Blame


1 3b0f3d61 2020-01-22 neels /* Generic infrastructure to implement various diff algorithms (implementation). */
2 3b0f3d61 2020-01-22 neels /*
3 3b0f3d61 2020-01-22 neels * Copyright (c) 2020 Neels Hofmeyr <neels@hofmeyr.de>
4 3b0f3d61 2020-01-22 neels *
5 3b0f3d61 2020-01-22 neels * Permission to use, copy, modify, and distribute this software for any
6 3b0f3d61 2020-01-22 neels * purpose with or without fee is hereby granted, provided that the above
7 3b0f3d61 2020-01-22 neels * copyright notice and this permission notice appear in all copies.
8 3b0f3d61 2020-01-22 neels *
9 3b0f3d61 2020-01-22 neels * THE SOFTWARE IS PROVIDED "AS IS" AND THE AUTHOR DISCLAIMS ALL WARRANTIES
10 3b0f3d61 2020-01-22 neels * WITH REGARD TO THIS SOFTWARE INCLUDING ALL IMPLIED WARRANTIES OF
11 3b0f3d61 2020-01-22 neels * MERCHANTABILITY AND FITNESS. IN NO EVENT SHALL THE AUTHOR BE LIABLE FOR
12 3b0f3d61 2020-01-22 neels * ANY SPECIAL, DIRECT, INDIRECT, OR CONSEQUENTIAL DAMAGES OR ANY DAMAGES
13 3b0f3d61 2020-01-22 neels * WHATSOEVER RESULTING FROM LOSS OF USE, DATA OR PROFITS, WHETHER IN AN
14 3b0f3d61 2020-01-22 neels * ACTION OF CONTRACT, NEGLIGENCE OR OTHER TORTIOUS ACTION, ARISING OUT OF
15 3b0f3d61 2020-01-22 neels * OR IN CONNECTION WITH THE USE OR PERFORMANCE OF THIS SOFTWARE.
16 3b0f3d61 2020-01-22 neels */
17 3b0f3d61 2020-01-22 neels
18 3b0f3d61 2020-01-22 neels
19 3b0f3d61 2020-01-22 neels #include <sys/queue.h>
20 732e8ee0 2020-09-20 stsp #include <ctype.h>
21 e10a628a 2020-09-16 stsp #include <errno.h>
22 3b0f3d61 2020-01-22 neels #include <stdint.h>
23 3b0f3d61 2020-01-22 neels #include <stdlib.h>
24 3b0f3d61 2020-01-22 neels #include <stdbool.h>
25 3b0f3d61 2020-01-22 neels #include <stdio.h>
26 3b0f3d61 2020-01-22 neels #include <string.h>
27 3b0f3d61 2020-01-22 neels #include <limits.h>
28 c6eecea3 2020-07-26 stsp #include <unistd.h>
29 3b0f3d61 2020-01-22 neels
30 3b0f3d61 2020-01-22 neels #include <assert.h>
31 3b0f3d61 2020-01-22 neels
32 1dfba055 2020-10-07 stsp #include <arraylist.h>
33 1dfba055 2020-10-07 stsp #include <diff_main.h>
34 3b0f3d61 2020-01-22 neels
35 85ab4559 2020-09-22 stsp #include "diff_internal.h"
36 2a1b94d0 2020-09-26 stsp #include "diff_debug.h"
37 c6eecea3 2020-07-26 stsp
38 b3fb4686 2020-09-20 neels static int
39 7a54ad3a 2020-09-20 stsp read_at(FILE *f, off_t at_pos, unsigned char *buf, size_t len)
40 c6eecea3 2020-07-26 stsp {
41 b3fb4686 2020-09-20 neels int r;
42 7a54ad3a 2020-09-20 stsp if (fseeko(f, at_pos, SEEK_SET) == -1)
43 b3fb4686 2020-09-20 neels return errno;
44 7a54ad3a 2020-09-20 stsp r = fread(buf, sizeof(char), len, f);
45 7a54ad3a 2020-09-20 stsp if ((r == 0 || r < len) && ferror(f))
46 b3fb4686 2020-09-20 neels return errno;
47 b3fb4686 2020-09-20 neels if (r != len)
48 b3fb4686 2020-09-20 neels return EIO;
49 b3fb4686 2020-09-20 neels return 0;
50 b3fb4686 2020-09-20 neels }
51 b3fb4686 2020-09-20 neels
52 b3fb4686 2020-09-20 neels static int
53 b3fb4686 2020-09-20 neels buf_cmp(const unsigned char *left, size_t left_len,
54 732e8ee0 2020-09-20 stsp const unsigned char *right, size_t right_len,
55 732e8ee0 2020-09-20 stsp bool ignore_whitespace)
56 b3fb4686 2020-09-20 neels {
57 732e8ee0 2020-09-20 stsp int cmp;
58 732e8ee0 2020-09-20 stsp
59 732e8ee0 2020-09-20 stsp if (ignore_whitespace) {
60 732e8ee0 2020-09-20 stsp int il = 0, ir = 0;
61 732e8ee0 2020-09-20 stsp while (il < left_len && ir < right_len) {
