2 404c43c4 2018-06-21 stsp * Copyright (c) 2018 Stefan Sperling <stsp@openbsd.org>
4 404c43c4 2018-06-21 stsp * Permission to use, copy, modify, and distribute this software for any
5 404c43c4 2018-06-21 stsp * purpose with or without fee is hereby granted, provided that the above
6 404c43c4 2018-06-21 stsp * copyright notice and this permission notice appear in all copies.
8 404c43c4 2018-06-21 stsp * THE SOFTWARE IS PROVIDED "AS IS" AND THE AUTHOR DISCLAIMS ALL WARRANTIES
9 404c43c4 2018-06-21 stsp * WITH REGARD TO THIS SOFTWARE INCLUDING ALL IMPLIED WARRANTIES OF
10 404c43c4 2018-06-21 stsp * MERCHANTABILITY AND FITNESS. IN NO EVENT SHALL THE AUTHOR BE LIABLE FOR
11 404c43c4 2018-06-21 stsp * ANY SPECIAL, DIRECT, INDIRECT, OR CONSEQUENTIAL DAMAGES OR ANY DAMAGES
12 404c43c4 2018-06-21 stsp * WHATSOEVER RESULTING FROM LOSS OF USE, DATA OR PROFITS, WHETHER IN AN
13 404c43c4 2018-06-21 stsp * ACTION OF CONTRACT, NEGLIGENCE OR OTHER TORTIOUS ACTION, ARISING OUT OF
14 404c43c4 2018-06-21 stsp * OR IN CONNECTION WITH THE USE OR PERFORMANCE OF THIS SOFTWARE.
17 404c43c4 2018-06-21 stsp #include <sys/queue.h>
18 404c43c4 2018-06-21 stsp #include <sys/stat.h>
20 404c43c4 2018-06-21 stsp #include <sha1.h>
21 404c43c4 2018-06-21 stsp #include <string.h>
22 404c43c4 2018-06-21 stsp #include <stdio.h>
23 404c43c4 2018-06-21 stsp #include <stdlib.h>
24 404c43c4 2018-06-21 stsp #include <time.h>
25 404c43c4 2018-06-21 stsp #include <util.h>
26 404c43c4 2018-06-21 stsp #include <zlib.h>
28 404c43c4 2018-06-21 stsp #include "got_error.h"
29 404c43c4 2018-06-21 stsp #include "got_object.h"
30 404c43c4 2018-06-21 stsp #include "got_blame.h"
31 404c43c4 2018-06-21 stsp #include "got_opentemp.h"
33 63581804 2018-07-09 stsp #include "got_lib_inflate.h"
34 404c43c4 2018-06-21 stsp #include "got_lib_delta.h"
35 404c43c4 2018-06-21 stsp #include "got_lib_object.h"
36 404c43c4 2018-06-21 stsp #include "got_lib_diff.h"
37 c35a7943 2018-07-12 stsp #include "got_lib_diffoffset.h"
39 404c43c4 2018-06-21 stsp struct got_blame_line {
40 404c43c4 2018-06-21 stsp int annotated;
41 9b94757a 2018-06-21 stsp struct got_object_id id;
44 c35a7943 2018-07-12 stsp struct got_blame_diff_offsets {
45 c35a7943 2018-07-12 stsp struct got_diffoffset_chunks *chunks;
46 c35a7943 2018-07-12 stsp struct got_object_id *commit_id;
47 c35a7943 2018-07-12 stsp SLIST_ENTRY(got_blame_diff_offsets) entry;
50 c35a7943 2018-07-12 stsp SLIST_HEAD(got_blame_diff_offsets_list, got_blame_diff_offsets);
52 404c43c4 2018-06-21 stsp struct got_blame {
54 404c43c4 2018-06-21 stsp size_t nlines;
55 404c43c4 2018-06-21 stsp struct got_blame_line *lines; /* one per line */
56 c35a7943 2018-07-12 stsp int ncommits;
57 c35a7943 2018-07-12 stsp struct got_blame_diff_offsets_list diff_offsets_list;
61 c35a7943 2018-07-12 stsp free_diff_offsets(struct got_blame_diff_offsets *diff_offsets)
63 c35a7943 2018-07-12 stsp if (diff_offsets->chunks)
64 c35a7943 2018-07-12 stsp got_diffoffset_free(diff_offsets->chunks);
