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 e10a628a 2020-09-16 stsp #include <diff/arraylist.h>
33 3b0f3d61 2020-01-22 neels #include <diff/diff_main.h>
34 3b0f3d61 2020-01-22 neels
35 e4464189 2020-09-20 stsp #include "diff_debug.h"
36 c6eecea3 2020-07-26 stsp
37 b3fb4686 2020-09-20 neels static int
38 b3fb4686 2020-09-20 neels read_at(int fd, int at_pos, unsigned char *buf, size_t len)
39 c6eecea3 2020-07-26 stsp {
40 b3fb4686 2020-09-20 neels int r;
41 b3fb4686 2020-09-20 neels if (lseek(fd, at_pos, SEEK_SET) == -1)
42 b3fb4686 2020-09-20 neels return errno;
43 b3fb4686 2020-09-20 neels r = read(fd, buf, len);
44 b3fb4686 2020-09-20 neels if (r == -1)
45 b3fb4686 2020-09-20 neels return errno;
46 b3fb4686 2020-09-20 neels if (r != len)
47 b3fb4686 2020-09-20 neels return EIO;
48 b3fb4686 2020-09-20 neels return 0;
49 b3fb4686 2020-09-20 neels }
50 b3fb4686 2020-09-20 neels
51 b3fb4686 2020-09-20 neels static int
52 b3fb4686 2020-09-20 neels buf_cmp(const unsigned char *left, size_t left_len,
53 732e8ee0 2020-09-20 stsp const unsigned char *right, size_t right_len,
54 732e8ee0 2020-09-20 stsp bool ignore_whitespace)
55 b3fb4686 2020-09-20 neels {
56 732e8ee0 2020-09-20 stsp int cmp;
57 732e8ee0 2020-09-20 stsp
58 732e8ee0 2020-09-20 stsp if (ignore_whitespace) {
59 732e8ee0 2020-09-20 stsp int il = 0, ir = 0;
60 732e8ee0 2020-09-20 stsp while (il < left_len && ir < right_len) {
61 732e8ee0 2020-09-20 stsp unsigned char cl = left[il];
62 732e8ee0 2020-09-20 stsp unsigned char cr = right[ir];
63 732e8ee0 2020-09-20 stsp
64 732e8ee0 2020-09-20 stsp if (isspace(cl) && il < left_len) {
65 732e8ee0 2020-09-20 stsp il++;
66 732e8ee0 2020-09-20 stsp continue;
67 732e8ee0 2020-09-20 stsp }
68 732e8ee0 2020-09-20 stsp if (isspace(cr) && ir < right_len) {
69 732e8ee0 2020-09-20 stsp ir++;
70 732e8ee0 2020-09-20 stsp continue;
71 732e8ee0 2020-09-20 stsp }
72 732e8ee0 2020-09-20 stsp
73 732e8ee0 2020-09-20 stsp if (cl > cr)
74 732e8ee0 2020-09-20 stsp return 1;
75 732e8ee0 2020-09-20 stsp if (cr > cl)
76 732e8ee0 2020-09-20 stsp return -1;
77 732e8ee0 2020-09-20 stsp il++;
78 732e8ee0 2020-09-20 stsp ir++;
79 732e8ee0 2020-09-20 stsp }
80 732e8ee0 2020-09-20 stsp while (il < left_len) {
81 732e8ee0 2020-09-20 stsp unsigned char cl = left[il++];
82 732e8ee0 2020-09-20 stsp if (!isspace(cl))
83 732e8ee0 2020-09-20 stsp return 1;
84 732e8ee0 2020-09-20 stsp }
85 732e8ee0 2020-09-20 stsp while (ir < right_len) {
86 732e8ee0 2020-09-20 stsp unsigned char cr = right[ir++];
87 732e8ee0 2020-09-20 stsp if (!isspace(cr))
88 732e8ee0 2020-09-20 stsp return -1;
89 732e8ee0 2020-09-20 stsp }
90 732e8ee0 2020-09-20 stsp
91 732e8ee0 2020-09-20 stsp return 0;
92 732e8ee0 2020-09-20 stsp }
93 732e8ee0 2020-09-20 stsp
