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 <string.h>
25 #include <stdio.h>
26 #include <stdlib.h>
27 #include <time.h>
28 #include <limits.h>
29 #include <util.h>
30 #include <zlib.h>
31 
32 #include "got_error.h"
33 #include "got_object.h"
34 #include "got_cancel.h"
35 #include "got_blame.h"
36 #include "got_commit_graph.h"
37 #include "got_opentemp.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 	unsigned char *map1;
69 	unsigned char *map2;
70 	off_t size1;
71 	off_t size2;
72 	int nlines1;
73 	int nlines2;
74 	off_t *line_offsets1;
75 	off_t *line_offsets2;
76 
77 	/*
78 	 * Map line numbers of an older version of the file to valid line
79 	 * numbers in the version of the file being blamed. This map is
80 	 * updated with each commit we traverse throughout the file's history.
81 	 * Lines mapped to -1 do not correspond to any line in the version
82 	 * being blamed.
83 	 */
84 	int *linemap1;
85 	int *linemap2;
86 
87 	struct diff_data *data1;
88 	struct diff_data *data2;
89 };
90 
91 static const struct got_error *
92 annotate_line(struct got_blame *blame, int lineno, struct got_object_id *id,
93     const struct got_error *(*cb)(void *, int, int, struct got_object_id *),
94     void *arg)
95 {
96 	const struct got_error *err = NULL;
97 	struct got_blame_line *line;
98 
99 	if (lineno < 0 || lineno >= blame->nlines)
100 		return NULL;
101 
102 	line = &blame->lines[lineno];
103 	if (line->annotated)
104 		return NULL;
105 
106 	memcpy(&line->id, id, sizeof(line->id));
107 	line->annotated = 1;
108 	blame->nannotated++;
109 	if (cb)
110 		err = cb(arg, blame->nlines, lineno + 1, id);
111 	return err;
112 }
113 
114 static const struct got_error *
115 blame_changes(struct got_blame *blame, struct diff_result *diff_result,
116     struct got_object_id *commit_id,
117     const struct got_error *(*cb)(void *, int, int, struct got_object_id *),
118     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_start, left_count;
128 		unsigned int right_start, 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_start = diff_chunk_get_left_start(c, diff_result, 0);
138 		left_count = diff_chunk_get_left_count(c);
139 		right_start = diff_chunk_get_right_start(c, diff_result, 0);
140 		right_count = diff_chunk_get_right_count(c);
141 
142 		if (left_count == right_count) {
143 			for (j = 0; j < left_count; j++) {
144 				blame->linemap1[idx1++] =
145 				    blame->linemap2[idx2++];
146 			}
147 			continue;
148 		}
149 
150 		if (right_count == 0) {
151 			for (j = 0; j < left_count; j++) {
152 				blame->linemap1[idx1++] = -1;
153 			}
154 			continue;
155 		}
156 
157 		for (j = 0; j < right_count; j++) {
158 			int ln = blame->linemap2[idx2++];
159 			err = annotate_line(blame, ln, commit_id, 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 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 	rc = diff_atomize_file(diff_data, cfg, f, *p, *size, 0);
190 	if (rc)
191 		return got_error_set_errno(rc, "diff_atomize_file");
192 
193 	return NULL;
194 }
195 
196 static const struct got_error *
197 blame_commit(struct got_blame *blame, struct got_object_id *id,
198     const char *path, struct got_repository *repo,
199     const struct got_error *(*cb)(void *, int, int, struct got_object_id *),
200     void *arg)
201 {
202 	const struct got_error *err = NULL;