65 c35a7943 2018-07-12 stsp free(diff_offsets->commit_id);
66 c35a7943 2018-07-12 stsp free(diff_offsets);
69 404c43c4 2018-06-21 stsp static const struct got_error *
70 c35a7943 2018-07-12 stsp alloc_diff_offsets(struct got_blame_diff_offsets **diff_offsets,
71 c35a7943 2018-07-12 stsp struct got_object_id *commit_id)
73 c35a7943 2018-07-12 stsp const struct got_error *err = NULL;
75 c35a7943 2018-07-12 stsp *diff_offsets = calloc(1, sizeof(**diff_offsets));
76 c35a7943 2018-07-12 stsp if (*diff_offsets == NULL)
77 c35a7943 2018-07-12 stsp return got_error_from_errno();
79 c35a7943 2018-07-12 stsp (*diff_offsets)->commit_id = got_object_id_dup(commit_id);
80 c35a7943 2018-07-12 stsp if ((*diff_offsets)->commit_id == NULL) {
81 c35a7943 2018-07-12 stsp err = got_error_from_errno();
82 c35a7943 2018-07-12 stsp free_diff_offsets(*diff_offsets);
83 c35a7943 2018-07-12 stsp *diff_offsets = NULL;
87 c35a7943 2018-07-12 stsp err = got_diffoffset_alloc(&(*diff_offsets)->chunks);
89 c35a7943 2018-07-12 stsp free_diff_offsets(*diff_offsets);
93 c35a7943 2018-07-12 stsp return NULL;
96 c35a7943 2018-07-12 stsp static const struct got_error *
97 84451b3e 2018-07-10 stsp annotate_line(struct got_blame *blame, int lineno, struct got_object_id *id,
98 84451b3e 2018-07-10 stsp const struct got_error *(*cb)(void *, int, int, struct got_object_id *),
101 404c43c4 2018-06-21 stsp const struct got_error *err = NULL;
102 404c43c4 2018-06-21 stsp struct got_blame_line *line;
104 404c43c4 2018-06-21 stsp if (lineno < 1 || lineno > blame->nlines)
105 84451b3e 2018-07-10 stsp return got_error(GOT_ERR_RANGE);
107 404c43c4 2018-06-21 stsp line = &blame->lines[lineno - 1];
108 404c43c4 2018-06-21 stsp if (line->annotated)
109 84451b3e 2018-07-10 stsp return NULL;
111 404c43c4 2018-06-21 stsp memcpy(&line->id, id, sizeof(line->id));
112 404c43c4 2018-06-21 stsp line->annotated = 1;
114 84451b3e 2018-07-10 stsp err = cb(arg, blame->nlines, lineno, id);
115 84451b3e 2018-07-10 stsp return err;
119 c35a7943 2018-07-12 stsp get_blamed_line(struct got_blame_diff_offsets_list *diff_offsets_list,
120 c35a7943 2018-07-12 stsp int lineno)
122 c35a7943 2018-07-12 stsp struct got_blame_diff_offsets *diff_offsets;
124 c35a7943 2018-07-12 stsp SLIST_FOREACH(diff_offsets, diff_offsets_list, entry)
125 c35a7943 2018-07-12 stsp lineno = got_diffoffset_get(diff_offsets->chunks, lineno);
127 c35a7943 2018-07-12 stsp return lineno;
130 404c43c4 2018-06-21 stsp static const struct got_error *
131 c35a7943 2018-07-12 stsp blame_changes(struct got_blame *blame, struct got_diff_changes *changes,
132 c35a7943 2018-07-12 stsp struct got_object_id *commit_id,
133 c35a7943 2018-07-12 stsp const struct got_error *(*cb)(void *, int, int, struct got_object_id *),
136 c35a7943 2018-07-12 stsp const struct got_error *err = NULL;
137 c35a7943 2018-07-12 stsp struct got_diff_change *change;
138 c35a7943 2018-07-12 stsp struct got_blame_diff_offsets *diff_offsets;
140 c35a7943 2018-07-12 stsp SIMPLEQ_FOREACH(change, &changes->entries, entry) {
