]>
Commit | Line | Data |
---|---|---|
92b7de93 JS |
1 | /* |
2 | * LibXDiff by Davide Libenzi ( File Differential Library ) | |
044fb190 | 3 | * Copyright (C) 2003-2016 Davide Libenzi, Johannes E. Schindelin |
92b7de93 JS |
4 | * |
5 | * This library is free software; you can redistribute it and/or | |
6 | * modify it under the terms of the GNU Lesser General Public | |
7 | * License as published by the Free Software Foundation; either | |
8 | * version 2.1 of the License, or (at your option) any later version. | |
9 | * | |
10 | * This library is distributed in the hope that it will be useful, | |
11 | * but WITHOUT ANY WARRANTY; without even the implied warranty of | |
12 | * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU | |
13 | * Lesser General Public License for more details. | |
14 | * | |
15 | * You should have received a copy of the GNU Lesser General Public | |
48425792 TZ |
16 | * License along with this library; if not, see |
17 | * <http://www.gnu.org/licenses/>. | |
92b7de93 JS |
18 | * |
19 | * Davide Libenzi <davidel@xmailserver.org> | |
20 | * | |
21 | */ | |
22 | #include "xinclude.h" | |
92b7de93 JS |
23 | |
24 | /* | |
25 | * The basic idea of patience diff is to find lines that are unique in | |
26 | * both files. These are intuitively the ones that we want to see as | |
27 | * common lines. | |
28 | * | |
29 | * The maximal ordered sequence of such line pairs (where ordered means | |
30 | * that the order in the sequence agrees with the order of the lines in | |
31 | * both files) naturally defines an initial set of common lines. | |
32 | * | |
33 | * Now, the algorithm tries to extend the set of common lines by growing | |
34 | * the line ranges where the files have identical lines. | |
35 | * | |
36 | * Between those common lines, the patience diff algorithm is applied | |
37 | * recursively, until no unique line pairs can be found; these line ranges | |
38 | * are handled by the well-known Myers algorithm. | |
39 | */ | |
40 | ||
41 | #define NON_UNIQUE ULONG_MAX | |
42 | ||
43 | /* | |
44 | * This is a hash mapping from line hash to line numbers in the first and | |
45 | * second file. | |
46 | */ | |
47 | struct hashmap { | |
48 | int nr, alloc; | |
49 | struct entry { | |
50 | unsigned long hash; | |
51 | /* | |
52 | * 0 = unused entry, 1 = first line, 2 = second, etc. | |
53 | * line2 is NON_UNIQUE if the line is not unique | |
54 | * in either the first or the second file. | |
55 | */ | |
56 | unsigned long line1, line2; | |
57 | /* | |
58 | * "next" & "previous" are used for the longest common | |
59 | * sequence; | |
60 | * initially, "next" reflects only the order in file1. | |
61 | */ | |
62 | struct entry *next, *previous; | |
2477ab2e JT |
63 | |
64 | /* | |
65 | * If 1, this entry can serve as an anchor. See | |
66 | * Documentation/diff-options.txt for more information. | |
67 | */ | |
68 | unsigned anchor : 1; | |
92b7de93 JS |
69 | } *entries, *first, *last; |
70 | /* were common records found? */ | |
71 | unsigned long has_matches; | |
72 | mmfile_t *file1, *file2; | |
73 | xdfenv_t *env; | |
74 | xpparam_t const *xpp; | |
75 | }; | |
76 | ||
2477ab2e JT |
77 | static int is_anchor(xpparam_t const *xpp, const char *line) |
78 | { | |
79 | int i; | |
80 | for (i = 0; i < xpp->anchors_nr; i++) { | |
81 | if (!strncmp(line, xpp->anchors[i], strlen(xpp->anchors[i]))) | |
82 | return 1; | |
83 | } | |
84 | return 0; | |
85 | } | |
86 | ||
92b7de93 | 87 | /* The argument "pass" is 1 for the first file, 2 for the second. */ |
