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 #include "commit.h"
  11 #include "quote.h"
  12
  13 typedef int (*pickaxe_fn)(mmfile_t *one, mmfile_t *two,
  14                           struct diff_options *o,
  15                           regex_t *regexp, kwset_t kws);
  16
  17 struct diffgrep_cb {
  18         regex_t *regexp;
  19         int hit;
  20 };
  21
  22 static int diffgrep_consume(void *priv, char *line, unsigned long len)
  23 {
  24         struct diffgrep_cb *data = priv;
  25         regmatch_t regmatch;
  26
  27         if (line[0] != '+' && line[0] != '-')
  28                 return 0;
  29         if (data->hit)
  30                 BUG("Already matched in diffgrep_consume! Broken xdiff_emit_line_fn?");
  31         if (!regexec_buf(data->regexp, line + 1, len - 1, 1,
  32                          &regmatch, 0)) {
  33                 data->hit = 1;
  34                 return 1;
  35         }
  36         return 0;
  37 }
  38
  39 static int diff_grep(mmfile_t *one, mmfile_t *two,
  40                      struct diff_options *o,
  41                      regex_t *regexp, kwset_t kws)
  42 {
  43         struct diffgrep_cb ecbdata;
  44         xpparam_t xpp;
  45         xdemitconf_t xecfg;
  46         int ret;
  47
  48         /*
  49          * We have both sides; need to run textual diff and see if
  50          * the pattern appears on added/deleted lines.
  51          */
  52         memset(&xpp, 0, sizeof(xpp));
  53         memset(&xecfg, 0, sizeof(xecfg));
  54         ecbdata.regexp = regexp;
  55         ecbdata.hit = 0;
  56         xecfg.flags = XDL_EMIT_NO_HUNK_HDR;
  57         xecfg.ctxlen = o->context;
  58         xecfg.interhunkctxlen = o->interhunkcontext;
  59
  60         /*
  61          * An xdiff error might be our "data->hit" from above. See the
  62          * comment for xdiff_emit_line_fn in xdiff-interface.h
  63          */
  64         ret = xdi_diff_outf(one, two, NULL, diffgrep_consume,
  65                             &ecbdata, &xpp, &xecfg);
  66         if (ecbdata.hit)
  67                 return 1;
  68         if (ret)
  69                 return ret;
  70         return 0;
  71 }
  72
  73 static unsigned int contains(mmfile_t *mf, regex_t *regexp, kwset_t kws,
  74                              unsigned int limit)
  75 {
  76         unsigned int cnt = 0;
  77         unsigned long sz = mf->size;
  78         const char *data = mf->ptr;
  79
  80         if (regexp) {
  81                 regmatch_t regmatch;
  82                 int flags = 0;
  83
  84                 while (sz &&
  85                        !regexec_buf(regexp, data, sz, 1, &regmatch, flags)) {
  86                         flags |= REG_NOTBOL;
  87                         data += regmatch.rm_eo;
  88                         sz -= regmatch.rm_eo;
  89                         if (sz && regmatch.rm_so == regmatch.rm_eo) {
  90                                 data++;
  91                                 sz--;
  92                         }
  93                         cnt++;
  94
  95                         if (limit && cnt == limit)
  96                                 return cnt;
  97                 }
  98
  99         } else { /* Classic exact string match */
 100                 while (sz) {
 101                         struct kwsmatch kwsm;
 102                         size_t offset = kwsexec(kws, data, sz, &kwsm);
 103                         if (offset == -1)
 104                                 break;
 105                         sz -= offset + kwsm.size[0];
 106                         data += offset + kwsm.size[0];
 107                         cnt++;
 108
 109                         if (limit && cnt == limit)
 110                                 return cnt;
 111                 }
 112         }
 113         return cnt;
 114 }
 115
 116 static int has_changes(mmfile_t *one, mmfile_t *two,
 117                        struct diff_options *o,
 118                        regex_t *regexp, kwset_t kws)
 119 {
 120         unsigned int c1 = one ? contains(one, regexp, kws, 0) : 0;
 121         unsigned int c2 = two ? contains(two, regexp, kws, c1 + 1) : 0;
 122         return c1 != c2;
 123 }
 124
 125 static int pickaxe_match(struct diff_filepair *p, struct diff_options *o,
 126                          regex_t *regexp, kwset_t kws, pickaxe_fn fn)
 127 {
 128         struct userdiff_driver *textconv_one = NULL;
 129         struct userdiff_driver *textconv_two = NULL;
 130         mmfile_t mf1, mf2;
 131         int ret;
 132
 133         /* ignore unmerged */
 134         if (!DIFF_FILE_VALID(p->one) && !DIFF_FILE_VALID(p->two))
 135                 return 0;
 136
 137         if (o->objfind) {
 138                 return  (DIFF_FILE_VALID(p->one) &&
 139                          oidset_contains(o->objfind, &p->one->oid)) ||
 140                         (DIFF_FILE_VALID(p->two) &&
 141                          oidset_contains(o->objfind, &p->two->oid));
 142         }
 143
 144         if (o->flags.allow_textconv) {
 145                 textconv_one = get_textconv(o->repo, p->one);
 146                 textconv_two = get_textconv(o->repo, p->two);
 147         }
 148
 149         /*
 150          * If we have an unmodified pair, we know that the count will be the
 151          * same and don't even have to load the blobs. Unless textconv is in
 152          * play, _and_ we are using two different textconv filters (e.g.,
 153          * because a pair is an exact rename with different textconv attributes
 154          * for each side, which might generate different content).
