xdiff/xpatience.c

   1 /*
   2  *  LibXDiff by Davide Libenzi ( File Differential Library )
   3  *  Copyright (C) 2003-2009 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, write to the Free Software
  17  *  Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA  02111-1307  USA
  18  *
  19  *  Davide Libenzi <davidel@xmailserver.org>
  20  *
  21  */
  22 #include "xinclude.h"
  23 #include "xtypes.h"
  24 #include "xdiff.h"
  25
  26 /*
  27  * The basic idea of patience diff is to find lines that are unique in
  28  * both files.  These are intuitively the ones that we want to see as
  29  * common lines.
  30  *
  31  * The maximal ordered sequence of such line pairs (where ordered means
  32  * that the order in the sequence agrees with the order of the lines in
  33  * both files) naturally defines an initial set of common lines.
  34  *
  35  * Now, the algorithm tries to extend the set of common lines by growing
  36  * the line ranges where the files have identical lines.
  37  *
  38  * Between those common lines, the patience diff algorithm is applied
  39  * recursively, until no unique line pairs can be found; these line ranges
  40  * are handled by the well-known Myers algorithm.
  41  */
  42
  43 #define NON_UNIQUE ULONG_MAX
  44
  45 /*
  46  * This is a hash mapping from line hash to line numbers in the first and
  47  * second file.
  48  */
  49 struct hashmap {
  50         int nr, alloc;
  51         struct entry {
  52                 unsigned long hash;
  53                 /*
  54                  * 0 = unused entry, 1 = first line, 2 = second, etc.
  55                  * line2 is NON_UNIQUE if the line is not unique
  56                  * in either the first or the second file.
  57                  */
  58                 unsigned long line1, line2;
  59                 /*
  60                  * "next" & "previous" are used for the longest common
  61                  * sequence;
  62                  * initially, "next" reflects only the order in file1.
  63                  */
  64                 struct entry *next, *previous;
  65         } *entries, *first, *last;
  66         /* were common records found? */
  67         unsigned long has_matches;
  68         mmfile_t *file1, *file2;
  69         xdfenv_t *env;
  70         xpparam_t const *xpp;
  71 };
  72
  73 /* The argument "pass" is 1 for the first file, 2 for the second. */
  74 static void insert_record(int line, struct hashmap *map, int pass)
  75 {
  76         xrecord_t **records = pass == 1 ?
  77                 map->env->xdf1.recs : map->env->xdf2.recs;
  78         xrecord_t *record = records[line - 1], *other;
  79         /*
  80          * After xdl_prepare_env() (or more precisely, due to
  81          * xdl_classify_record()), the "ha" member of the records (AKA lines)
  82          * is _not_ the hash anymore, but a linearized version of it.  In
  83          * other words, the "ha" member is guaranteed to start with 0 and
  84          * the second record's ha can only be 0 or 1, etc.
  85          *
  86          * So we multiply ha by 2 in the hope that the hashing was
  87          * "unique enough".
  88          */
  89         int index = (int)((record->ha << 1) % map->alloc);
  90
  91         while (map->entries[index].line1) {
  92                 other = map->env->xdf1.recs[map->entries[index].line1 - 1];
  93                 if (map->entries[index].hash != record->ha ||
  94                                 !xdl_recmatch(record->ptr, record->size,
  95                                         other->ptr, other->size,
  96                                         map->xpp->flags)) {
  97                         if (++index >= map->alloc)
  98                                 index = 0;
  99                         continue;
 100                 }
 101                 if (pass == 2)
 102                         map->has_matches = 1;
 103                 if (pass == 1 || map->entries[index].line2)
 104                         map->entries[index].line2 = NON_UNIQUE;
 105                 else
 106                         map->entries[index].line2 = line;
 107                 return;
 108         }
 109         if (pass == 2)
 110                 return;
 111         map->entries[index].line1 = line;
 112         map->entries[index].hash = record->ha;
 113         if (!map->first)
 114                 map->first = map->entries + index;
 115         if (map->last) {
 116                 map->last->next = map->entries + index;
 117                 map->entries[index].previous = map->last;
 118         }
 119         map->last = map->entries + index;
 120         map->nr++;
 121 }
 122
 123 /*
 124  * This function has to be called for each recursion into the inter-hunk
 125  * parts, as previously non-unique lines can become unique when being
 126  * restricted to a smaller part of the files.