2477ab2e JT |
88 | static void insert_record(xpparam_t const *xpp, int line, struct hashmap *map, |
89 | int pass) | |
92b7de93 JS |
90 | { |
91 | xrecord_t **records = pass == 1 ? | |
92 | map->env->xdf1.recs : map->env->xdf2.recs; | |
204aa2d2 | 93 | xrecord_t *record = records[line - 1]; |
92b7de93 JS |
94 | /* |
95 | * After xdl_prepare_env() (or more precisely, due to | |
96 | * xdl_classify_record()), the "ha" member of the records (AKA lines) | |
97 | * is _not_ the hash anymore, but a linearized version of it. In | |
98 | * other words, the "ha" member is guaranteed to start with 0 and | |
99 | * the second record's ha can only be 0 or 1, etc. | |
100 | * | |
101 | * So we multiply ha by 2 in the hope that the hashing was | |
102 | * "unique enough". | |
103 | */ | |
104 | int index = (int)((record->ha << 1) % map->alloc); | |
105 | ||
106 | while (map->entries[index].line1) { | |
204aa2d2 | 107 | if (map->entries[index].hash != record->ha) { |
92b7de93 JS |
108 | if (++index >= map->alloc) |
109 | index = 0; | |
110 | continue; | |
111 | } | |
112 | if (pass == 2) | |
113 | map->has_matches = 1; | |
114 | if (pass == 1 || map->entries[index].line2) | |
115 | map->entries[index].line2 = NON_UNIQUE; | |
116 | else | |
117 | map->entries[index].line2 = line; | |
118 | return; | |
119 | } | |
120 | if (pass == 2) | |
121 | return; | |
122 | map->entries[index].line1 = line; | |
123 | map->entries[index].hash = record->ha; | |
2477ab2e | 124 | map->entries[index].anchor = is_anchor(xpp, map->env->xdf1.recs[line - 1]->ptr); |
92b7de93 JS |
125 | if (!map->first) |
126 | map->first = map->entries + index; | |
127 | if (map->last) { | |
128 | map->last->next = map->entries + index; | |
129 | map->entries[index].previous = map->last; | |
130 | } | |
131 | map->last = map->entries + index; | |
132 | map->nr++; | |
133 | } | |
134 | ||
135 | /* | |
136 | * This function has to be called for each recursion into the inter-hunk | |
137 | * parts, as previously non-unique lines can become unique when being | |
138 | * restricted to a smaller part of the files. | |
139 | * | |
140 | * It is assumed that env has been prepared using xdl_prepare(). | |
141 | */ | |
142 | static int fill_hashmap(mmfile_t *file1, mmfile_t *file2, | |
143 | xpparam_t const *xpp, xdfenv_t *env, | |
144 | struct hashmap *result, | |
145 | int line1, int count1, int line2, int count2) | |
146 | { | |
147 | result->file1 = file1; | |
148 | result->file2 = file2; | |
149 | result->xpp = xpp; | |
150 | result->env = env; | |
151 | ||
152 | /* We know exactly how large we want the hash map */ | |
153 | result->alloc = count1 * 2; | |
154 | result->entries = (struct entry *) | |
155 | xdl_malloc(result->alloc * sizeof(struct entry)); | |
156 | if (!result->entries) | |
157 | return -1; | |
158 | memset(result->entries, 0, result->alloc * sizeof(struct entry)); | |
159 | ||
160 | /* First, fill with entries from the first file */ | |
161 | while (count1--) | |
2477ab2e | 162 | insert_record(xpp, line1++, result, 1); |
92b7de93 JS |
163 | |
164 | /* Then search for matches in the second file */ | |
165 | while (count2--) | |
2477ab2e | 166 | insert_record(xpp, line2++, result, 2); |
92b7de93 JS |
167 | |
168 | return 0; | |
169 | } | |
170 | ||
171 | /* | |
172 | * Find the longest sequence with a smaller last element (meaning a smaller | |
173 | * line2, as we construct the sequence with entries ordered by line1). | |
174 | */ | |
175 | static int binary_search(struct entry **sequence, int longest, | |
