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 there are no CR characters to filter out, then just pass */
 142                 if (!stats.cr)
 143                         return GIT_PASSTHROUGH;
 144
 145                 /* If safecrlf is enabled, sanity-check the result. */
 146                 if (stats.cr != stats.crlf || stats.lf != stats.crlf) {
 147                         switch (ca->safe_crlf) {
 148                         case GIT_SAFE_CRLF_FAIL:
 149                                 giterr_set(
 150                                         GITERR_FILTER, "LF would be replaced by CRLF in '%s'",
 151                                         git_filter_source_path(src));
 152                                 return -1;
 153                         case GIT_SAFE_CRLF_WARN:
 154                                 /* TODO: issue warning when warning API is available */;
 155                                 break;
 156                         default:
 157                                 break;
 158                         }
 159                 }
 160
 161                 /*
 162                  * We're currently not going to even try to convert stuff
 163                  * that has bare CR characters. Does anybody do that crazy
 164                  * stuff?
 165                  */
 166                 if (stats.cr != stats.crlf)
 167                         return GIT_PASSTHROUGH;
 168
 169                 if (ca->crlf_action == GIT_CRLF_GUESS) {
 170                         /*
 171                          * If the file in the index has any CR in it, do not convert.
 172                          * This is the new safer autocrlf handling.
 173                          */
 174                         if (has_cr_in_index(src))
 175                                 return GIT_PASSTHROUGH;
 176                 }
 177
 178                 if (!stats.cr)
 179                         return GIT_PASSTHROUGH;
 180         }
 181
 182         /* Actually drop the carriage returns */
 183         return git_buf_text_crlf_to_lf(to, from);
 184 }
 185
 186 static const char *line_ending(struct crlf_attrs *ca)
 187 {
 188         switch (ca->crlf_action) {
 189         case GIT_CRLF_BINARY:
 190         case GIT_CRLF_INPUT:
 191                 return "\n";
 192
 193         case GIT_CRLF_CRLF:
 194                 return "\r\n";
 195
 196         case GIT_CRLF_AUTO:
 197         case GIT_CRLF_TEXT:
 198         case GIT_CRLF_GUESS:
 199                 break;
 200
 201         default:
 202                 goto line_ending_error;
 203         }
 204
 205         switch (ca->eol) {
 206         case GIT_EOL_UNSET:
 207                 return GIT_EOL_NATIVE == GIT_EOL_CRLF ? "\r\n" : "\n";
 208
 209         case GIT_EOL_CRLF:
 210                 return "\r\n";
 211
 212         case GIT_EOL_LF:
 213                 return "\n";
 214
 215         default:
 216                 goto line_ending_error;
 217         }
 218
 219 line_ending_error:
 220         giterr_set(GITERR_INVALID, "Invalid input to line ending filter");
 221         return NULL;
 222 }
 223
 224 static int crlf_apply_to_workdir(
 225         struct crlf_attrs *ca, git_buf *to, const git_buf *from)
 226 {
 227         const char *workdir_ending = NULL;
 228
 229         /* Empty file? Nothing to do. */
 230         if (git_buf_len(from) == 0)
 231                 return 0;
 232
 233         /* Don't filter binary files */
 234         if (git_buf_text_is_binary(from))
 235                 return GIT_PASSTHROUGH;
 236
 237         /* Determine proper line ending */
 238         workdir_ending = line_ending(ca);
 239         if (!workdir_ending)
 240                 return -1;
 241
 242         /* only LF->CRLF conversion is supported, do nothing on LF platforms */
 243         if (strcmp(workdir_ending, "\r\n") != 0)
 244                 return GIT_PASSTHROUGH;
 245
 246         return git_buf_text_lf_to_crlf(to, from);
 247 }
 248
 249 static int crlf_check(
 250         git_filter        *self,
 251         void              **payload, /* points to NULL ptr on entry, may be set */
