Andrew's git - gitweb.git/blob - 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
  13typedef int (*pickaxe_fn)(mmfile_t *one, mmfile_t *two,
  14                          struct diff_options *o,
  15                          regex_t *regexp, kwset_t kws);
  16
  17struct diffgrep_cb {
  18        regex_t *regexp;
  19        int hit;
  20};
  21
  22static 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
  39static 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
  70static 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 (sz && *data &&
  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 && *data && regmatch.rm_so == regmatch.rm_eo) {
  90                                data++;
  91                                sz--;
  92                        }
  93                        cnt++;
  94                }
  95
  96        } else { /* Classic exact string match */
  97                while (sz) {
  98                        struct kwsmatch kwsm;
  99                        size_t offset = kwsexec(kws, data, sz, &kwsm);
 100                        if (offset == -1)
 101                                break;
 102                        sz -= offset + kwsm.size[0];
 103                        data += offset + kwsm.size[0];
 104                        cnt++;
 105                }
 106        }
 107        return cnt;
 108}
 109
 110static int has_changes(mmfile_t *one, mmfile_t *two,
 111                       struct diff_options *o,
 112                       regex_t *regexp, kwset_t kws)
 113{
 114        unsigned int one_contains = one ? contains(one, regexp, kws) : 0;
 115        unsigned int two_contains = two ? contains(two, regexp, kws) : 0;
 116        return one_contains != two_contains;
 117}
 118
 119static int pickaxe_match(struct diff_filepair *p, struct diff_options *o,
 120                         regex_t *regexp, kwset_t kws, pickaxe_fn fn)
 121{
 122        struct userdiff_driver *textconv_one = NULL;
 123        struct userdiff_driver *textconv_two = NULL;
 124        mmfile_t mf1, mf2;
 125        int ret;
 126
 127        if (!o->pickaxe[0])
 128                return 0;
 129
 130        /* ignore unmerged */
 131        if (!DIFF_FILE_VALID(p->one) && !DIFF_FILE_VALID(p->two))
 132                return 0;
 133
 134        if (DIFF_OPT_TST(o, ALLOW_TEXTCONV)) {
 135                textconv_one = get_textconv(p->one);
 136                textconv_two = get_textconv(p->two);
 137        }
 138
 139        /*
 140         * If we have an unmodified pair, we know that the count will be the
 141         * same and don't even have to load the blobs. Unless textconv is in
 142         * play, _and_ we are using two different textconv filters (e.g.,
 143         * because a pair is an exact rename with different textconv attributes
 144         * for each side, which might generate different content).
 145         */
 146        if (textconv_one == textconv_two && diff_unmodified_pair(p))
 147                return 0;
 148
 149        mf1.size = fill_textconv(textconv_one, p->one, &mf1.ptr);
 150        mf2.size = fill_textconv(textconv_two, p->two, &mf2.ptr);
 151
 152        ret = fn(DIFF_FILE_VALID(p->one) ? &mf1 : NULL,
 153                 DIFF_FILE_VALID(p->two) ? &mf2 : NULL,
 154                 o, regexp, kws);
 155
 156        if (textconv_one)
 157                free(mf1.ptr);
 158        if (textconv_two)
 159                free(mf2.ptr);
 160        diff_free_filespec_data(p->one);
 161        diff_free_filespec_data(p->two);
 162
 163        return ret;
 164}
 165
 166static void pickaxe(struct diff_queue_struct *q, struct diff_options *o,
 167                    regex_t *regexp, kwset_t kws, pickaxe_fn fn)
 168{
 169        int i;
 170        struct diff_queue_struct outq;
 171
 172        DIFF_QUEUE_CLEAR(&outq);
 173
 174        if (o->pickaxe_opts & DIFF_PICKAXE_ALL) {
 175                /* Showing the whole changeset if needle exists */
 176                for (i = 0; i < q->nr; i++) {
 177                        struct diff_filepair *p = q->queue[i];
 178                        if (pickaxe_match(p, o, regexp, kws, fn))
 179                                return; /* do not munge the queue */
 180                }
 181
 182                /*
 183                 * Otherwise we will clear the whole queue by copying
 184                 * the empty outq at the end of this function, but
 185                 * first clear the current entries in the queue.
 186                 */
 187                for (i = 0; i < q->nr; i++)
 188                        diff_free_filepair(q->queue[i]);
 189        } else {
 190                /* Showing only the filepairs that has the needle */
 191                for (i = 0; i < q->nr; i++) {
 192                        struct diff_filepair *p = q->queue[i];
 193                        if (pickaxe_match(p, o, regexp, kws, fn))
 194                                diff_q(&outq, p);
 195                        else
 196                                diff_free_filepair(p);
 197                }
 198        }
 199
 200        free(q->queue);
 201        *q = outq;
 202}
 203
 204static void regcomp_or_die(regex_t *regex, const char *needle, int cflags)
 205{
 206        int err = regcomp(regex, needle, cflags);
 207        if (err) {
 208                /* The POSIX.2 people are surely sick */
 209                char errbuf[1024];
 210                regerror(err, regex, errbuf, 1024);
 211                regfree(regex);
 212                die("invalid regex: %s", errbuf);
 213        }
 214}
 215
 216void diffcore_pickaxe(struct diff_options *o)
 217{
 218        const char *needle = o->pickaxe;
 219        int opts = o->pickaxe_opts;
 220        regex_t regex, *regexp = NULL;
 221        kwset_t kws = NULL;
 222
 223        if (opts & (DIFF_PICKAXE_REGEX | DIFF_PICKAXE_KIND_G)) {
 224                int cflags = REG_EXTENDED | REG_NEWLINE;
 225                if (DIFF_OPT_TST(o, PICKAXE_IGNORE_CASE))
 226                        cflags |= REG_ICASE;
 227                regcomp_or_die(&regex, needle, cflags);
 228                regexp = &regex;
 229        } else if (DIFF_OPT_TST(o, PICKAXE_IGNORE_CASE) &&
 230                   has_non_ascii(needle)) {
 231                struct strbuf sb = STRBUF_INIT;
 232                int cflags = REG_NEWLINE | REG_ICASE;
 233
 234                basic_regex_quote_buf(&sb, needle);
 235                regcomp_or_die(&regex, sb.buf, cflags);
 236                strbuf_release(&sb);
 237                regexp = &regex;
 238        } else {
 239                kws = kwsalloc(DIFF_OPT_TST(o, PICKAXE_IGNORE_CASE)
 240                               ? tolower_trans_tbl : NULL);
 241                kwsincr(kws, needle, strlen(needle));
 242                kwsprep(kws);
 243        }
 244
 245        /* Might want to warn when both S and G are on; I don't care... */
 246        pickaxe(&diff_queued_diff, o, regexp, kws,
 247                (opts & DIFF_PICKAXE_KIND_G) ? diff_grep : has_changes);
 248
 249        if (regexp)
 250                regfree(regexp);
 251        else
 252                kwsfree(kws);
 253        return;
 254}