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 3b0f3d61 2020-01-22 neels
233 8546b045 2020-09-20 neels if (!solved || state->temp_result.len) {
234 8546b045 2020-09-20 neels /* Append to temp_result */
235 8546b045 2020-09-20 neels result = &state->temp_result;
236 8546b045 2020-09-20 neels } else if (!state->result->chunks.len) {
237 8546b045 2020-09-20 neels /* Append to final result */
238 8546b045 2020-09-20 neels result = &state->result->chunks;
239 8546b045 2020-09-20 neels }
240 8546b045 2020-09-20 neels if (result) {
241 8546b045 2020-09-20 neels ARRAYLIST_ADD(new_chunk, *result);
242 8546b045 2020-09-20 neels if (!new_chunk)
243 8546b045 2020-09-20 neels return NULL;
244 8546b045 2020-09-20 neels *new_chunk = chunk;
245 8546b045 2020-09-20 neels goto chunk_added;
246 8546b045 2020-09-20 neels }
247 8546b045 2020-09-20 neels
248 8546b045 2020-09-20 neels /* Append to solved chunks; make sure that adjacent chunks of same type are combined, and that a minus chunk
249 8546b045 2020-09-20 neels * never directly follows a plus chunk. */
250 8546b045 2020-09-20 neels result = &state->result->chunks;
251 8546b045 2020-09-20 neels
252 8546b045 2020-09-20 neels prev_last_t = result->len > 1 ?
253 8546b045 2020-09-20 neels diff_chunk_type(&result->head[result->len - 2]) : CHUNK_EMPTY;
254 8546b045 2020-09-20 neels last_t = diff_chunk_type(&result->head[result->len - 1]);
255 8546b045 2020-09-20 neels new_t = diff_chunk_type(&chunk);
256 8546b045 2020-09-20 neels
257 8546b045 2020-09-20 neels if (new_t == last_t) {
258 8546b045 2020-09-20 neels new_chunk = &result->head[result->len - 1];
259 8546b045 2020-09-20 neels new_chunk->left_count += chunk.left_count;
260 8546b045 2020-09-20 neels new_chunk->right_count += chunk.right_count;
261 8546b045 2020-09-20 neels } else if (last_t == CHUNK_PLUS && new_t == CHUNK_MINUS) {
262 8546b045 2020-09-20 neels /* If a minus-chunk follows a plus-chunk, place it above the plus-chunk.
263 8546b045 2020-09-20 neels * Is the one before that also a minus? combine. */
264 8546b045 2020-09-20 neels if (prev_last_t == CHUNK_MINUS) {
265 8546b045 2020-09-20 neels new_chunk = &result->head[result->len - 2];
266 8546b045 2020-09-20 neels new_chunk->left_count += chunk.left_count;
267 8546b045 2020-09-20 neels new_chunk->right_count += chunk.right_count;
268 8546b045 2020-09-20 neels } else {
269 8546b045 2020-09-20 neels ARRAYLIST_INSERT(new_chunk, *result, result->len - 2);
270 8546b045 2020-09-20 neels if (!new_chunk)
271 8546b045 2020-09-20 neels return NULL;
272 8546b045 2020-09-20 neels *new_chunk = chunk;
273 8546b045 2020-09-20 neels }
274 8546b045 2020-09-20 neels } else {
275 8546b045 2020-09-20 neels ARRAYLIST_ADD(new_chunk, *result);
276 8546b045 2020-09-20 neels if (!new_chunk)
277 8546b045 2020-09-20 neels return NULL;
278 8546b045 2020-09-20 neels *new_chunk = chunk;
279 8546b045 2020-09-20 neels goto chunk_added;
280 8546b045 2020-09-20 neels }
281 8546b045 2020-09-20 neels
282 8546b045 2020-09-20 neels chunk_added:
283 8546b045 2020-09-20 neels debug("Add %s chunk:\n", new_chunk->solved ? "solved" : "UNSOLVED");
284 8546b045 2020-09-20 neels debug("L\n");
