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_blame.h"
35 #include "got_commit_graph.h"
36 #include "got_opentemp.h"
37 
38 #include "got_lib_inflate.h"
39 #include "got_lib_delta.h"
40 #include "got_lib_object.h"
41 #include "got_lib_diff.h"
42 
43 #ifndef MAX
44 #define	MAX(_a,_b) ((_a) > (_b) ? (_a) : (_b))
45 #endif
46 
47 struct got_blame_line {
48 	int annotated;
49 	struct got_object_id id;
50 };
51 
52 struct got_blame {
53 	struct diff_config *cfg;
54 	int nlines;	/* number of lines in file being blamed */
55 	int nannotated;	/* number of lines already annotated */
56 	struct got_blame_line *lines; /* one per line */
57 	int ncommits;
58 
59 	/*
60 	 * These change with every traversed commit. After diffing
61 	 * commits N:N-1, in preparation for diffing commits N-1:N-2,
62 	 * data for commit N is retained and flipped into data for N-1.
63 	 * 
64 	 */
65 	FILE *f1; /* older version from commit N-1. */
66 	FILE *f2; /* newer version from commit N. */
67 	unsigned char *map1;
68 	unsigned char *map2;
69 	off_t size1;
70 	off_t size2;
71 	int nlines1;
72 	int nlines2;
73 	off_t *line_offsets1;
74 	off_t *line_offsets2;
75 
76 	/*
77 	 * Map line numbers of an older version of the file to valid line
78 	 * numbers in the version of the file being blamed. This map is
79 	 * updated with each commit we traverse throughout the file's history.
80 	 * Lines mapped to -1 do not correspond to any line in the version
81 	 * being blamed.
82 	 */
83 	int *linemap1;
84 	int *linemap2;
85 
86 	struct diff_data *data1;
87 	struct diff_data *data2;
88 };
89 
90 static const struct got_error *
91 annotate_line(struct got_blame *blame, int lineno,
92     struct got_commit_object *commit, struct got_object_id *id,
93     got_blame_cb cb, void *arg)
94 {
95 	const struct got_error *err = NULL;
96 	struct got_blame_line *line;
97 
98 	if (lineno < 0 || lineno >= blame->nlines)
99 		return NULL;
100 
101 	line = &blame->lines[lineno];
102 	if (line->annotated)
103 		return NULL;
104 
105 	memcpy(&line->id, id, sizeof(line->id));
106 	line->annotated = 1;
107 	blame->nannotated++;
108 	if (cb)
109 		err = cb(arg, blame->nlines, lineno + 1, commit, id);
110 	return err;
111 }
112 
113 static const struct got_error *
114 blame_changes(struct got_blame *blame, struct diff_result *diff_result,
115     struct got_commit_object *commit, struct got_object_id *commit_id,
116     got_blame_cb cb, void *arg)
117 {
118 	const struct got_error *err = NULL;
119 	int i;
120 	int idx1 = 0, idx2 = 0;
121 
122 	for (i = 0; i < diff_result->chunks.len &&
123 	    blame->nannotated < blame->nlines; i++) {
124 		struct diff_chunk *c = diff_chunk_get(diff_result, i);
125 		unsigned int left_count, right_count;
126 		int j;
127 
128 		/*
129 		 * We do not need to worry about idx1/idx2 growing out
130 		 * of bounds because the diff implementation ensures
131 		 * that chunk ranges never exceed the number of lines
132 		 * in the left/right input files.