 127  *
 128  * It is assumed that env has been prepared using xdl_prepare().
 129  */
 130 static int fill_hashmap(mmfile_t *file1, mmfile_t *file2,
 131                 xpparam_t const *xpp, xdfenv_t *env,
 132                 struct hashmap *result,
 133                 int line1, int count1, int line2, int count2)
 134 {
 135         result->file1 = file1;
 136         result->file2 = file2;
 137         result->xpp = xpp;
 138         result->env = env;
 139
 140         /* We know exactly how large we want the hash map */
 141         result->alloc = count1 * 2;
 142         result->entries = (struct entry *)
 143                 xdl_malloc(result->alloc * sizeof(struct entry));
 144         if (!result->entries)
 145                 return -1;
 146         memset(result->entries, 0, result->alloc * sizeof(struct entry));
 147
 148         /* First, fill with entries from the first file */
 149         while (count1--)
 150                 insert_record(line1++, result, 1);
 151
 152         /* Then search for matches in the second file */
 153         while (count2--)
 154                 insert_record(line2++, result, 2);
 155
 156         return 0;
 157 }
 158
 159 /*
 160  * Find the longest sequence with a smaller last element (meaning a smaller
 161  * line2, as we construct the sequence with entries ordered by line1).
 162  */
 163 static int binary_search(struct entry **sequence, int longest,
 164                 struct entry *entry)
 165 {
 166         int left = -1, right = longest;
 167
 168         while (left + 1 < right) {
 169                 int middle = (left + right) / 2;
 170                 /* by construction, no two entries can be equal */
 171                 if (sequence[middle]->line2 > entry->line2)
 172                         right = middle;
 173                 else
 174                         left = middle;
 175         }
 176         /* return the index in "sequence", _not_ the sequence length */
 177         return left;
 178 }
 179
 180 /*
 181  * The idea is to start with the list of common unique lines sorted by
 182  * the order in file1.  For each of these pairs, the longest (partial)
 183  * sequence whose last element's line2 is smaller is determined.
 184  *
 185  * For efficiency, the sequences are kept in a list containing exactly one
 186  * item per sequence length: the sequence with the smallest last
 187  * element (in terms of line2).
 188  */
 189 static struct entry *find_longest_common_sequence(struct hashmap *map)
 190 {
 191         struct entry **sequence = xdl_malloc(map->nr * sizeof(struct entry *));
 192         int longest = 0, i;
 193         struct entry *entry;
 194
 195         for (entry = map->first; entry; entry = entry->next) {
 196                 if (!entry->line2 || entry->line2 == NON_UNIQUE)
 197                         continue;
 198                 i = binary_search(sequence, longest, entry);
 199                 entry->previous = i < 0 ? NULL : sequence[i];
 200                 sequence[++i] = entry;
 201                 if (i == longest)
 202                         longest++;
 203         }
 204
 205         /* No common unique lines were found */
 206         if (!longest) {
 207                 xdl_free(sequence);
 208                 return NULL;
 209         }
 210
 211         /* Iterate starting at the last element, adjusting the "next" members */
 212         entry = sequence[longest - 1];
 213         entry->next = NULL;
 214         while (entry->previous) {
 215                 entry->previous->next = entry;
 216                 entry = entry->previous;
 217         }
 218         xdl_free(sequence);
 219         return entry;
 220 }
 221
 222 static int match(struct hashmap *map, int line1, int line2)
 223 {
 224         xrecord_t *record1 = map->env->xdf1.recs[line1 - 1];
 225         xrecord_t *record2 = map->env->xdf2.recs[line2 - 1];
 226         return xdl_recmatch(record1->ptr, record1->size,
 227                 record2->ptr, record2->size, map->xpp->flags);
 228 }
 229
 230 static int patience_diff(mmfile_t *file1, mmfile_t *file2,
 231                 xpparam_t const *xpp, xdfenv_t *env,
 232                 int line1, int count1, int line2, int count2);
 233
 234 static int walk_common_sequence(struct hashmap *map, struct entry *first,
 235                 int line1, int count1, int line2, int count2)
 236 {
 237         int end1 = line1 + count1, end2 = line2 + count2;
 238         int next1, next2;
 239
 240         for (;;) {
 241                 /* Try to grow the line ranges of common lines */