285 8546b045 2020-09-20 neels debug_dump_atoms(&state->left, new_chunk->left_start, new_chunk->left_count);
286 3b0f3d61 2020-01-22 neels debug("R\n");
287 8546b045 2020-09-20 neels debug_dump_atoms(&state->right, new_chunk->right_start, new_chunk->right_count);
288 8546b045 2020-09-20 neels return new_chunk;
289 3b0f3d61 2020-01-22 neels }
290 3b0f3d61 2020-01-22 neels
291 61a7b578 2020-05-06 neels void
292 7a54ad3a 2020-09-20 stsp diff_data_init_root(struct diff_data *d, FILE *f, const uint8_t *data,
293 00d5652b 2020-09-22 stsp unsigned long long len, int diff_flags)
294 3b0f3d61 2020-01-22 neels {
295 3b0f3d61 2020-01-22 neels *d = (struct diff_data){
296 7a54ad3a 2020-09-20 stsp .f = f,
297 c6eecea3 2020-07-26 stsp .pos = 0,
298 3b0f3d61 2020-01-22 neels .data = data,
299 3b0f3d61 2020-01-22 neels .len = len,
300 3b0f3d61 2020-01-22 neels .root = d,
301 00d5652b 2020-09-22 stsp .diff_flags = diff_flags,
302 3b0f3d61 2020-01-22 neels };
303 3b0f3d61 2020-01-22 neels }
304 3b0f3d61 2020-01-22 neels
305 61a7b578 2020-05-06 neels void
306 61a7b578 2020-05-06 neels diff_data_init_subsection(struct diff_data *d, struct diff_data *parent,
307 61a7b578 2020-05-06 neels struct diff_atom *from_atom, unsigned int atoms_count)
308 3b0f3d61 2020-01-22 neels {
309 05b5f01f 2020-09-21 stsp struct diff_atom *last_atom;
310 05b5f01f 2020-09-21 stsp
311 05b5f01f 2020-09-21 stsp if (atoms_count == 0) {
312 05b5f01f 2020-09-21 stsp *d = (struct diff_data){
313 05b5f01f 2020-09-21 stsp .f = NULL,
314 05b5f01f 2020-09-21 stsp .pos = 0,
315 05b5f01f 2020-09-21 stsp .data = "",
316 05b5f01f 2020-09-21 stsp .len = 0,
317 05b5f01f 2020-09-21 stsp .root = parent->root,
318 05b5f01f 2020-09-21 stsp .atoms.head = NULL,
319 05b5f01f 2020-09-21 stsp .atoms.len = atoms_count,
320 05b5f01f 2020-09-21 stsp };
321 05b5f01f 2020-09-21 stsp
322 05b5f01f 2020-09-21 stsp return;
323 05b5f01f 2020-09-21 stsp }
324 05b5f01f 2020-09-21 stsp
325 05b5f01f 2020-09-21 stsp last_atom = from_atom + atoms_count - 1;
326 3b0f3d61 2020-01-22 neels *d = (struct diff_data){
327 7a54ad3a 2020-09-20 stsp .f = NULL,
328 c6eecea3 2020-07-26 stsp .pos = from_atom->pos,
329 3b0f3d61 2020-01-22 neels .data = from_atom->at,
330 c6eecea3 2020-07-26 stsp .len = (last_atom->pos + last_atom->len) - from_atom->pos,
331 3b0f3d61 2020-01-22 neels .root = parent->root,
332 3b0f3d61 2020-01-22 neels .atoms.head = from_atom,
333 3b0f3d61 2020-01-22 neels .atoms.len = atoms_count,
334 3b0f3d61 2020-01-22 neels };
335 3b0f3d61 2020-01-22 neels
336 3b0f3d61 2020-01-22 neels debug("subsection:\n");
337 3b0f3d61 2020-01-22 neels debug_dump(d);
338 3b0f3d61 2020-01-22 neels }
339 3b0f3d61 2020-01-22 neels
340 61a7b578 2020-05-06 neels void
341 61a7b578 2020-05-06 neels diff_data_free(struct diff_data *diff_data)
342 3b0f3d61 2020-01-22 neels {
343 3b0f3d61 2020-01-22 neels if (!diff_data)
344 3b0f3d61 2020-01-22 neels return;
345 3b0f3d61 2020-01-22 neels if (diff_data->atoms.allocated)
346 3b0f3d61 2020-01-22 neels ARRAYLIST_FREE(diff_data->atoms);
347 3b0f3d61 2020-01-22 neels }
348 3b0f3d61 2020-01-22 neels
349 3e6cba3a 2020-08-13 stsp int
350 0d27172a 2020-05-06 neels diff_algo_none(const struct diff_algo_config *algo_config,
