1 /* Generic infrastructure to implement various diff algorithms (implementation). */
2 /*
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 
19 #include <sys/queue.h>
20 #include <ctype.h>
21 #include <errno.h>
22 #include <stdint.h>
23 #include <stdlib.h>
24 #include <stdbool.h>
25 #include <stdio.h>
26 #include <string.h>
27 #include <limits.h>
28 #include <unistd.h>
29 
30 #include <assert.h>
31 
32 #include <arraylist.h>
33 #include <diff_main.h>
34 
35 #include "diff_internal.h"
36 #include "diff_debug.h"
37 
38 static int
39 read_at(FILE *f, off_t at_pos, unsigned char *buf, size_t len)
40 {
41 	int r;
42 	if (fseeko(f, at_pos, SEEK_SET) == -1)
43 		return errno;
44 	r = fread(buf, sizeof(char), len, f);
45 	if ((r == 0 || r < len) && ferror(f))
46 		return errno;
47 	if (r != len)
48 		return EIO;
49 	return 0;
50 }
51 
52 static int
53 buf_cmp(const unsigned char *left, size_t left_len,
54 	const unsigned char *right, size_t right_len,
55 	bool ignore_whitespace)
56 {
57 	int cmp;
58 
59 	if (ignore_whitespace) {
60 		int il = 0, ir = 0;
61 		while (il < left_len && ir < right_len) {
62 			unsigned char cl = left[il];
63 			unsigned char cr = right[ir];
64 
65 			if (isspace(cl) && il < left_len) {
66 				il++;
67 				continue;
68 			}
69 			if (isspace(cr) && ir < right_len) {
70 				ir++;
71 				continue;
72 			}
73 
74 			if (cl > cr)
75 				return 1;
76 			if (cr > cl)
77 				return -1;
78 			il++;
79 			ir++;
80 		}
81 		while (il < left_len) {
82 			unsigned char cl = left[il++];
83 			if (!isspace(cl))
84 				return 1;
85 		}
86 		while (ir < right_len) {
87 			unsigned char cr = right[ir++];
88 			if (!isspace(cr))
89 				return -1;
90 		}
91 
92 		return 0;
93 	}
94 
95 	cmp = memcmp(left, right, MIN(left_len, right_len));
96 	if (cmp)
97 		return cmp;
98 	if (left_len == right_len)
99 		return 0;
100 	return (left_len > right_len) ? 1 : -1;
101 }
102 
103 int
104 diff_atom_cmp(int *cmp,
105 	      const struct diff_atom *left,
106 	      const struct diff_atom *right)
107 {
108 	off_t remain_left, remain_right;
109 	int flags = (left->d->root->diff_flags | right->d->root->diff_flags);
110 	bool ignore_whitespace = (flags & DIFF_FLAG_IGNORE_WHITESPACE);
111 
112 	if (!ignore_whitespace) {
113 		if (!left->len && !right->len) {
114 			*cmp = 0;
115 			return 0;
116 		}
117 		if (!right->len) {
118 			*cmp = 1;
119 			return 0;
120 		}
121 		if (!left->len) {
122 			*cmp = -1;
123 			return 0;
124 		}
125 	}
126 
127 	if (left->at != NULL && right->at != NULL) {
128 		*cmp = buf_cmp(left->at, left->len, right->at, right->len,
129 		    ignore_whitespace);
130 		return 0;
131 	}
132 
133 	remain_left = left->len;
134 	remain_right = right->len;
135 	while (remain_left > 0 || remain_right > 0) {
136 		const size_t chunksz = 8192;
137 		unsigned char buf_left[chunksz], buf_right[chunksz];
138 		const uint8_t *p_left, *p_right;
139 		off_t n_left, n_right;
