src/mime/mime.c

   1 /* Functionality for handling mime types */
   2
   3 #ifndef _GNU_SOURCE
   4 #define _GNU_SOURCE /* XXX: we _WANT_ strcasestr() ! */
   5 #endif
   6
   7 #ifdef HAVE_CONFIG_H
   8 #include "config.h"
   9 #endif
  10
  11 #include <string.h>
  12
  13 #include "elinks.h"
  14
  15 #include "cache/cache.h"
  16 #include "config/options.h"
  17 #include "encoding/encoding.h"
  18 #include "intl/gettext/libintl.h"
  19 #include "main/module.h"
  20 #include "mime/backend/common.h"
  21 #include "mime/mime.h"
  22 #include "protocol/header.h"    /* For parse_header() */
  23 #include "protocol/uri.h"
  24 #include "util/conv.h"
  25 #include "util/file.h"
  26 #include "util/memory.h"
  27 #include "util/string.h"
  28
  29
  30 enum mime_options {
  31         MIME_TREE,
  32         MIME_DEFAULT_TYPE,
  33
  34         MIME_OPTIONS,
  35 };
  36
  37 static union option_info mime_options[] = {
  38         INIT_OPT_TREE("", N_("MIME"),
  39                 "mime", OPT_SORT,
  40                 N_("MIME-related options (handlers of various MIME types).")),
  41
  42         INIT_OPT_STRING("mime", N_("Default MIME-type"),
  43                 "default_type", 0, DEFAULT_MIME_TYPE,
  44                 N_("Document MIME-type to assume by default "
  45                 "(when we are unable to guess it properly "
  46                 "from known information about the document).")),
  47
  48         NULL_OPTION_INFO,
  49 };
  50
  51 #define get_opt_mime(which)     mime_options[(which)].option
  52 #define get_default_mime_type() get_opt_mime(MIME_DEFAULT_TYPE).value.string
  53
  54 /* Checks protocols headers for a suitable filename */
  55 static unsigned char *
  56 get_content_filename(struct uri *uri, struct cache_entry *cached)
  57 {
  58         unsigned char *filename, *pos;
  59
  60         if (!cached) cached = find_in_cache(uri);
  61
  62         if (!cached || !cached->head)
  63                 return NULL;
  64
  65         pos = parse_header(cached->head, "Content-Disposition", NULL);
  66         if (!pos) return NULL;
  67
  68         parse_header_param(pos, "filename", &filename);
  69         mem_free(pos);
  70         if (!filename) return NULL;
  71
  72         /* Remove start and ending quotes. */
  73         if (filename[0] == '"') {
  74                 int len = strlen(filename);
  75
  76                 if (len > 1 && filename[len - 1] == '"') {
  77                         filename[len - 1] = 0;
  78                         memmove(filename, filename + 1, len);
  79                 }
  80
  81                 /* It was an empty quotation: "" */
  82                 if (!filename[1]) {
  83                         mem_free(filename);
  84                         return NULL;
  85                 }
  86         }
  87
  88         /* We don't want to add any directories from the path so make sure we
  89          * only add the filename. */
  90         pos = get_filename_position(filename);
  91         if (!*pos) {
  92                 mem_free(filename);
  93                 return NULL;
  94         }
  95
  96         if (pos > filename)
  97                 memmove(filename, pos, strlen(pos) + 1);
  98
  99         return filename;
 100 }
 101
 102 /* Checks if application/x-<extension> has any handlers. */
 103 static inline unsigned char *
 104 check_extension_type(unsigned char *extension)
 105 {
 106         /* Trim the extension so only last .<extension> is used. */
 107         unsigned char *trimmed = strrchr(extension, '.');
 108         struct mime_handler *handler;
 109         unsigned char *content_type;
 110
 111         if (!trimmed)
 112                 return NULL;
 113
 114         content_type = straconcat("application/x-", trimmed + 1,
 115                                   (unsigned char *) NULL);
 116         if (!content_type)
 117                 return NULL;
 118
 119         handler = get_mime_type_handler(content_type, 1);
 120         if (handler) {
 121                 mem_free(handler);
 122                 return content_type;
 123         }
 124
 125         mem_free(content_type);
 126         return NULL;
 127 }
 128
