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 404c43c4 2018-06-21 stsp #include "got_lib_zbuf.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"
38 404c43c4 2018-06-21 stsp struct got_blame_line {
39 404c43c4 2018-06-21 stsp int annotated;
40 9b94757a 2018-06-21 stsp struct got_object_id id;
43 404c43c4 2018-06-21 stsp struct got_blame {
45 404c43c4 2018-06-21 stsp size_t nlines;
46 404c43c4 2018-06-21 stsp struct got_blame_line *lines; /* one per line */
49 404c43c4 2018-06-21 stsp static const struct got_error *
50 404c43c4 2018-06-21 stsp dump_blob_and_count_lines(size_t *nlines, FILE *outfile,
51 404c43c4 2018-06-21 stsp struct got_blob_object *blob)
53 404c43c4 2018-06-21 stsp const struct got_error *err = NULL;
54 404c43c4 2018-06-21 stsp size_t len, hdrlen;
55 404c43c4 2018-06-21 stsp const uint8_t *buf;
58 404c43c4 2018-06-21 stsp hdrlen = got_object_blob_get_hdrlen(blob);
59 404c43c4 2018-06-21 stsp *nlines = 0;
61 404c43c4 2018-06-21 stsp err = got_object_blob_read_block(&len, blob);
64 404c43c4 2018-06-21 stsp if (len == 0)
66 404c43c4 2018-06-21 stsp buf = got_object_blob_get_read_buf(blob);
67 404c43c4 2018-06-21 stsp for (i = 0; i < len; i++) {
68 404c43c4 2018-06-21 stsp if (buf[i] == '\n')
69 404c43c4 2018-06-21 stsp (*nlines)++;
71 404c43c4 2018-06-21 stsp /* Skip blob object header first time around. */
72 404c43c4 2018-06-21 stsp fwrite(buf + hdrlen, len - hdrlen, 1, outfile);
74 404c43c4 2018-06-21 stsp } while (len != 0);
77 404c43c4 2018-06-21 stsp fflush(outfile);
78 404c43c4 2018-06-21 stsp rewind(outfile);
80 404c43c4 2018-06-21 stsp return NULL;
84 404c43c4 2018-06-21 stsp annotate_line(struct got_blame *blame, int lineno, struct got_object_id *id)
86 404c43c4 2018-06-21 stsp struct got_blame_line *line;
88 404c43c4 2018-06-21 stsp if (lineno < 1 || lineno > blame->nlines)
91 404c43c4 2018-06-21 stsp line = &blame->lines[lineno - 1];
92 404c43c4 2018-06-21 stsp if (line->annotated)
95 404c43c4 2018-06-21 stsp memcpy(&line->id, id, sizeof(line->id));
96 404c43c4 2018-06-21 stsp line->annotated = 1;
99 404c43c4 2018-06-21 stsp static const struct got_error *
100 404c43c4 2018-06-21 stsp blame_commit(struct got_blame *blame, struct got_object_id *id,
101 404c43c4 2018-06-21 stsp struct got_object_id *pid, const char *path, struct got_repository *repo)
103 404c43c4 2018-06-21 stsp const struct got_error *err = NULL;
104 404c43c4 2018-06-21 stsp struct got_object *obj = NULL, *pobj = NULL;
105 404c43c4 2018-06-21 stsp struct got_blob_object *blob = NULL, *pblob = NULL;
106 404c43c4 2018-06-21 stsp struct got_diff_changes *changes = NULL;
108 404c43c4 2018-06-21 stsp err = got_object_open_by_path(&obj, repo, id, path);
111 404c43c4 2018-06-21 stsp if (got_object_get_type(obj) != GOT_OBJ_TYPE_BLOB) {
112 404c43c4 2018-06-21 stsp err = got_error(GOT_ERR_OBJ_TYPE);
116 404c43c4 2018-06-21 stsp err = got_object_open_by_path(&pobj, repo, pid, path);
118 404c43c4 2018-06-21 stsp if (err->code == GOT_ERR_NO_OBJ) {
119 404c43c4 2018-06-21 stsp /* Blob's history began in previous commit. */
120 404c43c4 2018-06-21 stsp err = got_error(GOT_ERR_ITER_COMPLETED);
124 404c43c4 2018-06-21 stsp if (got_object_get_type(pobj) != GOT_OBJ_TYPE_BLOB) {
126 404c43c4 2018-06-21 stsp * Encountered a non-blob at the path (probably a tree).
127 404c43c4 2018-06-21 stsp * Blob's history began in previous commit.
