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