62 732e8ee0 2020-09-20 stsp unsigned char cl = left[il];
63 732e8ee0 2020-09-20 stsp unsigned char cr = right[ir];
64 732e8ee0 2020-09-20 stsp
65 732e8ee0 2020-09-20 stsp if (isspace(cl) && il < left_len) {
66 732e8ee0 2020-09-20 stsp il++;
67 732e8ee0 2020-09-20 stsp continue;
68 732e8ee0 2020-09-20 stsp }
69 732e8ee0 2020-09-20 stsp if (isspace(cr) && ir < right_len) {
70 732e8ee0 2020-09-20 stsp ir++;
71 732e8ee0 2020-09-20 stsp continue;
72 732e8ee0 2020-09-20 stsp }
73 732e8ee0 2020-09-20 stsp
74 732e8ee0 2020-09-20 stsp if (cl > cr)
75 732e8ee0 2020-09-20 stsp return 1;
76 732e8ee0 2020-09-20 stsp if (cr > cl)
77 732e8ee0 2020-09-20 stsp return -1;
78 732e8ee0 2020-09-20 stsp il++;
79 732e8ee0 2020-09-20 stsp ir++;
80 732e8ee0 2020-09-20 stsp }
81 732e8ee0 2020-09-20 stsp while (il < left_len) {
82 732e8ee0 2020-09-20 stsp unsigned char cl = left[il++];
83 732e8ee0 2020-09-20 stsp if (!isspace(cl))
84 732e8ee0 2020-09-20 stsp return 1;
85 732e8ee0 2020-09-20 stsp }
86 732e8ee0 2020-09-20 stsp while (ir < right_len) {
87 732e8ee0 2020-09-20 stsp unsigned char cr = right[ir++];
88 732e8ee0 2020-09-20 stsp if (!isspace(cr))
89 732e8ee0 2020-09-20 stsp return -1;
90 732e8ee0 2020-09-20 stsp }
91 732e8ee0 2020-09-20 stsp
92 732e8ee0 2020-09-20 stsp return 0;
93 732e8ee0 2020-09-20 stsp }
94 732e8ee0 2020-09-20 stsp
95 732e8ee0 2020-09-20 stsp cmp = memcmp(left, right, MIN(left_len, right_len));
96 b3fb4686 2020-09-20 neels if (cmp)
97 b3fb4686 2020-09-20 neels return cmp;
98 b3fb4686 2020-09-20 neels if (left_len == right_len)
99 b3fb4686 2020-09-20 neels return 0;
100 b3fb4686 2020-09-20 neels return (left_len > right_len) ? 1 : -1;
101 b3fb4686 2020-09-20 neels }
102 b3fb4686 2020-09-20 neels
103 b3fb4686 2020-09-20 neels int
104 b3fb4686 2020-09-20 neels diff_atom_cmp(int *cmp,
105 b3fb4686 2020-09-20 neels const struct diff_atom *left,
106 b3fb4686 2020-09-20 neels const struct diff_atom *right)
107 b3fb4686 2020-09-20 neels {
108 c6eecea3 2020-07-26 stsp off_t remain_left, remain_right;
109 00d5652b 2020-09-22 stsp int flags = (left->d->root->diff_flags | right->d->root->diff_flags);
110 00d5652b 2020-09-22 stsp bool ignore_whitespace = (flags & DIFF_FLAG_IGNORE_WHITESPACE);
111 c6eecea3 2020-07-26 stsp
112 732e8ee0 2020-09-20 stsp if (!ignore_whitespace) {
113 732e8ee0 2020-09-20 stsp if (!left->len && !right->len) {
114 732e8ee0 2020-09-20 stsp *cmp = 0;
115 732e8ee0 2020-09-20 stsp return 0;
116 732e8ee0 2020-09-20 stsp }
117 732e8ee0 2020-09-20 stsp if (!right->len) {
118 732e8ee0 2020-09-20 stsp *cmp = 1;
119 732e8ee0 2020-09-20 stsp return 0;
120 732e8ee0 2020-09-20 stsp }
121 732e8ee0 2020-09-20 stsp if (!left->len) {
122 732e8ee0 2020-09-20 stsp *cmp = -1;
123 732e8ee0 2020-09-20 stsp return 0;
124 732e8ee0 2020-09-20 stsp }
125 b3fb4686 2020-09-20 neels }
126 c6eecea3 2020-07-26 stsp
127 b3fb4686 2020-09-20 neels if (left->at != NULL && right->at != NULL) {
128 732e8ee0 2020-09-20 stsp *cmp = buf_cmp(left->at, left->len, right->at, right->len,
129 732e8ee0 2020-09-20 stsp ignore_whitespace);
130 b3fb4686 2020-09-20 neels return 0;
131 b3fb4686 2020-09-20 neels }
132 c6eecea3 2020-07-26 stsp
133 c6eecea3 2020-07-26 stsp remain_left = left->len;
134 c6eecea3 2020-07-26 stsp remain_right = right->len;
135 c6eecea3 2020-07-26 stsp while (remain_left > 0 || remain_right > 0) {
136 c6eecea3 2020-07-26 stsp const size_t chunksz = 8192;
