pith/mimetype.c

   1 #if !defined(lint) && !defined(DOS)
   2 static char rcsid[] = "$Id: mimetype.c 955 2008-03-06 23:52:36Z hubert@u.washington.edu $";
   3 #endif
   4
   5 /*
   6  * ========================================================================
   7  * Copyright 2006-2008 University of Washington
   8  * Copyright 2013-2016 Eduardo Chappa
   9  *
  10  * Licensed under the Apache License, Version 2.0 (the "License");
  11  * you may not use this file except in compliance with the License.
  12  * You may obtain a copy of the License at
  13  *
  14  *     http://www.apache.org/licenses/LICENSE-2.0
  15  *
  16  * ========================================================================
  17  */
  18
  19 #include "../pith/headers.h"
  20 #include "../pith/mimetype.h"
  21 #include "../pith/state.h"
  22 #include "../pith/conf.h"
  23 #include "../pith/mailcap.h"
  24 #include "../pith/util.h"
  25
  26 /*
  27  * We've decided not to implement the RFC1524 standard minimum path, because
  28  * some of us think it is harder to debug a problem when you may be misled
  29  * into looking at the wrong mailcap entry.  Likewise for MIME.Types files.
  30  */
  31 #if defined(DOS) || defined(OS2)
  32 #define MT_PATH_SEPARATOR ';'
  33 #define MT_USER_FILE      "MIMETYPE"
  34 #define MT_STDPATH        NULL
  35 #else /* !DOS */
  36 #define MT_PATH_SEPARATOR ':'
  37 #define MT_USER_FILE      NULL
  38 #define MT_STDPATH         \
  39                 ".mime.types:/etc/mime.types:/usr/local/lib/mime.types"
  40 #endif /* !DOS */
  41
  42 #define LINE_BUF_SIZE      2000
  43
  44
  45 /*
  46  * Types used to pass parameters and operator functions to the
  47  * mime.types searching routines.
  48  */
  49 #define MT_MAX_FILE_EXTENSION 3
  50
  51
  52 /*
  53  * Internal prototypes
  54  */
  55 int  mt_browse_types_file(MT_OPERATORPROC, MT_MAP_T *, char *);
  56 int  mt_srch_by_type(MT_MAP_T *, FILE *);
  57
  58
  59
  60 /*
  61  * Exported function that does the work of sniffing the mime.types
  62  * files and filling in the body pointer if found.  Returns 1 (TRUE) if
  63  * extension found, and body pointer filled in, 0 (FALSE) otherwise.
  64  */
  65 int
  66 set_mime_type_by_extension(struct mail_bodystruct *body, char *filename)
  67 {
  68     MT_MAP_T  e2b;
  69
  70     if(mt_get_file_ext(filename, &e2b.from.ext)
  71        && mt_srch_mime_type(mt_srch_by_ext, &e2b)){
  72         body->type    = e2b.to.mime.type;
  73         body->subtype = e2b.to.mime.subtype; /* NOTE: subtype was malloc'd */
  74         return(1);
  75     }
  76
  77     return(0);
  78 }
  79
  80
  81 /*
  82  * Exported function that maps from mime types to file extensions.
  83  */
  84 int
  85 set_mime_extension_by_type (char *ext, char *mtype)
  86 {
  87     MT_MAP_T t2e;
  88
  89     t2e.from.mime_type = mtype;
  90     t2e.to.ext         = ext;
  91     return (mt_srch_mime_type (mt_srch_by_type, &t2e));
  92 }
  93
  94
  95
  96
  97 /*
  98  * Separate and return a pointer to the first character in the 'filename'
  99  * character buffer that comes after the rightmost '.' character in the
 100  * filename. (What I mean is a pointer to the filename - extension).
 101  *
 102  * Returns 1 if an extension is found, 0 otherwise.
 103  */
 104 int
 105 mt_get_file_ext(char *filename, char **extension)
 106 {
 107     dprint((5, "mt_get_file_ext : filename=\"%s\", ",
 108            filename ? filename : "?"));
 109
 110     for(*extension = NULL; filename && *filename; filename++)
 111       if(*filename == '.')
 112         *extension = filename + 1;
 113
 114     dprint((5, "extension=\"%s\"\n",
 115            (extension && *extension) ? *extension : "?"));
 116
 117     return(*extension ? 1 : 0);
 118 }
 119
 120
 121 /*
 122  * Build a list of possible mime.type files.  For each one that exists
 123  * call the mt_operator function.
 124  * Loop terminates when mt_operator returns non-zero.