94 732e8ee0 2020-09-20 stsp cmp = memcmp(left, right, MIN(left_len, right_len));
95 b3fb4686 2020-09-20 neels if (cmp)
96 b3fb4686 2020-09-20 neels return cmp;
97 b3fb4686 2020-09-20 neels if (left_len == right_len)
98 b3fb4686 2020-09-20 neels return 0;
99 b3fb4686 2020-09-20 neels return (left_len > right_len) ? 1 : -1;
100 b3fb4686 2020-09-20 neels }
101 b3fb4686 2020-09-20 neels
102 b3fb4686 2020-09-20 neels int
103 b3fb4686 2020-09-20 neels diff_atom_cmp(int *cmp,
104 b3fb4686 2020-09-20 neels const struct diff_atom *left,
105 b3fb4686 2020-09-20 neels const struct diff_atom *right)
106 b3fb4686 2020-09-20 neels {
107 c6eecea3 2020-07-26 stsp off_t remain_left, remain_right;
108 732e8ee0 2020-09-20 stsp bool ignore_whitespace;
109 732e8ee0 2020-09-20 stsp
110 732e8ee0 2020-09-20 stsp ignore_whitespace = (left->d->root->ignore_whitespace ||
111 732e8ee0 2020-09-20 stsp right->d->root->ignore_whitespace);
112 c6eecea3 2020-07-26 stsp
113 732e8ee0 2020-09-20 stsp if (!ignore_whitespace) {
114 732e8ee0 2020-09-20 stsp if (!left->len && !right->len) {
115 732e8ee0 2020-09-20 stsp *cmp = 0;
116 732e8ee0 2020-09-20 stsp return 0;
117 732e8ee0 2020-09-20 stsp }
118 732e8ee0 2020-09-20 stsp if (!right->len) {
119 732e8ee0 2020-09-20 stsp *cmp = 1;
120 732e8ee0 2020-09-20 stsp return 0;
121 732e8ee0 2020-09-20 stsp }
122 732e8ee0 2020-09-20 stsp if (!left->len) {
123 732e8ee0 2020-09-20 stsp *cmp = -1;
124 732e8ee0 2020-09-20 stsp return 0;
125 732e8ee0 2020-09-20 stsp }
126 b3fb4686 2020-09-20 neels }
127 c6eecea3 2020-07-26 stsp
128 b3fb4686 2020-09-20 neels if (left->at != NULL && right->at != NULL) {
129 732e8ee0 2020-09-20 stsp *cmp = buf_cmp(left->at, left->len, right->at, right->len,
130 732e8ee0 2020-09-20 stsp ignore_whitespace);
131 b3fb4686 2020-09-20 neels return 0;
132 b3fb4686 2020-09-20 neels }
133 c6eecea3 2020-07-26 stsp
134 c6eecea3 2020-07-26 stsp remain_left = left->len;
135 c6eecea3 2020-07-26 stsp remain_right = right->len;
136 c6eecea3 2020-07-26 stsp while (remain_left > 0 || remain_right > 0) {
137 c6eecea3 2020-07-26 stsp const size_t chunksz = 8192;
138 c6eecea3 2020-07-26 stsp unsigned char buf_left[chunksz], buf_right[chunksz];
139 c6eecea3 2020-07-26 stsp const uint8_t *p_left, *p_right;
140 c6eecea3 2020-07-26 stsp off_t n_left, n_right;
141 c6eecea3 2020-07-26 stsp ssize_t r;
142 c6eecea3 2020-07-26 stsp
143 b3fb4686 2020-09-20 neels if (!remain_right) {
144 b3fb4686 2020-09-20 neels *cmp = 1;
145 b3fb4686 2020-09-20 neels return 0;
146 b3fb4686 2020-09-20 neels }
147 b3fb4686 2020-09-20 neels if (!remain_left) {
148 b3fb4686 2020-09-20 neels *cmp = -1;
149 b3fb4686 2020-09-20 neels return 0;
150 b3fb4686 2020-09-20 neels }
151 b3fb4686 2020-09-20 neels
152 c6eecea3 2020-07-26 stsp n_left = MIN(chunksz, remain_left);