137 c6eecea3 2020-07-26 stsp unsigned char buf_left[chunksz], buf_right[chunksz];
138 c6eecea3 2020-07-26 stsp const uint8_t *p_left, *p_right;
139 c6eecea3 2020-07-26 stsp off_t n_left, n_right;
140 c6eecea3 2020-07-26 stsp ssize_t r;
141 c6eecea3 2020-07-26 stsp
142 b3fb4686 2020-09-20 neels if (!remain_right) {
143 b3fb4686 2020-09-20 neels *cmp = 1;
144 b3fb4686 2020-09-20 neels return 0;
145 b3fb4686 2020-09-20 neels }
146 b3fb4686 2020-09-20 neels if (!remain_left) {
147 b3fb4686 2020-09-20 neels *cmp = -1;
148 b3fb4686 2020-09-20 neels return 0;
149 b3fb4686 2020-09-20 neels }
150 b3fb4686 2020-09-20 neels
151 c6eecea3 2020-07-26 stsp n_left = MIN(chunksz, remain_left);
152 c6eecea3 2020-07-26 stsp n_right = MIN(chunksz, remain_right);
153 c6eecea3 2020-07-26 stsp
154 c6eecea3 2020-07-26 stsp if (left->at == NULL) {
155 7a54ad3a 2020-09-20 stsp r = read_at(left->d->root->f,
156 b3fb4686 2020-09-20 neels left->pos + (left->len - remain_left),
157 b3fb4686 2020-09-20 neels buf_left, n_left);
158 b3fb4686 2020-09-20 neels if (r) {
159 b3fb4686 2020-09-20 neels *cmp = 0;
160 b3fb4686 2020-09-20 neels return r;
161 b3fb4686 2020-09-20 neels }
162 c6eecea3 2020-07-26 stsp p_left = buf_left;
163 c6eecea3 2020-07-26 stsp } else {
164 c6eecea3 2020-07-26 stsp p_left = left->at + (left->len - remain_left);
165 c6eecea3 2020-07-26 stsp }
166 c6eecea3 2020-07-26 stsp
167 c6eecea3 2020-07-26 stsp if (right->at == NULL) {
168 7a54ad3a 2020-09-20 stsp r = read_at(right->d->root->f,
169 b3fb4686 2020-09-20 neels right->pos + (right->len - remain_right),
170 b3fb4686 2020-09-20 neels buf_right, n_right);
171 b3fb4686 2020-09-20 neels if (r) {
172 b3fb4686 2020-09-20 neels *cmp = 0;
173 b3fb4686 2020-09-20 neels return r;
174 b3fb4686 2020-09-20 neels }
175 c6eecea3 2020-07-26 stsp p_right = buf_right;
176 c6eecea3 2020-07-26 stsp } else {
177 c6eecea3 2020-07-26 stsp p_right = right->at + (right->len - remain_right);
178 c6eecea3 2020-07-26 stsp }
179 b3fb4686 2020-09-20 neels
180 732e8ee0 2020-09-20 stsp r = buf_cmp(p_left, n_left, p_right, n_right,
181 732e8ee0 2020-09-20 stsp ignore_whitespace);
182 b3fb4686 2020-09-20 neels if (r) {
183 b3fb4686 2020-09-20 neels *cmp = r;
184 b3fb4686 2020-09-20 neels return 0;
185 c6eecea3 2020-07-26 stsp }
186 c6eecea3 2020-07-26 stsp
187 c6eecea3 2020-07-26 stsp remain_left -= n_left;
188 c6eecea3 2020-07-26 stsp remain_right -= n_right;
189 c6eecea3 2020-07-26 stsp }
190 3b0f3d61 2020-01-22 neels
191 b3fb4686 2020-09-20 neels *cmp = 0;
192 b3fb4686 2020-09-20 neels return 0;
193 b3fb4686 2020-09-20 neels }
194 b3fb4686 2020-09-20 neels
195 b3fb4686 2020-09-20 neels int
196 b3fb4686 2020-09-20 neels diff_atom_same(bool *same,
197 b3fb4686 2020-09-20 neels const struct diff_atom *left,
198 b3fb4686 2020-09-20 neels const struct diff_atom *right)
199 b3fb4686 2020-09-20 neels {
200 b3fb4686 2020-09-20 neels int cmp;
201 b3fb4686 2020-09-20 neels int r = diff_atom_cmp(&cmp, left, right);
202 b3fb4686 2020-09-20 neels if (r) {
203 b3fb4686 2020-09-20 neels *same = true;
204 b3fb4686 2020-09-20 neels return r;
205 b3fb4686 2020-09-20 neels }
206 b3fb4686 2020-09-20 neels *same = (cmp == 0);
207 b3fb4686 2020-09-20 neels return 0;
208 c6eecea3 2020-07-26 stsp }
209 c6eecea3 2020-07-26 stsp
210 0d27172a 2020-05-06 neels /* Even if a left or right side is empty, diff output may need to know the
211 0d27172a 2020-05-06 neels * position in that file.