 125  */
 126 int
 127 mt_srch_mime_type(MT_OPERATORPROC mt_operator, MT_MAP_T *mt_map)
 128 {
 129     char *s, *pathcopy, *path;
 130     int   rv = 0;
 131
 132     dprint((5, "- mt_srch_mime_type -\n"));
 133
 134     pathcopy = mc_conf_path(ps_global->VAR_MIMETYPE_PATH, getenv("MIMETYPES"),
 135                             MT_USER_FILE, MT_PATH_SEPARATOR, MT_STDPATH);
 136
 137     path = pathcopy;                    /* overloaded "path" */
 138
 139     dprint((7, "mime_types: path: %s\n", path ? path : "?"));
 140     while(path){
 141         if((s = strindex(path, MT_PATH_SEPARATOR)) != NULL)
 142           *s++ = '\0';
 143
 144         if((rv = mt_browse_types_file(mt_operator, mt_map, path)) != 0)
 145           break;
 146
 147         path = s;
 148     }
 149
 150     if(pathcopy)
 151       fs_give((void **)&pathcopy);
 152
 153     if(!rv && mime_os_specific_access()){
 154         if(mt_operator == mt_srch_by_ext){
 155             char buf[256];
 156
 157             buf[0] = '\0';
 158             if(mime_get_os_mimetype_from_ext(mt_map->from.ext, buf, 256)){
 159                 if((s = strindex(buf, '/')) != NULL){
 160                     *s++ = '\0';
 161                     mt_map->to.mime.type = mt_translate_type(buf);
 162                     mt_map->to.mime.subtype = cpystr(s);
 163                     rv = 1;
 164                 }
 165             }
 166         }
 167         else if(mt_operator == mt_srch_by_type){
 168             if(mime_get_os_ext_from_mimetype(mt_map->from.mime_type,
 169                                   mt_map->to.ext, 32)){
 170                 /* the 32 comes from var ext[] in display_attachment() */
 171                 if(*(s = mt_map->to.ext) == '.')
 172                   while((*s = *(s+1)) != '\0')
 173                     s++;
 174
 175                 rv = 1;
 176             }
 177         }
 178         else
 179           alpine_panic("Unhandled mime type search");
 180     }
 181
 182     /* if we still can not find the type, but it is a .docx (or alike) extension
 183         set the type here. Do not use the grope function.
 184      */
 185     if(rv == 0){
 186         rv = 1; /* assume success */
 187         mt_map->to.mime.type = TYPEAPPLICATION;
 188         if(!strucmp(mt_map->from.ext, "docx"))
 189             mt_map->to.mime.subtype = cpystr("VND.OPENXMLFORMATS-OFFICEDOCUMENT.WORDPROCESSINGML.DOCUMENT");
 190         else if(!strucmp(mt_map->from.ext, "xslx"))
 191             mt_map->to.mime.subtype = cpystr("VND.OPENXMLFORMATS-OFFICEDOCUMENT.SPREADSHEETML.SHEET");
 192         else if(!strucmp(mt_map->from.ext, "xltx"))
 193             mt_map->to.mime.subtype = cpystr("VND.OPENXMLFORMATS-OFFICEDOCUMENT.SPREADSHEETML.TEMPLATE");
 194         else if(!strucmp(mt_map->from.ext, "potx"))
 195             mt_map->to.mime.subtype = cpystr("VND.OPENXMLFORMATS-OFFICEDOCUMENT.PRESENTATIONML.TEMPLATE");
 196         else if(!strucmp(mt_map->from.ext, "ppsx"))
 197             mt_map->to.mime.subtype = cpystr("VND.OPENXMLFORMATS-OFFICEDOCUMENT.PRESENTATIONML.SLIDESHOW");
 198         else if(!strucmp(mt_map->from.ext, "pptx"))
 199             mt_map->to.mime.subtype = cpystr("VND.OPENXMLFORMATS-OFFICEDOCUMENT.PRESENTATIONML.PRESENTATION");
 200         else if(!strucmp(mt_map->from.ext, "sldx"))
 201             mt_map->to.mime.subtype = cpystr("VND.OPENXMLFORMATS-OFFICEDOCUMENT.PRESENTATIONML.SLIDE");
 202         else if(!strucmp(mt_map->from.ext, "dotx"))
 203             mt_map->to.mime.subtype = cpystr("VND.OPENXMLFORMATS-OFFICEDOCUMENT.WORDPROCESSINGML.TEMPLATE");
 204         else if(!strucmp(mt_map->from.ext, "xlam"))
 205             mt_map->to.mime.subtype = cpystr("VND.MS-EXCEL.ADDIN.MACROENABLED.12");
 206         else if(!strucmp(mt_map->from.ext, "xslb"))
 207             mt_map->to.mime.subtype = cpystr("VND.MS-EXCEL.SHEET.BINARY.MACROENABLED.12");
 208         else rv = 0;    /* else, failure */
 209     }
 210
 211
 212     return(rv);
 213 }
 214
 215
 216 /*
 217  * Try to match a file extension against extensions found in the file
 218  * ``filename'' if that file exists. return 1 if a match
 219  * was found and 0 in all other cases.
 220  */
 221 int
 222 mt_browse_types_file(MT_OPERATORPROC mt_operator, MT_MAP_T *mt_map, char *filename)
 223 {
 224     int   rv = 0;
 225     FILE *file;
 226
 227     dprint((7, "mt_browse_types_file(%s)\n", filename ? filename : "?"));
 228     if((file = our_fopen(filename, "rb")) != NULL){
 229         rv = (*mt_operator)(mt_map, file);
 230         fclose(file);
 231     }
 232     else{
 233       dprint((1, "mt_browse: FAILED open(%s) : %s.\n",
 234                  filename ? filename : "?", error_description(errno)));
 235     }
 236
 237     return(rv);
 238 }
 239
 240
 241 /*
 242  * scan each line of the file. Treat each line as a mime type definition.