 129 /* Check if part of the extension coresponds to a supported encoding and if it
 130  * has any handlers. */
 131 static inline unsigned char *
 132 check_encoding_type(unsigned char *extension)
 133 {
 134         enum stream_encoding encoding = guess_encoding(extension);
 135         const unsigned char *const *extension_list;
 136         unsigned char *last_extension = strrchr(extension, '.');
 137
 138         if (encoding == ENCODING_NONE || !last_extension)
 139                 return NULL;
 140
 141         for (extension_list = listext_encoded(encoding);
 142              extension_list && *extension_list;
 143              extension_list++) {
 144                 unsigned char *content_type;
 145
 146                 if (strcmp(*extension_list, last_extension))
 147                         continue;
 148
 149                 *last_extension = '\0';
 150                 content_type = get_content_type_backends(extension);
 151                 *last_extension = '.';
 152
 153                 return content_type;
 154         }
 155
 156         return NULL;
 157 }
 158
 159 #if 0
 160 #define DEBUG_CONTENT_TYPE
 161 #endif
 162
 163 #ifdef DEBUG_CONTENT_TYPE
 164 #define debug_get_content_type_params(cached) \
 165         DBG("get_content_type(head, url)\n=== head ===\n%s\n=== url ===\n%s\n", (cached)->head, struri((cached)->uri))
 166 #define debug_ctype(ctype__) DBG("ctype= %s", (ctype__))
 167 #define debug_extension(extension__) DBG("extension= %s", (extension__))
 168 #else
 169 #define debug_get_content_type_params(cached)
 170 #define debug_ctype(ctype__)
 171 #define debug_extension(extension__)
 172 #endif
 173
 174 unsigned char *
 175 get_extension_content_type(unsigned char *extension)
 176 {
 177         unsigned char *ctype;
 178
 179         assert(extension && *extension);
 180
 181         ctype = get_content_type_backends(extension);
 182         debug_ctype(ctype);
 183         if (ctype) return ctype;
 184
 185         ctype = check_encoding_type(extension);
 186         debug_ctype(ctype);
 187         if (ctype) return ctype;
 188
 189         ctype = check_extension_type(extension);
 190         debug_ctype(ctype);
 191         return ctype;
 192 }
 193
 194 unsigned char *
 195 get_cache_header_content_type(struct cache_entry *cached)
 196 {
 197         unsigned char *extension, *ctype;
 198
 199         ctype = parse_header(cached->head, "Content-Type", NULL);
 200         if (ctype) {
 201                 unsigned char *end = strchr((const char *)ctype, ';');
 202                 int ctypelen;
 203
 204                 if (end) *end = '\0';
 205
 206                 ctypelen = strlen(ctype);
 207                 while (ctypelen && ctype[--ctypelen] <= ' ')
 208                         ctype[ctypelen] = '\0';
 209
 210                 debug_ctype(ctype);
 211
 212                 if (*ctype) {
 213                         return ctype;
 214                 }
 215
 216                 mem_free(ctype);
 217         }
 218
 219         /* This searches cached->head for filename so put here */
 220         extension = get_content_filename(cached->uri, cached);
 221         debug_extension(extension);
 222         if (extension) {
 223                 ctype = get_extension_content_type(extension);
 224                 mem_free(extension);
 225                 if (ctype) {
 226                         return ctype;
 227                 }
 228         }
 229
 230         return NULL;
 231 }
 232
 233 static unsigned char *
 234 get_fragment_content_type(struct cache_entry *cached)
 235 {
 236         struct fragment *fragment;
 237         size_t length;
 238         unsigned char *sample;
 239         unsigned char *ctype = NULL;
 240
 241         if (list_empty(cached->frag))
 242                 return NULL;
 243
 244         fragment = cached->frag.next;
 245         if (fragment->offset)
 246                 return NULL;
 247
 248         length = fragment->length > 1024 ? 1024 : fragment->length;
 249         sample = memacpy(fragment->data, length);
 250         if (!sample)
 251                 return NULL;
 252
 253         if (c_strcasestr((const char *)sample, "<html>"))
 254                 ctype = stracpy("text/html");
 255
 256         mem_free(sample);
