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 2013-2017 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/mimedesc.h"
  22 #include "../pith/state.h"
  23 #include "../pith/conf.h"
  24 #include "../pith/mailcap.h"
  25 #include "../pith/util.h"
  26
  27 /*
  28  * We've decided not to implement the RFC1524 standard minimum path, because
  29  * some of us think it is harder to debug a problem when you may be misled
  30  * into looking at the wrong mailcap entry.  Likewise for MIME.Types files.
  31  */
  32 #if defined(DOS) || defined(OS2)
  33 #define MT_PATH_SEPARATOR ';'
  34 #define MT_USER_FILE      "MIMETYPE"
  35 #define MT_STDPATH        NULL
  36 #else /* !DOS */
  37 #define MT_PATH_SEPARATOR ':'
  38 #define MT_USER_FILE      NULL
  39 #define MT_STDPATH         \
  40                 ".mime.types:/etc/mime.types:/usr/local/lib/mime.types"
  41 #endif /* !DOS */
  42
  43 #define LINE_BUF_SIZE      2000
  44
  45
  46 /*
  47  * Types used to pass parameters and operator functions to the
  48  * mime.types searching routines.
  49  */
  50 #define MT_MAX_FILE_EXTENSION 3
  51
  52
  53 /*
  54  * Internal prototypes
  55  */
  56 int  mt_browse_types_file(MT_OPERATORPROC, MT_MAP_T *, char *);
  57 int  mt_srch_by_type(MT_MAP_T *, FILE *);
  58
  59
  60
  61 /*
  62  * Exported function that does the work of sniffing the mime.types
  63  * files and filling in the body pointer if found.  Returns 1 (TRUE) if
  64  * extension found, and body pointer filled in, 0 (FALSE) otherwise.
  65  */
  66 int
  67 set_mime_type_by_extension(struct mail_bodystruct *body, char *filename)
  68 {
  69     MT_MAP_T  e2b;
  70
  71     if(mt_get_file_ext(filename, &e2b.from.ext)
  72        && mt_srch_mime_type(mt_srch_by_ext, &e2b)){
  73         body->type    = e2b.to.mime.type;
  74         body->subtype = e2b.to.mime.subtype; /* NOTE: subtype was malloc'd */
  75         return(1);
  76     }
  77
  78     return(0);
  79 }
  80
  81
  82 /*
  83  * Exported function that maps from mime types to file extensions.
  84  */
  85 int
  86 set_mime_extension_by_type (char *ext, char *mtype)
  87 {
  88     MT_MAP_T t2e;
  89
  90     t2e.from.mime_type = mtype;
  91     t2e.to.ext         = ext;
  92     return (mt_srch_mime_type (mt_srch_by_type, &t2e));
  93 }
  94
  95
  96 int
  97 check_mime_type_by_extension (char *ext, char *mtype)
  98 {
  99     MT_MAP_T  e2t;
 100     char mimet[128];
 101
 102     if((e2t.from.ext = ext) != NULL && *e2t.from.ext
 103        && mt_srch_mime_type(mt_srch_by_ext, &e2t)){
 104         snprintf(mimet, sizeof(mimet), "%s/%s",
 105                 body_type_names(e2t.to.mime.type), e2t.to.mime.subtype);
 106         mimet[sizeof(mimet) - 1] = '\0';
 107         fs_give((void **)& e2t.to.mime.subtype);
 108         return strucmp(mimet, mtype) ? 0 : 1;
 109     }
 110
 111     return(0);
 112 }
 113
 114
 115 /*
 116  * Separate and return a pointer to the first character in the 'filename'
 117  * character buffer that comes after the rightmost '.' character in the
 118  * filename. (What I mean is a pointer to the filename - extension).
 119  *
 120  * Returns 1 if an extension is found, 0 otherwise.
 121  */
 122 int
 123 mt_get_file_ext(char *filename, char **extension)
 124 {
 125     dprint((5, "mt_get_file_ext : filename=\"%s\", ",
 126            filename ? filename : "?"));
 127
 128     for(*extension = NULL; filename && *filename; filename++)
 129       if(*filename == '.')
 130         *extension = filename + 1;
 131
 132     dprint((5, "extension=\"%s\"\n",
 133            (extension && *extension) ? *extension : "?"));
 134
 135     return(*extension ? 1 : 0);
 136 }
 137
 138
 139 /*
 140  * Build a list of possible mime.type files.  For each one that exists
 141  * call the mt_operator function.
 142  * Loop terminates when mt_operator returns non-zero.
