diffcore-pickaxe.c

   1 /*
   2  * Copyright (C) 2005 Junio C Hamano
   3  * Copyright (C) 2010 Google Inc.
   4  */
   5 #include "cache.h"
   6 #include "diff.h"
   7 #include "diffcore.h"
   8 #include "xdiff-interface.h"
   9 #include "kwset.h"
  10
  11 typedef int (*pickaxe_fn)(mmfile_t *one, mmfile_t *two,
  12                           struct diff_options *o,
  13                           regex_t *regexp, kwset_t kws);
  14
  15 static int pickaxe_match(struct diff_filepair *p, struct diff_options *o,
  16                          regex_t *regexp, kwset_t kws, pickaxe_fn fn);
  17
  18 static void pickaxe(struct diff_queue_struct *q, struct diff_options *o,
  19                     regex_t *regexp, kwset_t kws, pickaxe_fn fn)
  20 {
  21         int i;
  22         struct diff_queue_struct outq;
  23
  24         DIFF_QUEUE_CLEAR(&outq);
  25
  26         if (o->pickaxe_opts & DIFF_PICKAXE_ALL) {
  27                 /* Showing the whole changeset if needle exists */
  28                 for (i = 0; i < q->nr; i++) {
  29                         struct diff_filepair *p = q->queue[i];
  30                         if (pickaxe_match(p, o, regexp, kws, fn))
  31                                 return; /* do not munge the queue */
  32                 }
  33
  34                 /*
  35                  * Otherwise we will clear the whole queue by copying
  36                  * the empty outq at the end of this function, but
  37                  * first clear the current entries in the queue.
  38                  */
  39                 for (i = 0; i < q->nr; i++)
  40                         diff_free_filepair(q->queue[i]);
  41         } else {
  42                 /* Showing only the filepairs that has the needle */
  43                 for (i = 0; i < q->nr; i++) {
  44                         struct diff_filepair *p = q->queue[i];
  45                         if (pickaxe_match(p, o, regexp, kws, fn))
  46                                 diff_q(&outq, p);
  47                         else
  48                                 diff_free_filepair(p);
  49                 }
  50         }
  51
  52         free(q->queue);
  53         *q = outq;
  54 }
  55
  56 struct diffgrep_cb {
  57         regex_t *regexp;
  58         int hit;
  59 };
  60
  61 static void diffgrep_consume(void *priv, char *line, unsigned long len)
  62 {
  63         struct diffgrep_cb *data = priv;
  64         regmatch_t regmatch;
  65         int hold;
  66
  67         if (line[0] != '+' && line[0] != '-')
  68                 return;
  69         if (data->hit)
  70                 /*
  71                  * NEEDSWORK: we should have a way to terminate the
  72                  * caller early.
  73                  */
  74                 return;
  75         /* Yuck -- line ought to be "const char *"! */
  76         hold = line[len];
  77         line[len] = '\0';
  78         data->hit = !regexec(data->regexp, line + 1, 1, &regmatch, 0);
  79         line[len] = hold;
  80 }
  81
  82 static int diff_grep(mmfile_t *one, mmfile_t *two,
  83                      struct diff_options *o,
  84                      regex_t *regexp, kwset_t kws)
  85 {
  86         regmatch_t regmatch;
  87         struct diffgrep_cb ecbdata;
  88         xpparam_t xpp;
  89         xdemitconf_t xecfg;
  90
  91         if (!one)
  92                 return !regexec(regexp, two->ptr, 1, &regmatch, 0);
  93         if (!two)
  94                 return !regexec(regexp, one->ptr, 1, &regmatch, 0);
  95
  96         /*
  97          * We have both sides; need to run textual diff and see if
  98          * the pattern appears on added/deleted lines.