203 	struct got_commit_object *commit = 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 = SIMPLEQ_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_id_by_path(&pblob_id, repo, pid->id, path);
220 	if (err) {
221 		if (err->code == GOT_ERR_NO_TREE_ENTRY)
222 			err = NULL;
223 		goto done;
224 	}
225 
226 	err = got_object_open_as_blob(&pblob, repo, pblob_id, 8192);
227 	if (err)
228 		goto done;
229 
230 	blame->f1 = got_opentemp();
231 	if (blame->f1 == NULL) {
232 		err = got_error_from_errno("got_opentemp");
233 		goto done;
234 	}
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, id, cb, arg);
261 		if (err)
262 			goto done;
263 	} else if (cb)
264 		err = cb(arg, blame->nlines, -1, id);
265 done:
266 	if (diff_result)
267 		diff_result_free(diff_result);
268 	if (commit)
269 		got_object_commit_close(commit);
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 	if (blame->f1 && fclose(blame->f1) != 0 && err == NULL)
294 		err = got_error_from_errno("fclose");
295 	if (blame->f2 && fclose(blame->f2) != 0 && err == NULL)
296 		err = got_error_from_errno("fclose");
297 	free(blame->lines);
298 	free(blame->line_offsets1);
299 	free(blame->line_offsets2);
300 	free(blame->linemap1);
301 	free(blame->linemap2);
302 	free(blame->cfg);
303 	free(blame);
304 	return err;
305 }
306 
307 static int
308 atomize_file(struct diff_data *d, FILE *f, off_t filesize, int nlines,
309     off_t *line_offsets)
310 {
311 	int i, rc = DIFF_RC_OK;
312 
313 	ARRAYLIST_INIT(d->atoms, nlines);
314 
315 	for (i = 0; i < nlines; i++) {
316 		struct diff_atom *atom;
317 		off_t len, pos = line_offsets[i];
318 		unsigned int hash = 0;
319 		int j;
320 
321 		ARRAYLIST_ADD(atom, d->atoms);
322 		if (atom == NULL) {
323 			rc = errno;
324 			break;
325 		}
326 
327 		if (i < nlines - 1)
328 			len = line_offsets[i + 1] - pos;
329 		else
330 			len = filesize - pos;
331 
332 		if (fseeko(f, pos, SEEK_SET) == -1) {
333 			rc = errno;
334 			break;
335 		}
336 		for (j = 0; j < len; j++) {
337 			int c = fgetc(f);
338 			if (c == EOF) {
339 				if (feof(f))
340 					rc = EIO; /* unexpected EOF */
341 				else
342 					rc = errno;
343 				goto done;
344 			}
345 
346 			hash = diff_atom_hash_update(hash, (unsigned char)c);
347 		}
348 		*atom = (struct diff_atom){
349 			.root = d,
350 			.pos = pos,
351 			.at = NULL,	/* atom data is not memory-mapped */
352 			.len = len,
353 			.hash = hash,
354 		};
355 	}
356 done:
357 	if (rc)
358 		ARRAYLIST_FREE(d->atoms);
359 
360 	return rc;
361 }
362 
363 static int
364 atomize_file_mmap(struct diff_data *d, unsigned char *p,
365     off_t filesize, int nlines, off_t *line_offsets)
366 {
367 	int i, rc = DIFF_RC_OK;
368 
369 	ARRAYLIST_INIT(d->atoms, nlines);
370 
371 	for (i = 0; i < nlines; i++) {
372 		struct diff_atom *atom;
373 		off_t len, pos = line_offsets[i];
374 		unsigned int hash = 0;
375 		int j;
376 
377 		ARRAYLIST_ADD(atom, d->atoms);
378 		if (atom == NULL) {
379 			rc = errno;
380 			break;
381 		}
382 
383 		if (i < nlines - 1)
384 			len = line_offsets[i + 1] - pos;
385 		else
386 			len = filesize - pos;
387 
388 		for (j = 0; j < len; j++)
389 			hash = diff_atom_hash_update(hash, p[pos + j]);
390 
391 		*atom = (struct diff_atom){
392 			.root = d,
393 			.pos = pos,
394 			.at = &p[pos],
395 			.len = len,
396 			.hash = hash,
397 		};
398 	}
399 
400 	if (rc)
401 		ARRAYLIST_FREE(d->atoms);
402 