141 c35a7943 2018-07-12 stsp int c = change->cv.c;
142 c35a7943 2018-07-12 stsp int d = change->cv.d;
143 c35a7943 2018-07-12 stsp int new_lineno = c;
144 c35a7943 2018-07-12 stsp int new_length = (c < d ? d - c + 1 : (c == d ? 1 : 0));
147 c35a7943 2018-07-12 stsp for (ln = new_lineno; ln < new_lineno + new_length; ln++) {
148 c35a7943 2018-07-12 stsp err = annotate_line(blame,
149 c35a7943 2018-07-12 stsp get_blamed_line(&blame->diff_offsets_list, ln),
150 c35a7943 2018-07-12 stsp commit_id, cb, arg);
152 c35a7943 2018-07-12 stsp return err;
156 c35a7943 2018-07-12 stsp err = alloc_diff_offsets(&diff_offsets, commit_id);
158 c35a7943 2018-07-12 stsp return err;
159 c35a7943 2018-07-12 stsp SIMPLEQ_FOREACH(change, &changes->entries, entry) {
160 c35a7943 2018-07-12 stsp int a = change->cv.a;
161 c35a7943 2018-07-12 stsp int b = change->cv.b;
162 c35a7943 2018-07-12 stsp int c = change->cv.c;
163 c35a7943 2018-07-12 stsp int d = change->cv.d;
164 c35a7943 2018-07-12 stsp int old_lineno = a;
165 c35a7943 2018-07-12 stsp int old_length = (a < b ? b - a + 1 : (a == b ? 1 : 0));
166 c35a7943 2018-07-12 stsp int new_lineno = c;
167 c35a7943 2018-07-12 stsp int new_length = (c < d ? d - c + 1 : (c == d ? 1 : 0));
169 c35a7943 2018-07-12 stsp err = got_diffoffset_add(diff_offsets->chunks,
170 c35a7943 2018-07-12 stsp old_lineno, old_length, new_lineno, new_length);
172 df9513f1 2018-07-13 stsp free_diff_offsets(diff_offsets);
173 c35a7943 2018-07-12 stsp return err;
176 c35a7943 2018-07-12 stsp SLIST_INSERT_HEAD(&blame->diff_offsets_list, diff_offsets, entry);
178 c35a7943 2018-07-12 stsp return NULL;
181 c35a7943 2018-07-12 stsp static const struct got_error *
182 404c43c4 2018-06-21 stsp blame_commit(struct got_blame *blame, struct got_object_id *id,
183 84451b3e 2018-07-10 stsp struct got_object_id *pid, const char *path, struct got_repository *repo,
184 84451b3e 2018-07-10 stsp const struct got_error *(*cb)(void *, int, int, struct got_object_id *),
187 404c43c4 2018-06-21 stsp const struct got_error *err = NULL;
188 404c43c4 2018-06-21 stsp struct got_object *obj = NULL, *pobj = NULL;
189 404c43c4 2018-06-21 stsp struct got_blob_object *blob = NULL, *pblob = NULL;
190 404c43c4 2018-06-21 stsp struct got_diff_changes *changes = NULL;
192 404c43c4 2018-06-21 stsp err = got_object_open_by_path(&obj, repo, id, path);
195 404c43c4 2018-06-21 stsp if (got_object_get_type(obj) != GOT_OBJ_TYPE_BLOB) {
196 404c43c4 2018-06-21 stsp err = got_error(GOT_ERR_OBJ_TYPE);
200 404c43c4 2018-06-21 stsp err = got_object_open_by_path(&pobj, repo, pid, path);
202 404c43c4 2018-06-21 stsp if (err->code == GOT_ERR_NO_OBJ) {
203 404c43c4 2018-06-21 stsp /* Blob's history began in previous commit. */
204 404c43c4 2018-06-21 stsp err = got_error(GOT_ERR_ITER_COMPLETED);
208 404c43c4 2018-06-21 stsp if (got_object_get_type(pobj) != GOT_OBJ_TYPE_BLOB) {
210 404c43c4 2018-06-21 stsp * Encountered a non-blob at the path (probably a tree).