133 		 */
134 		left_count = diff_chunk_get_left_count(c);
135 		right_count = diff_chunk_get_right_count(c);
136 
137 		if (left_count == right_count) {
138 			for (j = 0; j < left_count; j++) {
139 				blame->linemap1[idx1++] =
140 				    blame->linemap2[idx2++];
141 			}
142 			continue;
143 		}
144 
145 		if (right_count == 0) {
146 			for (j = 0; j < left_count; j++) {
147 				blame->linemap1[idx1++] = -1;
148 			}
149 			continue;
150 		}
151 
152 		for (j = 0; j < right_count; j++) {
153 			int ln = blame->linemap2[idx2++];
154 			err = annotate_line(blame, ln, commit, commit_id,
155 			    cb, arg);
156 			if (err)
157 				return err;
158 			if (blame->nlines == blame->nannotated)
159 				break;
160 		}
161 	}
162 
163 	return NULL;
164 }
165 
166 static const struct got_error *
167 blame_prepare_file(FILE *f, unsigned char **p, off_t *size,
168     int *nlines, off_t **line_offsets, struct diff_data *diff_data,
169     const struct diff_config *cfg, struct got_blob_object *blob)
170 {
171 	const struct got_error *err = NULL;
172 	int diff_flags = 0, rc;
173 
174 	err = got_object_blob_dump_to_file(size, nlines, line_offsets,
175 	    f, blob);
176 	if (err)
177 		return err;
178 
179 #ifndef GOT_DIFF_NO_MMAP
180 	*p = mmap(NULL, *size, PROT_READ, MAP_PRIVATE, fileno(f), 0);
181 	if (*p == MAP_FAILED)
182 #endif
183 		*p = NULL; /* fall back on file I/O */
184 
185 	/* Allow blaming lines in binary files even though it's useless. */
186 	diff_flags |= DIFF_FLAG_FORCE_TEXT_DATA;
187 
188 	rc = diff_atomize_file(diff_data, cfg, f, *p, *size, diff_flags);
189 	if (rc)
190 		return got_error_set_errno(rc, "diff_atomize_file");
191 
192 	return NULL;
193 }
194 
195 static const struct got_error *
196 blame_commit(struct got_blame *blame, struct got_object_id *id,
197     const char *path, struct got_repository *repo,
198     got_blame_cb cb, void *arg)
199 {
200 	const struct got_error *err = NULL;
201 	struct got_commit_object *commit = NULL, *pcommit = NULL;
202 	struct got_object_qid *pid = NULL;
203 	struct got_object_id *pblob_id = NULL;
204 	struct got_blob_object *pblob = NULL;
205 	struct diff_result *diff_result = NULL;
206 
207 	err = got_object_open_as_commit(&commit, repo, id);
208 	if (err)
209 		return err;
210 
211 	pid = STAILQ_FIRST(got_object_commit_get_parent_ids(commit));
212 	if (pid == NULL) {
213 		got_object_commit_close(commit);
214 		return NULL;
215 	}
216 
217 	err = got_object_open_as_commit(&pcommit, repo, pid->id);
218 	if (err)
219 		goto done;
220 
221 	err = got_object_id_by_path(&pblob_id, repo, pcommit, path);
222 	if (err) {
223 		if (err->code == GOT_ERR_NO_TREE_ENTRY)
224 			err = NULL;
225 		goto done;
226 	}
227 
228 	err = got_object_open_as_blob(&pblob, repo, pblob_id, 8192);
229 	if (err)
230 		goto done;
231 
232 	blame->f1 = got_opentemp();
233 	if (blame->f1 == NULL) {
234 		err = got_error_from_errno("got_opentemp");
235 		goto done;
236 	}
237 
238 	err = blame_prepare_file(blame->f1, &blame->map1, &blame->size1,
239 	    &blame->nlines1, &blame->line_offsets1, blame->data1,
240 	    blame->cfg, pblob);
241 	if (err)
242 		goto done;
243 
244 	diff_result = diff_main(blame->cfg, blame->data1, blame->data2);
245 	if (diff_result == NULL) {
246 		err = got_error_set_errno(ENOMEM, "malloc");
247 		goto done;
248 	}
249 	if (diff_result->rc != DIFF_RC_OK) {
250 		err = got_error_set_errno(diff_result->rc, "diff");
251 		goto done;
252 	}
253 	if (diff_result->chunks.len > 0) {
254 		if (blame->nlines1 > 0) {
255 			blame->linemap1 = calloc(blame->nlines1,
256 			    sizeof(*blame->linemap1));
257 			if (blame->linemap1 == NULL) {
258 				err = got_error_from_errno("malloc");