129 404c43c4 2018-06-21 stsp err = got_error(GOT_ERR_ITER_COMPLETED);
133 404c43c4 2018-06-21 stsp /* If blob hashes match then don't bother with diffing. */
134 404c43c4 2018-06-21 stsp if (got_object_id_cmp(&obj->id, &pobj->id) == 0)
137 404c43c4 2018-06-21 stsp err = got_object_blob_open(&blob, repo, obj, 8192);
141 404c43c4 2018-06-21 stsp err = got_object_blob_open(&pblob, repo, pobj, 8192);
145 404c43c4 2018-06-21 stsp err = got_diff_blob_lines_changed(&changes, blob, pblob);
149 404c43c4 2018-06-21 stsp if (changes) {
150 404c43c4 2018-06-21 stsp struct got_diff_change *change;
151 404c43c4 2018-06-21 stsp SIMPLEQ_FOREACH(change, &changes->entries, entry) {
152 404c43c4 2018-06-21 stsp int a = change->cv.a;
153 404c43c4 2018-06-21 stsp int b = change->cv.b;
154 404c43c4 2018-06-21 stsp int lineno;
155 d157810f 2018-06-21 stsp for (lineno = a; lineno <= b; lineno++)
156 d157810f 2018-06-21 stsp annotate_line(blame, lineno, id);
161 404c43c4 2018-06-21 stsp got_object_close(obj);
163 404c43c4 2018-06-21 stsp got_object_close(pobj);
165 404c43c4 2018-06-21 stsp got_object_blob_close(blob);
167 404c43c4 2018-06-21 stsp got_object_blob_close(pblob);
168 404c43c4 2018-06-21 stsp return err;
171 404c43c4 2018-06-21 stsp static void
172 404c43c4 2018-06-21 stsp blame_close(struct got_blame *blame)
174 404c43c4 2018-06-21 stsp if (blame->f)
175 404c43c4 2018-06-21 stsp fclose(blame->f);
176 404c43c4 2018-06-21 stsp free(blame->lines);
177 404c43c4 2018-06-21 stsp free(blame);
180 404c43c4 2018-06-21 stsp static const struct got_error *
181 404c43c4 2018-06-21 stsp blame_open(struct got_blame **blamep, const char *path,
182 404c43c4 2018-06-21 stsp struct got_object_id *start_commit_id, struct got_repository *repo)
184 404c43c4 2018-06-21 stsp const struct got_error *err = NULL;
185 404c43c4 2018-06-21 stsp struct got_object *obj = NULL;
186 404c43c4 2018-06-21 stsp struct got_blob_object *blob = NULL;
187 404c43c4 2018-06-21 stsp struct got_blame *blame = NULL;
188 404c43c4 2018-06-21 stsp struct got_commit_object *commit = NULL;
189 404c43c4 2018-06-21 stsp struct got_object_id *id = NULL;
190 404c43c4 2018-06-21 stsp int lineno;
192 404c43c4 2018-06-21 stsp *blamep = NULL;
194 404c43c4 2018-06-21 stsp err = got_object_open_by_path(&obj, repo, start_commit_id, path);
196 404c43c4 2018-06-21 stsp return err;
197 404c43c4 2018-06-21 stsp if (got_object_get_type(obj) != GOT_OBJ_TYPE_BLOB) {
198 404c43c4 2018-06-21 stsp err = got_error(GOT_ERR_OBJ_TYPE);
202 404c43c4 2018-06-21 stsp err = got_object_blob_open(&blob, repo, obj, 8192);
206 404c43c4 2018-06-21 stsp blame = calloc(1, sizeof(*blame));
207 404c43c4 2018-06-21 stsp if (blame == NULL)
208 404c43c4 2018-06-21 stsp return got_error_from_errno();
210 404c43c4 2018-06-21 stsp blame->f = got_opentemp();
211 404c43c4 2018-06-21 stsp if (blame->f == NULL) {
212 404c43c4 2018-06-21 stsp err = got_error_from_errno();
215 404c43c4 2018-06-21 stsp err = dump_blob_and_count_lines(&blame->nlines, blame->f, blob);
219 404c43c4 2018-06-21 stsp blame->lines = calloc(blame->nlines, sizeof(*blame->lines));
220 404c43c4 2018-06-21 stsp if (blame->lines == NULL) {
221 404c43c4 2018-06-21 stsp err = got_error_from_errno();
225 404c43c4 2018-06-21 stsp /* Loop over first-parent history and try to blame commits. */
226 404c43c4 2018-06-21 stsp err = got_object_open_as_commit(&commit, repo, start_commit_id);
229 404c43c4 2018-06-21 stsp id = got_object_id_dup(start_commit_id);
230 404c43c4 2018-06-21 stsp if (id == NULL) {
231 404c43c4 2018-06-21 stsp err = got_error_from_errno();
234 404c43c4 2018-06-21 stsp while (1) {
