Andrew's git - gitweb.git/blob

bfaababbe5d836d8b1fa53215bca3c0f405f1ecc
   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
  11typedef int (*pickaxe_fn)(struct diff_filepair *p, struct diff_options *o, regex_t *regexp, kwset_t kws);
  12
  13static 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
  51struct diffgrep_cb {
  52        regex_t *regexp;
  53        int hit;
  54};
  55
  56static 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
  77static int diff_grep(struct diff_filepair *p, struct diff_options *o,
  78                     regex_t *regexp, kwset_t kws)
  79{
  80        regmatch_t regmatch;
  81        struct userdiff_driver *textconv_one = NULL;
  82        struct userdiff_driver *textconv_two = NULL;
  83        mmfile_t mf1, mf2;
  84        int hit;
  85
  86        if (!o->pickaxe[0])
  87                return 0;
  88
  89        if (DIFF_OPT_TST(o, ALLOW_TEXTCONV)) {
  90                textconv_one = get_textconv(p->one);
  91                textconv_two = get_textconv(p->two);
  92        }
  93
  94        if (textconv_one == textconv_two && diff_unmodified_pair(p))
  95                return 0;
  96
  97        mf1.size = fill_textconv(textconv_one, p->one, &mf1.ptr);
  98        mf2.size = fill_textconv(textconv_two, p->two, &mf2.ptr);
  99
 100        if (!DIFF_FILE_VALID(p->one)) {
 101                if (!DIFF_FILE_VALID(p->two))
 102                        return 0; /* ignore unmerged */
 103                /* created "two" -- does it have what we are looking for? */
 104                hit = !regexec(regexp, mf2.ptr, 1, &regmatch, 0);
 105        } else if (!DIFF_FILE_VALID(p->two)) {
 106                /* removed "one" -- did it have what we are looking for? */
 107                hit = !regexec(regexp, mf1.ptr, 1, &regmatch, 0);
 108        } else {
 109                /*
 110                 * We have both sides; need to run textual diff and see if
 111                 * the pattern appears on added/deleted lines.
 112                 */
 113                struct diffgrep_cb ecbdata;
 114                xpparam_t xpp;
 115                xdemitconf_t xecfg;
 116
 117                memset(&xpp, 0, sizeof(xpp));
 118                memset(&xecfg, 0, sizeof(xecfg));
 119                ecbdata.regexp = regexp;
 120                ecbdata.hit = 0;
 121                xecfg.ctxlen = o->context;
 122                xecfg.interhunkctxlen = o->interhunkcontext;
 123                xdi_diff_outf(&mf1, &mf2, diffgrep_consume, &ecbdata,
 124                              &xpp, &xecfg);
 125                hit = ecbdata.hit;
 126        }
 127        if (textconv_one)
 128                free(mf1.ptr);
 129        if (textconv_two)
 130                free(mf2.ptr);
 131        diff_free_filespec_data(p->one);
 132        diff_free_filespec_data(p->two);
 133        return hit;
 134}
 135
 136static void diffcore_pickaxe_grep(struct diff_options *o)
 137{
 138        int err;
 139        regex_t regex;
 140        int cflags = REG_EXTENDED | REG_NEWLINE;
 141
 142        if (DIFF_OPT_TST(o, PICKAXE_IGNORE_CASE))
 143                cflags |= REG_ICASE;
 144
 145        err = regcomp(&regex, o->pickaxe, cflags);
 146        if (err) {
 147                char errbuf[1024];
 148                regerror(err, &regex, errbuf, 1024);
 149                regfree(&regex);
 150                die("invalid log-grep regex: %s", errbuf);
 151        }
 152
 153        pickaxe(&diff_queued_diff, o, &regex, NULL, diff_grep);
 154
 155        regfree(&regex);
 156        return;
 157}
 158
 159static unsigned int contains(mmfile_t *mf, struct diff_options *o,
 160                             regex_t *regexp, kwset_t kws)
 161{
 162        unsigned int cnt;
 163        unsigned long sz;
 164        const char *data;
 165
 166        sz = mf->size;
 167        data = mf->ptr;
 168        cnt = 0;
 169
 170        if (regexp) {
 171                regmatch_t regmatch;
 172                int flags = 0;
 173
 174                assert(data[sz] == '\0');
 175                while (*data && !regexec(regexp, data, 1, &regmatch, flags)) {
 176                        flags |= REG_NOTBOL;
 177                        data += regmatch.rm_eo;
 178                        if (*data && regmatch.rm_so == regmatch.rm_eo)
 179                                data++;
