1 /*
2  * Copyright (c) 2018, 2019, 2020 Stefan Sperling <stsp@openbsd.org>
3  * Copyright (c) 2020 Neels Hofmeyr <neels@hofmeyr.de>
4  *
5  * Permission to use, copy, modify, and distribute this software for any
6  * purpose with or without fee is hereby granted, provided that the above
7  * copyright notice and this permission notice appear in all copies.
8  *
9  * THE SOFTWARE IS PROVIDED "AS IS" AND THE AUTHOR DISCLAIMS ALL WARRANTIES
10  * WITH REGARD TO THIS SOFTWARE INCLUDING ALL IMPLIED WARRANTIES OF
11  * MERCHANTABILITY AND FITNESS. IN NO EVENT SHALL THE AUTHOR BE LIABLE FOR
12  * ANY SPECIAL, DIRECT, INDIRECT, OR CONSEQUENTIAL DAMAGES OR ANY DAMAGES
13  * WHATSOEVER RESULTING FROM LOSS OF USE, DATA OR PROFITS, WHETHER IN AN
14  * ACTION OF CONTRACT, NEGLIGENCE OR OTHER TORTIOUS ACTION, ARISING OUT OF
15  * OR IN CONNECTION WITH THE USE OR PERFORMANCE OF THIS SOFTWARE.
16  */
17 
18 #include <sys/queue.h>
19 #include <sys/mman.h>
20 #include <sys/stat.h>
21 
22 #include <errno.h>
23 #include <sha1.h>
24 #include <sha2.h>
25 #include <string.h>
26 #include <stdio.h>
27 #include <stdlib.h>
28 #include <time.h>
29 #include <limits.h>
30 #include <util.h>
31 #include <zlib.h>
32 
33 #include "got_error.h"
34 #include "got_object.h"
35 #include "got_cancel.h"
36 #include "got_commit_graph.h"
37 #include "got_opentemp.h"
38 #include "got_diff.h"
39 #include "got_blame.h"
40 
41 #include "got_lib_inflate.h"
42 #include "got_lib_delta.h"
43 #include "got_lib_object.h"
44 #include "got_lib_diff.h"
45 
46 #ifndef MAX
47 #define	MAX(_a,_b) ((_a) > (_b) ? (_a) : (_b))
48 #endif
49 
50 struct got_blame_line {
51 	int annotated;
52 	struct got_object_id id;
53 };
54 
55 struct got_blame {
56 	struct diff_config *cfg;
57 	int nlines;	/* number of lines in file being blamed */
58 	int nannotated;	/* number of lines already annotated */
59 	struct got_blame_line *lines; /* one per line */
60 	int ncommits;
61 
62 	/*
63 	 * These change with every traversed commit. After diffing
64 	 * commits N:N-1, in preparation for diffing commits N-1:N-2,
65 	 * data for commit N is retained and flipped into data for N-1.
66 	 *
67 	 */
68 	FILE *f1; /* older version from commit N-1. */
69 	FILE *f2; /* newer version from commit N. */
70 	int fd;
71 	unsigned char *map1;
72 	unsigned char *map2;
73 	off_t size1;
74 	off_t size2;
75 	int nlines1;
76 	int nlines2;
77 	off_t *line_offsets1;
78 	off_t *line_offsets2;
79 
80 	/*
81 	 * Map line numbers of an older version of the file to valid line
82 	 * numbers in the version of the file being blamed. This map is
83 	 * updated with each commit we traverse throughout the file's history.
84 	 * Lines mapped to -1 do not correspond to any line in the version
85 	 * being blamed.