259 				goto done;
260 			}
261 		}
262 		err = blame_changes(blame, diff_result, commit, id, cb, arg);
263 		if (err)
264 			goto done;
265 	} else if (cb)
266 		err = cb(arg, blame->nlines, -1, commit, id);
267 done:
268 	if (diff_result)
269 		diff_result_free(diff_result);
270 	if (commit)
271 		got_object_commit_close(commit);
272 	if (pcommit)
273 		got_object_commit_close(pcommit);
274 	free(pblob_id);
275 	if (pblob)
276 		got_object_blob_close(pblob);
277 	return err;
278 }
279 
280 static const struct got_error *
281 blame_close(struct got_blame *blame)
282 {
283 	const struct got_error *err = NULL;
284 
285 	diff_data_free(blame->data1);
286 	free(blame->data1);
287 	diff_data_free(blame->data2);
288 	free(blame->data2);
289 	if (blame->map1) {
290 		if (munmap(blame->map1, blame->size1) == -1 && err == NULL)
291 			err = got_error_from_errno("munmap");
292 	}
293 	if (blame->map2) {
294 		if (munmap(blame->map2, blame->size2) == -1 && err == NULL)
295 			err = got_error_from_errno("munmap");
296 	}
297 	if (blame->f1 && fclose(blame->f1) == EOF && err == NULL)
298 		err = got_error_from_errno("fclose");
299 	if (blame->f2 && fclose(blame->f2) == EOF && err == NULL)
300 		err = got_error_from_errno("fclose");
301 	free(blame->lines);
302 	free(blame->line_offsets1);
303 	free(blame->line_offsets2);
304 	free(blame->linemap1);
305 	free(blame->linemap2);
306 	free(blame->cfg);
307 	free(blame);
308 	return err;
309 }
310 
311 static int
312 atomize_file(struct diff_data *d, FILE *f, off_t filesize, int nlines,
313     off_t *line_offsets)
314 {
315 	int i, rc = DIFF_RC_OK;
316 	int embedded_nul = 0;
317 
318 	ARRAYLIST_INIT(d->atoms, nlines);
319 
320 	for (i = 0; i < nlines; i++) {
321 		struct diff_atom *atom;
322 		off_t len, pos = line_offsets[i];
323 		unsigned int hash = 0;
324 		int j;
325 
326 		ARRAYLIST_ADD(atom, d->atoms);
327 		if (atom == NULL) {
328 			rc = errno;
329 			break;
330 		}
331 
332 		if (i < nlines - 1)
333 			len = line_offsets[i + 1] - pos;
334 		else
335 			len = filesize - pos;
336 
337 		if (fseeko(f, pos, SEEK_SET) == -1) {
338 			rc = errno;
339 			break;
340 		}
341 		for (j = 0; j < len; j++) {
342 			int c = fgetc(f);
343 			if (c == EOF) {
344 				if (feof(f))
345 					rc = EIO; /* unexpected EOF */
346 				else
347 					rc = errno;
348 				goto done;
349 			}
350 
351 			hash = diff_atom_hash_update(hash, (unsigned char)c);
352 
353 			if (c == '\0')
354 				embedded_nul = 1;
355 
356 		}
357 		*atom = (struct diff_atom){
358 			.root = d,
359 			.pos = pos,
360 			.at = NULL,	/* atom data is not memory-mapped */
361 			.len = len,
362 			.hash = hash,
363 		};
364 	}
365 
366 	/* File are considered binary if they contain embedded '\0' bytes. */
367 	if (embedded_nul)
368 		d->atomizer_flags |= DIFF_ATOMIZER_FOUND_BINARY_DATA;
369 done:
370 	if (rc)
371 		ARRAYLIST_FREE(d->atoms);
372 
373 	return rc;
374 }
375 
376 static int
377 atomize_file_mmap(struct diff_data *d, unsigned char *p,
378     off_t filesize, int nlines, off_t *line_offsets)
379 {
380 	int i, rc = DIFF_RC_OK;
381 	int embedded_nul = 0;
382 
383 	ARRAYLIST_INIT(d->atoms, nlines);
384 
385 	for (i = 0; i < nlines; i++) {
386 		struct diff_atom *atom;
387 		off_t len, pos = line_offsets[i];
388 		unsigned int hash = 0;
389 		int j;
390 
391 		ARRAYLIST_ADD(atom, d->atoms);
392 		if (atom == NULL) {
393 			rc = errno;
394 			break;
395 		}
396 
397 		if (i < nlines - 1)
398 			len = line_offsets[i + 1] - pos;
399 		else
400 			len = filesize - pos;
401 
402 		for (j = 0; j < len; j++)
403 			hash = diff_atom_hash_update(hash, p[pos + j]);
404 
405 		if (!embedded_nul && memchr(&p[pos], '\0', len) != NULL)