212 0d27172a 2020-05-06 neels * So left_start or right_start must never be NULL -- pass left_count or
213 0d27172a 2020-05-06 neels * right_count as zero to indicate staying at that position without consuming
214 0d27172a 2020-05-06 neels * any lines. */
215 61a7b578 2020-05-06 neels struct diff_chunk *
216 61a7b578 2020-05-06 neels diff_state_add_chunk(struct diff_state *state, bool solved,
217 61a7b578 2020-05-06 neels struct diff_atom *left_start, unsigned int left_count,
218 61a7b578 2020-05-06 neels struct diff_atom *right_start, unsigned int right_count)
219 3b0f3d61 2020-01-22 neels {
220 8546b045 2020-09-20 neels diff_chunk_arraylist_t *result = NULL;
221 8546b045 2020-09-20 neels struct diff_chunk *new_chunk;
222 8546b045 2020-09-20 neels struct diff_chunk chunk = {
223 3b0f3d61 2020-01-22 neels .solved = solved,
224 3b0f3d61 2020-01-22 neels .left_start = left_start,
225 3b0f3d61 2020-01-22 neels .left_count = left_count,
226 3b0f3d61 2020-01-22 neels .right_start = right_start,
227 3b0f3d61 2020-01-22 neels .right_count = right_count,
228 3b0f3d61 2020-01-22 neels };
229 8546b045 2020-09-20 neels enum diff_chunk_type last_t;
230 8546b045 2020-09-20 neels enum diff_chunk_type prev_last_t;
231 8546b045 2020-09-20 neels enum diff_chunk_type new_t;
232 bb867e68 2020-10-11 neels
233 bb867e68 2020-10-11 neels debug("Add %s chunk:\n", chunk.solved ? "solved" : "UNSOLVED");
234 bb867e68 2020-10-11 neels debug("L\n");
235 bb867e68 2020-10-11 neels debug_dump_atoms(&state->left, chunk.left_start, chunk.left_count);
236 bb867e68 2020-10-11 neels debug("R\n");
237 bb867e68 2020-10-11 neels debug_dump_atoms(&state->right, chunk.right_start, chunk.right_count);
238 3b0f3d61 2020-01-22 neels
239 8546b045 2020-09-20 neels if (!solved || state->temp_result.len) {
240 8546b045 2020-09-20 neels /* Append to temp_result */
241 8546b045 2020-09-20 neels result = &state->temp_result;
242 8546b045 2020-09-20 neels } else if (!state->result->chunks.len) {
243 8546b045 2020-09-20 neels /* Append to final result */
244 8546b045 2020-09-20 neels result = &state->result->chunks;
245 8546b045 2020-09-20 neels }
246 8546b045 2020-09-20 neels if (result) {
247 8546b045 2020-09-20 neels ARRAYLIST_ADD(new_chunk, *result);
248 8546b045 2020-09-20 neels if (!new_chunk)
249 8546b045 2020-09-20 neels return NULL;
250 8546b045 2020-09-20 neels *new_chunk = chunk;
251 8546b045 2020-09-20 neels goto chunk_added;
252 8546b045 2020-09-20 neels }
253 8546b045 2020-09-20 neels
254 8546b045 2020-09-20 neels /* Append to solved chunks; make sure that adjacent chunks of same type are combined, and that a minus chunk
255 8546b045 2020-09-20 neels * never directly follows a plus chunk. */
256 8546b045 2020-09-20 neels result = &state->result->chunks;
257 8546b045 2020-09-20 neels
258 8546b045 2020-09-20 neels prev_last_t = result->len > 1 ?
259 8546b045 2020-09-20 neels diff_chunk_type(&result->head[result->len - 2]) : CHUNK_EMPTY;
260 8546b045 2020-09-20 neels last_t = diff_chunk_type(&result->head[result->len - 1]);
261 8546b045 2020-09-20 neels new_t = diff_chunk_type(&chunk);
262 8546b045 2020-09-20 neels
263 8546b045 2020-09-20 neels if (new_t == last_t) {
264 8546b045 2020-09-20 neels new_chunk = &result->head[result->len - 1];
265 8546b045 2020-09-20 neels new_chunk->left_count += chunk.left_count;
266 8546b045 2020-09-20 neels new_chunk->right_count += chunk.right_count;
267 bb867e68 2020-10-11 neels debug(" - added chunk touches previous one of same type, joined:\n");
268 bb867e68 2020-10-11 neels debug("L\n");
269 bb867e68 2020-10-11 neels debug_dump_atoms(&state->left, new_chunk->left_start, new_chunk->left_count);
270 bb867e68 2020-10-11 neels debug("R\n");
271 bb867e68 2020-10-11 neels debug_dump_atoms(&state->right, new_chunk->right_start, new_chunk->right_count);
272 8546b045 2020-09-20 neels } else if (last_t == CHUNK_PLUS && new_t == CHUNK_MINUS) {
273 8546b045 2020-09-20 neels /* If a minus-chunk follows a plus-chunk, place it above the plus-chunk.