140 		ssize_t r;
141 
142 		if (!remain_right) {
143 			*cmp = 1;
144 			return 0;
145 		}
146 		if (!remain_left) {
147 			*cmp = -1;
148 			return 0;
149 		}
150 
151 		n_left = MIN(chunksz, remain_left);
152 		n_right = MIN(chunksz, remain_right);
153 
154 		if (left->at == NULL) {
155 			r = read_at(left->d->root->f,
156 				    left->pos + (left->len - remain_left),
157 				    buf_left, n_left);
158 			if (r) {
159 				*cmp = 0;
160 				return r;
161 			}
162 			p_left = buf_left;
163 		} else {
164 			p_left = left->at + (left->len - remain_left);
165 		}
166 
167 		if (right->at == NULL) {
168 			r = read_at(right->d->root->f,
169 				    right->pos + (right->len - remain_right),
170 				    buf_right, n_right);
171 			if (r) {
172 				*cmp = 0;
173 				return r;
174 			}
175 			p_right = buf_right;
176 		} else {
177 			p_right = right->at + (right->len - remain_right);
178 		}
179 
180 		r = buf_cmp(p_left, n_left, p_right, n_right,
181 		    ignore_whitespace);
182 		if (r) {
183 			*cmp = r;
184 			return 0;
185 		}
186 
187 		remain_left -= n_left;
188 		remain_right -= n_right;
189 	}
190 
191 	*cmp = 0;
192 	return 0;
193 }
194 
195 int
196 diff_atom_same(bool *same,
197 	       const struct diff_atom *left,
198 	       const struct diff_atom *right)
199 {
200 	int cmp;
201 	int r = diff_atom_cmp(&cmp, left, right);
202 	if (r) {
203 		*same = true;
204 		return r;
205 	}
206 	*same = (cmp == 0);
207 	return 0;
208 }
209 
210 static struct diff_chunk *
211 diff_state_add_solved_chunk(struct diff_state *state,
212 			    const struct diff_chunk *chunk)
213 {
214 	diff_chunk_arraylist_t *result;
215 	struct diff_chunk *new_chunk;
216 	enum diff_chunk_type last_t;
217 	enum diff_chunk_type new_t;
218 
219 	/* Append to solved chunks; make sure that adjacent chunks of same type are combined, and that a minus chunk
220 	 * never directly follows a plus chunk. */
221 	result = &state->result->chunks;
222 
223 	last_t = diff_chunk_type(&result->head[result->len - 1]);
224 	new_t = diff_chunk_type(chunk);
225 
226 	debug("Add %s chunk:\n", chunk->solved ? "solved" : "UNSOLVED");
227 	debug("L\n");
228 	debug_dump_atoms(&state->left, chunk->left_start, chunk->left_count);
229 	debug("R\n");
230 	debug_dump_atoms(&state->right, chunk->right_start, chunk->right_count);
231 
232 	if (new_t == last_t) {
233 		new_chunk = &result->head[result->len - 1];
234 		new_chunk->left_count += chunk->left_count;
235 		new_chunk->right_count += chunk->right_count;
236 		debug("  - added chunk touches previous one of same type, joined:\n");
237 		debug("L\n");
238 		debug_dump_atoms(&state->left, new_chunk->left_start, new_chunk->left_count);
239 		debug("R\n");
240 		debug_dump_atoms(&state->right, new_chunk->right_start, new_chunk->right_count);
241 	} else if (last_t == CHUNK_PLUS && new_t == CHUNK_MINUS) {
242 		enum diff_chunk_type prev_last_t =
243 			result->len > 1 ?
