diff-helper.c

   1 /*
   2  * Copyright (C) 2005 Junio C Hamano
   3  */
   4 #include <limits.h>
   5 #include "cache.h"
   6 #include "strbuf.h"
   7 #include "diff.h"
   8
   9 static int matches_pathspec(const char *name, const char **spec, int cnt)
  10 {
  11         int i;
  12         int namelen = strlen(name);
  13         for (i = 0; i < cnt; i++) {
  14                 int speclen = strlen(spec[i]);
  15                 if (! strncmp(spec[i], name, speclen) &&
  16                     speclen <= namelen &&
  17                     (name[speclen] == 0 ||
  18                      name[speclen] == '/'))
  19                         return 1;
  20         }
  21         return 0;
  22 }
  23
  24 static int detect_rename = 0;
  25
  26 /*
  27  * We do not detect circular renames.  Just hold created and deleted
  28  * entries and later attempt to match them up.  If they do not match,
  29  * then spit them out as deletes or creates as original.
  30  */
  31
  32 static struct diff_spec_hold {
  33         struct diff_spec_hold *next;
  34         struct diff_spec_hold *matched;
  35         struct diff_spec old, new;
  36         char path[1];
  37 } *createdfile, *deletedfile;
  38
  39 static void hold_spec(const char *path,
  40                       struct diff_spec *old, struct diff_spec *new)
  41 {
  42         struct diff_spec_hold **list, *elem;
  43         list = (! old->file_valid) ? &createdfile : &deletedfile;
  44         elem = xmalloc(sizeof(*elem) + strlen(path));
  45         strcpy(elem->path, path);
  46         elem->next = *list;
  47         *list = elem;
  48         elem->old = *old;
  49         elem->new = *new;
  50         elem->matched = 0;
  51 }
  52
  53 #define MINIMUM_SCORE 7000
  54 int estimate_similarity(struct diff_spec *one, struct diff_spec *two)
  55 {
  56         /* Return how similar they are, representing the score as an
  57          * integer between 0 and 10000.
  58          *
  59          * This version is very dumb and detects exact matches only.
  60          * Wnen Nico's delta stuff gets in, I'll use the delta
  61          * algorithm to estimate the similarity score in core.
  62          */
  63
  64         if (one->sha1_valid && two->sha1_valid &&
  65             !memcmp(one->blob_sha1, two->blob_sha1, 20))
  66                 return 10000;
  67         return 0;
  68 }
  69
  70 static void flush_renames(const char **spec, int cnt, int reverse)
  71 {
  72         struct diff_spec_hold *rename_src, *rename_dst, *elem;
  73         struct diff_spec_hold *leftover = NULL;
  74         int score, best_score;
  75
  76         while (createdfile) {
  77                 rename_dst = createdfile;
  78                 createdfile = rename_dst->next;
  79                 best_score = MINIMUM_SCORE;
  80                 rename_src = NULL;
  81                 for (elem = deletedfile;
  82                      elem;
  83                      elem = elem->next) {
  84                         if (elem->matched)
  85                                 continue;
  86                         score = estimate_similarity(&elem->old,
  87                                                     &rename_dst->new);
  88                         if (best_score < score) {
  89                                 rename_src = elem;
  90                                 best_score = score;
  91                         }
  92                 }
  93                 if (rename_src) {
  94                         rename_src->matched = rename_dst;
  95                         rename_dst->matched = rename_src;
  96
  97                         if (!cnt ||
  98                             matches_pathspec(rename_src->path, spec, cnt) ||
  99                             matches_pathspec(rename_dst->path, spec, cnt)) {
 100                                 if (reverse)
 101                                         run_external_diff(rename_dst->path,
 102                                                           rename_src->path,
 103                                                           &rename_dst->new,
 104                                                           &rename_src->old);
 105                                 else
 106                                         run_external_diff(rename_src->path,
 107                                                           rename_dst->path,
 108                                                           &rename_src->old,
 109                                                           &rename_dst->new);
 110                         }
 111                 }
 112                 else {
 113                         rename_dst->next = leftover;
 114                         leftover = rename_dst;
 115                 }
 116         }
 117
 118         /* unmatched deletes */
 119         for (elem = deletedfile; elem; elem = elem->next) {
 120                 if (elem->matched)
 121                         continue;
 122                 if (!cnt ||
 123                     matches_pathspec(elem->path, spec, cnt)) {
 124                         if (reverse)
 125                                 run_external_diff(elem->path, NULL,
 126                                                   &elem->new, &elem->old);
 127                         else
 128                                 run_external_diff(elem->path, NULL,
 129                                                   &elem->old, &elem->new);
 130                 }
 131         }
