xdiff / xpatience.con commit Merge branch 'bc/hash-independent-tests-part-4' (fcdd9cc)
   1/*
   2 *  LibXDiff by Davide Libenzi ( File Differential Library )
   3 *  Copyright (C) 2003-2016 Davide Libenzi, Johannes E. Schindelin
   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
  16 *  License along with this library; if not, see
  17 *  <http://www.gnu.org/licenses/>.
  18 *
  19 *  Davide Libenzi <davidel@xmailserver.org>
  20 *
  21 */
  22#include "xinclude.h"
  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 */
  47struct 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;
  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;
  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
  77static 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
  87/* The argument "pass" is 1 for the first file, 2 for the second. */
  88static void insert_record(xpparam_t const *xpp, int line, struct hashmap *map,
  89                          int pass)
  90{
  91        xrecord_t **records = pass == 1 ?
  92                map->env->xdf1.recs : map->env->xdf2.recs;
  93        xrecord_t *record = records[line - 1], *other;
  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) {
 107                other = map->env->xdf1.recs[map->entries[index].line1 - 1];
 108                if (map->entries[index].hash != record->ha ||
 109                                !xdl_recmatch(record->ptr, record->size,
 110                                        other->ptr, other->size,
 111                                        map->xpp->flags)) {
 112                        if (++index >= map->alloc)
 113                                index = 0;
 114                        continue;
 115                }
 116                if (pass == 2)
 117                        map->has_matches = 1;
 118                if (pass == 1 || map->entries[index].line2)
 119                        map->entries[index].line2 = NON_UNIQUE;
 120                else
 121                        map->entries[index].line2 = line;
 122                return;
 123        }
 124        if (pass == 2)
 125                return;
 126        map->entries[index].line1 = line;
 127        map->entries[index].hash = record->ha;
 128        map->entries[index].anchor = is_anchor(xpp, map->env->xdf1.recs[line - 1]->ptr);
 129        if (!map->first)
 130                map->first = map->entries + index;
 131        if (map->last) {
 132                map->last->next = map->entries + index;
 133                map->entries[index].previous = map->last;
 134        }
 135        map->last = map->entries + index;
 136        map->nr++;
 137}
 138
 139/*
 140 * This function has to be called for each recursion into the inter-hunk
 141 * parts, as previously non-unique lines can become unique when being
 142 * restricted to a smaller part of the files.
 143 *
 144 * It is assumed that env has been prepared using xdl_prepare().
 145 */
 146static int fill_hashmap(mmfile_t *file1, mmfile_t *file2,
 147                xpparam_t const *xpp, xdfenv_t *env,
 148                struct hashmap *result,
 149                int line1, int count1, int line2, int count2)
 150{
 151        result->file1 = file1;
 152        result->file2 = file2;
 153        result->xpp = xpp;
 154        result->env = env;
 155
 156        /* We know exactly how large we want the hash map */
 157        result->alloc = count1 * 2;
 158        result->entries = (struct entry *)
 159                xdl_malloc(result->alloc * sizeof(struct entry));
 160        if (!result->entries)
 161                return -1;
 162        memset(result->entries, 0, result->alloc * sizeof(struct entry));
 163
 164        /* First, fill with entries from the first file */
 165        while (count1--)
 166                insert_record(xpp, line1++, result, 1);
 167
 168        /* Then search for matches in the second file */
 169        while (count2--)
 170                insert_record(xpp, line2++, result, 2);
 171
 172        return 0;
 173}
 174
 175/*
 176 * Find the longest sequence with a smaller last element (meaning a smaller
 177 * line2, as we construct the sequence with entries ordered by line1).
 178 */
 179static int binary_search(struct entry **sequence, int longest,
 180                struct entry *entry)
 181{
 182        int left = -1, right = longest;
 183
 184        while (left + 1 < right) {
 185                int middle = left + (right - left) / 2;
 186                /* by construction, no two entries can be equal */
 187                if (sequence[middle]->line2 > entry->line2)
 188                        right = middle;
 189                else
 190                        left = middle;
 191        }
 192        /* return the index in "sequence", _not_ the sequence length */
 193        return left;
 194}
 195
 196/*
 197 * The idea is to start with the list of common unique lines sorted by
 198 * the order in file1.  For each of these pairs, the longest (partial)
 199 * sequence whose last element's line2 is smaller is determined.
 200 *
 201 * For efficiency, the sequences are kept in a list containing exactly one
 202 * item per sequence length: the sequence with the smallest last
 203 * element (in terms of line2).
 204 */
 205static struct entry *find_longest_common_sequence(struct hashmap *map)
 206{
 207        struct entry **sequence = xdl_malloc(map->nr * sizeof(struct entry *));
 208        int longest = 0, i;
 209        struct entry *entry;
 210
 211        /*
 212         * If not -1, this entry in sequence must never be overridden.
 213         * Therefore, overriding entries before this has no effect, so
 214         * do not do that either.
