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, struct got_object_id *id,
92     const struct got_error *(*cb)(void *, int, int, struct got_object_id *),
93     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, 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_object_id *commit_id,
116     const struct got_error *(*cb)(void *, int, int, struct got_object_id *),
117     void *arg)
118 {
119 	const struct got_error *err = NULL;
120 	int i;
121 	int idx1 = 0, idx2 = 0;
122 
123 	for (i = 0; i < diff_result->chunks.len &&
124 	    blame->nannotated < blame->nlines; i++) {
125 		struct diff_chunk *c = diff_chunk_get(diff_result, i);
126 		unsigned int left_start, left_count;
127 		unsigned int right_start, 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_start = diff_chunk_get_left_start(c, diff_result, 0);
137 		left_count = diff_chunk_get_left_count(c);
138 		right_start = diff_chunk_get_right_start(c, diff_result, 0);
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_id, 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     const struct got_error *(*cb)(void *, int, int, struct got_object_id *),
202     void *arg)
203 {
204 	const struct got_error *err = NULL;
205 	struct got_commit_object *commit = 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_id_by_path(&pblob_id, repo, pid->id, 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, id, cb, arg);
263 		if (err)
264 			goto done;
265 	} else if (cb)
266 		err = cb(arg, blame->nlines, -1, id);
267 done:
268 	if (diff_result)
269 		diff_result_free(diff_result);
270 	if (commit)
271 		got_object_commit_close(commit);
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 	if (blame->f1 && fclose(blame->f1) == EOF && err == NULL)
296 		err = got_error_from_errno("fclose");
297 	if (blame->f2 && fclose(blame->f2) == EOF && err == NULL)
298 		err = got_error_from_errno("fclose");
299 	free(blame->lines);
300 	free(blame->line_offsets1);
301 	free(blame->line_offsets2);
302 	free(blame->linemap1);
303 	free(blame->linemap2);
304 	free(blame->cfg);
305 	free(blame);
306 	return err;
307 }
308 
309 static int
310 atomize_file(struct diff_data *d, FILE *f, off_t filesize, int nlines,
311     off_t *line_offsets)
312 {
313 	int i, rc = DIFF_RC_OK;
314 	int embedded_nul = 0;
315 
316 	ARRAYLIST_INIT(d->atoms, nlines);
317 
318 	for (i = 0; i < nlines; i++) {
319 		struct diff_atom *atom;
320 		off_t len, pos = line_offsets[i];
321 		unsigned int hash = 0;
322 		int j;
323 
324 		ARRAYLIST_ADD(atom, d->atoms);
325 		if (atom == NULL) {
326 			rc = errno;
327 			break;
328 		}
329 
330 		if (i < nlines - 1)
331 			len = line_offsets[i + 1] - pos;
332 		else
333 			len = filesize - pos;
334 
335 		if (fseeko(f, pos, SEEK_SET) == -1) {
336 			rc = errno;
337 			break;
338 		}
339 		for (j = 0; j < len; j++) {
340 			int c = fgetc(f);
341 			if (c == EOF) {
342 				if (feof(f))
343 					rc = EIO; /* unexpected EOF */
344 				else
345 					rc = errno;
346 				goto done;
347 			}
348 
349 			hash = diff_atom_hash_update(hash, (unsigned char)c);
350 
351 			if (c == '\0')
352 				embedded_nul = 1;
353 
354 		}
355 		*atom = (struct diff_atom){
356 			.root = d,
357 			.pos = pos,
358 			.at = NULL,	/* atom data is not memory-mapped */
359 			.len = len,
360 			.hash = hash,
361 		};
362 	}
363 
364 	/* File are considered binary if they contain embedded '\0' bytes. */
365 	if (embedded_nul)
366 		d->atomizer_flags |= DIFF_ATOMIZER_FOUND_BINARY_DATA;
367 done:
368 	if (rc)
369 		ARRAYLIST_FREE(d->atoms);
370 
371 	return rc;
372 }
373 
374 static int
375 atomize_file_mmap(struct diff_data *d, unsigned char *p,
376     off_t filesize, int nlines, off_t *line_offsets)
377 {
378 	int i, rc = DIFF_RC_OK;
379 	int embedded_nul = 0;