351 0d27172a 2020-05-06 neels struct diff_state *state)
352 3b0f3d61 2020-01-22 neels {
353 3b0f3d61 2020-01-22 neels debug("\n** %s\n", __func__);
354 3b0f3d61 2020-01-22 neels debug("left:\n");
355 3b0f3d61 2020-01-22 neels debug_dump(&state->left);
356 3b0f3d61 2020-01-22 neels debug("right:\n");
357 3b0f3d61 2020-01-22 neels debug_dump(&state->right);
358 0d27172a 2020-05-06 neels debug_dump_myers_graph(&state->left, &state->right, NULL, NULL, 0, NULL,
359 0d27172a 2020-05-06 neels 0);
360 3b0f3d61 2020-01-22 neels
361 3b0f3d61 2020-01-22 neels /* Add a chunk of equal lines, if any */
362 3b0f3d61 2020-01-22 neels unsigned int equal_atoms = 0;
363 0d27172a 2020-05-06 neels while (equal_atoms < state->left.atoms.len
364 b3fb4686 2020-09-20 neels && equal_atoms < state->right.atoms.len) {
365 b3fb4686 2020-09-20 neels int r;
366 b3fb4686 2020-09-20 neels bool same;
367 b3fb4686 2020-09-20 neels r = diff_atom_same(&same, &state->left.atoms.head[equal_atoms],
368 b3fb4686 2020-09-20 neels &state->right.atoms.head[equal_atoms]);
369 b3fb4686 2020-09-20 neels if (r)
370 b3fb4686 2020-09-20 neels return r;
371 b3fb4686 2020-09-20 neels if (!same)
372 b3fb4686 2020-09-20 neels break;
373 3b0f3d61 2020-01-22 neels equal_atoms++;
374 b3fb4686 2020-09-20 neels }
375 3b0f3d61 2020-01-22 neels if (equal_atoms) {
376 3b0f3d61 2020-01-22 neels if (!diff_state_add_chunk(state, true,
377 0d27172a 2020-05-06 neels &state->left.atoms.head[0],
378 0d27172a 2020-05-06 neels equal_atoms,
379 0d27172a 2020-05-06 neels &state->right.atoms.head[0],
380 0d27172a 2020-05-06 neels equal_atoms))
381 3e6cba3a 2020-08-13 stsp return ENOMEM;
382 3b0f3d61 2020-01-22 neels }
383 3b0f3d61 2020-01-22 neels
384 3b0f3d61 2020-01-22 neels /* Add a "minus" chunk with all lines from the left. */
385 3b0f3d61 2020-01-22 neels if (equal_atoms < state->left.atoms.len) {
386 3b0f3d61 2020-01-22 neels if (!diff_state_add_chunk(state, true,
387 3b0f3d61 2020-01-22 neels &state->left.atoms.head[equal_atoms],
388 3b0f3d61 2020-01-22 neels state->left.atoms.len - equal_atoms,
389 3b0f3d61 2020-01-22 neels NULL, 0))
390 3e6cba3a 2020-08-13 stsp return ENOMEM;
391 3b0f3d61 2020-01-22 neels }
392 3b0f3d61 2020-01-22 neels
393 3b0f3d61 2020-01-22 neels /* Add a "plus" chunk with all lines from the right. */
394 3b0f3d61 2020-01-22 neels if (equal_atoms < state->right.atoms.len) {
395 3b0f3d61 2020-01-22 neels if (!diff_state_add_chunk(state, true,
396 3b0f3d61 2020-01-22 neels NULL, 0,
397 3b0f3d61 2020-01-22 neels &state->right.atoms.head[equal_atoms],
398 3b0f3d61 2020-01-22 neels state->right.atoms.len - equal_atoms))
399 3e6cba3a 2020-08-13 stsp return ENOMEM;
400 3b0f3d61 2020-01-22 neels }
401 3b0f3d61 2020-01-22 neels return DIFF_RC_OK;
402 3b0f3d61 2020-01-22 neels }
403 3b0f3d61 2020-01-22 neels
404 3e6cba3a 2020-08-13 stsp int
405 0d27172a 2020-05-06 neels diff_run_algo(const struct diff_algo_config *algo_config,
406 0d27172a 2020-05-06 neels struct diff_state *state)
407 3b0f3d61 2020-01-22 neels {
408 3e6cba3a 2020-08-13 stsp int rc;
409 3b0f3d61 2020-01-22 neels ARRAYLIST_FREE(state->temp_result);
410 3b0f3d61 2020-01-22 neels