153 c6eecea3 2020-07-26 stsp n_right = MIN(chunksz, remain_right);
154 c6eecea3 2020-07-26 stsp
155 c6eecea3 2020-07-26 stsp if (left->at == NULL) {
156 b3fb4686 2020-09-20 neels r = read_at(left->d->root->fd,
157 b3fb4686 2020-09-20 neels left->pos + (left->len - remain_left),
158 b3fb4686 2020-09-20 neels buf_left, n_left);
159 b3fb4686 2020-09-20 neels if (r) {
160 b3fb4686 2020-09-20 neels *cmp = 0;
161 b3fb4686 2020-09-20 neels return r;
162 b3fb4686 2020-09-20 neels }
163 c6eecea3 2020-07-26 stsp p_left = buf_left;
164 c6eecea3 2020-07-26 stsp } else {
165 c6eecea3 2020-07-26 stsp p_left = left->at + (left->len - remain_left);
166 c6eecea3 2020-07-26 stsp }
167 c6eecea3 2020-07-26 stsp
168 c6eecea3 2020-07-26 stsp if (right->at == NULL) {
169 b3fb4686 2020-09-20 neels r = read_at(right->d->root->fd,
170 b3fb4686 2020-09-20 neels right->pos + (right->len - remain_right),
171 b3fb4686 2020-09-20 neels buf_right, n_right);
172 b3fb4686 2020-09-20 neels if (r) {
173 b3fb4686 2020-09-20 neels *cmp = 0;
174 b3fb4686 2020-09-20 neels return r;
175 b3fb4686 2020-09-20 neels }
176 c6eecea3 2020-07-26 stsp p_right = buf_right;
177 c6eecea3 2020-07-26 stsp } else {
178 c6eecea3 2020-07-26 stsp p_right = right->at + (right->len - remain_right);
179 c6eecea3 2020-07-26 stsp }
180 b3fb4686 2020-09-20 neels
181 732e8ee0 2020-09-20 stsp r = buf_cmp(p_left, n_left, p_right, n_right,
182 732e8ee0 2020-09-20 stsp ignore_whitespace);
183 b3fb4686 2020-09-20 neels if (r) {
184 b3fb4686 2020-09-20 neels *cmp = r;
185 b3fb4686 2020-09-20 neels return 0;
186 c6eecea3 2020-07-26 stsp }
187 c6eecea3 2020-07-26 stsp
188 c6eecea3 2020-07-26 stsp remain_left -= n_left;
189 c6eecea3 2020-07-26 stsp remain_right -= n_right;
190 c6eecea3 2020-07-26 stsp }
191 3b0f3d61 2020-01-22 neels
192 b3fb4686 2020-09-20 neels *cmp = 0;
193 b3fb4686 2020-09-20 neels return 0;
194 b3fb4686 2020-09-20 neels }
195 b3fb4686 2020-09-20 neels
196 b3fb4686 2020-09-20 neels int
197 b3fb4686 2020-09-20 neels diff_atom_same(bool *same,
198 b3fb4686 2020-09-20 neels const struct diff_atom *left,
199 b3fb4686 2020-09-20 neels const struct diff_atom *right)
200 b3fb4686 2020-09-20 neels {
201 b3fb4686 2020-09-20 neels int cmp;
202 b3fb4686 2020-09-20 neels int r = diff_atom_cmp(&cmp, left, right);
203 b3fb4686 2020-09-20 neels if (r) {
204 b3fb4686 2020-09-20 neels *same = true;
205 b3fb4686 2020-09-20 neels return r;
206 b3fb4686 2020-09-20 neels }
207 b3fb4686 2020-09-20 neels *same = (cmp == 0);
208 b3fb4686 2020-09-20 neels return 0;
209 c6eecea3 2020-07-26 stsp }
210 c6eecea3 2020-07-26 stsp
211 0d27172a 2020-05-06 neels /* Even if a left or right side is empty, diff output may need to know the
212 0d27172a 2020-05-06 neels * position in that file.