176 | struct entry *entry) | |
177 | { | |
178 | int left = -1, right = longest; | |
179 | ||
180 | while (left + 1 < right) { | |
19716b21 | 181 | int middle = left + (right - left) / 2; |
92b7de93 JS |
182 | /* by construction, no two entries can be equal */ |
183 | if (sequence[middle]->line2 > entry->line2) | |
184 | right = middle; | |
185 | else | |
186 | left = middle; | |
187 | } | |
188 | /* return the index in "sequence", _not_ the sequence length */ | |
189 | return left; | |
190 | } | |
191 | ||
192 | /* | |
193 | * The idea is to start with the list of common unique lines sorted by | |
194 | * the order in file1. For each of these pairs, the longest (partial) | |
195 | * sequence whose last element's line2 is smaller is determined. | |
196 | * | |
197 | * For efficiency, the sequences are kept in a list containing exactly one | |
198 | * item per sequence length: the sequence with the smallest last | |
199 | * element (in terms of line2). | |
200 | */ | |
201 | static struct entry *find_longest_common_sequence(struct hashmap *map) | |
202 | { | |
203 | struct entry **sequence = xdl_malloc(map->nr * sizeof(struct entry *)); | |
204 | int longest = 0, i; | |
205 | struct entry *entry; | |
206 | ||
2477ab2e JT |
207 | /* |
208 | * If not -1, this entry in sequence must never be overridden. | |
209 | * Therefore, overriding entries before this has no effect, so | |
210 | * do not do that either. | |
211 | */ | |
212 | int anchor_i = -1; | |
213 | ||
92b7de93 JS |
214 | for (entry = map->first; entry; entry = entry->next) { |
215 | if (!entry->line2 || entry->line2 == NON_UNIQUE) | |
216 | continue; | |
217 | i = binary_search(sequence, longest, entry); | |
218 | entry->previous = i < 0 ? NULL : sequence[i]; | |
2477ab2e JT |
219 | ++i; |
220 | if (i <= anchor_i) | |
221 | continue; | |
222 | sequence[i] = entry; | |
223 | if (entry->anchor) { | |
224 | anchor_i = i; | |
225 | longest = anchor_i + 1; | |
226 | } else if (i == longest) { | |
92b7de93 | 227 | longest++; |
2477ab2e | 228 | } |
92b7de93 JS |
229 | } |
230 | ||
231 | /* No common unique lines were found */ | |
232 | if (!longest) { | |
233 | xdl_free(sequence); | |
234 | return NULL; | |
235 | } | |
236 | ||
237 | /* Iterate starting at the last element, adjusting the "next" members */ | |
238 | entry = sequence[longest - 1]; | |
239 | entry->next = NULL; | |
240 | while (entry->previous) { | |
241 | entry->previous->next = entry; | |
242 | entry = entry->previous; | |
243 | } | |
244 | xdl_free(sequence); | |
245 | return entry; | |
246 | } | |
247 | ||
248 | static int match(struct hashmap *map, int line1, int line2) | |
249 | { | |
250 | xrecord_t *record1 = map->env->xdf1.recs[line1 - 1]; | |
251 | xrecord_t *record2 = map->env->xdf2.recs[line2 - 1]; | |
204aa2d2 | 252 | return record1->ha == record2->ha; |
92b7de93 JS |
253 | } |
254 | ||
255 | static int patience_diff(mmfile_t *file1, mmfile_t *file2, | |
256 | xpparam_t const *xpp, xdfenv_t *env, | |
257 | int line1, int count1, int line2, int count2); | |
258 | ||
259 | static int walk_common_sequence(struct hashmap *map, struct entry *first, | |
260 | int line1, int count1, int line2, int count2) | |
261 | { | |
262 | int end1 = line1 + count1, end2 = line2 + count2; | |
263 | int next1, next2; | |
264 | ||
265 | for (;;) { | |
266 | /* Try to grow the line ranges of common lines */ | |
267 | if (first) { | |
268 | next1 = first->line1; | |
269 | next2 = first->line2; | |
270 | while (next1 > line1 && next2 > line2 && | |