411 0d27172a 2020-05-06 neels if (!algo_config || !algo_config->impl
412 0d27172a 2020-05-06 neels || !state->recursion_depth_left) {
413 3b0f3d61 2020-01-22 neels debug("MAX RECURSION REACHED, just dumping diff chunks\n");
414 3b0f3d61 2020-01-22 neels return diff_algo_none(algo_config, state);
415 3b0f3d61 2020-01-22 neels }
416 3b0f3d61 2020-01-22 neels
417 3b0f3d61 2020-01-22 neels ARRAYLIST_INIT(state->temp_result, DIFF_RESULT_ALLOC_BLOCKSIZE);
418 3b0f3d61 2020-01-22 neels rc = algo_config->impl(algo_config, state);
419 3b0f3d61 2020-01-22 neels switch (rc) {
420 3b0f3d61 2020-01-22 neels case DIFF_RC_USE_DIFF_ALGO_FALLBACK:
421 0d27172a 2020-05-06 neels debug("Got DIFF_RC_USE_DIFF_ALGO_FALLBACK (%p)\n",
422 0d27172a 2020-05-06 neels algo_config->fallback_algo);
423 3b0f3d61 2020-01-22 neels rc = diff_run_algo(algo_config->fallback_algo, state);
424 3b0f3d61 2020-01-22 neels goto return_rc;
425 3b0f3d61 2020-01-22 neels
426 3b0f3d61 2020-01-22 neels case DIFF_RC_OK:
427 3b0f3d61 2020-01-22 neels /* continue below */
428 3b0f3d61 2020-01-22 neels break;
429 3b0f3d61 2020-01-22 neels
430 3b0f3d61 2020-01-22 neels default:
431 3b0f3d61 2020-01-22 neels /* some error happened */
432 3b0f3d61 2020-01-22 neels goto return_rc;
433 3b0f3d61 2020-01-22 neels }
434 3b0f3d61 2020-01-22 neels
435 0d27172a 2020-05-06 neels /* Pick up any diff chunks that are still unsolved and feed to
436 0d27172a 2020-05-06 neels * inner_algo. inner_algo will solve unsolved chunks and append to
437 0d27172a 2020-05-06 neels * result, and subsequent solved chunks on this level are then appended
438 0d27172a 2020-05-06 neels * to result afterwards. */
439 3b0f3d61 2020-01-22 neels int i;
440 3b0f3d61 2020-01-22 neels for (i = 0; i < state->temp_result.len; i++) {
441 3b0f3d61 2020-01-22 neels struct diff_chunk *c = &state->temp_result.head[i];
442 3b0f3d61 2020-01-22 neels if (c->solved) {
443 3b0f3d61 2020-01-22 neels struct diff_chunk *final_c;
444 3b0f3d61 2020-01-22 neels ARRAYLIST_ADD(final_c, state->result->chunks);
445 3b0f3d61 2020-01-22 neels if (!final_c) {
446 3e6cba3a 2020-08-13 stsp rc = ENOMEM;
447 3b0f3d61 2020-01-22 neels goto return_rc;
448 3b0f3d61 2020-01-22 neels }
449 3b0f3d61 2020-01-22 neels *final_c = *c;
450 3b0f3d61 2020-01-22 neels continue;
451 3b0f3d61 2020-01-22 neels }
452 3b0f3d61 2020-01-22 neels
453 3b0f3d61 2020-01-22 neels /* c is an unsolved chunk, feed to inner_algo */
454 3b0f3d61 2020-01-22 neels struct diff_state inner_state = {
455 3b0f3d61 2020-01-22 neels .result = state->result,
456 3b0f3d61 2020-01-22 neels .recursion_depth_left = state->recursion_depth_left - 1,
457 3b0f3d61 2020-01-22 neels };
458 0d27172a 2020-05-06 neels diff_data_init_subsection(&inner_state.left, &state->left,
459 0d27172a 2020-05-06 neels c->left_start, c->left_count);
460 0d27172a 2020-05-06 neels diff_data_init_subsection(&inner_state.right, &state->right,
461 0d27172a 2020-05-06 neels c->right_start, c->right_count);
462 3b0f3d61 2020-01-22 neels
463 3b0f3d61 2020-01-22 neels rc = diff_run_algo(algo_config->inner_algo, &inner_state);
464 3b0f3d61 2020-01-22 neels
465 3b0f3d61 2020-01-22 neels if (rc != DIFF_RC_OK)