274 8546b045 2020-09-20 neels * Is the one before that also a minus? combine. */
275 8546b045 2020-09-20 neels if (prev_last_t == CHUNK_MINUS) {
276 8546b045 2020-09-20 neels new_chunk = &result->head[result->len - 2];
277 8546b045 2020-09-20 neels new_chunk->left_count += chunk.left_count;
278 8546b045 2020-09-20 neels new_chunk->right_count += chunk.right_count;
279 bb867e68 2020-10-11 neels
280 bb867e68 2020-10-11 neels debug(" - added minus-chunk follows plus-chunk,"
281 bb867e68 2020-10-11 neels " put before that plus-chunk and joined"
282 bb867e68 2020-10-11 neels " with preceding minus-chunk:\n");
283 bb867e68 2020-10-11 neels debug("L\n");
284 bb867e68 2020-10-11 neels debug_dump_atoms(&state->left, new_chunk->left_start, new_chunk->left_count);
285 bb867e68 2020-10-11 neels debug("R\n");
286 bb867e68 2020-10-11 neels debug_dump_atoms(&state->right, new_chunk->right_start, new_chunk->right_count);
287 8546b045 2020-09-20 neels } else {
288 724967e9 2020-10-11 neels ARRAYLIST_INSERT(new_chunk, *result, result->len - 1);
289 8546b045 2020-09-20 neels if (!new_chunk)
290 8546b045 2020-09-20 neels return NULL;
291 8546b045 2020-09-20 neels *new_chunk = chunk;
292 bb867e68 2020-10-11 neels
293 bb867e68 2020-10-11 neels debug(" - added minus-chunk follows plus-chunk,"
294 bb867e68 2020-10-11 neels " put before that plus-chunk\n");
295 8546b045 2020-09-20 neels }
296 8546b045 2020-09-20 neels } else {
297 8546b045 2020-09-20 neels ARRAYLIST_ADD(new_chunk, *result);
298 8546b045 2020-09-20 neels if (!new_chunk)
299 8546b045 2020-09-20 neels return NULL;
300 8546b045 2020-09-20 neels *new_chunk = chunk;
301 8546b045 2020-09-20 neels goto chunk_added;
302 8546b045 2020-09-20 neels }
303 8546b045 2020-09-20 neels
304 8546b045 2020-09-20 neels chunk_added:
305 8546b045 2020-09-20 neels return new_chunk;
306 3b0f3d61 2020-01-22 neels }
307 3b0f3d61 2020-01-22 neels
308 61a7b578 2020-05-06 neels void
309 7a54ad3a 2020-09-20 stsp diff_data_init_root(struct diff_data *d, FILE *f, const uint8_t *data,
310 00d5652b 2020-09-22 stsp unsigned long long len, int diff_flags)
311 3b0f3d61 2020-01-22 neels {
312 3b0f3d61 2020-01-22 neels *d = (struct diff_data){
313 7a54ad3a 2020-09-20 stsp .f = f,
314 c6eecea3 2020-07-26 stsp .pos = 0,
315 3b0f3d61 2020-01-22 neels .data = data,
316 3b0f3d61 2020-01-22 neels .len = len,
317 3b0f3d61 2020-01-22 neels .root = d,
318 00d5652b 2020-09-22 stsp .diff_flags = diff_flags,
319 3b0f3d61 2020-01-22 neels };
320 3b0f3d61 2020-01-22 neels }
321 3b0f3d61 2020-01-22 neels
322 61a7b578 2020-05-06 neels void
323 61a7b578 2020-05-06 neels diff_data_init_subsection(struct diff_data *d, struct diff_data *parent,
324 61a7b578 2020-05-06 neels struct diff_atom *from_atom, unsigned int atoms_count)
325 3b0f3d61 2020-01-22 neels {
326 05b5f01f 2020-09-21 stsp struct diff_atom *last_atom;
327 05b5f01f 2020-09-21 stsp
328 05b5f01f 2020-09-21 stsp if (atoms_count == 0) {
329 05b5f01f 2020-09-21 stsp *d = (struct diff_data){
330 05b5f01f 2020-09-21 stsp .f = NULL,
331 05b5f01f 2020-09-21 stsp .pos = 0,
332 05b5f01f 2020-09-21 stsp .data = "",
333 05b5f01f 2020-09-21 stsp .len = 0,
334 05b5f01f 2020-09-21 stsp .root = parent->root,
335 05b5f01f 2020-09-21 stsp .atoms.head = NULL,
336 05b5f01f 2020-09-21 stsp .atoms.len = atoms_count,
337 05b5f01f 2020-09-21 stsp };
338 05b5f01f 2020-09-21 stsp
339 05b5f01f 2020-09-21 stsp return;
340 05b5f01f 2020-09-21 stsp }
341 05b5f01f 2020-09-21 stsp
342 05b5f01f 2020-09-21 stsp last_atom = from_atom + atoms_count - 1;
343 3b0f3d61 2020-01-22 neels *d = (struct diff_data){
344 7a54ad3a 2020-09-20 stsp .f = NULL,
345 c6eecea3 2020-07-26 stsp .pos = from_atom->pos,
346 3b0f3d61 2020-01-22 neels .data = from_atom->at,
347 c6eecea3 2020-07-26 stsp .len = (last_atom->pos + last_atom->len) - from_atom->pos,
348 3b0f3d61 2020-01-22 neels .root = parent->root,
349 3b0f3d61 2020-01-22 neels .atoms.head = from_atom,
350 3b0f3d61 2020-01-22 neels .atoms.len = atoms_count,
351 3b0f3d61 2020-01-22 neels };
352 3b0f3d61 2020-01-22 neels