403 	return rc;
404 }
405 
406 /* Implements diff_atomize_func_t */
407 static int
408 blame_atomize_file(void *arg, struct diff_data *d)
409 {
410 	struct got_blame *blame = arg;
411 
412 	if (d->f == blame->f1) {
413 		if (blame->map1)
414 			return atomize_file_mmap(d, blame->map1,
415 			    blame->size1, blame->nlines1,
416 			    blame->line_offsets1);
417 		else
418 			return atomize_file(d, blame->f1, blame->size1,
419 			    blame->nlines1, blame->line_offsets1);
420 	} else if (d->f == blame->f2) {
421 		if (d->atoms.len > 0) {
422 			/* Re-use data from previous commit. */
423 			return DIFF_RC_OK;
424 		}
425 		if (blame->map2)
426 			return atomize_file_mmap(d, blame->map2,
427 			    blame->size2, blame->nlines2,
428 			    blame->line_offsets2);
429 		else
430 			return atomize_file(d, blame->f2, blame->size2,
431 			    blame->nlines2, blame->line_offsets2);
432 	}
433 
434 	return DIFF_RC_OK;
435 }
436 
437 static const struct got_error *
438 close_file2_and_reuse_file1(struct got_blame *blame)
439 {
440 	struct diff_data *d;
441 
442 	free(blame->line_offsets2);
443 	blame->line_offsets2 = blame->line_offsets1;
444 	blame->line_offsets1 = NULL;
445 
446 	free(blame->linemap2);
447 	blame->linemap2 = blame->linemap1;
448 	blame->linemap1 = NULL;
449 
450 	if (blame->map2) {
451 		if (munmap(blame->map2, blame->size2) == -1)
452 			return got_error_from_errno("munmap");
453 		blame->map2 = blame->map1;
454 		blame->map2 = NULL;
455 
456 	}
457 	blame->size2 = blame->size1;
458 	blame->size1 = 0;
459 
460 	if (fclose(blame->f2) == EOF)
461 		return got_error_from_errno("fclose");
462 	blame->f2 = blame->f1;
463 	blame->f1 = NULL;
464 
465 	blame->nlines2 = blame->nlines1;
466 	blame->nlines1 = 0;
467 
468 	free(blame->line_offsets2);
469 	blame->line_offsets2 = blame->line_offsets1;
470 	blame->line_offsets2 = NULL;
471 
472 	diff_data_free(blame->data2); /* does not free pointer itself */
473 	memset(blame->data2, 0, sizeof(*blame->data2));
474 	d = blame->data2;
475 	blame->data2 = blame->data1;
476 	blame->data1 = d;
477 
478 	return NULL;
479 }
480 
481 static const struct got_error *
482 blame_open(struct got_blame **blamep, const char *path,
483     struct got_object_id *start_commit_id, struct got_repository *repo,
484     const struct got_error *(*cb)(void *, int, int, struct got_object_id *),
485     void *arg, got_cancel_cb cancel_cb, void *cancel_arg)
486 {
487 	const struct got_error *err = NULL;
488 	struct got_object_id *obj_id = NULL;
489 	struct got_blob_object *blob = NULL;
490 	struct got_blame *blame = NULL;
491 	struct got_object_id *id = NULL;
492 	int lineno;
493 	struct got_commit_graph *graph = NULL;
494 
495 	*blamep = NULL;
496 
497 	err = got_object_id_by_path(&obj_id, repo, start_commit_id, path);
498 	if (err)
499 		goto done;
500 
501 	err = got_object_open_as_blob(&blob, repo, obj_id, 8192);
502 	if (err)
503 		goto done;
504 
505 	blame = calloc(1, sizeof(*blame));
506 	if (blame == NULL) {
507 		err = got_error_from_errno("calloc");
508 		goto done;
509 	}
510 
511 	blame->data1 = calloc(1, sizeof(*blame->data1));
512 	if (blame->data1 == NULL) {
513 		err = got_error_from_errno("calloc");
514 		goto done;
515 	}
516 	blame->data2 = calloc(1, sizeof(*blame->data2));
517 	if (blame->data2 == NULL) {
518 		err = got_error_from_errno("calloc");
519 		goto done;
520 	}
521 