86 	 */
87 	int *linemap1;
88 	int *linemap2;
89 
90 	struct diff_data *data1;
91 	struct diff_data *data2;
92 };
93 
94 static const struct got_error *
95 annotate_line(struct got_blame *blame, int lineno,
96     struct got_commit_object *commit, struct got_object_id *id,
97     got_blame_cb cb, void *arg)
98 {
99 	const struct got_error *err = NULL;
100 	struct got_blame_line *line;
101 
102 	if (lineno < 0 || lineno >= blame->nlines)
103 		return NULL;
104 
105 	line = &blame->lines[lineno];
106 	if (line->annotated)
107 		return NULL;
108 
109 	memcpy(&line->id, id, sizeof(line->id));
110 	line->annotated = 1;
111 	blame->nannotated++;
112 	if (cb)
113 		err = cb(arg, blame->nlines, lineno + 1, commit, id);
114 	return err;
115 }
116 
117 static const struct got_error *
118 blame_changes(struct got_blame *blame, struct diff_result *diff_result,
119     struct got_commit_object *commit, struct got_object_id *commit_id,
120     got_blame_cb cb, void *arg)
121 {
122 	const struct got_error *err = NULL;
123 	int i;
124 	int idx1 = 0, idx2 = 0;
125 
126 	for (i = 0; i < diff_result->chunks.len &&
127 	    blame->nannotated < blame->nlines; i++) {
128 		struct diff_chunk *c = diff_chunk_get(diff_result, i);
129 		unsigned int left_count, right_count;
130 		int j;
131 
132 		/*
133 		 * We do not need to worry about idx1/idx2 growing out
134 		 * of bounds because the diff implementation ensures
135 		 * that chunk ranges never exceed the number of lines
136 		 * in the left/right input files.
137 		 */
138 		left_count = diff_chunk_get_left_count(c);
139 		right_count = diff_chunk_get_right_count(c);
140 
141 		if (left_count == right_count) {
142 			for (j = 0; j < left_count; j++) {
143 				blame->linemap1[idx1++] =
144 				    blame->linemap2[idx2++];
145 			}
146 			continue;
147 		}
148 
149 		if (right_count == 0) {
150 			for (j = 0; j < left_count; j++) {
151 				blame->linemap1[idx1++] = -1;
152 			}
153 			continue;
154 		}
155 
156 		for (j = 0; j < right_count; j++) {
157 			int ln = blame->linemap2[idx2++];
158 			err = annotate_line(blame, ln, commit, commit_id,
159 			    cb, arg);
160 			if (err)
161 				return err;
162 			if (blame->nlines == blame->nannotated)
163 				break;
164 		}
165 	}
166 
167 	return NULL;
168 }
169 
170 static const struct got_error *
171 blame_prepare_file(FILE *f, unsigned char **p, off_t *size,
172     int *nlines, off_t **line_offsets, struct diff_data *diff_data,
173     const struct diff_config *cfg, struct got_blob_object *blob)
174 {
175 	const struct got_error *err = NULL;
176 	int diff_flags = 0, rc;
177 
178 	err = got_object_blob_dump_to_file(size, nlines, line_offsets,
179 	    f, blob);
180 	if (err)
181 		return err;
182 
183 #ifndef GOT_DIFF_NO_MMAP
184 	*p = mmap(NULL, *size, PROT_READ, MAP_PRIVATE, fileno(f), 0);
185 	if (*p == MAP_FAILED)
186 #endif
187 		*p = NULL; /* fall back on file I/O */
188 
189 	/* Allow blaming lines in binary files even though it's useless. */
190 	diff_flags |= DIFF_FLAG_FORCE_TEXT_DATA;
191 
192 	rc = diff_atomize_file(diff_data, cfg, f, *p, *size, diff_flags);
193 	if (rc)
194 		return got_error_set_errno(rc, "diff_atomize_file");
195 
196 	return NULL;
