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         git_buf safe = GIT_BUF_INIT;
 128         int error = 0;
 129
 130         /* Empty file? Nothing to do */
 131         if (!git_buf_len(from))
 132                 return 0;
 133
 134         /* Heuristics to see if we can skip the conversion.
 135          * Straight from Core Git.
 136          */
 137         if (ca->crlf_action == GIT_CRLF_AUTO || ca->crlf_action == GIT_CRLF_GUESS) {
 138                 git_buf_text_stats stats;
 139
 140                 /* Check heuristics for binary vs text - returns true if binary */
 141                 if (git_buf_text_gather_stats(&stats, from, false))
 142                         return GIT_PASSTHROUGH;
 143
 144                 /*
 145                  * We're currently not going to even try to convert stuff
 146                  * that has bare CR characters. Does anybody do that crazy
 147                  * stuff?
 148                  */
 149                 if (stats.cr != stats.crlf)
 150                         return GIT_PASSTHROUGH;
 151
 152                 if (ca->crlf_action == GIT_CRLF_GUESS) {
 153                         /*
 154                          * If the file in the index has any CR in it, do not convert.
 155                          * This is the new safer autocrlf handling.
 156                          */
 157                         if (has_cr_in_index(src))
 158                                 return GIT_PASSTHROUGH;
 159                 }
 160
 161                 if (!stats.cr && !ca->safe_crlf)
 162                         return GIT_PASSTHROUGH;
 163         }
 164
 165         /* Actually drop the carriage returns */
 166         if ((error = git_buf_text_crlf_to_lf(to, from)) < 0)
 167                 return error;
 168
 169         /* If safecrlf is enabled, sanity-check the result. */
 170         if (ca->safe_crlf) {
 171                 if ((error = git_buf_grow(&safe, max(from->size, to->size))) < 0 ||
 172                         (error = git_buf_text_lf_to_crlf(&safe, to)) < 0)
 173                         goto done;
 174
 175                 if (git_buf_cmp(from, &safe) != 0) {
 176                         giterr_set(GITERR_FILTER, "LF would be replaced by CRLF in '%s'",
 177                                 git_filter_source_path(src));
 178                         error = -1;
 179                 }
 180         }
 181
 182 done:
 183         git_buf_free(&safe);
 184
 185         return error;
 186 }
 187
 188 static const char *line_ending(struct crlf_attrs *ca)
 189 {
 190         switch (ca->crlf_action) {
 191         case GIT_CRLF_BINARY:
 192         case GIT_CRLF_INPUT:
 193                 return "\n";
 194
 195         case GIT_CRLF_CRLF:
 196                 return "\r\n";
 197
 198         case GIT_CRLF_AUTO:
 199         case GIT_CRLF_TEXT:
 200         case GIT_CRLF_GUESS:
 201                 break;
 202
 203         default:
 204                 goto line_ending_error;
 205         }
 206
 207         switch (ca->eol) {
 208         case GIT_EOL_UNSET:
 209                 return GIT_EOL_NATIVE == GIT_EOL_CRLF ? "\r\n" : "\n";
 210
 211         case GIT_EOL_CRLF:
 212                 return "\r\n";
 213
 214         case GIT_EOL_LF:
 215                 return "\n";
 216
 217         default:
 218                 goto line_ending_error;
 219         }
 220
 221 line_ending_error:
 222         giterr_set(GITERR_INVALID, "Invalid input to line ending filter");
 223         return NULL;
 224 }
 225
 226 static int crlf_apply_to_workdir(
 227         struct crlf_attrs *ca, git_buf *to, const git_buf *from)
 228 {
 229         const char *workdir_ending = NULL;
 230
 231         /* Empty file? Nothing to do. */
 232         if (git_buf_len(from) == 0)
 233                 return 0;
 234
 235         /* Don't filter binary files */
 236         if (git_buf_text_is_binary(from))
 237                 return GIT_PASSTHROUGH;
 238
 239         /* Determine proper line ending */
 240         workdir_ending = line_ending(ca);
 241         if (!workdir_ending)
 242                 return -1;
 243
 244         /* only LF->CRLF conversion is supported, do nothing on LF platforms */
 245         if (strcmp(workdir_ending, "\r\n") != 0)
 246                 return GIT_PASSTHROUGH;
 247
 248         return git_buf_text_lf_to_crlf(to, from);
 249 }
 250
 251 static int crlf_check(
 252         git_filter        *self,
 253         void              **payload, /* points to NULL ptr on entry, may be set */
 254         const git_filter_source *src,
 255         const char **attr_values)
 256 {
 257         int error;
 258         struct crlf_attrs ca;
 259
 260         GIT_UNUSED(self);
 261
 262         if (!attr_values) {
 263                 ca.crlf_action = GIT_CRLF_GUESS;
 264                 ca.eol = GIT_EOL_UNSET;
 265         } else {
 266                 ca.crlf_action = check_crlf(attr_values[2]); /* text */
 267                 if (ca.crlf_action == GIT_CRLF_GUESS)
 268                         ca.crlf_action = check_crlf(attr_values[0]); /* clrf */
 269                 ca.eol = check_eol(attr_values[1]); /* eol */
 270         }
 271         ca.auto_crlf = GIT_AUTO_CRLF_DEFAULT;
 272
 273         /*
 274          * Use the core Git logic to see if we should perform CRLF for this file
 275          * based on its attributes & the value of `core.autocrlf`
 276          */
 277         ca.crlf_action = crlf_input_action(&ca);
 278
 279         if (ca.crlf_action == GIT_CRLF_BINARY)
 280                 return GIT_PASSTHROUGH;
 281
 282         if (ca.crlf_action == GIT_CRLF_GUESS ||
 283                 (ca.crlf_action == GIT_CRLF_AUTO &&
 284                 git_filter_source_mode(src) == GIT_FILTER_SMUDGE)) {
 285                 error = git_repository__cvar(
 286                         &ca.auto_crlf, git_filter_source_repo(src), GIT_CVAR_AUTO_CRLF);
 287                 if (error < 0)
 288                         return error;
 289
 290                 if (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
 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 }