213 0d27172a 2020-05-06 neels * So left_start or right_start must never be NULL -- pass left_count or
214 0d27172a 2020-05-06 neels * right_count as zero to indicate staying at that position without consuming
215 0d27172a 2020-05-06 neels * any lines. */
216 61a7b578 2020-05-06 neels struct diff_chunk *
217 61a7b578 2020-05-06 neels diff_state_add_chunk(struct diff_state *state, bool solved,
218 61a7b578 2020-05-06 neels struct diff_atom *left_start, unsigned int left_count,
219 61a7b578 2020-05-06 neels struct diff_atom *right_start, unsigned int right_count)
220 3b0f3d61 2020-01-22 neels {
221 8546b045 2020-09-20 neels diff_chunk_arraylist_t *result = NULL;
222 8546b045 2020-09-20 neels struct diff_chunk *new_chunk;
223 8546b045 2020-09-20 neels struct diff_chunk chunk = {
224 3b0f3d61 2020-01-22 neels .solved = solved,
225 3b0f3d61 2020-01-22 neels .left_start = left_start,
226 3b0f3d61 2020-01-22 neels .left_count = left_count,
227 3b0f3d61 2020-01-22 neels .right_start = right_start,
228 3b0f3d61 2020-01-22 neels .right_count = right_count,
229 3b0f3d61 2020-01-22 neels };
230 8546b045 2020-09-20 neels enum diff_chunk_type last_t;
231 8546b045 2020-09-20 neels enum diff_chunk_type prev_last_t;
232 8546b045 2020-09-20 neels enum diff_chunk_type new_t;
233 3b0f3d61 2020-01-22 neels
234 8546b045 2020-09-20 neels if (!solved || state->temp_result.len) {
235 8546b045 2020-09-20 neels /* Append to temp_result */
236 8546b045 2020-09-20 neels result = &state->temp_result;
237 8546b045 2020-09-20 neels } else if (!state->result->chunks.len) {
238 8546b045 2020-09-20 neels /* Append to final result */
239 8546b045 2020-09-20 neels result = &state->result->chunks;
240 8546b045 2020-09-20 neels }
241 8546b045 2020-09-20 neels if (result) {
242 8546b045 2020-09-20 neels ARRAYLIST_ADD(new_chunk, *result);
243 8546b045 2020-09-20 neels if (!new_chunk)
244 8546b045 2020-09-20 neels return NULL;
245 8546b045 2020-09-20 neels *new_chunk = chunk;
246 8546b045 2020-09-20 neels goto chunk_added;
247 8546b045 2020-09-20 neels }
248 8546b045 2020-09-20 neels
249 8546b045 2020-09-20 neels /* Append to solved chunks; make sure that adjacent chunks of same type are combined, and that a minus chunk
250 8546b045 2020-09-20 neels * never directly follows a plus chunk. */
251 8546b045 2020-09-20 neels result = &state->result->chunks;
252 8546b045 2020-09-20 neels
253 8546b045 2020-09-20 neels prev_last_t = result->len > 1 ?
254 8546b045 2020-09-20 neels diff_chunk_type(&result->head[result->len - 2]) : CHUNK_EMPTY;
255 8546b045 2020-09-20 neels last_t = diff_chunk_type(&result->head[result->len - 1]);
256 8546b045 2020-09-20 neels new_t = diff_chunk_type(&chunk);
257 8546b045 2020-09-20 neels
258 8546b045 2020-09-20 neels if (new_t == last_t) {
259 8546b045 2020-09-20 neels new_chunk = &result->head[result->len - 1];
260 8546b045 2020-09-20 neels new_chunk->left_count += chunk.left_count;
261 8546b045 2020-09-20 neels new_chunk->right_count += chunk.right_count;
262 8546b045 2020-09-20 neels } else if (last_t == CHUNK_PLUS && new_t == CHUNK_MINUS) {
263 8546b045 2020-09-20 neels /* If a minus-chunk follows a plus-chunk, place it above the plus-chunk.