211 404c43c4 2018-06-21 stsp * Blob's history began in previous commit.
213 404c43c4 2018-06-21 stsp err = got_error(GOT_ERR_ITER_COMPLETED);
217 404c43c4 2018-06-21 stsp /* If blob hashes match then don't bother with diffing. */
218 89a4e64f 2018-07-11 stsp if (got_object_id_cmp(&obj->id, &pobj->id) == 0) {
220 89a4e64f 2018-07-11 stsp err = cb(arg, blame->nlines, -1, id);
224 404c43c4 2018-06-21 stsp err = got_object_blob_open(&blob, repo, obj, 8192);
228 404c43c4 2018-06-21 stsp err = got_object_blob_open(&pblob, repo, pobj, 8192);
232 c35a7943 2018-07-12 stsp err = got_diff_blob_lines_changed(&changes, pblob, blob);
236 404c43c4 2018-06-21 stsp if (changes) {
237 c35a7943 2018-07-12 stsp err = blame_changes(blame, changes, id, cb, arg);
238 ce7f1bfe 2018-07-13 stsp got_diff_free_changes(changes);
239 d68a0a7d 2018-07-10 stsp } else if (cb)
240 3bf198ba 2018-07-10 stsp err = cb(arg, blame->nlines, -1, id);
243 404c43c4 2018-06-21 stsp got_object_close(obj);
245 404c43c4 2018-06-21 stsp got_object_close(pobj);
247 404c43c4 2018-06-21 stsp got_object_blob_close(blob);
249 404c43c4 2018-06-21 stsp got_object_blob_close(pblob);
250 404c43c4 2018-06-21 stsp return err;
253 404c43c4 2018-06-21 stsp static void
254 404c43c4 2018-06-21 stsp blame_close(struct got_blame *blame)
256 c35a7943 2018-07-12 stsp struct got_blame_diff_offsets *diff_offsets;
258 404c43c4 2018-06-21 stsp if (blame->f)
259 404c43c4 2018-06-21 stsp fclose(blame->f);
260 404c43c4 2018-06-21 stsp free(blame->lines);
261 c35a7943 2018-07-12 stsp while (!SLIST_EMPTY(&blame->diff_offsets_list)) {
262 c35a7943 2018-07-12 stsp diff_offsets = SLIST_FIRST(&blame->diff_offsets_list);
263 c35a7943 2018-07-12 stsp SLIST_REMOVE_HEAD(&blame->diff_offsets_list, entry);
264 c35a7943 2018-07-12 stsp free_diff_offsets(diff_offsets);
266 404c43c4 2018-06-21 stsp free(blame);
269 404c43c4 2018-06-21 stsp static const struct got_error *
270 404c43c4 2018-06-21 stsp blame_open(struct got_blame **blamep, const char *path,
271 84451b3e 2018-07-10 stsp struct got_object_id *start_commit_id, struct got_repository *repo,
272 84451b3e 2018-07-10 stsp const struct got_error *(*cb)(void *, int, int, struct got_object_id *),
275 404c43c4 2018-06-21 stsp const struct got_error *err = NULL;
276 404c43c4 2018-06-21 stsp struct got_object *obj = NULL;
277 404c43c4 2018-06-21 stsp struct got_blob_object *blob = NULL;
278 404c43c4 2018-06-21 stsp struct got_blame *blame = NULL;
279 404c43c4 2018-06-21 stsp struct got_commit_object *commit = NULL;
280 404c43c4 2018-06-21 stsp struct got_object_id *id = NULL;
281 404c43c4 2018-06-21 stsp int lineno;
283 404c43c4 2018-06-21 stsp *blamep = NULL;
285 404c43c4 2018-06-21 stsp err = got_object_open_by_path(&obj, repo, start_commit_id, path);
287 404c43c4 2018-06-21 stsp return err;
288 404c43c4 2018-06-21 stsp if (got_object_get_type(obj) != GOT_OBJ_TYPE_BLOB) {
289 404c43c4 2018-06-21 stsp err = got_error(GOT_ERR_OBJ_TYPE);