380 
381 	ARRAYLIST_INIT(d->atoms, nlines);
382 
383 	for (i = 0; i < nlines; i++) {
384 		struct diff_atom *atom;
385 		off_t len, pos = line_offsets[i];
386 		unsigned int hash = 0;
387 		int j;
388 
389 		ARRAYLIST_ADD(atom, d->atoms);
390 		if (atom == NULL) {
391 			rc = errno;
392 			break;
393 		}
394 
395 		if (i < nlines - 1)
396 			len = line_offsets[i + 1] - pos;
397 		else
398 			len = filesize - pos;
399 
400 		for (j = 0; j < len; j++)
401 			hash = diff_atom_hash_update(hash, p[pos + j]);
402 
403 		if (!embedded_nul && memchr(&p[pos], '\0', len) != NULL)
404 			embedded_nul = 1;
405 
406 		*atom = (struct diff_atom){
407 			.root = d,
408 			.pos = pos,
409 			.at = &p[pos],
410 			.len = len,
411 			.hash = hash,
412 		};
413 	}
414 
415 	/* File are considered binary if they contain embedded '\0' bytes. */
416 	if (embedded_nul)
417 		d->atomizer_flags |= DIFF_ATOMIZER_FOUND_BINARY_DATA;
418 
419 	if (rc)
420 		ARRAYLIST_FREE(d->atoms);
421 
422 	return rc;
423 }
424 
425 /* Implements diff_atomize_func_t */
426 static int
427 blame_atomize_file(void *arg, struct diff_data *d)
428 {
429 	struct got_blame *blame = arg;
430 
431 	if (d->f == blame->f1) {
432 		if (blame->map1)
433 			return atomize_file_mmap(d, blame->map1,
434 			    blame->size1, blame->nlines1,
435 			    blame->line_offsets1);
436 		else
437 			return atomize_file(d, blame->f1, blame->size1,
438 			    blame->nlines1, blame->line_offsets1);
439 	} else if (d->f == blame->f2) {
440 		if (d->atoms.len > 0) {
441 			/* Re-use data from previous commit. */
442 			return DIFF_RC_OK;
443 		}
444 		if (blame->map2)
445 			return atomize_file_mmap(d, blame->map2,
446 			    blame->size2, blame->nlines2,
447 			    blame->line_offsets2);
448 		else
449 			return atomize_file(d, blame->f2, blame->size2,
450 			    blame->nlines2, blame->line_offsets2);
451 	}
452 
453 	return DIFF_RC_OK;
454 }
455 
456 static const struct got_error *
457 close_file2_and_reuse_file1(struct got_blame *blame)
458 {
459 	struct diff_data *d;
460 
461 	free(blame->line_offsets2);
462 	blame->line_offsets2 = blame->line_offsets1;
463 	blame->line_offsets1 = NULL;
464 
465 	free(blame->linemap2);
466 	blame->linemap2 = blame->linemap1;
467 	blame->linemap1 = NULL;
468 
469 	if (blame->map2) {
470 		if (munmap(blame->map2, blame->size2) == -1)
471 			return got_error_from_errno("munmap");
472 		blame->map2 = blame->map1;
473 		blame->map1 = NULL;
474 
475 	}
476 	blame->size2 = blame->size1;
477 	blame->size1 = 0;
478 
479 	if (fclose(blame->f2) == EOF)
480 		return got_error_from_errno("fclose");
481 	blame->f2 = blame->f1;
482 	blame->f1 = NULL;
483 
484 	blame->nlines2 = blame->nlines1;
485 	blame->nlines1 = 0;
486 
487 	diff_data_free(blame->data2); /* does not free pointer itself */
488 	memset(blame->data2, 0, sizeof(*blame->data2));
489 	d = blame->data2;
490 	blame->data2 = blame->data1;
491 	blame->data1 = d;
492 
493 	return NULL;
494 }
495 
496 static const struct got_error *
497 blame_open(struct got_blame **blamep, const char *path,
498     struct got_object_id *start_commit_id, struct got_repository *repo,
499     const struct got_error *(*cb)(void *, int, int, struct got_object_id *),
500     void *arg, got_cancel_cb cancel_cb, void *cancel_arg)
501 {
502 	const struct got_error *err = 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 = NULL;
507 	int lineno;
508 	struct got_commit_graph *graph = NULL;
509 
510 	*blamep = NULL;
511 
512 	err = got_object_id_by_path(&obj_id, repo, start_commit_id, path);
513 	if (err)
514 		goto done;
515 
516 	err = got_object_open_as_blob(&blob, repo, obj_id, 8192);
517 	if (err)
518 		goto done;
519 
520 	blame = calloc(1, sizeof(*blame));
521 	if (blame == NULL) {