197 }
198 
199 static const struct got_error *
200 blame_commit(struct got_blame *blame, struct got_object_id *id,
201     const char *path, struct got_repository *repo,
202     got_blame_cb cb, void *arg)
203 {
204 	const struct got_error *err = NULL;
205 	struct got_commit_object *commit = NULL, *pcommit = NULL;
206 	struct got_object_qid *pid = NULL;
207 	struct got_object_id *pblob_id = NULL;
208 	struct got_blob_object *pblob = NULL;
209 	struct diff_result *diff_result = NULL;
210 
211 	err = got_object_open_as_commit(&commit, repo, id);
212 	if (err)
213 		return err;
214 
215 	pid = STAILQ_FIRST(got_object_commit_get_parent_ids(commit));
216 	if (pid == NULL) {
217 		got_object_commit_close(commit);
218 		return NULL;
219 	}
220 
221 	err = got_object_open_as_commit(&pcommit, repo, &pid->id);
222 	if (err)
223 		goto done;
224 
225 	err = got_object_id_by_path(&pblob_id, repo, pcommit, path);
226 	if (err) {
227 		if (err->code == GOT_ERR_NO_TREE_ENTRY)
228 			err = NULL;
229 		goto done;
230 	}
231 
232 	err = got_object_open_as_blob(&pblob, repo, pblob_id, 8192, blame->fd);
233 	if (err)
234 		goto done;
235 
236 	err = blame_prepare_file(blame->f1, &blame->map1, &blame->size1,
237 	    &blame->nlines1, &blame->line_offsets1, blame->data1,
238 	    blame->cfg, pblob);
239 	if (err)
240 		goto done;
241 
242 	diff_result = diff_main(blame->cfg, blame->data1, blame->data2);
243 	if (diff_result == NULL) {
244 		err = got_error_set_errno(ENOMEM, "malloc");
245 		goto done;
246 	}
247 	if (diff_result->rc != DIFF_RC_OK) {
248 		err = got_error_set_errno(diff_result->rc, "diff");
249 		goto done;
250 	}
251 	if (diff_result->chunks.len > 0) {
252 		if (blame->nlines1 > 0) {
253 			blame->linemap1 = calloc(blame->nlines1,
254 			    sizeof(*blame->linemap1));
255 			if (blame->linemap1 == NULL) {
256 				err = got_error_from_errno("malloc");
257 				goto done;
258 			}
259 		}
260 		err = blame_changes(blame, diff_result, commit, id, cb, arg);
261 		if (err)
262 			goto done;
263 	} else if (cb)
264 		err = cb(arg, blame->nlines, -1, commit, id);
265 done:
266 	if (diff_result)
267 		diff_result_free(diff_result);
268 	if (commit)
269 		got_object_commit_close(commit);
270 	if (pcommit)
271 		got_object_commit_close(pcommit);
272 	free(pblob_id);
273 	if (pblob)
274 		got_object_blob_close(pblob);
275 	return err;
276 }
277 
278 static const struct got_error *
279 blame_close(struct got_blame *blame)
280 {
281 	const struct got_error *err = NULL;
282 
283 	diff_data_free(blame->data1);
284 	free(blame->data1);
285 	diff_data_free(blame->data2);
286 	free(blame->data2);
287 	if (blame->map1) {
288 		if (munmap(blame->map1, blame->size1) == -1 && err == NULL)
289 			err = got_error_from_errno("munmap");
290 	}
291 	if (blame->map2) {
292 		if (munmap(blame->map2, blame->size2) == -1 && err == NULL)
293 			err = got_error_from_errno("munmap");
294 	}
295 	free(blame->lines);
296 	free(blame->line_offsets1);
297 	free(blame->line_offsets2);
298 	free(blame->linemap1);
299 	free(blame->linemap2);
300 	free(blame->cfg);
301 	free(blame);
302 	return err;
303 }
304 
305 static int
306 atomize_file(struct diff_data *d, FILE *f, off_t filesize, int nlines,