244 				diff_chunk_type(&result->head[result->len - 2])
245 				: CHUNK_EMPTY;
246 		/* If a minus-chunk follows a plus-chunk, place it above the plus-chunk->
247 		 * Is the one before that also a minus? combine. */
248 		if (prev_last_t == CHUNK_MINUS) {
249 			new_chunk = &result->head[result->len - 2];
250 			new_chunk->left_count += chunk->left_count;
251 			new_chunk->right_count += chunk->right_count;
252 
253 			debug("  - added minus-chunk follows plus-chunk,"
254 			      " put before that plus-chunk and joined"
255 			      " with preceding minus-chunk:\n");
256 			debug("L\n");
257 			debug_dump_atoms(&state->left, new_chunk->left_start, new_chunk->left_count);
258 			debug("R\n");
259 			debug_dump_atoms(&state->right, new_chunk->right_start, new_chunk->right_count);
260 		} else {
261 			ARRAYLIST_INSERT(new_chunk, *result, result->len - 1);
262 			if (!new_chunk)
263 				return NULL;
264 			*new_chunk = *chunk;
265 
266 			debug("  - added minus-chunk follows plus-chunk,"
267 			      " put before that plus-chunk\n");
268 		}
269 	} else {
270 		ARRAYLIST_ADD(new_chunk, *result);
271 		if (!new_chunk)
272 			return NULL;
273 		*new_chunk = *chunk;
274 	}
275 	return new_chunk;
276 }
277 
278 /* Even if a left or right side is empty, diff output may need to know the
279  * position in that file.
280  * So left_start or right_start must never be NULL -- pass left_count or
281  * right_count as zero to indicate staying at that position without consuming
282  * any lines. */
283 struct diff_chunk *
284 diff_state_add_chunk(struct diff_state *state, bool solved,
285 		     struct diff_atom *left_start, unsigned int left_count,
286 		     struct diff_atom *right_start, unsigned int right_count)
287 {
288 	diff_chunk_arraylist_t *result = NULL;
289 	struct diff_chunk *new_chunk;
290 	struct diff_chunk chunk = {
291 		.solved = solved,
292 		.left_start = left_start,
293 		.left_count = left_count,
294 		.right_start = right_start,
295 		.right_count = right_count,
296 	};
297 
298 	if (!solved || state->temp_result.len) {
299 		/* Append to temp_result */
300 		debug("append to temp_result\n");
301 		result = &state->temp_result;
302 	} else if (!state->result->chunks.len) {
303 		/* Append to final result */
304 		debug("first chunk\n");
305 		result = &state->result->chunks;
306 	}
307 	if (result) {
308 		ARRAYLIST_ADD(new_chunk, *result);
309 		if (!new_chunk)
310 			return NULL;
311 		*new_chunk = chunk;
312 		return new_chunk;
313 	}
314 
315 	return diff_state_add_solved_chunk(state, &chunk);
316 }
317 
318 void
319 diff_data_init_root(struct diff_data *d, FILE *f, const uint8_t *data,
320     unsigned long long len, int diff_flags)
321 {
322 	*d = (struct diff_data){
323 		.f = f,
324 		.pos = 0,
325 		.data = data,
326 		.len = len,
327 		.root = d,
328 		.diff_flags = diff_flags,
329 	};
330 }
331 
332 void
333 diff_data_init_subsection(struct diff_data *d, struct diff_data *parent,
334 			  struct diff_atom *from_atom, unsigned int atoms_count)
335 {
336 	struct diff_atom *last_atom;
337 
338 	if (atoms_count == 0) {
339 		*d = (struct diff_data){
340 			.f = NULL,
341 			.pos = 0,
342 			.data = "",
343 			.len = 0,
344 			.root = parent->root,
345 			.atoms.head = NULL,
346 			.atoms.len = atoms_count,
347 		};
348 
349 		return;
350 	}
351 
352 	last_atom = from_atom + atoms_count - 1;
353 	*d = (struct diff_data){
354 		.f = NULL,
