src/crlf.c

   1 /*
   2  * Copyright (C) the libgit2 contributors. All rights reserved.
   3  *
   4  * This file is part of libgit2, distributed under the GNU GPL v2 with
   5  * a Linking Exception. For full terms see the included COPYING file.
   6  */
   7
   8 #include "git2/attr.h"
   9 #include "git2/blob.h"
  10 #include "git2/index.h"
  11 #include "git2/sys/filter.h"
  12
  13 #include "common.h"
  14 #include "fileops.h"
  15 #include "hash.h"
  16 #include "filter.h"
  17 #include "buf_text.h"
  18 #include "repository.h"
  19
  20 struct crlf_attrs {
  21         int crlf_action;
  22         int eol;
  23         int auto_crlf;
  24         int safe_crlf;
  25 };
  26
  27 struct crlf_filter {
  28         git_filter f;
  29 };
  30
  31 static int check_crlf(const char *value)
  32 {
  33         if (GIT_ATTR_TRUE(value))
  34                 return GIT_CRLF_TEXT;
  35
  36         if (GIT_ATTR_FALSE(value))
  37                 return GIT_CRLF_BINARY;
  38
  39         if (GIT_ATTR_UNSPECIFIED(value))
  40                 return GIT_CRLF_GUESS;
  41
  42         if (strcmp(value, "input") == 0)
  43                 return GIT_CRLF_INPUT;
  44
  45         if (strcmp(value, "auto") == 0)
  46                 return GIT_CRLF_AUTO;
  47
  48         return GIT_CRLF_GUESS;
  49 }
  50
  51 static int check_eol(const char *value)
  52 {
  53         if (GIT_ATTR_UNSPECIFIED(value))
  54                 return GIT_EOL_UNSET;
  55
  56         if (strcmp(value, "lf") == 0)
  57                 return GIT_EOL_LF;
  58
  59         if (strcmp(value, "crlf") == 0)
  60                 return GIT_EOL_CRLF;
  61
  62         return GIT_EOL_UNSET;
  63 }
  64
  65 static int crlf_input_action(struct crlf_attrs *ca)
  66 {
  67         if (ca->crlf_action == GIT_CRLF_BINARY)
  68                 return GIT_CRLF_BINARY;
  69
  70         if (ca->eol == GIT_EOL_LF)
  71                 return GIT_CRLF_INPUT;
  72
  73         if (ca->eol == GIT_EOL_CRLF)
  74                 return GIT_CRLF_CRLF;
  75
  76         return ca->crlf_action;
  77 }
  78
  79 static int has_cr_in_index(const git_filter_source *src)
  80 {
  81         git_repository *repo = git_filter_source_repo(src);
  82         const char *path = git_filter_source_path(src);
  83         git_index *index;
  84         const git_index_entry *entry;
  85         git_blob *blob;
  86         const void *blobcontent;
  87         git_off_t blobsize;
  88         bool found_cr;
  89
  90         if (!path)
  91                 return false;
  92
  93         if (git_repository_index__weakptr(&index, repo) < 0) {
  94                 giterr_clear();
  95                 return false;
  96         }
  97
  98         if (!(entry = git_index_get_bypath(index, path, 0)) &&
  99                 !(entry = git_index_get_bypath(index, path, 1)))
 100                 return false;
 101
 102         if (!S_ISREG(entry->mode)) /* don't crlf filter non-blobs */
 103                 return true;
 104
 105         if (git_blob_lookup(&blob, repo, &entry->id) < 0)
 106                 return false;
 107
 108         blobcontent = git_blob_rawcontent(blob);
 109         blobsize    = git_blob_rawsize(blob);
 110         if (!git__is_sizet(blobsize))
 111                 blobsize = (size_t)-1;
 112
 113         found_cr = (blobcontent != NULL &&
 114                 blobsize > 0 &&
 115                 memchr(blobcontent, '\r', (size_t)blobsize) != NULL);
 116
 117         git_blob_free(blob);
 118         return found_cr;
 119 }
 120
 121 static int crlf_apply_to_odb(
 122         struct crlf_attrs *ca,
 123         git_buf *to,
 124         const git_buf *from,
 125         const git_filter_source *src)
 126 {
 127         /* Empty file? Nothing to do */
 128         if (!git_buf_len(from))
 129                 return 0;
 130
 131         /* Heuristics to see if we can skip the conversion.