 252         const git_filter_source *src,
 253         const char **attr_values)
 254 {
 255         int error;
 256         struct crlf_attrs ca;
 257
 258         GIT_UNUSED(self);
 259
 260         if (!attr_values) {
 261                 ca.crlf_action = GIT_CRLF_GUESS;
 262                 ca.eol = GIT_EOL_UNSET;
 263         } else {
 264                 ca.crlf_action = check_crlf(attr_values[2]); /* text */
 265                 if (ca.crlf_action == GIT_CRLF_GUESS)
 266                         ca.crlf_action = check_crlf(attr_values[0]); /* clrf */
 267                 ca.eol = check_eol(attr_values[1]); /* eol */
 268         }
 269         ca.auto_crlf = GIT_AUTO_CRLF_DEFAULT;
 270
 271         /*
 272          * Use the core Git logic to see if we should perform CRLF for this file
 273          * based on its attributes & the value of `core.autocrlf`
 274          */
 275         ca.crlf_action = crlf_input_action(&ca);
 276
 277         if (ca.crlf_action == GIT_CRLF_BINARY)
 278                 return GIT_PASSTHROUGH;
 279
 280         if (ca.crlf_action == GIT_CRLF_GUESS ||
 281                 (ca.crlf_action == GIT_CRLF_AUTO &&
 282                 git_filter_source_mode(src) == GIT_FILTER_SMUDGE)) {
 283
 284                 error = git_repository__cvar(
 285                         &ca.auto_crlf, git_filter_source_repo(src), GIT_CVAR_AUTO_CRLF);
 286                 if (error < 0)
 287                         return error;
 288
 289                 if (ca.crlf_action == GIT_CRLF_GUESS &&
 290                         ca.auto_crlf == GIT_AUTO_CRLF_FALSE)
 291                         return GIT_PASSTHROUGH;
 292
 293                 if (ca.auto_crlf == GIT_AUTO_CRLF_INPUT &&
 294                         git_filter_source_mode(src) == GIT_FILTER_SMUDGE)
 295                         return GIT_PASSTHROUGH;
 296         }
 297
 298         if (git_filter_source_mode(src) == GIT_FILTER_CLEAN) {
 299                 error = git_repository__cvar(
 300                         &ca.safe_crlf, git_filter_source_repo(src), GIT_CVAR_SAFE_CRLF);
 301                 if (error < 0)
 302                         return error;
 303
 304                 /* downgrade FAIL to WARN if ALLOW_UNSAFE option is used */
 305                 if ((git_filter_source_options(src) & GIT_FILTER_OPT_ALLOW_UNSAFE) &&
 306                         ca.safe_crlf == GIT_SAFE_CRLF_FAIL)
 307                         ca.safe_crlf = GIT_SAFE_CRLF_WARN;
 308         }
 309
 310         *payload = git__malloc(sizeof(ca));
 311         GITERR_CHECK_ALLOC(*payload);
 312         memcpy(*payload, &ca, sizeof(ca));
 313
 314         return 0;
 315 }
 316
 317 static int crlf_apply(
 318         git_filter    *self,
 319         void         **payload, /* may be read and/or set */
 320         git_buf       *to,
 321         const git_buf *from,
 322         const git_filter_source *src)
 323 {
 324         /* initialize payload in case `check` was bypassed */
 325         if (!*payload) {
 326                 int error = crlf_check(self, payload, src, NULL);
 327                 if (error < 0 && error != GIT_PASSTHROUGH)
 328                         return error;
 329         }
 330
 331         if (git_filter_source_mode(src) == GIT_FILTER_SMUDGE)
 332                 return crlf_apply_to_workdir(*payload, to, from);
 333         else
 334                 return crlf_apply_to_odb(*payload, to, from, src);
 335 }
 336
 337 static void crlf_cleanup(
 338         git_filter *self,
 339         void       *payload)
 340 {
 341         GIT_UNUSED(self);
 342         git__free(payload);
 343 }
 344
 345 git_filter *git_crlf_filter_new(void)
 346 {
 347         struct crlf_filter *f = git__calloc(1, sizeof(struct crlf_filter));
 348
 349         f->f.version = GIT_FILTER_VERSION;
 350         f->f.attributes = "crlf eol text";
 351         f->f.initialize = NULL;
 352         f->f.shutdown = git_filter_free;
 353         f->f.check    = crlf_check;
 354         f->f.apply    = crlf_apply;
 355         f->f.cleanup  = crlf_cleanup;
 356
 357         return (git_filter *)f;
 358 }