353 3b0f3d61 2020-01-22 neels debug("subsection:\n");
354 3b0f3d61 2020-01-22 neels debug_dump(d);
355 3b0f3d61 2020-01-22 neels }
356 3b0f3d61 2020-01-22 neels
357 61a7b578 2020-05-06 neels void
358 61a7b578 2020-05-06 neels diff_data_free(struct diff_data *diff_data)
359 3b0f3d61 2020-01-22 neels {
360 3b0f3d61 2020-01-22 neels if (!diff_data)
361 3b0f3d61 2020-01-22 neels return;
362 3b0f3d61 2020-01-22 neels if (diff_data->atoms.allocated)
363 3b0f3d61 2020-01-22 neels ARRAYLIST_FREE(diff_data->atoms);
364 3b0f3d61 2020-01-22 neels }
365 3b0f3d61 2020-01-22 neels
366 3e6cba3a 2020-08-13 stsp int
367 0d27172a 2020-05-06 neels diff_algo_none(const struct diff_algo_config *algo_config,
368 0d27172a 2020-05-06 neels struct diff_state *state)
369 3b0f3d61 2020-01-22 neels {
370 3b0f3d61 2020-01-22 neels debug("\n** %s\n", __func__);
371 3b0f3d61 2020-01-22 neels debug("left:\n");
372 3b0f3d61 2020-01-22 neels debug_dump(&state->left);
373 3b0f3d61 2020-01-22 neels debug("right:\n");
374 3b0f3d61 2020-01-22 neels debug_dump(&state->right);
375 0d27172a 2020-05-06 neels debug_dump_myers_graph(&state->left, &state->right, NULL, NULL, 0, NULL,
376 0d27172a 2020-05-06 neels 0);
377 3b0f3d61 2020-01-22 neels
378 3b0f3d61 2020-01-22 neels /* Add a chunk of equal lines, if any */
379 467a993d 2020-10-11 neels struct diff_atom *l = state->left.atoms.head;
380 467a993d 2020-10-11 neels unsigned int l_len = state->left.atoms.len;
381 467a993d 2020-10-11 neels struct diff_atom *r = state->right.atoms.head;
382 467a993d 2020-10-11 neels unsigned int r_len = state->right.atoms.len;
383 41ff30f3 2020-10-11 neels unsigned int equal_atoms_start = 0;
384 2146cf12 2020-10-11 neels unsigned int equal_atoms_end = 0;
385 13497fff 2020-10-11 neels unsigned int l_idx = 0;
386 13497fff 2020-10-11 neels unsigned int r_idx = 0;
387 467a993d 2020-10-11 neels
388 41ff30f3 2020-10-11 neels while (equal_atoms_start < l_len
389 41ff30f3 2020-10-11 neels && equal_atoms_start < r_len) {
390 467a993d 2020-10-11 neels int err;
391 b3fb4686 2020-09-20 neels bool same;
392 41ff30f3 2020-10-11 neels err = diff_atom_same(&same, &l[equal_atoms_start],
393 41ff30f3 2020-10-11 neels &r[equal_atoms_start]);
394 467a993d 2020-10-11 neels if (err)
395 467a993d 2020-10-11 neels return err;
396 b3fb4686 2020-09-20 neels if (!same)
397 b3fb4686 2020-09-20 neels break;
398 41ff30f3 2020-10-11 neels equal_atoms_start++;
399 b3fb4686 2020-09-20 neels }
400 2146cf12 2020-10-11 neels while (equal_atoms_end < (l_len - equal_atoms_start)
401 2146cf12 2020-10-11 neels && equal_atoms_end < (r_len - equal_atoms_start)) {
402 2146cf12 2020-10-11 neels int err;
403 2146cf12 2020-10-11 neels bool same;
404 2146cf12 2020-10-11 neels err = diff_atom_same(&same, &l[l_len - 1 - equal_atoms_end],
405 2146cf12 2020-10-11 neels &r[r_len - 1 - equal_atoms_end]);
406 2146cf12 2020-10-11 neels if (err)
407 2146cf12 2020-10-11 neels return err;
408 2146cf12 2020-10-11 neels if (!same)
409 2146cf12 2020-10-11 neels break;
410 2146cf12 2020-10-11 neels equal_atoms_end++;
411 2146cf12 2020-10-11 neels }
412 2146cf12 2020-10-11 neels
413 2146cf12 2020-10-11 neels /* Add a chunk of equal lines at the start */
414 41ff30f3 2020-10-11 neels if (equal_atoms_start) {
415 3b0f3d61 2020-01-22 neels if (!diff_state_add_chunk(state, true,
416 467a993d 2020-10-11 neels &l[0],
417 41ff30f3 2020-10-11 neels equal_atoms_start,
418 467a993d 2020-10-11 neels &r[0],
419 41ff30f3 2020-10-11 neels equal_atoms_start))
420 3e6cba3a 2020-08-13 stsp return ENOMEM;
421 13497fff 2020-10-11 neels l_idx += equal_atoms_start;
422 13497fff 2020-10-11 neels r_idx += equal_atoms_start;
423 3b0f3d61 2020-01-22 neels }
424 3b0f3d61 2020-01-22 neels
425 3b0f3d61 2020-01-22 neels /* Add a "minus" chunk with all lines from the left. */
426 2146cf12 2020-10-11 neels if (equal_atoms_start + equal_atoms_end < l_len) {
427 13497fff 2020-10-11 neels unsigned int add_len = l_len - equal_atoms_start - equal_atoms_end;
428 3b0f3d61 2020-01-22 neels if (!diff_state_add_chunk(state, true,