406 			embedded_nul = 1;
407 
408 		*atom = (struct diff_atom){
409 			.root = d,
410 			.pos = pos,
411 			.at = &p[pos],
412 			.len = len,
413 			.hash = hash,
414 		};
415 	}
416 
417 	/* File are considered binary if they contain embedded '\0' bytes. */
418 	if (embedded_nul)
419 		d->atomizer_flags |= DIFF_ATOMIZER_FOUND_BINARY_DATA;
420 
421 	if (rc)
422 		ARRAYLIST_FREE(d->atoms);
423 
424 	return rc;
425 }
426 
427 /* Implements diff_atomize_func_t */
428 static int
429 blame_atomize_file(void *arg, struct diff_data *d)
430 {
431 	struct got_blame *blame = arg;
432 
433 	if (d->f == blame->f1) {
434 		if (blame->map1)
435 			return atomize_file_mmap(d, blame->map1,
436 			    blame->size1, blame->nlines1,
437 			    blame->line_offsets1);
438 		else
439 			return atomize_file(d, blame->f1, blame->size1,
440 			    blame->nlines1, blame->line_offsets1);
441 	} else if (d->f == blame->f2) {
442 		if (d->atoms.len > 0) {
443 			/* Re-use data from previous commit. */
444 			return DIFF_RC_OK;
445 		}
446 		if (blame->map2)
447 			return atomize_file_mmap(d, blame->map2,
448 			    blame->size2, blame->nlines2,
449 			    blame->line_offsets2);
450 		else
451 			return atomize_file(d, blame->f2, blame->size2,
452 			    blame->nlines2, blame->line_offsets2);
453 	}
454 
455 	return DIFF_RC_OK;
456 }
457 
458 static const struct got_error *
459 close_file2_and_reuse_file1(struct got_blame *blame)
460 {
461 	struct diff_data *d;
462 
463 	free(blame->line_offsets2);
464 	blame->line_offsets2 = blame->line_offsets1;
465 	blame->line_offsets1 = NULL;
466 
467 	free(blame->linemap2);
468 	blame->linemap2 = blame->linemap1;
469 	blame->linemap1 = NULL;
470 
471 	if (blame->map2) {
472 		if (munmap(blame->map2, blame->size2) == -1)
473 			return got_error_from_errno("munmap");
474 		blame->map2 = blame->map1;
475 		blame->map1 = NULL;
476 
477 	}
478 	blame->size2 = blame->size1;
479 	blame->size1 = 0;
480 
481 	if (fclose(blame->f2) == EOF)
482 		return got_error_from_errno("fclose");
483 	blame->f2 = blame->f1;
484 	blame->f1 = NULL;
485 
486 	blame->nlines2 = blame->nlines1;
487 	blame->nlines1 = 0;
488 
489 	diff_data_free(blame->data2); /* does not free pointer itself */
490 	memset(blame->data2, 0, sizeof(*blame->data2));
491 	d = blame->data2;
492 	blame->data2 = blame->data1;
493 	blame->data1 = d;
494 
495 	return NULL;
496 }
497 
498 static const struct got_error *
499 blame_open(struct got_blame **blamep, const char *path,
500     struct got_object_id *start_commit_id, struct got_repository *repo,
501     got_blame_cb cb, void *arg, got_cancel_cb cancel_cb, void *cancel_arg)
502 {
503 	const struct got_error *err = NULL;
504 	struct got_commit_object *start_commit = NULL, *last_commit = NULL;
505 	struct got_object_id *obj_id = NULL;
506 	struct got_blob_object *blob = NULL;
507 	struct got_blame *blame = NULL;
508 	struct got_object_id *id = NULL;
509 	int lineno;
510 	struct got_commit_graph *graph = NULL;
511 
512 	*blamep = NULL;
513 
514 	err = got_object_open_as_commit(&start_commit, repo, start_commit_id);
515 	if (err)
516 		goto done;
517 
518 	err = got_object_id_by_path(&obj_id, repo, start_commit, path);
519 	if (err)
520 		goto done;
521 
522 	err = got_object_open_as_blob(&blob, repo, obj_id, 8192);
523 	if (err)
524 		goto done;
525 
526 	blame = calloc(1, sizeof(*blame));
527 	if (blame == NULL) {
528 		err = got_error_from_errno("calloc");
529 		goto done;
530 	}
531 
532 	blame->data1 = calloc(1, sizeof(*blame->data1));
533 	if (blame->data1 == NULL) {
534 		err = got_error_from_errno("calloc");
535 		goto done;
536 	}
537 	blame->data2 = calloc(1, sizeof(*blame->data2));