355 		.pos = from_atom->pos,
356 		.data = from_atom->at,
357 		.len = (last_atom->pos + last_atom->len) - from_atom->pos,
358 		.root = parent->root,
359 		.atoms.head = from_atom,
360 		.atoms.len = atoms_count,
361 	};
362 
363 	debug("subsection:\n");
364 	debug_dump(d);
365 }
366 
367 void
368 diff_data_free(struct diff_data *diff_data)
369 {
370 	if (!diff_data)
371 		return;
372 	if (diff_data->atoms.allocated)
373 		ARRAYLIST_FREE(diff_data->atoms);
374 }
375 
376 int
377 diff_algo_none(const struct diff_algo_config *algo_config,
378 	       struct diff_state *state)
379 {
380 	debug("\n** %s\n", __func__);
381 	debug("left:\n");
382 	debug_dump(&state->left);
383 	debug("right:\n");
384 	debug_dump(&state->right);
385 	debug_dump_myers_graph(&state->left, &state->right, NULL, NULL, 0, NULL,
386 			       0);
387 
388 	/* Add a chunk of equal lines, if any */
389 	struct diff_atom *l = state->left.atoms.head;
390 	unsigned int l_len = state->left.atoms.len;
391 	struct diff_atom *r = state->right.atoms.head;
392 	unsigned int r_len = state->right.atoms.len;
393 	unsigned int equal_atoms_start = 0;
394 	unsigned int equal_atoms_end = 0;
395 	unsigned int l_idx = 0;
396 	unsigned int r_idx = 0;
397 
398 	while (equal_atoms_start < l_len
399 	       && equal_atoms_start < r_len) {
400 		int err;
401 		bool same;
402 		err = diff_atom_same(&same, &l[equal_atoms_start],
403 				   &r[equal_atoms_start]);
404 		if (err)
405 			return err;
406 		if (!same)
407 			break;
408 		equal_atoms_start++;
409 	}
410 	while (equal_atoms_end < (l_len - equal_atoms_start)
411 	       && equal_atoms_end < (r_len - equal_atoms_start)) {
412 		int err;
413 		bool same;
414 		err = diff_atom_same(&same, &l[l_len - 1 - equal_atoms_end],
415 				   &r[r_len - 1 - equal_atoms_end]);
416 		if (err)
417 			return err;
418 		if (!same)
419 			break;
420 		equal_atoms_end++;
421 	}
422 
423 	/* Add a chunk of equal lines at the start */
424 	if (equal_atoms_start) {
425 		if (!diff_state_add_chunk(state, true,
426 					  &l[0],
427 					  equal_atoms_start,
428 					  &r[0],
429 					  equal_atoms_start))
430 			return ENOMEM;
431 		l_idx += equal_atoms_start;
432 		r_idx += equal_atoms_start;
433 	}
434 
435 	/* Add a "minus" chunk with all lines from the left. */
436 	if (equal_atoms_start + equal_atoms_end < l_len) {
437 		unsigned int add_len = l_len - equal_atoms_start - equal_atoms_end;
438 		if (!diff_state_add_chunk(state, true,
439 					  &l[l_idx], add_len,
440 					  &r[r_idx], 0))
441 			return ENOMEM;
442 		l_idx += add_len;
443 	}
444 
445 	/* Add a "plus" chunk with all lines from the right. */
446 	if (equal_atoms_start + equal_atoms_end < r_len) {
447 		unsigned int add_len = r_len - equal_atoms_start - equal_atoms_end;
448 		if (!diff_state_add_chunk(state, true,
449 					  &l[equal_atoms_start], 0,
450 					  &r[equal_atoms_start], add_len))
451 			return ENOMEM;
452 		r_idx += add_len;
453 	}
454 
455 	/* Add a chunk of equal lines at the end */
456 	if (equal_atoms_end) {
457 		if (!diff_state_add_chunk(state, true,
458 					  &l[l_idx], equal_atoms_end,
459 					  &r[r_idx], equal_atoms_end))
460 			return ENOMEM;
461 	}
462 
463 	return DIFF_RC_OK;
464 }
465 
466 int
467 diff_run_algo(const struct diff_algo_config *algo_config,
468 	      struct diff_state *state)
469 {
470 	int rc;
471 	ARRAYLIST_FREE(state->temp_result);
472 
