 | find_alignment(&sb, &output_option); | |
2438 | ||
2439 | output(&sb, output_option); | |
2440 | free((void *)sb.final_buf); | |
2441 | for (ent = sb.ent; ent; ) { | |
2442 | struct blame_entry *e = ent->next; | |
2443 | free(ent); | |
2444 | ent = e; | |
2445 | } | |
c2e525d9 | 2446 | |
870b39c1 | 2447 | if (show_stats) { |
c2e525d9 JH |
2448 | printf("num read blob: %d\n", num_read_blob); |
2449 | printf("num get patch: %d\n", num_get_patch); | |
2450 | printf("num commits: %d\n", num_commits); | |
2451 | } | |
cee7f245 JH |
2452 | return 0; |
2453 | } |