293 404c43c4 2018-06-21 stsp err = got_object_blob_open(&blob, repo, obj, 8192);
297 404c43c4 2018-06-21 stsp blame = calloc(1, sizeof(*blame));
298 404c43c4 2018-06-21 stsp if (blame == NULL)
299 404c43c4 2018-06-21 stsp return got_error_from_errno();
301 404c43c4 2018-06-21 stsp blame->f = got_opentemp();
302 404c43c4 2018-06-21 stsp if (blame->f == NULL) {
303 404c43c4 2018-06-21 stsp err = got_error_from_errno();
306 84451b3e 2018-07-10 stsp err = got_object_blob_dump_to_file(NULL, &blame->nlines, blame->f,
311 404c43c4 2018-06-21 stsp blame->lines = calloc(blame->nlines, sizeof(*blame->lines));
312 404c43c4 2018-06-21 stsp if (blame->lines == NULL) {
313 404c43c4 2018-06-21 stsp err = got_error_from_errno();
317 404c43c4 2018-06-21 stsp /* Loop over first-parent history and try to blame commits. */
318 404c43c4 2018-06-21 stsp err = got_object_open_as_commit(&commit, repo, start_commit_id);
321 404c43c4 2018-06-21 stsp id = got_object_id_dup(start_commit_id);
322 404c43c4 2018-06-21 stsp if (id == NULL) {
323 404c43c4 2018-06-21 stsp err = got_error_from_errno();
326 404c43c4 2018-06-21 stsp while (1) {
327 404c43c4 2018-06-21 stsp struct got_object_qid *pid;
329 404c43c4 2018-06-21 stsp pid = SIMPLEQ_FIRST(&commit->parent_ids);
330 404c43c4 2018-06-21 stsp if (pid == NULL)
333 84451b3e 2018-07-10 stsp err = blame_commit(blame, id, pid->id, path, repo, cb, arg);
335 404c43c4 2018-06-21 stsp if (err->code == GOT_ERR_ITER_COMPLETED)
336 404c43c4 2018-06-21 stsp err = NULL;
341 404c43c4 2018-06-21 stsp id = got_object_id_dup(pid->id);
342 404c43c4 2018-06-21 stsp if (id == NULL) {
343 404c43c4 2018-06-21 stsp err = got_error_from_errno();
346 ed77f2ae 2018-06-21 stsp got_object_commit_close(commit);
347 ed77f2ae 2018-06-21 stsp err = got_object_open_as_commit(&commit, repo, id);
352 404c43c4 2018-06-21 stsp /* Annotate remaining non-annotated lines with last commit. */
353 84451b3e 2018-07-10 stsp for (lineno = 1; lineno <= blame->nlines; lineno++) {
354 84451b3e 2018-07-10 stsp err = annotate_line(blame, lineno, id, cb, arg);
362 404c43c4 2018-06-21 stsp got_object_close(obj);
364 404c43c4 2018-06-21 stsp got_object_blob_close(blob);
365 404c43c4 2018-06-21 stsp if (commit)
366 404c43c4 2018-06-21 stsp got_object_commit_close(commit);
369 1828273a 2018-07-09 stsp blame_close(blame);
371 404c43c4 2018-06-21 stsp *blamep = blame;
373 404c43c4 2018-06-21 stsp return err;
376 404c43c4 2018-06-21 stsp static const struct got_error *
377 404c43c4 2018-06-21 stsp blame_line(struct got_object_id **id, struct got_blame *blame, int lineno)
379 404c43c4 2018-06-21 stsp if (lineno < 1 || lineno > blame->nlines)
380 404c43c4 2018-06-21 stsp return got_error(GOT_ERR_RANGE);
381 404c43c4 2018-06-21 stsp *id = &blame->lines[lineno - 1].id;
382 404c43c4 2018-06-21 stsp return NULL;
385 404c43c4 2018-06-21 stsp static char *
386 404c43c4 2018-06-21 stsp parse_next_line(FILE *f, size_t *len)