235 404c43c4 2018-06-21 stsp struct got_object_qid *pid;
237 404c43c4 2018-06-21 stsp pid = SIMPLEQ_FIRST(&commit->parent_ids);
238 404c43c4 2018-06-21 stsp if (pid == NULL)
241 404c43c4 2018-06-21 stsp err = blame_commit(blame, id, pid->id, path, repo);
243 404c43c4 2018-06-21 stsp if (err->code == GOT_ERR_ITER_COMPLETED)
244 404c43c4 2018-06-21 stsp err = NULL;
249 404c43c4 2018-06-21 stsp id = got_object_id_dup(pid->id);
250 404c43c4 2018-06-21 stsp if (id == NULL) {
251 404c43c4 2018-06-21 stsp err = got_error_from_errno();
254 ed77f2ae 2018-06-21 stsp got_object_commit_close(commit);
255 ed77f2ae 2018-06-21 stsp err = got_object_open_as_commit(&commit, repo, id);
260 404c43c4 2018-06-21 stsp /* Annotate remaining non-annotated lines with last commit. */
261 d157810f 2018-06-21 stsp for (lineno = 1; lineno < blame->nlines; lineno++)
262 d157810f 2018-06-21 stsp annotate_line(blame, lineno, id);
267 404c43c4 2018-06-21 stsp got_object_close(obj);
269 404c43c4 2018-06-21 stsp got_object_blob_close(blob);
270 404c43c4 2018-06-21 stsp if (commit)
271 404c43c4 2018-06-21 stsp got_object_commit_close(commit);
273 404c43c4 2018-06-21 stsp blame_close(blame);
275 404c43c4 2018-06-21 stsp *blamep = blame;
277 404c43c4 2018-06-21 stsp return err;
280 404c43c4 2018-06-21 stsp static const struct got_error *
281 404c43c4 2018-06-21 stsp blame_line(struct got_object_id **id, struct got_blame *blame, int lineno)
283 404c43c4 2018-06-21 stsp if (lineno < 1 || lineno > blame->nlines)
284 404c43c4 2018-06-21 stsp return got_error(GOT_ERR_RANGE);
285 404c43c4 2018-06-21 stsp *id = &blame->lines[lineno - 1].id;
286 404c43c4 2018-06-21 stsp return NULL;
289 404c43c4 2018-06-21 stsp static char *
290 404c43c4 2018-06-21 stsp parse_next_line(FILE *f, size_t *len)
292 404c43c4 2018-06-21 stsp char *line;
293 404c43c4 2018-06-21 stsp size_t linelen;
294 404c43c4 2018-06-21 stsp size_t lineno;
295 404c43c4 2018-06-21 stsp const char delim[3] = { '\0', '\0', '\0'};
297 404c43c4 2018-06-21 stsp line = fparseln(f, &linelen, &lineno, delim, 0);
299 404c43c4 2018-06-21 stsp *len = linelen;
300 404c43c4 2018-06-21 stsp return line;
303 404c43c4 2018-06-21 stsp const struct got_error *
304 404c43c4 2018-06-21 stsp got_blame(const char *path, struct got_object_id *start_commit_id,
305 404c43c4 2018-06-21 stsp struct got_repository *repo, FILE *outfile)
307 404c43c4 2018-06-21 stsp const struct got_error *err = NULL;
308 404c43c4 2018-06-21 stsp struct got_blame *blame;
309 404c43c4 2018-06-21 stsp int lineno;
310 404c43c4 2018-06-21 stsp char *abspath;
312 404c43c4 2018-06-21 stsp if (asprintf(&abspath, "%s%s", path[0] == '/' ? "" : "/", path) == -1)
313 404c43c4 2018-06-21 stsp return got_error_from_errno();
315 404c43c4 2018-06-21 stsp err = blame_open(&blame, abspath, start_commit_id, repo);
317 404c43c4 2018-06-21 stsp free(abspath);
318 404c43c4 2018-06-21 stsp return err;
321 404c43c4 2018-06-21 stsp for (lineno = 1; lineno < blame->nlines; lineno++) {
322 404c43c4 2018-06-21 stsp struct got_object_id *id;
323 404c43c4 2018-06-21 stsp char *line, *id_str;
325 404c43c4 2018-06-21 stsp line = parse_next_line(blame->f, NULL);
326 404c43c4 2018-06-21 stsp if (line == NULL)
329 404c43c4 2018-06-21 stsp err = blame_line(&id, blame, lineno);
333 404c43c4 2018-06-21 stsp err = got_object_id_str(&id_str, id);
335 404c43c4 2018-06-21 stsp free(line);
339 404c43c4 2018-06-21 stsp fprintf(outfile, "%.8s %s\n", id_str, line);
340 404c43c4 2018-06-21 stsp free(line);
341 404c43c4 2018-06-21 stsp free(id_str);
344 404c43c4 2018-06-21 stsp blame_close(blame);
345 404c43c4 2018-06-21 stsp free(abspath);
346 404c43c4 2018-06-21 stsp return err;