 257
 258         return ctype;
 259 }
 260
 261 unsigned char *
 262 get_content_type(struct cache_entry *cached)
 263 {
 264         unsigned char *extension, *ctype;
 265
 266         debug_get_content_type_params(cached);
 267
 268         if (cached->content_type)
 269                 return cached->content_type;
 270
 271         /* If there's one in header, it's simple.. */
 272         if (cached->head) {
 273                 ctype = get_cache_header_content_type(cached);
 274                 if (ctype && *ctype) {
 275                         cached->content_type = ctype;
 276                         return ctype;
 277                 }
 278                 mem_free_if(ctype);
 279         }
 280
 281         /* We can't use the extension string we are getting below, because we
 282          * want to support also things like "ps.gz" - that'd never work, as we
 283          * would always compare only to "gz". */
 284         /* Guess type accordingly to the extension */
 285         extension = get_extension_from_uri(cached->uri);
 286         debug_extension(extension);
 287
 288         if (extension) {
 289                 /* XXX: A little hack for making extension handling case
 290                  * insensitive. We could probably do it better by making
 291                  * guess_encoding() case independent the real problem however
 292                  * is with default (via option system) and mimetypes resolving
 293                  * doing that option and hash lookup will not be easy to
 294                  * convert. --jonas */
 295                 convert_to_lowercase_locale_indep(extension, strlen(extension));
 296
 297                 ctype = get_extension_content_type(extension);
 298                 mem_free(extension);
 299                 if (ctype && *ctype) {
 300                         cached->content_type = ctype;
 301                         return ctype;
 302                 }
 303                 mem_free_if(ctype);
 304         }
 305
 306         ctype = get_fragment_content_type(cached);
 307         if (ctype && *ctype) {
 308                 cached->content_type = ctype;
 309                 return ctype;
 310         }
 311
 312         debug_ctype(get_default_mime_type());
 313
 314         /* text/plain for pager mode */
 315         if (cached->uri && cached->uri->string
 316             && !strcmp(cached->uri->string, "file:///dev/stdin")) {
 317                 cached->content_type = stracpy("text/plain");
 318         } else
 319                 /* Fallback.. use some hardwired default */
 320                 cached->content_type = stracpy(get_default_mime_type());
 321
 322         return cached->content_type;
 323 }
 324
 325 struct mime_handler *
 326 get_mime_type_handler(unsigned char *content_type, int xwin)
 327 {
 328         return get_mime_handler_backends(content_type, xwin);
 329 }
 330
 331 struct string *
 332 add_mime_filename_to_string(struct string *string, struct uri *uri)
 333 {
 334         unsigned char *filename = get_content_filename(uri, NULL);
 335
 336         assert(uri->data);
 337
 338         if (filename) {
 339                 add_shell_safe_to_string(string, filename, strlen(filename));
 340                 mem_free(filename);
 341
 342                 return string;
 343         }
 344
 345         return add_uri_to_string(string, uri, URI_FILENAME);
 346 }
 347
 348 /* Backends dynamic area: */
 349
 350 #include "mime/backend/default.h"
 351 #include "mime/backend/mailcap.h"
 352 #include "mime/backend/mimetypes.h"
 353
 354 static struct module *mime_submodules[] = {
 355         &default_mime_module,
 356 #ifdef CONFIG_MAILCAP
 357         &mailcap_mime_module,
 358 #endif
 359 #ifdef CONFIG_MIMETYPES
 360         &mimetypes_mime_module,
 361 #endif
 362         NULL,
 363 };
 364
 365 struct module mime_module = struct_module(
 366         /* name: */             N_("MIME"),
 367         /* options: */          mime_options,
 368         /* hooks: */            NULL,
 369         /* submodules: */       mime_submodules,
 370         /* data: */             NULL,
 371         /* init: */             NULL,
 372         /* done: */             NULL
 373 );