522 	blame->f2 = got_opentemp();
523 	if (blame->f2 == NULL) {
524 		err = got_error_from_errno("got_opentemp");
525 		goto done;
526 	}
527 	err = got_diff_get_config(&blame->cfg, GOT_DIFF_ALGORITHM_PATIENCE,
528 	    blame_atomize_file, blame);
529 	if (err)
530 		goto done;
531 
532 	err = blame_prepare_file(blame->f2, &blame->map2, &blame->size2,
533 	    &blame->nlines2, &blame->line_offsets2, blame->data2,
534 	    blame->cfg, blob);
535 	blame->nlines = blame->nlines2;
536 	if (err || blame->nlines == 0)
537 		goto done;
538 
539 	got_object_blob_close(blob);
540 	blob = NULL;
541 
542 	/* Don't include \n at EOF in the blame line count. */
543 	if (blame->line_offsets2[blame->nlines - 1] == blame->size2)
544 		blame->nlines--;
545 
546 	blame->lines = calloc(blame->nlines, sizeof(*blame->lines));
547 	if (blame->lines == NULL) {
548 		err = got_error_from_errno("calloc");
549 		goto done;
550 	}
551 
552 	blame->linemap2 = calloc(blame->nlines2, sizeof(*blame->linemap2));
553 	if (blame->linemap2 == NULL) {
554 		err = got_error_from_errno("calloc");
555 		goto done;
556 	}
557 	for (lineno = 0; lineno < blame->nlines2; lineno++)
558 		blame->linemap2[lineno] = lineno;
559 
560 	err = got_commit_graph_open(&graph, path, 1);
561 	if (err)
562 		goto done;
563 
564 	err = got_commit_graph_iter_start(graph, start_commit_id, repo,
565 	    cancel_cb, cancel_arg);
566 	if (err)
567 		goto done;
568 	for (;;) {
569 		struct got_object_id *next_id;
570 		err = got_commit_graph_iter_next(&next_id, graph, repo,
571 		    cancel_cb, cancel_arg);
572 		if (err) {
573 			if (err->code == GOT_ERR_ITER_COMPLETED) {
574 				err = NULL;
575 				break;
576 			}
577 			goto done;
578 		}
579 		if (next_id) {
580 			id = next_id;
581 			err = blame_commit(blame, id, path, repo, cb, arg);
582 			if (err) {
583 				if (err->code == GOT_ERR_ITER_COMPLETED)
584 					err = NULL;
585 				goto done;
586 			}
587 			if (blame->nannotated == blame->nlines)
588 				break;
589 
590 			err = close_file2_and_reuse_file1(blame);
591 			if (err)
592 				goto done;
593 		}
594 	}
595 
596 	if (id && blame->nannotated < blame->nlines) {
597 		/* Annotate remaining non-annotated lines with last commit. */
598 		for (lineno = 0; lineno < blame->nlines; lineno++) {
599 			err = annotate_line(blame, lineno, id, cb, arg);
600 			if (err)
601 				goto done;
602 		}
603 	}
604 
605 done:
606 	if (graph)
607 		got_commit_graph_close(graph);
608 	free(obj_id);
609 	if (blob)
610 		got_object_blob_close(blob);
611 	if (err) {
612 		if (blame)
613 			blame_close(blame);
614 	} else
615 		*blamep = blame;
616 
617 	return err;
618 }
619 
620 const struct got_error *
621 got_blame(const char *path, struct got_object_id *commit_id,
622     struct got_repository *repo,
623     const struct got_error *(*cb)(void *, int, int, struct got_object_id *),
624     void *arg, got_cancel_cb cancel_cb, void* cancel_arg)
625 {
626 	const struct got_error *err = NULL, *close_err = NULL;
627 	struct got_blame *blame;
628 	char *abspath;
629 
630 	if (asprintf(&abspath, "%s%s", path[0] == '/' ? "" : "/", path) == -1)
631 		return got_error_from_errno2("asprintf", path);
632 
633 	err = blame_open(&blame, abspath, commit_id, repo, cb, arg,
634 	    cancel_cb, cancel_arg);
635 	free(abspath);
636 	if (blame)
637 		close_err = blame_close(blame);
638 	return err ? err : close_err;
639 }