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)(struct diff_filepair *p, struct diff_options *o, regex_t *regexp, kwset_t kws);
  12
  13 static void pickaxe(struct diff_queue_struct *q, struct diff_options *o,
  14                     regex_t *regexp, kwset_t kws, pickaxe_fn fn)
  15 {
  16         int i;
  17         struct diff_queue_struct outq;
  18
  19         DIFF_QUEUE_CLEAR(&outq);
  20
  21         if (o->pickaxe_opts & DIFF_PICKAXE_ALL) {
  22                 /* Showing the whole changeset if needle exists */
  23                 for (i = 0; i < q->nr; i++) {
  24                         struct diff_filepair *p = q->queue[i];
  25                         if (fn(p, o, regexp, kws))
  26                                 return; /* do not munge the queue */
  27                 }
  28
  29                 /*
  30                  * Otherwise we will clear the whole queue by copying
  31                  * the empty outq at the end of this function, but
  32                  * first clear the current entries in the queue.
  33                  */
  34                 for (i = 0; i < q->nr; i++)
  35                         diff_free_filepair(q->queue[i]);
  36         } else {
  37                 /* Showing only the filepairs that has the needle */
  38                 for (i = 0; i < q->nr; i++) {
  39                         struct diff_filepair *p = q->queue[i];
  40                         if (fn(p, o, regexp, kws))
  41                                 diff_q(&outq, p);
  42                         else
  43                                 diff_free_filepair(p);
  44                 }
  45         }
  46
  47         free(q->queue);
  48         *q = outq;
  49 }
  50
  51 struct diffgrep_cb {
  52         regex_t *regexp;
  53         int hit;
  54 };
  55
  56 static void diffgrep_consume(void *priv, char *line, unsigned long len)
  57 {
  58         struct diffgrep_cb *data = priv;
  59         regmatch_t regmatch;
  60         int hold;
  61
  62         if (line[0] != '+' && line[0] != '-')
  63                 return;
  64         if (data->hit)
  65                 /*
  66                  * NEEDSWORK: we should have a way to terminate the
  67                  * caller early.
  68                  */
  69                 return;
  70         /* Yuck -- line ought to be "const char *"! */
  71         hold = line[len];
  72         line[len] = '\0';
  73         data->hit = !regexec(data->regexp, line + 1, 1, &regmatch, 0);
  74         line[len] = hold;
  75 }
  76
  77 static int diff_grep(struct diff_filepair *p, struct diff_options *o,
  78                      regex_t *regexp, kwset_t kws)
  79 {
  80         regmatch_t regmatch;
  81         struct userdiff_driver *textconv_one = NULL;
  82         struct userdiff_driver *textconv_two = NULL;
  83         mmfile_t mf1, mf2;
  84         int hit;
  85
  86         if (diff_unmodified_pair(p))
  87                 return 0;
  88
  89         if (DIFF_OPT_TST(o, ALLOW_TEXTCONV)) {
  90                 textconv_one = get_textconv(p->one);
  91                 textconv_two = get_textconv(p->two);
  92         }
  93
  94         mf1.size = fill_textconv(textconv_one, p->one, &mf1.ptr);
  95         mf2.size = fill_textconv(textconv_two, p->two, &mf2.ptr);
  96
  97         if (!DIFF_FILE_VALID(p->one)) {
  98                 if (!DIFF_FILE_VALID(p->two))
  99                         return 0; /* ignore unmerged */
 100                 /* created "two" -- does it have what we are looking for? */
 101                 hit = !regexec(regexp, mf2.ptr, 1, &regmatch, 0);
 102         } else if (!DIFF_FILE_VALID(p->two)) {
 103                 /* removed "one" -- did it have what we are looking for? */
 104                 hit = !regexec(regexp, mf1.ptr, 1, &regmatch, 0);
 105         } else {
 106                 /*
 107                  * We have both sides; need to run textual diff and see if
 108                  * the pattern appears on added/deleted lines.