388 404c43c4 2018-06-21 stsp char *line;
389 404c43c4 2018-06-21 stsp size_t linelen;
390 404c43c4 2018-06-21 stsp size_t lineno;
391 404c43c4 2018-06-21 stsp const char delim[3] = { '\0', '\0', '\0'};
393 404c43c4 2018-06-21 stsp line = fparseln(f, &linelen, &lineno, delim, 0);
395 404c43c4 2018-06-21 stsp *len = linelen;
396 404c43c4 2018-06-21 stsp return line;
399 404c43c4 2018-06-21 stsp const struct got_error *
400 404c43c4 2018-06-21 stsp got_blame(const char *path, struct got_object_id *start_commit_id,
401 404c43c4 2018-06-21 stsp struct got_repository *repo, FILE *outfile)
403 404c43c4 2018-06-21 stsp const struct got_error *err = NULL;
404 404c43c4 2018-06-21 stsp struct got_blame *blame;
405 404c43c4 2018-06-21 stsp int lineno;
406 404c43c4 2018-06-21 stsp char *abspath;
408 404c43c4 2018-06-21 stsp if (asprintf(&abspath, "%s%s", path[0] == '/' ? "" : "/", path) == -1)
409 404c43c4 2018-06-21 stsp return got_error_from_errno();
411 84451b3e 2018-07-10 stsp err = blame_open(&blame, abspath, start_commit_id, repo, NULL, NULL);
413 404c43c4 2018-06-21 stsp free(abspath);
414 404c43c4 2018-06-21 stsp return err;
417 dd031dc0 2018-07-04 stsp for (lineno = 1; lineno <= blame->nlines; lineno++) {
418 404c43c4 2018-06-21 stsp struct got_object_id *id;
419 404c43c4 2018-06-21 stsp char *line, *id_str;
421 404c43c4 2018-06-21 stsp line = parse_next_line(blame->f, NULL);
422 404c43c4 2018-06-21 stsp if (line == NULL)
425 404c43c4 2018-06-21 stsp err = blame_line(&id, blame, lineno);
427 d4af3990 2018-07-13 stsp free(line);
431 404c43c4 2018-06-21 stsp err = got_object_id_str(&id_str, id);
432 58c811c5 2018-07-13 stsp /* Do not free id; It points into blame->lines. */
434 404c43c4 2018-06-21 stsp free(line);
438 404c43c4 2018-06-21 stsp fprintf(outfile, "%.8s %s\n", id_str, line);
439 404c43c4 2018-06-21 stsp free(line);
440 404c43c4 2018-06-21 stsp free(id_str);
443 404c43c4 2018-06-21 stsp blame_close(blame);
444 404c43c4 2018-06-21 stsp free(abspath);
445 404c43c4 2018-06-21 stsp return err;
448 84451b3e 2018-07-10 stsp const struct got_error *
449 84451b3e 2018-07-10 stsp got_blame_incremental(const char *path, struct got_object_id *commit_id,
450 84451b3e 2018-07-10 stsp struct got_repository *repo,
451 84451b3e 2018-07-10 stsp const struct got_error *(*cb)(void *, int, int, struct got_object_id *),
454 84451b3e 2018-07-10 stsp const struct got_error *err = NULL;
455 84451b3e 2018-07-10 stsp struct got_blame *blame;
456 84451b3e 2018-07-10 stsp char *abspath;
458 84451b3e 2018-07-10 stsp if (asprintf(&abspath, "%s%s", path[0] == '/' ? "" : "/", path) == -1)
459 84451b3e 2018-07-10 stsp return got_error_from_errno();
461 84451b3e 2018-07-10 stsp err = blame_open(&blame, abspath, commit_id, repo, cb, arg);
462 84451b3e 2018-07-10 stsp free(abspath);
464 75b7a700 2018-07-10 stsp blame_close(blame);
465 84451b3e 2018-07-10 stsp return err;