429 13497fff 2020-10-11 neels &l[l_idx], add_len,
430 13497fff 2020-10-11 neels &r[r_idx], 0))
431 13497fff 2020-10-11 neels return ENOMEM;
432 13497fff 2020-10-11 neels l_idx += add_len;
433 3b0f3d61 2020-01-22 neels }
434 3b0f3d61 2020-01-22 neels
435 3b0f3d61 2020-01-22 neels /* Add a "plus" chunk with all lines from the right. */
436 2146cf12 2020-10-11 neels if (equal_atoms_start + equal_atoms_end < r_len) {
437 13497fff 2020-10-11 neels unsigned int add_len = r_len - equal_atoms_start - equal_atoms_end;
438 3b0f3d61 2020-01-22 neels if (!diff_state_add_chunk(state, true,
439 13497fff 2020-10-11 neels &l[equal_atoms_start], 0,
440 13497fff 2020-10-11 neels &r[equal_atoms_start], add_len))
441 13497fff 2020-10-11 neels return ENOMEM;
442 13497fff 2020-10-11 neels r_idx += add_len;
443 3b0f3d61 2020-01-22 neels }
444 2146cf12 2020-10-11 neels
445 2146cf12 2020-10-11 neels /* Add a chunk of equal lines at the end */
446 2146cf12 2020-10-11 neels if (equal_atoms_end) {
447 2146cf12 2020-10-11 neels if (!diff_state_add_chunk(state, true,
448 13497fff 2020-10-11 neels &l[l_idx], equal_atoms_end,
449 13497fff 2020-10-11 neels &r[r_idx], equal_atoms_end))
450 2146cf12 2020-10-11 neels return ENOMEM;
451 2146cf12 2020-10-11 neels }
452 2146cf12 2020-10-11 neels
453 3b0f3d61 2020-01-22 neels return DIFF_RC_OK;
454 3b0f3d61 2020-01-22 neels }
455 3b0f3d61 2020-01-22 neels
456 3e6cba3a 2020-08-13 stsp int
457 0d27172a 2020-05-06 neels diff_run_algo(const struct diff_algo_config *algo_config,
458 0d27172a 2020-05-06 neels struct diff_state *state)
459 3b0f3d61 2020-01-22 neels {
460 3e6cba3a 2020-08-13 stsp int rc;
461 3b0f3d61 2020-01-22 neels ARRAYLIST_FREE(state->temp_result);
462 3b0f3d61 2020-01-22 neels
463 0d27172a 2020-05-06 neels if (!algo_config || !algo_config->impl
464 0d27172a 2020-05-06 neels || !state->recursion_depth_left) {
465 3b0f3d61 2020-01-22 neels debug("MAX RECURSION REACHED, just dumping diff chunks\n");
466 3b0f3d61 2020-01-22 neels return diff_algo_none(algo_config, state);
467 3b0f3d61 2020-01-22 neels }
468 3b0f3d61 2020-01-22 neels
469 3b0f3d61 2020-01-22 neels ARRAYLIST_INIT(state->temp_result, DIFF_RESULT_ALLOC_BLOCKSIZE);
470 3b0f3d61 2020-01-22 neels rc = algo_config->impl(algo_config, state);
471 3b0f3d61 2020-01-22 neels switch (rc) {
472 3b0f3d61 2020-01-22 neels case DIFF_RC_USE_DIFF_ALGO_FALLBACK:
473 0d27172a 2020-05-06 neels debug("Got DIFF_RC_USE_DIFF_ALGO_FALLBACK (%p)\n",
474 0d27172a 2020-05-06 neels algo_config->fallback_algo);
475 3b0f3d61 2020-01-22 neels rc = diff_run_algo(algo_config->fallback_algo, state);
476 3b0f3d61 2020-01-22 neels goto return_rc;
477 3b0f3d61 2020-01-22 neels
478 3b0f3d61 2020-01-22 neels case DIFF_RC_OK:
479 3b0f3d61 2020-01-22 neels /* continue below */
480 3b0f3d61 2020-01-22 neels break;
481 3b0f3d61 2020-01-22 neels
482 3b0f3d61 2020-01-22 neels default:
483 3b0f3d61 2020-01-22 neels /* some error happened */
484 3b0f3d61 2020-01-22 neels goto return_rc;
485 3b0f3d61 2020-01-22 neels }
486 3b0f3d61 2020-01-22 neels
487 0d27172a 2020-05-06 neels /* Pick up any diff chunks that are still unsolved and feed to
488 0d27172a 2020-05-06 neels * inner_algo. inner_algo will solve unsolved chunks and append to
489 0d27172a 2020-05-06 neels * result, and subsequent solved chunks on this level are then appended
490 0d27172a 2020-05-06 neels * to result afterwards. */
491 3b0f3d61 2020-01-22 neels int i;
492 3b0f3d61 2020-01-22 neels for (i = 0; i < state->temp_result.len; i++) {
493 3b0f3d61 2020-01-22 neels struct diff_chunk *c = &state->temp_result.head[i];
494 3b0f3d61 2020-01-22 neels if (c->solved) {
495 3b0f3d61 2020-01-22 neels struct diff_chunk *final_c;
496 3b0f3d61 2020-01-22 neels ARRAYLIST_ADD(final_c, state->result->chunks);
497 3b0f3d61 2020-01-22 neels if (!final_c) {
498 3e6cba3a 2020-08-13 stsp rc = ENOMEM;
499 3b0f3d61 2020-01-22 neels goto return_rc;
500 3b0f3d61 2020-01-22 neels }
501 3b0f3d61 2020-01-22 neels *final_c = *c;
502 3b0f3d61 2020-01-22 neels continue;
503 3b0f3d61 2020-01-22 neels }