 155          */
 156         if (textconv_one == textconv_two && diff_unmodified_pair(p))
 157                 return 0;
 158
 159         if ((o->pickaxe_opts & DIFF_PICKAXE_KIND_G) &&
 160             !o->flags.text &&
 161             ((!textconv_one && diff_filespec_is_binary(o->repo, p->one)) ||
 162              (!textconv_two && diff_filespec_is_binary(o->repo, p->two))))
 163                 return 0;
 164
 165         mf1.size = fill_textconv(o->repo, textconv_one, p->one, &mf1.ptr);
 166         mf2.size = fill_textconv(o->repo, textconv_two, p->two, &mf2.ptr);
 167
 168         ret = fn(&mf1, &mf2, o, regexp, kws);
 169
 170         if (textconv_one)
 171                 free(mf1.ptr);
 172         if (textconv_two)
 173                 free(mf2.ptr);
 174         diff_free_filespec_data(p->one);
 175         diff_free_filespec_data(p->two);
 176
 177         return ret;
 178 }
 179
 180 static void pickaxe(struct diff_queue_struct *q, struct diff_options *o,
 181                     regex_t *regexp, kwset_t kws, pickaxe_fn fn)
 182 {
 183         int i;
 184         struct diff_queue_struct outq;
 185
 186         DIFF_QUEUE_CLEAR(&outq);
 187
 188         if (o->pickaxe_opts & DIFF_PICKAXE_ALL) {
 189                 /* Showing the whole changeset if needle exists */
 190                 for (i = 0; i < q->nr; i++) {
 191                         struct diff_filepair *p = q->queue[i];
 192                         if (pickaxe_match(p, o, regexp, kws, fn))
 193                                 return; /* do not munge the queue */
 194                 }
 195
 196                 /*
 197                  * Otherwise we will clear the whole queue by copying
 198                  * the empty outq at the end of this function, but
 199                  * first clear the current entries in the queue.
 200                  */
 201                 for (i = 0; i < q->nr; i++)
 202                         diff_free_filepair(q->queue[i]);
 203         } else {
 204                 /* Showing only the filepairs that has the needle */
 205                 for (i = 0; i < q->nr; i++) {
 206                         struct diff_filepair *p = q->queue[i];
 207                         if (pickaxe_match(p, o, regexp, kws, fn))
 208                                 diff_q(&outq, p);
 209                         else
 210                                 diff_free_filepair(p);
 211                 }
 212         }
 213
 214         free(q->queue);
 215         *q = outq;
 216 }
 217
 218 static void regcomp_or_die(regex_t *regex, const char *needle, int cflags)
 219 {
 220         int err = regcomp(regex, needle, cflags);
 221         if (err) {
 222                 /* The POSIX.2 people are surely sick */
 223                 char errbuf[1024];
 224                 regerror(err, regex, errbuf, 1024);
 225                 die("invalid regex: %s", errbuf);
 226         }
 227 }
 228
 229 void diffcore_pickaxe(struct diff_options *o)
 230 {
 231         const char *needle = o->pickaxe;
 232         int opts = o->pickaxe_opts;
 233         regex_t regex, *regexp = NULL;
 234         kwset_t kws = NULL;
 235         pickaxe_fn fn;
 236
 237         if (opts & ~DIFF_PICKAXE_KIND_OBJFIND &&
 238             (!needle || !*needle))
 239                 BUG("should have needle under -G or -S");
 240         if (opts & (DIFF_PICKAXE_REGEX | DIFF_PICKAXE_KIND_G)) {
 241                 int cflags = REG_EXTENDED | REG_NEWLINE;
 242                 if (o->pickaxe_opts & DIFF_PICKAXE_IGNORE_CASE)
 243                         cflags |= REG_ICASE;
 244                 regcomp_or_die(&regex, needle, cflags);
 245                 regexp = &regex;
 246
 247                 if (opts & DIFF_PICKAXE_KIND_G)
 248                         fn = diff_grep;
 249                 else if (opts & DIFF_PICKAXE_REGEX)
 250                         fn = has_changes;
 251                 else
 252                         /*
 253                          * We don't need to check the combination of
 254                          * -G and --pickaxe-regex, by the time we get
 255                          * here diff.c has already died if they're
 256                          * combined. See the usage tests in
 257                          * t4209-log-pickaxe.sh.
 258                          */
 259                         BUG("unreachable");
 260         } else if (opts & DIFF_PICKAXE_KIND_S) {
 261                 if (o->pickaxe_opts & DIFF_PICKAXE_IGNORE_CASE &&
 262                     has_non_ascii(needle)) {
 263                         struct strbuf sb = STRBUF_INIT;
 264                         int cflags = REG_NEWLINE | REG_ICASE;
 265
 266                         basic_regex_quote_buf(&sb, needle);
 267                         regcomp_or_die(&regex, sb.buf, cflags);
 268                         strbuf_release(&sb);
 269                         regexp = &regex;
 270                 } else {
 271                         kws = kwsalloc(o->pickaxe_opts & DIFF_PICKAXE_IGNORE_CASE
 272                                        ? tolower_trans_tbl : NULL);
 273                         kwsincr(kws, needle, strlen(needle));
 274                         kwsprep(kws);
 275                 }
 276                 fn = has_changes;
 277         } else if (opts & DIFF_PICKAXE_KIND_OBJFIND) {
 278                 fn = NULL;
 279         } else {
 280                 BUG("unknown pickaxe_opts flag");
 281         }
 282
 283         pickaxe(&diff_queued_diff, o, regexp, kws, fn);
 284
 285         if (regexp)
 286                 regfree(regexp);
 287         if (kws)
 288                 kwsfree(kws);
 289         return;
 290 }