538 	if (blame->data2 == NULL) {
539 		err = got_error_from_errno("calloc");
540 		goto done;
541 	}
542 
543 	blame->f2 = got_opentemp();
544 	if (blame->f2 == NULL) {
545 		err = got_error_from_errno("got_opentemp");
546 		goto done;
547 	}
548 	err = got_diff_get_config(&blame->cfg, GOT_DIFF_ALGORITHM_PATIENCE,
549 	    blame_atomize_file, blame);
550 	if (err)
551 		goto done;
552 
553 	err = blame_prepare_file(blame->f2, &blame->map2, &blame->size2,
554 	    &blame->nlines2, &blame->line_offsets2, blame->data2,
555 	    blame->cfg, blob);
556 	blame->nlines = blame->nlines2;
557 	if (err || blame->nlines == 0)
558 		goto done;
559 
560 	got_object_blob_close(blob);
561 	blob = NULL;
562 
563 	/* Don't include \n at EOF in the blame line count. */
564 	if (blame->line_offsets2[blame->nlines - 1] == blame->size2)
565 		blame->nlines--;
566 
567 	blame->lines = calloc(blame->nlines, sizeof(*blame->lines));
568 	if (blame->lines == NULL) {
569 		err = got_error_from_errno("calloc");
570 		goto done;
571 	}
572 
573 	blame->linemap2 = calloc(blame->nlines2, sizeof(*blame->linemap2));
574 	if (blame->linemap2 == NULL) {
575 		err = got_error_from_errno("calloc");
576 		goto done;
577 	}
578 	for (lineno = 0; lineno < blame->nlines2; lineno++)
579 		blame->linemap2[lineno] = lineno;
580 
581 	err = got_commit_graph_open(&graph, path, 1);
582 	if (err)
583 		goto done;
584 
585 	err = got_commit_graph_iter_start(graph, start_commit_id, repo,
586 	    cancel_cb, cancel_arg);
587 	if (err)
588 		goto done;
589 	for (;;) {
590 		struct got_object_id *next_id;
591 		err = got_commit_graph_iter_next(&next_id, graph, repo,
592 		    cancel_cb, cancel_arg);
593 		if (err) {
594 			if (err->code == GOT_ERR_ITER_COMPLETED) {
595 				err = NULL;
596 				break;
597 			}
598 			goto done;
599 		}
600 		if (next_id) {
601 			id = next_id;
602 			err = blame_commit(blame, id, path, repo, cb, arg);
603 			if (err) {
604 				if (err->code == GOT_ERR_ITER_COMPLETED)
605 					err = NULL;
606 				goto done;
607 			}
608 			if (blame->nannotated == blame->nlines)
609 				break;
610 
611 			err = close_file2_and_reuse_file1(blame);
612 			if (err)
613 				goto done;
614 		}
615 	}
616 
617 	if (id && blame->nannotated < blame->nlines) {
618 		/* Annotate remaining non-annotated lines with last commit. */
619 		err = got_object_open_as_commit(&last_commit, repo, id);
620 		if (err)
621 			goto done;
622 		for (lineno = 0; lineno < blame->nlines; lineno++) {
623 			err = annotate_line(blame, lineno, last_commit, id,
624 			    cb, arg);
625 			if (err)
626 				goto done;
627 		}
628 	}
629 
630 done:
631 	if (graph)
632 		got_commit_graph_close(graph);
633 	free(obj_id);
634 	if (blob)
635 		got_object_blob_close(blob);
636 	if (start_commit)
637 		got_object_commit_close(start_commit);
638 	if (last_commit)
639 		got_object_commit_close(last_commit);
640 	if (err) {
641 		if (blame)
642 			blame_close(blame);
643 	} else
644 		*blamep = blame;
645 
646 	return err;
647 }
648 
649 const struct got_error *
650 got_blame(const char *path, struct got_object_id *commit_id,
651     struct got_repository *repo, got_blame_cb cb, void *arg,
652     got_cancel_cb cancel_cb, void* cancel_arg)
653 {
654 	const struct got_error *err = NULL, *close_err = NULL;
655 	struct got_blame *blame;
656 	char *abspath;
657 
658 	if (asprintf(&abspath, "%s%s", path[0] == '/' ? "" : "/", path) == -1)
659 		return got_error_from_errno2("asprintf", path);
660 
661 	err = blame_open(&blame, abspath, commit_id, repo, cb, arg,
662 	    cancel_cb, cancel_arg);
663 	free(abspath);
664 	if (blame)
665 		close_err = blame_close(blame);
666 	return err ? err : close_err;
667 }