466 3b0f3d61 2020-01-22 neels goto return_rc;
467 3b0f3d61 2020-01-22 neels }
468 3b0f3d61 2020-01-22 neels
469 3b0f3d61 2020-01-22 neels rc = DIFF_RC_OK;
470 3b0f3d61 2020-01-22 neels return_rc:
471 3b0f3d61 2020-01-22 neels ARRAYLIST_FREE(state->temp_result);
472 3b0f3d61 2020-01-22 neels return rc;
473 3b0f3d61 2020-01-22 neels }
474 3b0f3d61 2020-01-22 neels
475 61a7b578 2020-05-06 neels struct diff_result *
476 61a7b578 2020-05-06 neels diff_main(const struct diff_config *config,
477 7a54ad3a 2020-09-20 stsp FILE *left_f, const uint8_t *left_data, off_t left_len,
478 7a54ad3a 2020-09-20 stsp FILE *right_f, const uint8_t *right_data, off_t right_len,
479 00d5652b 2020-09-22 stsp int diff_flags)
480 3b0f3d61 2020-01-22 neels {
481 3b0f3d61 2020-01-22 neels struct diff_result *result = malloc(sizeof(struct diff_result));
482 3b0f3d61 2020-01-22 neels if (!result)
483 3b0f3d61 2020-01-22 neels return NULL;
484 3b0f3d61 2020-01-22 neels
485 3b0f3d61 2020-01-22 neels *result = (struct diff_result){};
486 7a54ad3a 2020-09-20 stsp diff_data_init_root(&result->left, left_f, left_data, left_len,
487 00d5652b 2020-09-22 stsp diff_flags);
488 7a54ad3a 2020-09-20 stsp diff_data_init_root(&result->right, right_f, right_data, right_len,
489 00d5652b 2020-09-22 stsp diff_flags);
490 3b0f3d61 2020-01-22 neels
491 3b0f3d61 2020-01-22 neels if (!config->atomize_func) {
492 3e6cba3a 2020-08-13 stsp result->rc = EINVAL;
493 3b0f3d61 2020-01-22 neels return result;
494 3b0f3d61 2020-01-22 neels }
495 3b0f3d61 2020-01-22 neels
496 0d27172a 2020-05-06 neels result->rc = config->atomize_func(config->atomize_func_data,
497 0d27172a 2020-05-06 neels &result->left, &result->right);
498 3b0f3d61 2020-01-22 neels if (result->rc != DIFF_RC_OK)
499 3b0f3d61 2020-01-22 neels return result;
500 3b0f3d61 2020-01-22 neels
501 3b0f3d61 2020-01-22 neels struct diff_state state = {
502 3b0f3d61 2020-01-22 neels .result = result,
503 00961134 2020-09-20 stsp .recursion_depth_left = config->max_recursion_depth ? : 32,
504 3b0f3d61 2020-01-22 neels };
505 0d27172a 2020-05-06 neels diff_data_init_subsection(&state.left, &result->left,
506 0d27172a 2020-05-06 neels result->left.atoms.head,
507 0d27172a 2020-05-06 neels result->left.atoms.len);
508 0d27172a 2020-05-06 neels diff_data_init_subsection(&state.right, &result->right,
509 0d27172a 2020-05-06 neels result->right.atoms.head,
510 0d27172a 2020-05-06 neels result->right.atoms.len);
511 3b0f3d61 2020-01-22 neels
512 3b0f3d61 2020-01-22 neels result->rc = diff_run_algo(config->algo, &state);
513 3b0f3d61 2020-01-22 neels
514 3b0f3d61 2020-01-22 neels return result;
515 3b0f3d61 2020-01-22 neels }
516 3b0f3d61 2020-01-22 neels
517 61a7b578 2020-05-06 neels void
518 61a7b578 2020-05-06 neels diff_result_free(struct diff_result *result)
519 3b0f3d61 2020-01-22 neels {
520 3b0f3d61 2020-01-22 neels if (!result)
521 3b0f3d61 2020-01-22 neels return;
522 3b0f3d61 2020-01-22 neels diff_data_free(&result->left);
523 3b0f3d61 2020-01-22 neels diff_data_free(&result->right);
524 3b0f3d61 2020-01-22 neels ARRAYLIST_FREE(result->chunks);
525 3b0f3d61 2020-01-22 neels free(result);
526 3b0f3d61 2020-01-22 neels }