522 		err = got_error_from_errno("calloc");
523 		goto done;
524 	}
525 
526 	blame->data1 = calloc(1, sizeof(*blame->data1));
527 	if (blame->data1 == NULL) {
528 		err = got_error_from_errno("calloc");
529 		goto done;
530 	}
531 	blame->data2 = calloc(1, sizeof(*blame->data2));
532 	if (blame->data2 == NULL) {
533 		err = got_error_from_errno("calloc");
534 		goto done;
535 	}
536 
537 	blame->f2 = got_opentemp();
538 	if (blame->f2 == NULL) {
539 		err = got_error_from_errno("got_opentemp");
540 		goto done;
541 	}
542 	err = got_diff_get_config(&blame->cfg, GOT_DIFF_ALGORITHM_PATIENCE,
543 	    blame_atomize_file, blame);
544 	if (err)
545 		goto done;
546 
547 	err = blame_prepare_file(blame->f2, &blame->map2, &blame->size2,
548 	    &blame->nlines2, &blame->line_offsets2, blame->data2,
549 	    blame->cfg, blob);
550 	blame->nlines = blame->nlines2;
551 	if (err || blame->nlines == 0)
552 		goto done;
553 
554 	got_object_blob_close(blob);
555 	blob = NULL;
556 
557 	/* Don't include \n at EOF in the blame line count. */
558 	if (blame->line_offsets2[blame->nlines - 1] == blame->size2)
559 		blame->nlines--;
560 
561 	blame->lines = calloc(blame->nlines, sizeof(*blame->lines));
562 	if (blame->lines == NULL) {
563 		err = got_error_from_errno("calloc");
564 		goto done;
565 	}
566 
567 	blame->linemap2 = calloc(blame->nlines2, sizeof(*blame->linemap2));
568 	if (blame->linemap2 == NULL) {
569 		err = got_error_from_errno("calloc");
570 		goto done;
571 	}
572 	for (lineno = 0; lineno < blame->nlines2; lineno++)
573 		blame->linemap2[lineno] = lineno;
574 
575 	err = got_commit_graph_open(&graph, path, 1);
576 	if (err)
577 		goto done;
578 
579 	err = got_commit_graph_iter_start(graph, start_commit_id, repo,
580 	    cancel_cb, cancel_arg);
581 	if (err)
582 		goto done;
583 	for (;;) {
584 		struct got_object_id *next_id;
585 		err = got_commit_graph_iter_next(&next_id, graph, repo,
586 		    cancel_cb, cancel_arg);
587 		if (err) {
588 			if (err->code == GOT_ERR_ITER_COMPLETED) {
589 				err = NULL;
590 				break;
591 			}
592 			goto done;
593 		}
594 		if (next_id) {
595 			id = next_id;
596 			err = blame_commit(blame, id, path, repo, cb, arg);
597 			if (err) {
598 				if (err->code == GOT_ERR_ITER_COMPLETED)
599 					err = NULL;
600 				goto done;
601 			}
602 			if (blame->nannotated == blame->nlines)
603 				break;
604 
605 			err = close_file2_and_reuse_file1(blame);
606 			if (err)
607 				goto done;
608 		}
609 	}
610 
611 	if (id && blame->nannotated < blame->nlines) {
612 		/* Annotate remaining non-annotated lines with last commit. */
613 		for (lineno = 0; lineno < blame->nlines; lineno++) {
614 			err = annotate_line(blame, lineno, id, cb, arg);
615 			if (err)
616 				goto done;
617 		}
618 	}
619 
620 done:
621 	if (graph)
622 		got_commit_graph_close(graph);
623 	free(obj_id);
624 	if (blob)
625 		got_object_blob_close(blob);
626 	if (err) {
627 		if (blame)
628 			blame_close(blame);
629 	} else
630 		*blamep = blame;
631 
632 	return err;
633 }
634 
635 const struct got_error *
636 got_blame(const char *path, struct got_object_id *commit_id,
637     struct got_repository *repo,
638     const struct got_error *(*cb)(void *, int, int, struct got_object_id *),
639     void *arg, got_cancel_cb cancel_cb, void* cancel_arg)
640 {
641 	const struct got_error *err = NULL, *close_err = NULL;
642 	struct got_blame *blame;
643 	char *abspath;
644 
645 	if (asprintf(&abspath, "%s%s", path[0] == '/' ? "" : "/", path) == -1)
646 		return got_error_from_errno2("asprintf", path);
647 
648 	err = blame_open(&blame, abspath, commit_id, repo, cb, arg,
649 	    cancel_cb, cancel_arg);
650 	free(abspath);
651 	if (blame)
652 		close_err = blame_close(blame);
653 	return err ? err : close_err;
654 }