 132
 133         /* unmatched creates */
 134         for (elem = leftover; elem; elem = elem->next) {
 135                 if (!cnt ||
 136                     matches_pathspec(elem->path, spec, cnt)) {
 137                         if (reverse)
 138                                 run_external_diff(elem->path, NULL,
 139                                                   &elem->new, &elem->old);
 140                         else
 141                                 run_external_diff(elem->path, NULL,
 142                                                   &elem->old, &elem->new);
 143                 }
 144         }
 145 }
 146
 147 static int parse_oneside_change(const char *cp, struct diff_spec *one,
 148                                 char *path)
 149 {
 150         int ch;
 151
 152         one->file_valid = one->sha1_valid = 1;
 153         one->mode = 0;
 154         while ((ch = *cp) && '0' <= ch && ch <= '7') {
 155                 one->mode = (one->mode << 3) | (ch - '0');
 156                 cp++;
 157         }
 158
 159         if (strncmp(cp, "\tblob\t", 6))
 160                 return -1;
 161         cp += 6;
 162         if (get_sha1_hex(cp, one->blob_sha1))
 163                 return -1;
 164         cp += 40;
 165         if (*cp++ != '\t')
 166                 return -1;
 167         strcpy(path, cp);
 168         return 0;
 169 }
 170
 171 static int parse_diff_raw_output(const char *buf,
 172                                  const char **spec, int cnt, int reverse)
 173 {
 174         struct diff_spec old, new;
 175         char path[PATH_MAX];
 176         const char *cp = buf;
 177         int ch;
 178
 179         switch (*cp++) {
 180         case 'U':
 181                 if (!cnt || matches_pathspec(cp + 1, spec, cnt))
 182                         diff_unmerge(cp + 1);
 183                 return 0;
 184         case '+':
 185                 old.file_valid = 0;
 186                 parse_oneside_change(cp, &new, path);
 187                 break;
 188         case '-':
 189                 new.file_valid = 0;
 190                 parse_oneside_change(cp, &old, path);
 191                 break;
 192         case '*':
 193                 old.file_valid = old.sha1_valid =
 194                         new.file_valid = new.sha1_valid = 1;
 195                 old.mode = new.mode = 0;
 196                 while ((ch = *cp) && ('0' <= ch && ch <= '7')) {
 197                         old.mode = (old.mode << 3) | (ch - '0');
 198                         cp++;
 199                 }
 200                 if (strncmp(cp, "->", 2))
 201                         return -1;
 202                 cp += 2;
 203                 while ((ch = *cp) && ('0' <= ch && ch <= '7')) {
 204                         new.mode = (new.mode << 3) | (ch - '0');
 205                         cp++;
 206                 }
 207                 if (strncmp(cp, "\tblob\t", 6))
 208                         return -1;
 209                 cp += 6;
 210                 if (get_sha1_hex(cp, old.blob_sha1))
 211                         return -1;
 212                 cp += 40;
 213                 if (strncmp(cp, "->", 2))
 214                         return -1;
 215                 cp += 2;
 216                 if (get_sha1_hex(cp, new.blob_sha1))
 217                         return -1;
 218                 cp += 40;
 219                 if (*cp++ != '\t')
 220                         return -1;
 221                 strcpy(path, cp);
 222                 break;
 223         default:
 224                 return -1;
 225         }
 226
 227         if (detect_rename && old.file_valid != new.file_valid) {
 228                 /* hold these */
 229                 hold_spec(path, &old, &new);
 230                 return 0;
 231         }
 232
 233         if (!cnt || matches_pathspec(path, spec, cnt)) {
 234                 if (reverse)
 235                         run_external_diff(path, NULL, &new, &old);
 236                 else
 237                         run_external_diff(path, NULL, &old, &new);
 238         }
 239         return 0;
 240 }
 241
 242 static const char *diff_helper_usage =
 243         "git-diff-helper [-r] [-R] [-z] paths...";
 244
 245 int main(int ac, const char **av) {
 246         struct strbuf sb;
 247         int reverse = 0;
 248         int line_termination = '\n';
 249
 250         strbuf_init(&sb);
 251
 252         while (1 < ac && av[1][0] == '-') {
 253                 if (av[1][1] == 'R')
 254                         reverse = 1;
 255                 else if (av[1][1] == 'z')
 256                         line_termination = 0;
 257                 else if (av[1][1] == 'r')
 258                         detect_rename = 1;
 259                 else
 260                         usage(diff_helper_usage);
 261                 ac--; av++;
 262         }
 263         /* the remaining parameters are paths patterns */
 264
 265         while (1) {
 266                 int status;
 267                 read_line(&sb, stdin, line_termination);
 268                 if (sb.eof)
 269                         break;
 270                 status = parse_diff_raw_output(sb.buf, av+1, ac-1, reverse);
 271                 if (status) {
 272                         flush_renames(av+1, ac-1, reverse);
 273                         printf("%s%c", sb.buf, line_termination);
 274                 }
 275         }
 276
 277         flush_renames(av+1, ac-1, reverse);
 278         return 0;
 279 }