 243  * The first word is a type/subtype specification. All following words
 244  * are file extensions belonging to that type/subtype. Words are separated
 245  * bij whitespace characters.
 246  * If a file extension occurs more than once, then the first definition
 247  * determines the file type and subtype.
 248  */
 249 int
 250 mt_srch_by_ext(MT_MAP_T *e2b, FILE *file)
 251 {
 252     char buffer[LINE_BUF_SIZE];
 253
 254     /* construct a loop reading the file line by line. Then check each
 255      * line for a matching definition.
 256      */
 257     while(fgets(buffer,LINE_BUF_SIZE,file) != NULL){
 258         char *typespec;
 259         char *try_extension;
 260
 261         if(buffer[0] == '#')
 262           continue;             /* comment */
 263
 264         /* divide the input buffer into words separated by whitespace.
 265          * The first words is the type and subtype. All following words
 266          * are file extensions.
 267          */
 268         dprint((5, "traverse: buffer=\"%s\"\n", buffer));
 269         typespec = strtok(buffer," \t");        /* extract type,subtype  */
 270         if(!typespec)
 271           continue;
 272
 273         dprint((5, "typespec=\"%s\"\n", typespec ? typespec : "?"));
 274         while((try_extension = strtok(NULL, " \t\n\r")) != NULL){
 275             /* compare the extensions, and assign the type if a match
 276              * is found.
 277              */
 278             dprint((5,"traverse: trying ext \"%s\"\n",try_extension));
 279             if(strucmp(try_extension, e2b->from.ext) == 0){
 280                 /* split the 'type/subtype' specification */
 281                 char *type, *subtype = NULL;
 282
 283                 type = strtok(typespec,"/");
 284                 if(type)
 285                   subtype = strtok(NULL,"/");
 286
 287                 dprint((5, "traverse: type=%s, subtype=%s.\n",
 288                            type ? type : "<null>",
 289                            subtype ? subtype : "<null>"));
 290                 /* The type is encoded as a small integer. we have to
 291                  * translate the character string naming the type into
 292                  * the corresponding number.
 293                  */
 294                 e2b->to.mime.type    = mt_translate_type(type);
 295                 e2b->to.mime.subtype = cpystr(subtype ? subtype : "x-unknown");
 296                 return 1; /* a match has been found */
 297             }
 298         }
 299     }
 300
 301     dprint((5, "traverse: search failed.\n"));
 302     return 0;
 303 }
 304
 305
 306 /*
 307  * scan each line of the file. Treat each line as a mime type definition.
 308  * Here we are looking for a matching type.  When that is found return the
 309  * first extension that is three chars or less.
 310  */
 311 int
 312 mt_srch_by_type(MT_MAP_T *t2e, FILE *file)
 313 {
 314     char buffer[LINE_BUF_SIZE];
 315
 316     /* construct a loop reading the file line by line. Then check each
 317      * line for a matching definition.
 318      */
 319     while(fgets(buffer,LINE_BUF_SIZE,file) != NULL){
 320         char *typespec;
 321         char *try_extension;
 322
 323         if(buffer[0] == '#')
 324           continue;             /* comment */
 325
 326         /* divide the input buffer into words separated by whitespace.
 327          * The first words is the type and subtype. All following words
 328          * are file extensions.
 329          */
 330         dprint((5, "traverse: buffer=%s.\n", buffer));
 331         typespec = strtok(buffer," \t");        /* extract type,subtype  */
 332         dprint((5, "typespec=%s.\n", typespec ? typespec : "?"));
 333         if (strucmp (typespec, t2e->from.mime_type) == 0) {
 334             while((try_extension = strtok(NULL, " \t\n\r")) != NULL){
 335                 if (strlen (try_extension) <= MT_MAX_FILE_EXTENSION) {
 336                     strncpy (t2e->to.ext, try_extension, 32);
 337                     /*
 338                      * not sure of the 32, so don't write to byte 32
 339                      * on purpose with
 340                      *     t2e->to.ext[31] = '\0';
 341                      * in case that breaks something
 342                      */
 343
 344                     return (1);
 345                 }
 346             }
 347         }
 348     }
 349
 350     dprint((5, "traverse: search failed.\n"));
 351     return 0;
 352 }
 353
 354
 355 /*
 356  * Translate a character string representing a content type into a short
 357  * integer number, according to the coding described in c-client/mail.h
 358  * List of content types taken from rfc1521, September 1993.
 359  */
 360 int
 361 mt_translate_type(char *type)
 362 {
 363     int i;
 364
 365     for (i=0;(i<=TYPEMAX) && body_types[i] && strucmp(type,body_types[i]);i++)
 366       ;
 367
 368     if (i > TYPEMAX)
 369       i = TYPEOTHER;
 370     else if (!body_types[i])    /* if empty slot, assign it to this type */
 371       body_types[i] = cpystr (type);
 372
 373     return(i);
 374 }