264 8546b045 2020-09-20 neels * Is the one before that also a minus? combine. */
265 8546b045 2020-09-20 neels if (prev_last_t == CHUNK_MINUS) {
266 8546b045 2020-09-20 neels new_chunk = &result->head[result->len - 2];
267 8546b045 2020-09-20 neels new_chunk->left_count += chunk.left_count;
268 8546b045 2020-09-20 neels new_chunk->right_count += chunk.right_count;
269 8546b045 2020-09-20 neels } else {
270 8546b045 2020-09-20 neels ARRAYLIST_INSERT(new_chunk, *result, result->len - 2);
271 8546b045 2020-09-20 neels if (!new_chunk)
272 8546b045 2020-09-20 neels return NULL;
273 8546b045 2020-09-20 neels *new_chunk = chunk;
274 8546b045 2020-09-20 neels }
275 8546b045 2020-09-20 neels } else {
276 8546b045 2020-09-20 neels ARRAYLIST_ADD(new_chunk, *result);
277 8546b045 2020-09-20 neels if (!new_chunk)
278 8546b045 2020-09-20 neels return NULL;
279 8546b045 2020-09-20 neels *new_chunk = chunk;
280 8546b045 2020-09-20 neels goto chunk_added;
281 8546b045 2020-09-20 neels }
282 8546b045 2020-09-20 neels
283 8546b045 2020-09-20 neels chunk_added:
284 8546b045 2020-09-20 neels debug("Add %s chunk:\n", new_chunk->solved ? "solved" : "UNSOLVED");
285 8546b045 2020-09-20 neels debug("L\n");
286 8546b045 2020-09-20 neels debug_dump_atoms(&state->left, new_chunk->left_start, new_chunk->left_count);
287 3b0f3d61 2020-01-22 neels debug("R\n");
288 8546b045 2020-09-20 neels debug_dump_atoms(&state->right, new_chunk->right_start, new_chunk->right_count);
289 8546b045 2020-09-20 neels return new_chunk;
290 3b0f3d61 2020-01-22 neels }
291 3b0f3d61 2020-01-22 neels
292 61a7b578 2020-05-06 neels void
293 c6eecea3 2020-07-26 stsp diff_data_init_root(struct diff_data *d, int fd, const uint8_t *data,
294 732e8ee0 2020-09-20 stsp unsigned long long len, bool ignore_whitespace)
295 3b0f3d61 2020-01-22 neels {
296 3b0f3d61 2020-01-22 neels *d = (struct diff_data){
297 c6eecea3 2020-07-26 stsp .fd = fd,
298 c6eecea3 2020-07-26 stsp .pos = 0,
299 3b0f3d61 2020-01-22 neels .data = data,
300 3b0f3d61 2020-01-22 neels .len = len,
301 3b0f3d61 2020-01-22 neels .root = d,
302 732e8ee0 2020-09-20 stsp .ignore_whitespace = ignore_whitespace,
303 3b0f3d61 2020-01-22 neels };
304 3b0f3d61 2020-01-22 neels }
305 3b0f3d61 2020-01-22 neels
306 61a7b578 2020-05-06 neels void
307 61a7b578 2020-05-06 neels diff_data_init_subsection(struct diff_data *d, struct diff_data *parent,
308 61a7b578 2020-05-06 neels struct diff_atom *from_atom, unsigned int atoms_count)
309 3b0f3d61 2020-01-22 neels {
310 3b0f3d61 2020-01-22 neels struct diff_atom *last_atom = from_atom + atoms_count - 1;
311 3b0f3d61 2020-01-22 neels *d = (struct diff_data){
312 c6eecea3 2020-07-26 stsp .fd = -1,
313 c6eecea3 2020-07-26 stsp .pos = from_atom->pos,
314 3b0f3d61 2020-01-22 neels .data = from_atom->at,
315 c6eecea3 2020-07-26 stsp .len = (last_atom->pos + last_atom->len) - from_atom->pos,
316 3b0f3d61 2020-01-22 neels .root = parent->root,
317 3b0f3d61 2020-01-22 neels .atoms.head = from_atom,
318 3b0f3d61 2020-01-22 neels .atoms.len = atoms_count,
319 3b0f3d61 2020-01-22 neels };
320 3b0f3d61 2020-01-22 neels
321 3b0f3d61 2020-01-22 neels debug("subsection:\n");
322 3b0f3d61 2020-01-22 neels debug_dump(d);
323 3b0f3d61 2020-01-22 neels }
324 3b0f3d61 2020-01-22 neels
325 61a7b578 2020-05-06 neels void