473 	if (!algo_config || !algo_config->impl
474 	    || !state->recursion_depth_left) {
475 		debug("MAX RECURSION REACHED, just dumping diff chunks\n");
476 		return diff_algo_none(algo_config, state);
477 	}
478 
479 	ARRAYLIST_INIT(state->temp_result, DIFF_RESULT_ALLOC_BLOCKSIZE);
480 	rc = algo_config->impl(algo_config, state);
481 	switch (rc) {
482 	case DIFF_RC_USE_DIFF_ALGO_FALLBACK:
483 		debug("Got DIFF_RC_USE_DIFF_ALGO_FALLBACK (%p)\n",
484 		      algo_config->fallback_algo);
485 		rc = diff_run_algo(algo_config->fallback_algo, state);
486 		goto return_rc;
487 
488 	case DIFF_RC_OK:
489 		/* continue below */
490 		break;
491 
492 	default:
493 		/* some error happened */
494 		goto return_rc;
495 	}
496 
497 	/* Pick up any diff chunks that are still unsolved and feed to
498 	 * inner_algo.  inner_algo will solve unsolved chunks and append to
499 	 * result, and subsequent solved chunks on this level are then appended
500 	 * to result afterwards. */
501 	int i;
502 	for (i = 0; i < state->temp_result.len; i++) {
503 		struct diff_chunk *c = &state->temp_result.head[i];
504 		if (c->solved) {
505 			diff_state_add_solved_chunk(state, c);
506 			continue;
507 		}
508 
509 		/* c is an unsolved chunk, feed to inner_algo */
510 		struct diff_state inner_state = {
511 			.result = state->result,
512 			.recursion_depth_left = state->recursion_depth_left - 1,
513 		};
514 		diff_data_init_subsection(&inner_state.left, &state->left,
515 					  c->left_start, c->left_count);
516 		diff_data_init_subsection(&inner_state.right, &state->right,
517 					  c->right_start, c->right_count);
518 
519 		rc = diff_run_algo(algo_config->inner_algo, &inner_state);
520 
521 		if (rc != DIFF_RC_OK)
522 			goto return_rc;
523 	}
524 
525 	rc = DIFF_RC_OK;
526 return_rc:
527 	ARRAYLIST_FREE(state->temp_result);
528 	return rc;
529 }
530 
531 struct diff_result *
532 diff_main(const struct diff_config *config,
533 	  FILE *left_f, const uint8_t *left_data, off_t left_len,
534 	  FILE *right_f, const uint8_t *right_data, off_t right_len,
535 	  int diff_flags)
536 {
537 	struct diff_result *result = malloc(sizeof(struct diff_result));
538 	if (!result)
539 		return NULL;
540 
541 	*result = (struct diff_result){};
542 	diff_data_init_root(&result->left, left_f, left_data, left_len,
543 	    diff_flags);
544 	diff_data_init_root(&result->right, right_f, right_data, right_len,
545 	    diff_flags);
546 
547 	if (!config->atomize_func) {
548 		result->rc = EINVAL;
549 		return result;
550 	}
551 
552 	result->rc = config->atomize_func(config->atomize_func_data,
553 					  &result->left, &result->right);
554 	if (result->rc != DIFF_RC_OK)
555 		return result;
556 
557 	struct diff_state state = {
558 		.result = result,
559 		.recursion_depth_left = config->max_recursion_depth ? : 32,
560 	};
561 	diff_data_init_subsection(&state.left, &result->left,
562 				  result->left.atoms.head,
563 				  result->left.atoms.len);
564 	diff_data_init_subsection(&state.right, &result->right,
565 				  result->right.atoms.head,
566 				  result->right.atoms.len);
567 
568 	result->rc = diff_run_algo(config->algo, &state);
569 
570 	return result;
571 }
572 
573 void
574 diff_result_free(struct diff_result *result)
575 {
576 	if (!result)
577 		return;
578 	diff_data_free(&result->left);
579 	diff_data_free(&result->right);
580 	ARRAYLIST_FREE(result->chunks);
581 	free(result);
582 }