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