326 61a7b578 2020-05-06 neels diff_data_free(struct diff_data *diff_data)
327 3b0f3d61 2020-01-22 neels {
328 3b0f3d61 2020-01-22 neels if (!diff_data)
329 3b0f3d61 2020-01-22 neels return;
330 3b0f3d61 2020-01-22 neels if (diff_data->atoms.allocated)
331 3b0f3d61 2020-01-22 neels ARRAYLIST_FREE(diff_data->atoms);
332 3b0f3d61 2020-01-22 neels }
333 3b0f3d61 2020-01-22 neels
334 3e6cba3a 2020-08-13 stsp int
335 0d27172a 2020-05-06 neels diff_algo_none(const struct diff_algo_config *algo_config,
336 0d27172a 2020-05-06 neels struct diff_state *state)
337 3b0f3d61 2020-01-22 neels {
338 3b0f3d61 2020-01-22 neels debug("\n** %s\n", __func__);
339 3b0f3d61 2020-01-22 neels debug("left:\n");
340 3b0f3d61 2020-01-22 neels debug_dump(&state->left);
341 3b0f3d61 2020-01-22 neels debug("right:\n");
342 3b0f3d61 2020-01-22 neels debug_dump(&state->right);
343 0d27172a 2020-05-06 neels debug_dump_myers_graph(&state->left, &state->right, NULL, NULL, 0, NULL,
344 0d27172a 2020-05-06 neels 0);
345 3b0f3d61 2020-01-22 neels
346 3b0f3d61 2020-01-22 neels /* Add a chunk of equal lines, if any */
347 3b0f3d61 2020-01-22 neels unsigned int equal_atoms = 0;
348 0d27172a 2020-05-06 neels while (equal_atoms < state->left.atoms.len
349 b3fb4686 2020-09-20 neels && equal_atoms < state->right.atoms.len) {
350 b3fb4686 2020-09-20 neels int r;
351 b3fb4686 2020-09-20 neels bool same;
352 b3fb4686 2020-09-20 neels r = diff_atom_same(&same, &state->left.atoms.head[equal_atoms],
353 b3fb4686 2020-09-20 neels &state->right.atoms.head[equal_atoms]);
354 b3fb4686 2020-09-20 neels if (r)
355 b3fb4686 2020-09-20 neels return r;
356 b3fb4686 2020-09-20 neels if (!same)
357 b3fb4686 2020-09-20 neels break;
358 3b0f3d61 2020-01-22 neels equal_atoms++;
359 b3fb4686 2020-09-20 neels }
360 3b0f3d61 2020-01-22 neels if (equal_atoms) {
361 3b0f3d61 2020-01-22 neels if (!diff_state_add_chunk(state, true,
362 0d27172a 2020-05-06 neels &state->left.atoms.head[0],
363 0d27172a 2020-05-06 neels equal_atoms,
364 0d27172a 2020-05-06 neels &state->right.atoms.head[0],
365 0d27172a 2020-05-06 neels equal_atoms))
366 3e6cba3a 2020-08-13 stsp return ENOMEM;
367 3b0f3d61 2020-01-22 neels }
368 3b0f3d61 2020-01-22 neels
369 3b0f3d61 2020-01-22 neels /* Add a "minus" chunk with all lines from the left. */
370 3b0f3d61 2020-01-22 neels if (equal_atoms < state->left.atoms.len) {
371 3b0f3d61 2020-01-22 neels if (!diff_state_add_chunk(state, true,
372 3b0f3d61 2020-01-22 neels &state->left.atoms.head[equal_atoms],
373 3b0f3d61 2020-01-22 neels state->left.atoms.len - equal_atoms,
374 3b0f3d61 2020-01-22 neels NULL, 0))
375 3e6cba3a 2020-08-13 stsp return ENOMEM;
376 3b0f3d61 2020-01-22 neels }
377 3b0f3d61 2020-01-22 neels
378 3b0f3d61 2020-01-22 neels /* Add a "plus" chunk with all lines from the right. */
379 3b0f3d61 2020-01-22 neels if (equal_atoms < state->right.atoms.len) {
380 3b0f3d61 2020-01-22 neels if (!diff_state_add_chunk(state, true,
381 3b0f3d61 2020-01-22 neels NULL, 0,
382 3b0f3d61 2020-01-22 neels &state->right.atoms.head[equal_atoms],
383 3b0f3d61 2020-01-22 neels state->right.atoms.len - equal_atoms))
384 3e6cba3a 2020-08-13 stsp return ENOMEM;
385 3b0f3d61 2020-01-22 neels }
386 3b0f3d61 2020-01-22 neels return DIFF_RC_OK;
387 3b0f3d61 2020-01-22 neels }
388 3b0f3d61 2020-01-22 neels
389 3e6cba3a 2020-08-13 stsp int