 242                 if (first) {
 243                         next1 = first->line1;
 244                         next2 = first->line2;
 245                         while (next1 > line1 && next2 > line2 &&
 246                                         match(map, next1 - 1, next2 - 1)) {
 247                                 next1--;
 248                                 next2--;
 249                         }
 250                 } else {
 251                         next1 = end1;
 252                         next2 = end2;
 253                 }
 254                 while (line1 < next1 && line2 < next2 &&
 255                                 match(map, line1, line2)) {
 256                         line1++;
 257                         line2++;
 258                 }
 259
 260                 /* Recurse */
 261                 if (next1 > line1 || next2 > line2) {
 262                         struct hashmap submap;
 263
 264                         memset(&submap, 0, sizeof(submap));
 265                         if (patience_diff(map->file1, map->file2,
 266                                         map->xpp, map->env,
 267                                         line1, next1 - line1,
 268                                         line2, next2 - line2))
 269                                 return -1;
 270                 }
 271
 272                 if (!first)
 273                         return 0;
 274
 275                 while (first->next &&
 276                                 first->next->line1 == first->line1 + 1 &&
 277                                 first->next->line2 == first->line2 + 1)
 278                         first = first->next;
 279
 280                 line1 = first->line1 + 1;
 281                 line2 = first->line2 + 1;
 282
 283                 first = first->next;
 284         }
 285 }
 286
 287 static int fall_back_to_classic_diff(struct hashmap *map,
 288                 int line1, int count1, int line2, int count2)
 289 {
 290         xpparam_t xpp;
 291         xpp.flags = map->xpp->flags & ~XDF_DIFF_ALGORITHM_MASK;
 292
 293         return xdl_fall_back_diff(map->env, &xpp,
 294                                   line1, count1, line2, count2);
 295 }
 296
 297 /*
 298  * Recursively find the longest common sequence of unique lines,
 299  * and if none was found, ask xdl_do_diff() to do the job.
 300  *
 301  * This function assumes that env was prepared with xdl_prepare_env().
 302  */
 303 static int patience_diff(mmfile_t *file1, mmfile_t *file2,
 304                 xpparam_t const *xpp, xdfenv_t *env,
 305                 int line1, int count1, int line2, int count2)
 306 {
 307         struct hashmap map;
 308         struct entry *first;
 309         int result = 0;
 310
 311         /* trivial case: one side is empty */
 312         if (!count1) {
 313                 while(count2--)
 314                         env->xdf2.rchg[line2++ - 1] = 1;
 315                 return 0;
 316         } else if (!count2) {
 317                 while(count1--)
 318                         env->xdf1.rchg[line1++ - 1] = 1;
 319                 return 0;
 320         }
 321
 322         memset(&map, 0, sizeof(map));
 323         if (fill_hashmap(file1, file2, xpp, env, &map,
 324                         line1, count1, line2, count2))
 325                 return -1;
 326
 327         /* are there any matching lines at all? */
 328         if (!map.has_matches) {
 329                 while(count1--)
 330                         env->xdf1.rchg[line1++ - 1] = 1;
 331                 while(count2--)
 332                         env->xdf2.rchg[line2++ - 1] = 1;
 333                 xdl_free(map.entries);
 334                 return 0;
 335         }
 336
 337         first = find_longest_common_sequence(&map);
 338         if (first)
 339                 result = walk_common_sequence(&map, first,
 340                         line1, count1, line2, count2);
 341         else
 342                 result = fall_back_to_classic_diff(&map,
 343                         line1, count1, line2, count2);
 344
 345         xdl_free(map.entries);
 346         return result;
 347 }
 348
 349 int xdl_do_patience_diff(mmfile_t *file1, mmfile_t *file2,
 350                 xpparam_t const *xpp, xdfenv_t *env)
 351 {
 352         if (xdl_prepare_env(file1, file2, xpp, env) < 0)
 353                 return -1;
 354
 355         /* environment is cleaned up in xdl_diff() */
 356         return patience_diff(file1, file2, xpp, env,
 357                         1, env->xdf1.nrec, 1, env->xdf2.nrec);
 358 }