 132          * Straight from Core Git.
 133          */
 134         if (ca->crlf_action == GIT_CRLF_AUTO || ca->crlf_action == GIT_CRLF_GUESS) {
 135                 git_buf_text_stats stats;
 136
 137                 /* Check heuristics for binary vs text - returns true if binary */
 138                 if (git_buf_text_gather_stats(&stats, from, false))
 139                         return GIT_PASSTHROUGH;
 140
 141                 /* If safecrlf is enabled, sanity-check the result. */
 142                 if (stats.cr != stats.crlf || stats.lf != stats.crlf) {
 143                         switch (ca->safe_crlf) {
 144                         case GIT_SAFE_CRLF_FAIL:
 145                                 giterr_set(
 146                                         GITERR_FILTER, "LF would be replaced by CRLF in '%s'",
 147                                         git_filter_source_path(src));
 148                                 return -1;
 149                         case GIT_SAFE_CRLF_WARN:
 150                                 /* TODO: issue warning when warning API is available */;
 151                                 break;
 152                         default:
 153                                 break;
 154                         }
 155                 }
 156
 157                 /*
 158                  * We're currently not going to even try to convert stuff
 159                  * that has bare CR characters. Does anybody do that crazy
 160                  * stuff?
 161                  */
 162                 if (stats.cr != stats.crlf)
 163                         return GIT_PASSTHROUGH;
 164
 165                 if (ca->crlf_action == GIT_CRLF_GUESS) {
 166                         /*
 167                          * If the file in the index has any CR in it, do not convert.
 168                          * This is the new safer autocrlf handling.
 169                          */
 170                         if (has_cr_in_index(src))
 171                                 return GIT_PASSTHROUGH;
 172                 }
 173
 174                 if (!stats.cr)
 175                         return GIT_PASSTHROUGH;
 176         }
 177
 178         /* Actually drop the carriage returns */
 179         return git_buf_text_crlf_to_lf(to, from);
 180 }
 181
 182 static const char *line_ending(struct crlf_attrs *ca)
 183 {
 184         switch (ca->crlf_action) {
 185         case GIT_CRLF_BINARY:
 186         case GIT_CRLF_INPUT:
 187                 return "\n";
 188
 189         case GIT_CRLF_CRLF:
 190                 return "\r\n";
 191
 192         case GIT_CRLF_AUTO:
 193         case GIT_CRLF_TEXT:
 194         case GIT_CRLF_GUESS:
 195                 break;
 196
 197         default:
 198                 goto line_ending_error;
 199         }
 200
 201         switch (ca->eol) {
 202         case GIT_EOL_UNSET:
 203                 return GIT_EOL_NATIVE == GIT_EOL_CRLF ? "\r\n" : "\n";
 204
 205         case GIT_EOL_CRLF:
 206                 return "\r\n";
 207
 208         case GIT_EOL_LF:
 209                 return "\n";
 210
 211         default:
 212                 goto line_ending_error;
 213         }
 214
 215 line_ending_error:
 216         giterr_set(GITERR_INVALID, "Invalid input to line ending filter");
 217         return NULL;
 218 }
 219
 220 static int crlf_apply_to_workdir(
 221         struct crlf_attrs *ca, git_buf *to, const git_buf *from)
 222 {
 223         const char *workdir_ending = NULL;
 224
 225         /* Empty file? Nothing to do. */
 226         if (git_buf_len(from) == 0)
 227                 return 0;
 228
 229         /* Don't filter binary files */
 230         if (git_buf_text_is_binary(from))
 231                 return GIT_PASSTHROUGH;
 232
 233         /* Determine proper line ending */
 234         workdir_ending = line_ending(ca);
 235         if (!workdir_ending)
 236                 return -1;
 237
 238         /* only LF->CRLF conversion is supported, do nothing on LF platforms */
 239         if (strcmp(workdir_ending, "\r\n") != 0)
 240                 return GIT_PASSTHROUGH;
 241
 242         return git_buf_text_lf_to_crlf(to, from);
 243 }
 244
 245 static int crlf_check(