 215         */
 216        int anchor_i = -1;
 217
 218        for (entry = map->first; entry; entry = entry->next) {
 219                if (!entry->line2 || entry->line2 == NON_UNIQUE)
 220                        continue;
 221                i = binary_search(sequence, longest, entry);
 222                entry->previous = i < 0 ? NULL : sequence[i];
 223                ++i;
 224                if (i <= anchor_i)
 225                        continue;
 226                sequence[i] = entry;
 227                if (entry->anchor) {
 228                        anchor_i = i;
 229                        longest = anchor_i + 1;
 230                } else if (i == longest) {
 231                        longest++;
 232                }
 233        }
 234
 235        /* No common unique lines were found */
 236        if (!longest) {
 237                xdl_free(sequence);
 238                return NULL;
 239        }
 240
 241        /* Iterate starting at the last element, adjusting the "next" members */
 242        entry = sequence[longest - 1];
 243        entry->next = NULL;
 244        while (entry->previous) {
 245                entry->previous->next = entry;
 246                entry = entry->previous;
 247        }
 248        xdl_free(sequence);
 249        return entry;
 250}
 251
 252static int match(struct hashmap *map, int line1, int line2)
 253{
 254        xrecord_t *record1 = map->env->xdf1.recs[line1 - 1];
 255        xrecord_t *record2 = map->env->xdf2.recs[line2 - 1];
 256        return xdl_recmatch(record1->ptr, record1->size,
 257                record2->ptr, record2->size, map->xpp->flags);
 258}
 259
 260static int patience_diff(mmfile_t *file1, mmfile_t *file2,
 261                xpparam_t const *xpp, xdfenv_t *env,
 262                int line1, int count1, int line2, int count2);
 263
 264static int walk_common_sequence(struct hashmap *map, struct entry *first,
 265                int line1, int count1, int line2, int count2)
 266{
 267        int end1 = line1 + count1, end2 = line2 + count2;
 268        int next1, next2;
 269
 270        for (;;) {
 271                /* Try to grow the line ranges of common lines */
 272                if (first) {
 273                        next1 = first->line1;
 274                        next2 = first->line2;
 275                        while (next1 > line1 && next2 > line2 &&
 276                                        match(map, next1 - 1, next2 - 1)) {
 277                                next1--;
 278                                next2--;
 279                        }
 280                } else {
 281                        next1 = end1;
 282                        next2 = end2;
 283                }
 284                while (line1 < next1 && line2 < next2 &&
 285                                match(map, line1, line2)) {
 286                        line1++;
 287                        line2++;
 288                }
 289
 290                /* Recurse */
 291                if (next1 > line1 || next2 > line2) {
 292                        struct hashmap submap;
 293
 294                        memset(&submap, 0, sizeof(submap));
 295                        if (patience_diff(map->file1, map->file2,
 296                                        map->xpp, map->env,
 297                                        line1, next1 - line1,
 298                                        line2, next2 - line2))
 299                                return -1;
 300                }
 301
 302                if (!first)
 303                        return 0;
 304
 305                while (first->next &&
 306                                first->next->line1 == first->line1 + 1 &&
 307                                first->next->line2 == first->line2 + 1)
 308                        first = first->next;
 309
 310                line1 = first->line1 + 1;
 311                line2 = first->line2 + 1;
 312
 313                first = first->next;
 314        }
 315}
 316
 317static int fall_back_to_classic_diff(struct hashmap *map,
 318                int line1, int count1, int line2, int count2)
 319{
 320        xpparam_t xpp;
 321        xpp.flags = map->xpp->flags & ~XDF_DIFF_ALGORITHM_MASK;
 322
 323        return xdl_fall_back_diff(map->env, &xpp,
 324                                  line1, count1, line2, count2);
 325}
 326
 327/*
 328 * Recursively find the longest common sequence of unique lines,
 329 * and if none was found, ask xdl_do_diff() to do the job.
 330 *
 331 * This function assumes that env was prepared with xdl_prepare_env().
 332 */
 333static int patience_diff(mmfile_t *file1, mmfile_t *file2,
 334                xpparam_t const *xpp, xdfenv_t *env,
 335                int line1, int count1, int line2, int count2)
 336{
 337        struct hashmap map;
 338        struct entry *first;
 339        int result = 0;
 340
 341        /* trivial case: one side is empty */
 342        if (!count1) {
 343                while(count2--)
 344                        env->xdf2.rchg[line2++ - 1] = 1;
 345                return 0;
 346        } else if (!count2) {
 347                while(count1--)
 348                        env->xdf1.rchg[line1++ - 1] = 1;
 349                return 0;
 350        }
 351
 352        memset(&map, 0, sizeof(map));
 353        if (fill_hashmap(file1, file2, xpp, env, &map,
 354                        line1, count1, line2, count2))
 355                return -1;
 356
 357        /* are there any matching lines at all? */
 358        if (!map.has_matches) {
 359                while(count1--)
 360                        env->xdf1.rchg[line1++ - 1] = 1;
 361                while(count2--)
 362                        env->xdf2.rchg[line2++ - 1] = 1;
 363                xdl_free(map.entries);
 364                return 0;
 365        }
 366
 367        first = find_longest_common_sequence(&map);
 368        if (first)
 369                result = walk_common_sequence(&map, first,
 370                        line1, count1, line2, count2);
 371        else
 372                result = fall_back_to_classic_diff(&map,
 373                        line1, count1, line2, count2);
 374
 375        xdl_free(map.entries);
 376        return result;
 377}
 378
 379int xdl_do_patience_diff(mmfile_t *file1, mmfile_t *file2,
 380                xpparam_t const *xpp, xdfenv_t *env)
 381{
 382        if (xdl_prepare_env(file1, file2, xpp, env) < 0)
 383                return -1;
 384
 385        /* environment is cleaned up in xdl_diff() */
 386        return patience_diff(file1, file2, xpp, env,
 387                        1, env->xdf1.nrec, 1, env->xdf2.nrec);
 388}