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 void fill_one(struct diff_filespec *one,
  78                      mmfile_t *mf, struct userdiff_driver *textconv)
  79 {
  80         if (DIFF_FILE_VALID(one)) {
  81                 mf->size = fill_textconv(textconv, one, &mf->ptr);
  82         } else {
  83                 memset(mf, 0, sizeof(*mf));
  84         }
  85 }
  86
  87 static int diff_grep(struct diff_filepair *p, struct diff_options *o,
  88                      regex_t *regexp, kwset_t kws)
  89 {
  90         regmatch_t regmatch;
  91         struct userdiff_driver *textconv_one = NULL;
  92         struct userdiff_driver *textconv_two = NULL;
  93         mmfile_t mf1, mf2;
  94         int hit;
  95
  96         if (diff_unmodified_pair(p))
  97                 return 0;
  98
  99         textconv_one = get_textconv(p->one);
 100         textconv_two = get_textconv(p->two);
 101
 102         fill_one(p->one, &mf1, textconv_one);
 103         fill_one(p->two, &mf2, textconv_two);
 104
 105         if (!mf1.ptr) {
 106                 if (!mf2.ptr)
 107                         return 0; /* ignore unmerged */
 108                 /* created "two" -- does it have what we are looking for? */
 109                 hit = !regexec(regexp, mf2.ptr, 1, &regmatch, 0);
 110         } else if (!mf2.ptr) {
 111                 /* removed "one" -- did it have what we are looking for? */
 112                 hit = !regexec(regexp, mf1.ptr, 1, &regmatch, 0);
 113         } else {
 114                 /*
 115                  * We have both sides; need to run textual diff and see if
 116                  * the pattern appears on added/deleted lines.
 117                  */
 118                 struct diffgrep_cb ecbdata;
 119                 xpparam_t xpp;
 120                 xdemitconf_t xecfg;
 121
 122                 memset(&xpp, 0, sizeof(xpp));
 123                 memset(&xecfg, 0, sizeof(xecfg));
 124                 ecbdata.regexp = regexp;
 125                 ecbdata.hit = 0;
 126                 xecfg.ctxlen = o->context;
 127                 xecfg.interhunkctxlen = o->interhunkcontext;
 128                 xdi_diff_outf(&mf1, &mf2, diffgrep_consume, &ecbdata,
 129                               &xpp, &xecfg);
 130                 hit = ecbdata.hit;
 131         }
 132         if (textconv_one)
 133                 free(mf1.ptr);
 134         if (textconv_two)
 135                 free(mf2.ptr);
 136         return hit;
 137 }
 138
 139 static void diffcore_pickaxe_grep(struct diff_options *o)
 140 {
 141         int err;
 142         regex_t regex;
 143         int cflags = REG_EXTENDED | REG_NEWLINE;
 144
 145         if (DIFF_OPT_TST(o, PICKAXE_IGNORE_CASE))
 146                 cflags |= REG_ICASE;
 147
 148         err = regcomp(&regex, o->pickaxe, cflags);
 149         if (err) {
 150                 char errbuf[1024];
 151                 regerror(err, &regex, errbuf, 1024);
 152                 regfree(&regex);
 153                 die("invalid log-grep regex: %s", errbuf);
 154         }
 155
 156         pickaxe(&diff_queued_diff, o, &regex, NULL, diff_grep);
 157
 158         regfree(&regex);
 159         return;
 160 }
 161
 162 static unsigned int contains(mmfile_t *mf, struct diff_options *o,
 163                              regex_t *regexp, kwset_t kws)
 164 {
 165         unsigned int cnt;
 166         unsigned long sz;
 167         const char *data;
 168
 169         sz = mf->size;
 170         data = mf->ptr;
 171         cnt = 0;
 172
 173         if (regexp) {
 174                 regmatch_t regmatch;
 175                 int flags = 0;
 176
 177                 assert(data[sz] == '\0');
 178                 while (*data && !regexec(regexp, data, 1, &regmatch, flags)) {
 179                         flags |= REG_NOTBOL;
 180                         data += regmatch.rm_eo;
 181                         if (*data && regmatch.rm_so == regmatch.rm_eo)
 182                                 data++;
 183                         cnt++;
 184                 }
 185
 186         } else { /* Classic exact string match */
 187                 while (sz) {