307     off_t *line_offsets)
308 {
309 	int i, rc = DIFF_RC_OK;
310 	int embedded_nul = 0;
311 
312 	ARRAYLIST_INIT(d->atoms, nlines);
313 
314 	for (i = 0; i < nlines; i++) {
315 		struct diff_atom *atom;
316 		off_t len, pos = line_offsets[i];
317 		unsigned int hash = 0;
318 		int j;
319 
320 		ARRAYLIST_ADD(atom, d->atoms);
321 		if (atom == NULL) {
322 			rc = errno;
323 			break;
324 		}
325 
326 		if (i < nlines - 1)
327 			len = line_offsets[i + 1] - pos;
328 		else
329 			len = filesize - pos;
330 
331 		if (fseeko(f, pos, SEEK_SET) == -1) {
332 			rc = errno;
333 			break;
334 		}
335 		for (j = 0; j < len; j++) {
336 			int c = fgetc(f);
337 			if (c == EOF) {
338 				if (feof(f))
339 					rc = EIO; /* unexpected EOF */
340 				else
341 					rc = errno;
342 				goto done;
343 			}
344 
345 			hash = diff_atom_hash_update(hash, (unsigned char)c);
346 
347 			if (c == '\0')
348 				embedded_nul = 1;
349 
350 		}
351 		*atom = (struct diff_atom){
352 			.root = d,
353 			.pos = pos,
354 			.at = NULL,	/* atom data is not memory-mapped */
355 			.len = len,
356 			.hash = hash,
357 		};
358 	}
359 
360 	/* File are considered binary if they contain embedded '\0' bytes. */
361 	if (embedded_nul)
362 		d->atomizer_flags |= DIFF_ATOMIZER_FOUND_BINARY_DATA;
363 done:
364 	if (rc)
365 		ARRAYLIST_FREE(d->atoms);
366 
367 	return rc;
368 }
369 
370 static int
371 atomize_file_mmap(struct diff_data *d, unsigned char *p,
372     off_t filesize, int nlines, off_t *line_offsets)
373 {
374 	int i, rc = DIFF_RC_OK;
375 	int embedded_nul = 0;
376 
377 	ARRAYLIST_INIT(d->atoms, nlines);
378 
379 	for (i = 0; i < nlines; i++) {
380 		struct diff_atom *atom;
381 		off_t len, pos = line_offsets[i];
382 		unsigned int hash = 0;
383 		int j;
384 
385 		ARRAYLIST_ADD(atom, d->atoms);
386 		if (atom == NULL) {
387 			rc = errno;
388 			break;
389 		}
390 
391 		if (i < nlines - 1)
392 			len = line_offsets[i + 1] - pos;
393 		else
394 			len = filesize - pos;
395 
396 		for (j = 0; j < len; j++)
397 			hash = diff_atom_hash_update(hash, p[pos + j]);
398 
399 		if (!embedded_nul && memchr(&p[pos], '\0', len) != NULL)
400 			embedded_nul = 1;
401 
402 		*atom = (struct diff_atom){
403 			.root = d,
404 			.pos = pos,
405 			.at = &p[pos],
406 			.len = len,
407 			.hash = hash,
408 		};
409 	}
410 
411 	/* File are considered binary if they contain embedded '\0' bytes. */
412 	if (embedded_nul)
413 		d->atomizer_flags |= DIFF_ATOMIZER_FOUND_BINARY_DATA;
414 
415 	if (rc)
416 		ARRAYLIST_FREE(d->atoms);
417 
418 	return rc;
419 }
420 
421 /* Implements diff_atomize_func_t */
422 static int
423 blame_atomize_file(void *arg, struct diff_data *d)
424 {
425 	struct got_blame *blame = arg;
426 
427 	if (d->f == blame->f1) {
428 		if (blame->map1)
429 			return atomize_file_mmap(d, blame->map1,
430 			    blame->size1, blame->nlines1,
431 			    blame->line_offsets1);
432 		else
433 			return atomize_file(d, blame->f1, blame->size1,
434 			    blame->nlines1, blame->line_offsets1);
435 	} else if (d->f == blame->f2) {
436 		if (d->atoms.len > 0) {
437 			/* Re-use data from previous commit. */
438 			return DIFF_RC_OK;
439 		}
440 		if (blame->map2)
