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