 188                         struct kwsmatch kwsm;
 189                         size_t offset = kwsexec(kws, data, sz, &kwsm);
 190                         const char *found;
 191                         if (offset == -1)
 192                                 break;
 193                         else
 194                                 found = data + offset;
 195                         sz -= found - data + kwsm.size[0];
 196                         data = found + kwsm.size[0];
 197                         cnt++;
 198                 }
 199         }
 200         return cnt;
 201 }
 202
 203 static int has_changes(struct diff_filepair *p, struct diff_options *o,
 204                        regex_t *regexp, kwset_t kws)
 205 {
 206         struct userdiff_driver *textconv_one = NULL;
 207         struct userdiff_driver *textconv_two = NULL;
 208         mmfile_t mf1, mf2;
 209         int ret;
 210
 211         if (!o->pickaxe[0])
 212                 return 0;
 213
 214         textconv_one = get_textconv(p->one);
 215         textconv_two = get_textconv(p->two);
 216
 217         /*
 218          * If we have an unmodified pair, we know that the count will be the
 219          * same and don't even have to load the blobs. Unless textconv is in
 220          * play, _and_ we are using two different textconv filters (e.g.,
 221          * because a pair is an exact rename with different textconv attributes
 222          * for each side, which might generate different content).
 223          */
 224         if (textconv_one == textconv_two && diff_unmodified_pair(p))
 225                 return 0;
 226
 227         fill_one(p->one, &mf1, textconv_one);
 228         fill_one(p->two, &mf2, textconv_two);
 229
 230         if (!mf1.ptr) {
 231                 if (!mf2.ptr)
 232                         ret = 0; /* ignore unmerged */
 233                 /* created */
 234                 ret = contains(&mf2, o, regexp, kws) != 0;
 235         }
 236         else if (!mf2.ptr) /* removed */
 237                 ret = contains(&mf1, o, regexp, kws) != 0;
 238         else
 239                 ret = contains(&mf1, o, regexp, kws) !=
 240                       contains(&mf2, o, regexp, kws);
 241
 242         if (textconv_one)
 243                 free(mf1.ptr);
 244         if (textconv_two)
 245                 free(mf2.ptr);
 246         diff_free_filespec_data(p->one);
 247         diff_free_filespec_data(p->two);
 248
 249         return ret;
 250 }
 251
 252 static void diffcore_pickaxe_count(struct diff_options *o)
 253 {
 254         const char *needle = o->pickaxe;
 255         int opts = o->pickaxe_opts;
 256         unsigned long len = strlen(needle);
 257         regex_t regex, *regexp = NULL;
 258         kwset_t kws = NULL;
 259
 260         if (opts & DIFF_PICKAXE_REGEX) {
 261                 int err;
 262                 err = regcomp(&regex, needle, REG_EXTENDED | REG_NEWLINE);
 263                 if (err) {
 264                         /* The POSIX.2 people are surely sick */
 265                         char errbuf[1024];
 266                         regerror(err, &regex, errbuf, 1024);
 267                         regfree(&regex);
 268                         die("invalid pickaxe regex: %s", errbuf);
 269                 }
 270                 regexp = &regex;
 271         } else {
 272                 kws = kwsalloc(DIFF_OPT_TST(o, PICKAXE_IGNORE_CASE)
 273                                ? tolower_trans_tbl : NULL);
 274                 kwsincr(kws, needle, len);
 275                 kwsprep(kws);
 276         }
 277
 278         pickaxe(&diff_queued_diff, o, regexp, kws, has_changes);
 279
 280         if (opts & DIFF_PICKAXE_REGEX)
 281                 regfree(&regex);
 282         else
 283                 kwsfree(kws);
 284         return;
 285 }
 286
 287 void diffcore_pickaxe(struct diff_options *o)
 288 {
 289         /* Might want to warn when both S and G are on; I don't care... */
 290         if (o->pickaxe_opts & DIFF_PICKAXE_KIND_G)
 291                 diffcore_pickaxe_grep(o);
 292         else
 293                 diffcore_pickaxe_count(o);
 294 }