441 			return atomize_file_mmap(d, blame->map2,
442 			    blame->size2, blame->nlines2,
443 			    blame->line_offsets2);
444 		else
445 			return atomize_file(d, blame->f2, blame->size2,
446 			    blame->nlines2, blame->line_offsets2);
447 	}
448 
449 	return DIFF_RC_OK;
450 }
451 
452 static const struct got_error *
453 flip_files(struct got_blame *blame)
454 {
455 	const struct got_error *err = NULL;
456 	struct diff_data *d;
457 	FILE *tmp;
458 
459 	free(blame->line_offsets2);
460 	blame->line_offsets2 = blame->line_offsets1;
461 	blame->line_offsets1 = NULL;
462 
463 	free(blame->linemap2);
464 	blame->linemap2 = blame->linemap1;
465 	blame->linemap1 = NULL;
466 
467 	if (blame->map2) {
468 		if (munmap(blame->map2, blame->size2) == -1)
469 			return got_error_from_errno("munmap");
470 		blame->map2 = blame->map1;
471 		blame->map1 = NULL;
472 	}
473 	blame->size2 = blame->size1;
474 
475 	err = got_opentemp_truncate(blame->f2);
476 	if (err)
477 		return err;
478 	tmp = blame->f2;
479 	blame->f2 = blame->f1;
480 	blame->f1 = tmp;
481 	blame->size1 = 0;
482 
483 	blame->nlines2 = blame->nlines1;
484 	blame->nlines1 = 0;
485 
486 	diff_data_free(blame->data2); /* does not free pointer itself */
487 	memset(blame->data2, 0, sizeof(*blame->data2));
488 	d = blame->data2;
489 	blame->data2 = blame->data1;
490 	blame->data1 = d;
491 
492 	return NULL;
493 }
494 
495 static const struct got_error *
496 blame_open(struct got_blame **blamep, const char *path,
497     struct got_object_id *start_commit_id, struct got_repository *repo,
498     enum got_diff_algorithm diff_algo, got_blame_cb cb, void *arg,
499     got_cancel_cb cancel_cb, void *cancel_arg,
500     int fd1, int fd2, FILE *f1, FILE *f2)
501 {
502 	const struct got_error *err = NULL;
503 	struct got_commit_object *start_commit = NULL, *last_commit = NULL;
504 	struct got_object_id *obj_id = NULL;
505 	struct got_blob_object *blob = NULL;
506 	struct got_blame *blame = NULL;
507 	struct got_object_id id;
508 	int lineno, have_id = 0;
509 	struct got_commit_graph *graph = NULL;
510 
511 	*blamep = NULL;
512 
513 	err = got_object_open_as_commit(&start_commit, repo, start_commit_id);
514 	if (err)
515 		goto done;
516 
517 	err = got_object_id_by_path(&obj_id, repo, start_commit, path);
518 	if (err)
519 		goto done;
520 
521 	err = got_object_open_as_blob(&blob, repo, obj_id, 8192, fd1);
522 	if (err)
523 		goto done;
524 
525 	blame = calloc(1, sizeof(*blame));
526 	if (blame == NULL) {
527 		err = got_error_from_errno("calloc");
528 		goto done;
529 	}
530 
531 	blame->data1 = calloc(1, sizeof(*blame->data1));
532 	if (blame->data1 == NULL) {
533 		err = got_error_from_errno("calloc");
534 		goto done;
535 	}
536 	blame->data2 = calloc(1, sizeof(*blame->data2));
537 	if (blame->data2 == NULL) {
538 		err = got_error_from_errno("calloc");
539 		goto done;
540 	}
541 
542 	blame->f1 = f1;
543 	blame->f2 = f2;
544 	blame->fd = fd2;
545 
546 	err = got_diff_get_config(&blame->cfg, diff_algo, blame_atomize_file,
547 	    blame);
548 	if (err)
549 		goto done;
550 
551 	err = blame_prepare_file(blame->f2, &blame->map2, &blame->size2,
552 	    &blame->nlines2, &blame->line_offsets2, blame->data2,