504 3b0f3d61 2020-01-22 neels
505 3b0f3d61 2020-01-22 neels /* c is an unsolved chunk, feed to inner_algo */
506 3b0f3d61 2020-01-22 neels struct diff_state inner_state = {
507 3b0f3d61 2020-01-22 neels .result = state->result,
508 3b0f3d61 2020-01-22 neels .recursion_depth_left = state->recursion_depth_left - 1,
509 3b0f3d61 2020-01-22 neels };
510 0d27172a 2020-05-06 neels diff_data_init_subsection(&inner_state.left, &state->left,
511 0d27172a 2020-05-06 neels c->left_start, c->left_count);
512 0d27172a 2020-05-06 neels diff_data_init_subsection(&inner_state.right, &state->right,
513 0d27172a 2020-05-06 neels c->right_start, c->right_count);
514 3b0f3d61 2020-01-22 neels
515 3b0f3d61 2020-01-22 neels rc = diff_run_algo(algo_config->inner_algo, &inner_state);
516 3b0f3d61 2020-01-22 neels
517 3b0f3d61 2020-01-22 neels if (rc != DIFF_RC_OK)
518 3b0f3d61 2020-01-22 neels goto return_rc;
519 3b0f3d61 2020-01-22 neels }
520 3b0f3d61 2020-01-22 neels
521 3b0f3d61 2020-01-22 neels rc = DIFF_RC_OK;
522 3b0f3d61 2020-01-22 neels return_rc:
523 3b0f3d61 2020-01-22 neels ARRAYLIST_FREE(state->temp_result);
524 3b0f3d61 2020-01-22 neels return rc;
525 3b0f3d61 2020-01-22 neels }
526 3b0f3d61 2020-01-22 neels
527 61a7b578 2020-05-06 neels struct diff_result *
528 61a7b578 2020-05-06 neels diff_main(const struct diff_config *config,
529 7a54ad3a 2020-09-20 stsp FILE *left_f, const uint8_t *left_data, off_t left_len,
530 7a54ad3a 2020-09-20 stsp FILE *right_f, const uint8_t *right_data, off_t right_len,
531 00d5652b 2020-09-22 stsp int diff_flags)
532 3b0f3d61 2020-01-22 neels {
533 3b0f3d61 2020-01-22 neels struct diff_result *result = malloc(sizeof(struct diff_result));
534 3b0f3d61 2020-01-22 neels if (!result)
535 3b0f3d61 2020-01-22 neels return NULL;
536 3b0f3d61 2020-01-22 neels
537 3b0f3d61 2020-01-22 neels *result = (struct diff_result){};
538 7a54ad3a 2020-09-20 stsp diff_data_init_root(&result->left, left_f, left_data, left_len,
539 00d5652b 2020-09-22 stsp diff_flags);
540 7a54ad3a 2020-09-20 stsp diff_data_init_root(&result->right, right_f, right_data, right_len,
541 00d5652b 2020-09-22 stsp diff_flags);
542 3b0f3d61 2020-01-22 neels
543 3b0f3d61 2020-01-22 neels if (!config->atomize_func) {
544 3e6cba3a 2020-08-13 stsp result->rc = EINVAL;
545 3b0f3d61 2020-01-22 neels return result;
546 3b0f3d61 2020-01-22 neels }
547 3b0f3d61 2020-01-22 neels
548 0d27172a 2020-05-06 neels result->rc = config->atomize_func(config->atomize_func_data,
549 0d27172a 2020-05-06 neels &result->left, &result->right);
550 3b0f3d61 2020-01-22 neels if (result->rc != DIFF_RC_OK)
551 3b0f3d61 2020-01-22 neels return result;
552 3b0f3d61 2020-01-22 neels
553 3b0f3d61 2020-01-22 neels struct diff_state state = {
554 3b0f3d61 2020-01-22 neels .result = result,
555 00961134 2020-09-20 stsp .recursion_depth_left = config->max_recursion_depth ? : 32,
556 3b0f3d61 2020-01-22 neels };
557 0d27172a 2020-05-06 neels diff_data_init_subsection(&state.left, &result->left,
558 0d27172a 2020-05-06 neels result->left.atoms.head,
559 0d27172a 2020-05-06 neels result->left.atoms.len);
560 0d27172a 2020-05-06 neels diff_data_init_subsection(&state.right, &result->right,
561 0d27172a 2020-05-06 neels result->right.atoms.head,
562 0d27172a 2020-05-06 neels result->right.atoms.len);
563 3b0f3d61 2020-01-22 neels
564 3b0f3d61 2020-01-22 neels result->rc = diff_run_algo(config->algo, &state);
565 3b0f3d61 2020-01-22 neels
566 3b0f3d61 2020-01-22 neels return result;
567 3b0f3d61 2020-01-22 neels }
568 3b0f3d61 2020-01-22 neels
569 61a7b578 2020-05-06 neels void
570 61a7b578 2020-05-06 neels diff_result_free(struct diff_result *result)
571 3b0f3d61 2020-01-22 neels {
572 3b0f3d61 2020-01-22 neels if (!result)
573 3b0f3d61 2020-01-22 neels return;
574 3b0f3d61 2020-01-22 neels diff_data_free(&result->left);
575 3b0f3d61 2020-01-22 neels diff_data_free(&result->right);
576 3b0f3d61 2020-01-22 neels ARRAYLIST_FREE(result->chunks);
577 3b0f3d61 2020-01-22 neels free(result);
578 3b0f3d61 2020-01-22 neels }