  99          */
 100         memset(&xpp, 0, sizeof(xpp));
 101         memset(&xecfg, 0, sizeof(xecfg));
 102         ecbdata.regexp = regexp;
 103         ecbdata.hit = 0;
 104         xecfg.ctxlen = o->context;
 105         xecfg.interhunkctxlen = o->interhunkcontext;
 106         xdi_diff_outf(one, two, diffgrep_consume, &ecbdata,
 107                       &xpp, &xecfg);
 108         return ecbdata.hit;
 109 }
 110
 111 static void diffcore_pickaxe_grep(struct diff_options *o)
 112 {
 113         int err;
 114         regex_t regex;
 115         int cflags = REG_EXTENDED | REG_NEWLINE;
 116
 117         if (DIFF_OPT_TST(o, PICKAXE_IGNORE_CASE))
 118                 cflags |= REG_ICASE;
 119
 120         err = regcomp(&regex, o->pickaxe, cflags);
 121         if (err) {
 122                 char errbuf[1024];
 123                 regerror(err, &regex, errbuf, 1024);
 124                 regfree(&regex);
 125                 die("invalid regex: %s", errbuf);
 126         }
 127
 128         pickaxe(&diff_queued_diff, o, &regex, NULL, diff_grep);
 129
 130         regfree(&regex);
 131         return;
 132 }
 133
 134 static unsigned int contains(mmfile_t *mf, regex_t *regexp, kwset_t kws)
 135 {
 136         unsigned int cnt;
 137         unsigned long sz;
 138         const char *data;
 139
 140         sz = mf->size;
 141         data = mf->ptr;
 142         cnt = 0;
 143
 144         if (regexp) {
 145                 regmatch_t regmatch;
 146                 int flags = 0;
 147
 148                 assert(data[sz] == '\0');
 149                 while (*data && !regexec(regexp, data, 1, &regmatch, flags)) {
 150                         flags |= REG_NOTBOL;
 151                         data += regmatch.rm_eo;
 152                         if (*data && regmatch.rm_so == regmatch.rm_eo)
 153                                 data++;
 154                         cnt++;
 155                 }
 156
 157         } else { /* Classic exact string match */
 158                 while (sz) {
 159                         struct kwsmatch kwsm;
 160                         size_t offset = kwsexec(kws, data, sz, &kwsm);
 161                         const char *found;
 162                         if (offset == -1)
 163                                 break;
 164                         else
 165                                 found = data + offset;
 166                         sz -= found - data + kwsm.size[0];
 167                         data = found + kwsm.size[0];
 168                         cnt++;
 169                 }
 170         }
 171         return cnt;
 172 }
 173
 174 static int has_changes(mmfile_t *one, mmfile_t *two,
 175                        struct diff_options *o,
 176                        regex_t *regexp, kwset_t kws)
 177 {
 178         unsigned int one_contains = one ? contains(one, regexp, kws) : 0;
 179         unsigned int two_contains = two ? contains(two, regexp, kws) : 0;
 180         return one_contains != two_contains;
 181 }
 182
 183 static int pickaxe_match(struct diff_filepair *p, struct diff_options *o,
 184                          regex_t *regexp, kwset_t kws, pickaxe_fn fn)
 185 {
 186         struct userdiff_driver *textconv_one = NULL;
 187         struct userdiff_driver *textconv_two = NULL;
 188         mmfile_t mf1, mf2;
 189         int ret;
 190
 191         if (!o->pickaxe[0])
 192                 return 0;
 193
 194         /* ignore unmerged */
 195         if (!DIFF_FILE_VALID(p->one) && !DIFF_FILE_VALID(p->two))
 196                 return 0;
 197
 198         if (DIFF_OPT_TST(o, ALLOW_TEXTCONV)) {
 199                 textconv_one = get_textconv(p->one);
 200                 textconv_two = get_textconv(p->two);
 201         }
 202
 203         /*
 204          * If we have an unmodified pair, we know that the count will be the
 205          * same and don't even have to load the blobs. Unless textconv is in
 206          * play, _and_ we are using two different textconv filters (e.g.,
 207          * because a pair is an exact rename with different textconv attributes
 208          * for each side, which might generate different content).
 209          */
 210         if (textconv_one == textconv_two && diff_unmodified_pair(p))
 211                 return 0;
 212
 213         mf1.size = fill_textconv(textconv_one, p->one, &mf1.ptr);
 214         mf2.size = fill_textconv(textconv_two, p->two, &mf2.ptr);
 215
 216         ret = fn(DIFF_FILE_VALID(p->one) ? &mf1 : NULL,
 217                  DIFF_FILE_VALID(p->two) ? &mf2 : NULL,
 218                  o, regexp, kws);
 219
 220         if (textconv_one)
 221                 free(mf1.ptr);
 222         if (textconv_two)
 223                 free(mf2.ptr);
 224         diff_free_filespec_data(p->one);
 225         diff_free_filespec_data(p->two);
 226
 227         return ret;
 228 }
 229
 230 static void diffcore_pickaxe_count(struct diff_options *o)
 231 {
 232         const char *needle = o->pickaxe;
 233         int opts = o->pickaxe_opts;
 234         unsigned long len = strlen(needle);
 235         regex_t regex, *regexp = NULL;
 236         kwset_t kws = NULL;
 237
 238         if (opts & DIFF_PICKAXE_REGEX) {
 239                 int err;
 240                 err = regcomp(&regex, needle, REG_EXTENDED | REG_NEWLINE);
 241                 if (err) {
 242                         /* The POSIX.2 people are surely sick */
 243                         char errbuf[1024];
 244                         regerror(err, &regex, errbuf, 1024);
 245                         regfree(&regex);
 246                         die("invalid regex: %s", errbuf);
 247                 }
 248                 regexp = &regex;
 249         } else {
 250                 kws = kwsalloc(DIFF_OPT_TST(o, PICKAXE_IGNORE_CASE)
 251                                ? tolower_trans_tbl : NULL);
 252                 kwsincr(kws, needle, len);
 253                 kwsprep(kws);
 254         }
 255
 256         pickaxe(&diff_queued_diff, o, regexp, kws, has_changes);
 257
 258         if (opts & DIFF_PICKAXE_REGEX)
 259                 regfree(&regex);
 260         else
 261                 kwsfree(kws);
 262         return;
 263 }
 264
 265 void diffcore_pickaxe(struct diff_options *o)
 266 {
 267         /* Might want to warn when both S and G are on; I don't care... */
 268         if (o->pickaxe_opts & DIFF_PICKAXE_KIND_G)
 269                 diffcore_pickaxe_grep(o);
 270         else
 271                 diffcore_pickaxe_count(o);
 272 }