271 | match(map, next1 - 1, next2 - 1)) { | |
272 | next1--; | |
273 | next2--; | |
274 | } | |
275 | } else { | |
276 | next1 = end1; | |
277 | next2 = end2; | |
278 | } | |
279 | while (line1 < next1 && line2 < next2 && | |
280 | match(map, line1, line2)) { | |
281 | line1++; | |
282 | line2++; | |
283 | } | |
284 | ||
285 | /* Recurse */ | |
286 | if (next1 > line1 || next2 > line2) { | |
92b7de93 JS |
287 | if (patience_diff(map->file1, map->file2, |
288 | map->xpp, map->env, | |
289 | line1, next1 - line1, | |
290 | line2, next2 - line2)) | |
291 | return -1; | |
292 | } | |
293 | ||
294 | if (!first) | |
295 | return 0; | |
296 | ||
297 | while (first->next && | |
298 | first->next->line1 == first->line1 + 1 && | |
299 | first->next->line2 == first->line2 + 1) | |
300 | first = first->next; | |
301 | ||
302 | line1 = first->line1 + 1; | |
303 | line2 = first->line2 + 1; | |
304 | ||
305 | first = first->next; | |
306 | } | |
307 | } | |
308 | ||
309 | static int fall_back_to_classic_diff(struct hashmap *map, | |
310 | int line1, int count1, int line2, int count2) | |
311 | { | |
92b7de93 | 312 | xpparam_t xpp; |
ec7967cf MK |
313 | |
314 | memset(&xpp, 0, sizeof(xpp)); | |
307ab20b | 315 | xpp.flags = map->xpp->flags & ~XDF_DIFF_ALGORITHM_MASK; |
92b7de93 | 316 | |
1d26b252 TRC |
317 | return xdl_fall_back_diff(map->env, &xpp, |
318 | line1, count1, line2, count2); | |
92b7de93 JS |
319 | } |
320 | ||
321 | /* | |
322 | * Recursively find the longest common sequence of unique lines, | |
323 | * and if none was found, ask xdl_do_diff() to do the job. | |
324 | * | |
325 | * This function assumes that env was prepared with xdl_prepare_env(). | |
326 | */ | |
327 | static int patience_diff(mmfile_t *file1, mmfile_t *file2, | |
328 | xpparam_t const *xpp, xdfenv_t *env, | |
329 | int line1, int count1, int line2, int count2) | |
330 | { | |
331 | struct hashmap map; | |
332 | struct entry *first; | |
333 | int result = 0; | |
334 | ||
335 | /* trivial case: one side is empty */ | |
336 | if (!count1) { | |
337 | while(count2--) | |
338 | env->xdf2.rchg[line2++ - 1] = 1; | |
339 | return 0; | |
340 | } else if (!count2) { | |
341 | while(count1--) | |
342 | env->xdf1.rchg[line1++ - 1] = 1; | |
343 | return 0; | |
344 | } | |
345 | ||
346 | memset(&map, 0, sizeof(map)); | |
347 | if (fill_hashmap(file1, file2, xpp, env, &map, | |
348 | line1, count1, line2, count2)) | |
349 | return -1; | |
350 | ||
351 | /* are there any matching lines at all? */ | |
352 | if (!map.has_matches) { | |
353 | while(count1--) | |
354 | env->xdf1.rchg[line1++ - 1] = 1; | |
355 | while(count2--) | |
356 | env->xdf2.rchg[line2++ - 1] = 1; | |
357 | xdl_free(map.entries); | |
358 | return 0; | |
359 | } | |
360 | ||
361 | first = find_longest_common_sequence(&map); | |
362 | if (first) | |
363 | result = walk_common_sequence(&map, first, | |
364 | line1, count1, line2, count2); | |
365 | else | |
366 | result = fall_back_to_classic_diff(&map, | |
367 | line1, count1, line2, count2); | |
368 | ||
369 | xdl_free(map.entries); | |
370 | return result; | |
371 | } | |
372 | ||
373 | int xdl_do_patience_diff(mmfile_t *file1, mmfile_t *file2, | |
374 | xpparam_t const *xpp, xdfenv_t *env) | |
375 | { | |
376 | if (xdl_prepare_env(file1, file2, xpp, env) < 0) | |
377 | return -1; | |
378 | ||
379 | /* environment is cleaned up in xdl_diff() */ | |
380 | return patience_diff(file1, file2, xpp, env, | |
381 | 1, env->xdf1.nrec, 1, env->xdf2.nrec); | |
382 | } |