 246         git_filter        *self,
 247         void              **payload, /* points to NULL ptr on entry, may be set */
 248         const git_filter_source *src,
 249         const char **attr_values)
 250 {
 251         int error;
 252         struct crlf_attrs ca;
 253
 254         GIT_UNUSED(self);
 255
 256         if (!attr_values) {
 257                 ca.crlf_action = GIT_CRLF_GUESS;
 258                 ca.eol = GIT_EOL_UNSET;
 259         } else {
 260                 ca.crlf_action = check_crlf(attr_values[2]); /* text */
 261                 if (ca.crlf_action == GIT_CRLF_GUESS)
 262                         ca.crlf_action = check_crlf(attr_values[0]); /* clrf */
 263                 ca.eol = check_eol(attr_values[1]); /* eol */
 264         }
 265         ca.auto_crlf = GIT_AUTO_CRLF_DEFAULT;
 266
 267         /*
 268          * Use the core Git logic to see if we should perform CRLF for this file
 269          * based on its attributes & the value of `core.autocrlf`
 270          */
 271         ca.crlf_action = crlf_input_action(&ca);
 272
 273         if (ca.crlf_action == GIT_CRLF_BINARY)
 274                 return GIT_PASSTHROUGH;
 275
 276         if (ca.crlf_action == GIT_CRLF_GUESS ||
 277                 (ca.crlf_action == GIT_CRLF_AUTO &&
 278                 git_filter_source_mode(src) == GIT_FILTER_SMUDGE)) {
 279
 280                 error = git_repository__cvar(
 281                         &ca.auto_crlf, git_filter_source_repo(src), GIT_CVAR_AUTO_CRLF);
 282                 if (error < 0)
 283                         return error;
 284
 285                 if (ca.auto_crlf == GIT_AUTO_CRLF_FALSE)
 286                         return GIT_PASSTHROUGH;
 287
 288                 if (ca.auto_crlf == GIT_AUTO_CRLF_INPUT &&
 289                         git_filter_source_mode(src) == GIT_FILTER_SMUDGE)
 290                         return GIT_PASSTHROUGH;
 291         }
 292
 293         if (git_filter_source_mode(src) == GIT_FILTER_CLEAN) {
 294                 error = git_repository__cvar(
 295                         &ca.safe_crlf, git_filter_source_repo(src), GIT_CVAR_SAFE_CRLF);
 296                 if (error < 0)
 297                         return error;
 298
 299                 /* downgrade FAIL to WARN if ALLOW_UNSAFE option is used */
 300                 if ((git_filter_source_options(src) & GIT_FILTER_OPT_ALLOW_UNSAFE) &&
 301                         ca.safe_crlf == GIT_SAFE_CRLF_FAIL)
 302                         ca.safe_crlf = GIT_SAFE_CRLF_WARN;
 303         }
 304
 305         *payload = git__malloc(sizeof(ca));
 306         GITERR_CHECK_ALLOC(*payload);
 307         memcpy(*payload, &ca, sizeof(ca));
 308
 309         return 0;
 310 }
 311
 312 static int crlf_apply(
 313         git_filter    *self,
 314         void         **payload, /* may be read and/or set */
 315         git_buf       *to,
 316         const git_buf *from,
 317         const git_filter_source *src)
 318 {
 319         /* initialize payload in case `check` was bypassed */
 320         if (!*payload) {
 321                 int error = crlf_check(self, payload, src, NULL);
 322                 if (error < 0 && error != GIT_PASSTHROUGH)
 323                         return error;
 324         }
 325
 326         if (git_filter_source_mode(src) == GIT_FILTER_SMUDGE)
 327                 return crlf_apply_to_workdir(*payload, to, from);
 328         else
 329                 return crlf_apply_to_odb(*payload, to, from, src);
 330 }
 331
 332 static void crlf_cleanup(
 333         git_filter *self,
 334         void       *payload)
 335 {
 336         GIT_UNUSED(self);
 337         git__free(payload);
 338 }
 339
 340 git_filter *git_crlf_filter_new(void)
 341 {
 342         struct crlf_filter *f = git__calloc(1, sizeof(struct crlf_filter));
 343
 344         f->f.version = GIT_FILTER_VERSION;
 345         f->f.attributes = "crlf eol text";
 346         f->f.initialize = NULL;
 347         f->f.shutdown = git_filter_free;
 348         f->f.check    = crlf_check;
 349         f->f.apply    = crlf_apply;
 350         f->f.cleanup  = crlf_cleanup;
 351
 352         return (git_filter *)f;
 353 }