390 0d27172a 2020-05-06 neels diff_run_algo(const struct diff_algo_config *algo_config,
391 0d27172a 2020-05-06 neels struct diff_state *state)
392 3b0f3d61 2020-01-22 neels {
393 3e6cba3a 2020-08-13 stsp int rc;
394 3b0f3d61 2020-01-22 neels ARRAYLIST_FREE(state->temp_result);
395 3b0f3d61 2020-01-22 neels
396 0d27172a 2020-05-06 neels if (!algo_config || !algo_config->impl
397 0d27172a 2020-05-06 neels || !state->recursion_depth_left) {
398 3b0f3d61 2020-01-22 neels debug("MAX RECURSION REACHED, just dumping diff chunks\n");
399 3b0f3d61 2020-01-22 neels return diff_algo_none(algo_config, state);
400 3b0f3d61 2020-01-22 neels }
401 3b0f3d61 2020-01-22 neels
402 3b0f3d61 2020-01-22 neels ARRAYLIST_INIT(state->temp_result, DIFF_RESULT_ALLOC_BLOCKSIZE);
403 3b0f3d61 2020-01-22 neels rc = algo_config->impl(algo_config, state);
404 3b0f3d61 2020-01-22 neels switch (rc) {
405 3b0f3d61 2020-01-22 neels case DIFF_RC_USE_DIFF_ALGO_FALLBACK:
406 0d27172a 2020-05-06 neels debug("Got DIFF_RC_USE_DIFF_ALGO_FALLBACK (%p)\n",
407 0d27172a 2020-05-06 neels algo_config->fallback_algo);
408 3b0f3d61 2020-01-22 neels rc = diff_run_algo(algo_config->fallback_algo, state);
409 3b0f3d61 2020-01-22 neels goto return_rc;
410 3b0f3d61 2020-01-22 neels
411 3b0f3d61 2020-01-22 neels case DIFF_RC_OK:
412 3b0f3d61 2020-01-22 neels /* continue below */
413 3b0f3d61 2020-01-22 neels break;
414 3b0f3d61 2020-01-22 neels
415 3b0f3d61 2020-01-22 neels default:
416 3b0f3d61 2020-01-22 neels /* some error happened */
417 3b0f3d61 2020-01-22 neels goto return_rc;
418 3b0f3d61 2020-01-22 neels }
419 3b0f3d61 2020-01-22 neels
420 0d27172a 2020-05-06 neels /* Pick up any diff chunks that are still unsolved and feed to
421 0d27172a 2020-05-06 neels * inner_algo. inner_algo will solve unsolved chunks and append to
422 0d27172a 2020-05-06 neels * result, and subsequent solved chunks on this level are then appended
423 0d27172a 2020-05-06 neels * to result afterwards. */
424 3b0f3d61 2020-01-22 neels int i;
425 3b0f3d61 2020-01-22 neels for (i = 0; i < state->temp_result.len; i++) {
426 3b0f3d61 2020-01-22 neels struct diff_chunk *c = &state->temp_result.head[i];
427 3b0f3d61 2020-01-22 neels if (c->solved) {
428 3b0f3d61 2020-01-22 neels struct diff_chunk *final_c;
429 3b0f3d61 2020-01-22 neels ARRAYLIST_ADD(final_c, state->result->chunks);
430 3b0f3d61 2020-01-22 neels if (!final_c) {
431 3e6cba3a 2020-08-13 stsp rc = ENOMEM;
432 3b0f3d61 2020-01-22 neels goto return_rc;
433 3b0f3d61 2020-01-22 neels }
434 3b0f3d61 2020-01-22 neels *final_c = *c;
435 3b0f3d61 2020-01-22 neels continue;
436 3b0f3d61 2020-01-22 neels }
437 3b0f3d61 2020-01-22 neels
438 3b0f3d61 2020-01-22 neels /* c is an unsolved chunk, feed to inner_algo */
439 3b0f3d61 2020-01-22 neels struct diff_state inner_state = {
440 3b0f3d61 2020-01-22 neels .result = state->result,
441 3b0f3d61 2020-01-22 neels .recursion_depth_left = state->recursion_depth_left - 1,
442 3b0f3d61 2020-01-22 neels };
443 0d27172a 2020-05-06 neels diff_data_init_subsection(&inner_state.left, &state->left,
444 0d27172a 2020-05-06 neels c->left_start, c->left_count);
445 0d27172a 2020-05-06 neels diff_data_init_subsection(&inner_state.right, &state->right,
446 0d27172a 2020-05-06 neels c->right_start, c->right_count);
447 3b0f3d61 2020-01-22 neels