 109                  */
 110                 struct diffgrep_cb ecbdata;
 111                 xpparam_t xpp;
 112                 xdemitconf_t xecfg;
 113
 114                 memset(&xpp, 0, sizeof(xpp));
 115                 memset(&xecfg, 0, sizeof(xecfg));
 116                 ecbdata.regexp = regexp;
 117                 ecbdata.hit = 0;
 118                 xecfg.ctxlen = o->context;
 119                 xecfg.interhunkctxlen = o->interhunkcontext;
 120                 xdi_diff_outf(&mf1, &mf2, diffgrep_consume, &ecbdata,
 121                               &xpp, &xecfg);
 122                 hit = ecbdata.hit;
 123         }
 124         if (textconv_one)
 125                 free(mf1.ptr);
 126         if (textconv_two)
 127                 free(mf2.ptr);
 128         return hit;
 129 }
 130
 131 static void diffcore_pickaxe_grep(struct diff_options *o)
 132 {
 133         int err;
 134         regex_t regex;
 135         int cflags = REG_EXTENDED | REG_NEWLINE;
 136
 137         if (DIFF_OPT_TST(o, PICKAXE_IGNORE_CASE))
 138                 cflags |= REG_ICASE;
 139
 140         err = regcomp(&regex, o->pickaxe, cflags);
 141         if (err) {
 142                 char errbuf[1024];
 143                 regerror(err, &regex, errbuf, 1024);
 144                 regfree(&regex);
 145                 die("invalid log-grep regex: %s", errbuf);
 146         }
 147
 148         pickaxe(&diff_queued_diff, o, &regex, NULL, diff_grep);
 149
 150         regfree(&regex);
 151         return;
 152 }
 153
 154 static unsigned int contains(mmfile_t *mf, struct diff_options *o,
 155                              regex_t *regexp, kwset_t kws)
 156 {
 157         unsigned int cnt;
 158         unsigned long sz;
 159         const char *data;
 160
 161         sz = mf->size;
 162         data = mf->ptr;
 163         cnt = 0;
 164
 165         if (regexp) {
 166                 regmatch_t regmatch;
 167                 int flags = 0;
 168
 169                 assert(data[sz] == '\0');
 170                 while (*data && !regexec(regexp, data, 1, &regmatch, flags)) {
 171                         flags |= REG_NOTBOL;
 172                         data += regmatch.rm_eo;
 173                         if (*data && regmatch.rm_so == regmatch.rm_eo)
 174                                 data++;
 175                         cnt++;
 176                 }
 177
 178         } else { /* Classic exact string match */
 179                 while (sz) {
 180                         struct kwsmatch kwsm;
 181                         size_t offset = kwsexec(kws, data, sz, &kwsm);
 182                         const char *found;
 183                         if (offset == -1)
 184                                 break;
 185                         else
 186                                 found = data + offset;
 187                         sz -= found - data + kwsm.size[0];
 188                         data = found + kwsm.size[0];
 189                         cnt++;
 190                 }
 191         }
 192         return cnt;
 193 }
 194
 195 static int has_changes(struct diff_filepair *p, struct diff_options *o,
 196                        regex_t *regexp, kwset_t kws)
 197 {
 198         struct userdiff_driver *textconv_one = NULL;
 199         struct userdiff_driver *textconv_two = NULL;
 200         mmfile_t mf1, mf2;
 201         int ret;
 202
 203         if (!o->pickaxe[0])
 204                 return 0;
 205
 206         if (DIFF_OPT_TST(o, ALLOW_TEXTCONV)) {
 207                 textconv_one = get_textconv(p->one);
 208                 textconv_two = get_textconv(p->two);
 209         }
 210
 211         /*
 212          * If we have an unmodified pair, we know that the count will be the
 213          * same and don't even have to load the blobs. Unless textconv is in
 214          * play, _and_ we are using two different textconv filters (e.g.,
 215          * because a pair is an exact rename with different textconv attributes
 216          * for each side, which might generate different content).
 217          */
 218         if (textconv_one == textconv_two && diff_unmodified_pair(p))
 219                 return 0;
 220
 221         mf1.size = fill_textconv(textconv_one, p->one, &mf1.ptr);
 222         mf2.size = fill_textconv(textconv_two, p->two, &mf2.ptr);
 223
 224         if (!DIFF_FILE_VALID(p->one)) {
 225                 if (!DIFF_FILE_VALID(p->two))
 226                         ret = 0; /* ignore unmerged */
 227                 /* created */
 228                 ret = contains(&mf2, o, regexp, kws) != 0;
 229         }
 230         else if (!DIFF_FILE_VALID(p->two)) /* removed */
 231                 ret = contains(&mf1, o, regexp, kws) != 0;
 232         else
 233                 ret = contains(&mf1, o, regexp, kws) !=
 234                       contains(&mf2, o, regexp, kws);
 235
 236         if (textconv_one)
 237                 free(mf1.ptr);
 238         if (textconv_two)
 239                 free(mf2.ptr);
 240         diff_free_filespec_data(p->one);
 241         diff_free_filespec_data(p->two);
 242
 243         return ret;
 244 }
 245
 246 static void diffcore_pickaxe_count(struct diff_options *o)
 247 {
 248         const char *needle = o->pickaxe;
 249         int opts = o->pickaxe_opts;
 250         unsigned long len = strlen(needle);
 251         regex_t regex, *regexp = NULL;
 252         kwset_t kws = NULL;
 253
 254         if (opts & DIFF_PICKAXE_REGEX) {
 255                 int err;
 256                 err = regcomp(&regex, needle, REG_EXTENDED | REG_NEWLINE);
 257                 if (err) {
 258                         /* The POSIX.2 people are surely sick */
 259                         char errbuf[1024];
 260                         regerror(err, &regex, errbuf, 1024);
 261                         regfree(&regex);
 262                         die("invalid pickaxe regex: %s", errbuf);
 263                 }
 264                 regexp = &regex;
 265         } else {
 266                 kws = kwsalloc(DIFF_OPT_TST(o, PICKAXE_IGNORE_CASE)
 267                                ? tolower_trans_tbl : NULL);
 268                 kwsincr(kws, needle, len);
 269                 kwsprep(kws);
 270         }
 271
 272         pickaxe(&diff_queued_diff, o, regexp, kws, has_changes);
 273
 274         if (opts & DIFF_PICKAXE_REGEX)
 275                 regfree(&regex);
 276         else
 277                 kwsfree(kws);
 278         return;
 279 }
 280
 281 void diffcore_pickaxe(struct diff_options *o)
 282 {
 283         /* Might want to warn when both S and G are on; I don't care... */
 284         if (o->pickaxe_opts & DIFF_PICKAXE_KIND_G)
 285                 diffcore_pickaxe_grep(o);
 286         else
 287                 diffcore_pickaxe_count(o);
 288 }