 143  */
 144 int
 145 mt_srch_mime_type(MT_OPERATORPROC mt_operator, MT_MAP_T *mt_map)
 146 {
 147     char *s, *pathcopy, *path;
 148     int   rv = 0;
 149
 150     dprint((5, "- mt_srch_mime_type -\n"));
 151
 152     pathcopy = mc_conf_path(ps_global->VAR_MIMETYPE_PATH, getenv("MIMETYPES"),
 153                             MT_USER_FILE, MT_PATH_SEPARATOR, MT_STDPATH);
 154
 155     path = pathcopy;                    /* overloaded "path" */
 156
 157     dprint((7, "mime_types: path: %s\n", path ? path : "?"));
 158     while(path){
 159         if((s = strindex(path, MT_PATH_SEPARATOR)) != NULL)
 160           *s++ = '\0';
 161
 162         if((rv = mt_browse_types_file(mt_operator, mt_map, path)) != 0)
 163           break;
 164
 165         path = s;
 166     }
 167
 168     if(pathcopy)
 169       fs_give((void **)&pathcopy);
 170
 171     if(!rv && mime_os_specific_access()){
 172         if(mt_operator == mt_srch_by_ext){
 173             char buf[256];
 174
 175             buf[0] = '\0';
 176             if(mime_get_os_mimetype_from_ext(mt_map->from.ext, buf, 256)){
 177                 if((s = strindex(buf, '/')) != NULL){
 178                     *s++ = '\0';
 179                     mt_map->to.mime.type = mt_translate_type(buf);
 180                     mt_map->to.mime.subtype = cpystr(s);
 181                     rv = 1;
 182                 }
 183             }
 184         }
 185         else if(mt_operator == mt_srch_by_type){
 186             if(mime_get_os_ext_from_mimetype(mt_map->from.mime_type,
 187                                   mt_map->to.ext, 32)){
 188                 /* the 32 comes from var ext[] in display_attachment() */
 189                 if(*(s = mt_map->to.ext) == '.')
 190                   while((*s = *(s+1)) != '\0')
 191                     s++;
 192
 193                 rv = 1;
 194             }
 195         }
 196         else
 197           alpine_panic("Unhandled mime type search");
 198     }
 199
 200     /* if we still can not find the type, but it is a .docx (or alike) extension
 201         set the type here. Do not use the grope function.
 202      */
 203     if(rv == 0){
 204         rv = 1; /* assume success */
 205         mt_map->to.mime.type = TYPEAPPLICATION;
 206         if(!strucmp(mt_map->from.ext, "docx"))
 207             mt_map->to.mime.subtype = cpystr("VND.OPENXMLFORMATS-OFFICEDOCUMENT.WORDPROCESSINGML.DOCUMENT");
 208         else if(!strucmp(mt_map->from.ext, "xslx"))
 209             mt_map->to.mime.subtype = cpystr("VND.OPENXMLFORMATS-OFFICEDOCUMENT.SPREADSHEETML.SHEET");
 210         else if(!strucmp(mt_map->from.ext, "xltx"))
 211             mt_map->to.mime.subtype = cpystr("VND.OPENXMLFORMATS-OFFICEDOCUMENT.SPREADSHEETML.TEMPLATE");
 212         else if(!strucmp(mt_map->from.ext, "potx"))
 213             mt_map->to.mime.subtype = cpystr("VND.OPENXMLFORMATS-OFFICEDOCUMENT.PRESENTATIONML.TEMPLATE");
 214         else if(!strucmp(mt_map->from.ext, "ppsx"))
 215             mt_map->to.mime.subtype = cpystr("VND.OPENXMLFORMATS-OFFICEDOCUMENT.PRESENTATIONML.SLIDESHOW");
 216         else if(!strucmp(mt_map->from.ext, "pptx"))
 217             mt_map->to.mime.subtype = cpystr("VND.OPENXMLFORMATS-OFFICEDOCUMENT.PRESENTATIONML.PRESENTATION");
 218         else if(!strucmp(mt_map->from.ext, "sldx"))
 219             mt_map->to.mime.subtype = cpystr("VND.OPENXMLFORMATS-OFFICEDOCUMENT.PRESENTATIONML.SLIDE");
 220         else if(!strucmp(mt_map->from.ext, "dotx"))
 221             mt_map->to.mime.subtype = cpystr("VND.OPENXMLFORMATS-OFFICEDOCUMENT.WORDPROCESSINGML.TEMPLATE");
 222         else if(!strucmp(mt_map->from.ext, "xlam"))
 223             mt_map->to.mime.subtype = cpystr("VND.MS-EXCEL.ADDIN.MACROENABLED.12");
 224         else if(!strucmp(mt_map->from.ext, "xslb"))
 225             mt_map->to.mime.subtype = cpystr("VND.MS-EXCEL.SHEET.BINARY.MACROENABLED.12");
 226         else rv = 0;    /* else, failure */
 227     }
 228
 229
 230     return(rv);
 231 }
 232
 233
 234 /*
 235  * Try to match a file extension against extensions found in the file
 236  * ``filename'' if that file exists. return 1 if a match
 237  * was found and 0 in all other cases.