448 3b0f3d61 2020-01-22 neels rc = diff_run_algo(algo_config->inner_algo, &inner_state);
449 3b0f3d61 2020-01-22 neels
450 3b0f3d61 2020-01-22 neels if (rc != DIFF_RC_OK)
451 3b0f3d61 2020-01-22 neels goto return_rc;
452 3b0f3d61 2020-01-22 neels }
453 3b0f3d61 2020-01-22 neels
454 3b0f3d61 2020-01-22 neels rc = DIFF_RC_OK;
455 3b0f3d61 2020-01-22 neels return_rc:
456 3b0f3d61 2020-01-22 neels ARRAYLIST_FREE(state->temp_result);
457 3b0f3d61 2020-01-22 neels return rc;
458 3b0f3d61 2020-01-22 neels }
459 3b0f3d61 2020-01-22 neels
460 61a7b578 2020-05-06 neels struct diff_result *
461 61a7b578 2020-05-06 neels diff_main(const struct diff_config *config,
462 c6eecea3 2020-07-26 stsp int left_fd, const uint8_t *left_data, off_t left_len,
463 732e8ee0 2020-09-20 stsp int right_fd, const uint8_t *right_data, off_t right_len,
464 732e8ee0 2020-09-20 stsp bool ignore_whitespace)
465 3b0f3d61 2020-01-22 neels {
466 3b0f3d61 2020-01-22 neels struct diff_result *result = malloc(sizeof(struct diff_result));
467 3b0f3d61 2020-01-22 neels if (!result)
468 3b0f3d61 2020-01-22 neels return NULL;
469 3b0f3d61 2020-01-22 neels
470 3b0f3d61 2020-01-22 neels *result = (struct diff_result){};
471 732e8ee0 2020-09-20 stsp diff_data_init_root(&result->left, left_fd, left_data, left_len,
472 732e8ee0 2020-09-20 stsp ignore_whitespace);
473 732e8ee0 2020-09-20 stsp diff_data_init_root(&result->right, right_fd, right_data, right_len,
474 732e8ee0 2020-09-20 stsp ignore_whitespace);
475 3b0f3d61 2020-01-22 neels
476 3b0f3d61 2020-01-22 neels if (!config->atomize_func) {
477 3e6cba3a 2020-08-13 stsp result->rc = EINVAL;
478 3b0f3d61 2020-01-22 neels return result;
479 3b0f3d61 2020-01-22 neels }
480 3b0f3d61 2020-01-22 neels
481 0d27172a 2020-05-06 neels result->rc = config->atomize_func(config->atomize_func_data,
482 0d27172a 2020-05-06 neels &result->left, &result->right);
483 3b0f3d61 2020-01-22 neels if (result->rc != DIFF_RC_OK)
484 3b0f3d61 2020-01-22 neels return result;
485 3b0f3d61 2020-01-22 neels
486 3b0f3d61 2020-01-22 neels struct diff_state state = {
487 3b0f3d61 2020-01-22 neels .result = result,
488 00961134 2020-09-20 stsp .recursion_depth_left = config->max_recursion_depth ? : 32,
489 3b0f3d61 2020-01-22 neels };
490 0d27172a 2020-05-06 neels diff_data_init_subsection(&state.left, &result->left,
491 0d27172a 2020-05-06 neels result->left.atoms.head,
492 0d27172a 2020-05-06 neels result->left.atoms.len);
493 0d27172a 2020-05-06 neels diff_data_init_subsection(&state.right, &result->right,
494 0d27172a 2020-05-06 neels result->right.atoms.head,
495 0d27172a 2020-05-06 neels result->right.atoms.len);
496 3b0f3d61 2020-01-22 neels
497 3b0f3d61 2020-01-22 neels result->rc = diff_run_algo(config->algo, &state);
498 3b0f3d61 2020-01-22 neels
499 3b0f3d61 2020-01-22 neels return result;
500 3b0f3d61 2020-01-22 neels }
501 3b0f3d61 2020-01-22 neels
502 61a7b578 2020-05-06 neels void
503 61a7b578 2020-05-06 neels diff_result_free(struct diff_result *result)
504 3b0f3d61 2020-01-22 neels {
505 3b0f3d61 2020-01-22 neels if (!result)
506 3b0f3d61 2020-01-22 neels return;
507 3b0f3d61 2020-01-22 neels diff_data_free(&result->left);
508 3b0f3d61 2020-01-22 neels diff_data_free(&result->right);
509 3b0f3d61 2020-01-22 neels ARRAYLIST_FREE(result->chunks);
510 3b0f3d61 2020-01-22 neels free(result);
511 3b0f3d61 2020-01-22 neels }