553 	    blame->cfg, blob);
554 	blame->nlines = blame->nlines2;
555 	if (err || blame->nlines == 0)
556 		goto done;
557 
558 	got_object_blob_close(blob);
559 	blob = NULL;
560 
561 	/* Don't include \n at EOF in the blame line count. */
562 	if (blame->line_offsets2[blame->nlines - 1] == blame->size2)
563 		blame->nlines--;
564 
565 	blame->lines = calloc(blame->nlines, sizeof(*blame->lines));
566 	if (blame->lines == NULL) {
567 		err = got_error_from_errno("calloc");
568 		goto done;
569 	}
570 
571 	blame->linemap2 = calloc(blame->nlines2, sizeof(*blame->linemap2));
572 	if (blame->linemap2 == NULL) {
573 		err = got_error_from_errno("calloc");
574 		goto done;
575 	}
576 	for (lineno = 0; lineno < blame->nlines2; lineno++)
577 		blame->linemap2[lineno] = lineno;
578 
579 	err = got_commit_graph_open(&graph, path, 1);
580 	if (err)
581 		goto done;
582 
583 	err = got_commit_graph_iter_start(graph, start_commit_id, repo,
584 	    cancel_cb, cancel_arg);
585 	if (err)
586 		goto done;
587 	for (;;) {
588 		err = got_commit_graph_iter_next(&id, graph, repo,
589 		    cancel_cb, cancel_arg);
590 		if (err) {
591 			if (err->code == GOT_ERR_ITER_COMPLETED) {
592 				err = NULL;
593 				break;
594 			}
595 			goto done;
596 		}
597 		have_id = 1;
598 
599 		err = blame_commit(blame, &id, path, repo, cb, arg);
600 		if (err) {
601 			if (err->code == GOT_ERR_ITER_COMPLETED)
602 				err = NULL;
603 			goto done;
604 		}
605 		if (blame->nannotated == blame->nlines)
606 			break;
607 
608 		err = flip_files(blame);
609 		if (err)
610 			goto done;
611 	}
612 
613 	if (have_id && blame->nannotated < blame->nlines) {
614 		/* Annotate remaining non-annotated lines with last commit. */
615 		err = got_object_open_as_commit(&last_commit, repo, &id);
616 		if (err)
617 			goto done;
618 		for (lineno = 0; lineno < blame->nlines; lineno++) {
619 			err = annotate_line(blame, lineno, last_commit, &id,
620 			    cb, arg);
621 			if (err)
622 				goto done;
623 		}
624 	}
625 
626 done:
627 	if (graph)
628 		got_commit_graph_close(graph);
629 	free(obj_id);
630 	if (blob)
631 		got_object_blob_close(blob);
632 	if (start_commit)
633 		got_object_commit_close(start_commit);
634 	if (last_commit)
635 		got_object_commit_close(last_commit);
636 	if (err) {
637 		if (blame)
638 			blame_close(blame);
639 	} else
640 		*blamep = blame;
641 
642 	return err;
643 }
644 
645 const struct got_error *
646 got_blame(const char *path, struct got_object_id *commit_id,
647     struct got_repository *repo, enum got_diff_algorithm diff_algo,
648     got_blame_cb cb, void *arg, got_cancel_cb cancel_cb, void* cancel_arg,
649     int fd1, int fd2, FILE *f1, FILE *f2)
650 {
651 	const struct got_error *err = NULL, *close_err = NULL;
652 	struct got_blame *blame;
653 	char *abspath;
654 
655 	if (asprintf(&abspath, "%s%s", path[0] == '/' ? "" : "/", path) == -1)
656 		return got_error_from_errno2("asprintf", path);
657 
658 	err = blame_open(&blame, abspath, commit_id, repo, diff_algo,
659 	    cb, arg, cancel_cb, cancel_arg, fd1, fd2, f1, f2);
660 	free(abspath);
661 	if (blame)
662 		close_err = blame_close(blame);
663 	return err ? err : close_err;
664 }