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