 180                        cnt++;
 181                }
 182
 183        } else { /* Classic exact string match */
 184                while (sz) {
 185                        struct kwsmatch kwsm;
 186                        size_t offset = kwsexec(kws, data, sz, &kwsm);
 187                        const char *found;
 188                        if (offset == -1)
 189                                break;
 190                        else
 191                                found = data + offset;
 192                        sz -= found - data + kwsm.size[0];
 193                        data = found + kwsm.size[0];
 194                        cnt++;
 195                }
 196        }
 197        return cnt;
 198}
 199
 200static int has_changes(struct diff_filepair *p, struct diff_options *o,
 201                       regex_t *regexp, kwset_t kws)
 202{
 203        struct userdiff_driver *textconv_one = NULL;
 204        struct userdiff_driver *textconv_two = NULL;
 205        mmfile_t mf1, mf2;
 206        int ret;
 207
 208        if (!o->pickaxe[0])
 209                return 0;
 210
 211        if (DIFF_OPT_TST(o, ALLOW_TEXTCONV)) {
 212                textconv_one = get_textconv(p->one);
 213                textconv_two = get_textconv(p->two);
 214        }
 215
 216        /*
 217         * If we have an unmodified pair, we know that the count will be the
 218         * same and don't even have to load the blobs. Unless textconv is in
 219         * play, _and_ we are using two different textconv filters (e.g.,
 220         * because a pair is an exact rename with different textconv attributes
 221         * for each side, which might generate different content).
 222         */
 223        if (textconv_one == textconv_two && diff_unmodified_pair(p))
 224                return 0;
 225
 226        mf1.size = fill_textconv(textconv_one, p->one, &mf1.ptr);
 227        mf2.size = fill_textconv(textconv_two, p->two, &mf2.ptr);
 228
 229        if (!DIFF_FILE_VALID(p->one)) {
 230                if (!DIFF_FILE_VALID(p->two))
 231                        ret = 0; /* ignore unmerged */
 232                /* created */
 233                ret = contains(&mf2, o, regexp, kws) != 0;
 234        }
 235        else if (!DIFF_FILE_VALID(p->two)) /* removed */
 236                ret = contains(&mf1, o, regexp, kws) != 0;
 237        else
 238                ret = contains(&mf1, o, regexp, kws) !=
 239                      contains(&mf2, o, regexp, kws);
 240
 241        if (textconv_one)
 242                free(mf1.ptr);
 243        if (textconv_two)
 244                free(mf2.ptr);
 245        diff_free_filespec_data(p->one);
 246        diff_free_filespec_data(p->two);
 247
 248        return ret;
 249}
 250
 251static void diffcore_pickaxe_count(struct diff_options *o)
 252{
 253        const char *needle = o->pickaxe;
 254        int opts = o->pickaxe_opts;
 255        unsigned long len = strlen(needle);
 256        regex_t regex, *regexp = NULL;
 257        kwset_t kws = NULL;
 258
 259        if (opts & DIFF_PICKAXE_REGEX) {
 260                int err;
 261                err = regcomp(&regex, needle, REG_EXTENDED | REG_NEWLINE);
 262                if (err) {
 263                        /* The POSIX.2 people are surely sick */
 264                        char errbuf[1024];
 265                        regerror(err, &regex, errbuf, 1024);
 266                        regfree(&regex);
 267                        die("invalid pickaxe regex: %s", errbuf);
 268                }
 269                regexp = &regex;
 270        } else {
 271                kws = kwsalloc(DIFF_OPT_TST(o, PICKAXE_IGNORE_CASE)
 272                               ? tolower_trans_tbl : NULL);
 273                kwsincr(kws, needle, len);
 274                kwsprep(kws);
 275        }
 276
 277        pickaxe(&diff_queued_diff, o, regexp, kws, has_changes);
 278
 279        if (opts & DIFF_PICKAXE_REGEX)
 280                regfree(&regex);
 281        else
 282                kwsfree(kws);
 283        return;
 284}
 285
 286void diffcore_pickaxe(struct diff_options *o)
 287{
 288        /* Might want to warn when both S and G are on; I don't care... */
 289        if (o->pickaxe_opts & DIFF_PICKAXE_KIND_G)
 290                diffcore_pickaxe_grep(o);
 291        else
 292                diffcore_pickaxe_count(o);
 293}