 238  */
 239 int
 240 mt_browse_types_file(MT_OPERATORPROC mt_operator, MT_MAP_T *mt_map, char *filename)
 241 {
 242     int   rv = 0;
 243     FILE *file;
 244
 245     dprint((7, "mt_browse_types_file(%s)\n", filename ? filename : "?"));
 246     if((file = our_fopen(filename, "rb")) != NULL){
 247         rv = (*mt_operator)(mt_map, file);
 248         fclose(file);
 249     }
 250     else{
 251       dprint((1, "mt_browse: FAILED open(%s) : %s.\n",
 252                  filename ? filename : "?", error_description(errno)));
 253     }
 254
 255     return(rv);
 256 }
 257
 258
 259 /*
 260  * scan each line of the file. Treat each line as a mime type definition.
 261  * The first word is a type/subtype specification. All following words
 262  * are file extensions belonging to that type/subtype. Words are separated
 263  * bij whitespace characters.
 264  * If a file extension occurs more than once, then the first definition
 265  * determines the file type and subtype.
 266  */
 267 int
 268 mt_srch_by_ext(MT_MAP_T *e2b, FILE *file)
 269 {
 270     char buffer[LINE_BUF_SIZE];
 271
 272     /* construct a loop reading the file line by line. Then check each
 273      * line for a matching definition.
 274      */
 275     while(fgets(buffer,LINE_BUF_SIZE,file) != NULL){
 276         char *typespec;
 277         char *try_extension;
 278
 279         if(buffer[0] == '#')
 280           continue;             /* comment */
 281
 282         /* divide the input buffer into words separated by whitespace.
 283          * The first words is the type and subtype. All following words
 284          * are file extensions.
 285          */
 286         dprint((5, "traverse: buffer=\"%s\"\n", buffer));
 287         typespec = strtok(buffer," \t");        /* extract type,subtype  */
 288         if(!typespec)
 289           continue;
 290
 291         dprint((5, "typespec=\"%s\"\n", typespec ? typespec : "?"));
 292         while((try_extension = strtok(NULL, " \t\n\r")) != NULL){
 293             /* compare the extensions, and assign the type if a match
 294              * is found.
 295              */
 296             dprint((5,"traverse: trying ext \"%s\"\n",try_extension));
 297             if(strucmp(try_extension, e2b->from.ext) == 0){
 298                 /* split the 'type/subtype' specification */
 299                 char *type, *subtype = NULL;
 300
 301                 type = strtok(typespec,"/");
 302                 if(type)
 303                   subtype = strtok(NULL,"/");
 304
 305                 dprint((5, "traverse: type=%s, subtype=%s.\n",
 306                            type ? type : "<null>",
 307                            subtype ? subtype : "<null>"));
 308                 /* The type is encoded as a small integer. we have to
 309                  * translate the character string naming the type into
 310                  * the corresponding number.
 311                  */
 312                 e2b->to.mime.type    = mt_translate_type(type);
 313                 e2b->to.mime.subtype = cpystr(subtype ? subtype : "x-unknown");
 314                 return 1; /* a match has been found */
 315             }
 316         }
 317     }
 318
 319     dprint((5, "traverse: search failed.\n"));
 320     return 0;
 321 }
 322
 323
 324 /*
 325  * scan each line of the file. Treat each line as a mime type definition.
 326  * Here we are looking for a matching type.  When that is found return the
 327  * first extension that is three chars or less.
 328  */
 329 int
 330 mt_srch_by_type(MT_MAP_T *t2e, FILE *file)
 331 {
 332     char buffer[LINE_BUF_SIZE];
 333
 334     /* construct a loop reading the file line by line. Then check each
 335      * line for a matching definition.
 336      */
 337     while(fgets(buffer,LINE_BUF_SIZE,file) != NULL){
 338         char *typespec;
 339         char *try_extension;
 340
 341         if(buffer[0] == '#')
 342           continue;             /* comment */
 343
 344         /* divide the input buffer into words separated by whitespace.
 345          * The first words is the type and subtype. All following words
 346          * are file extensions.
 347          */
 348         dprint((5, "traverse: buffer=%s.\n", buffer));
 349         typespec = strtok(buffer," \t");        /* extract type,subtype  */
 350         dprint((5, "typespec=%s.\n", typespec ? typespec : "?"));
 351         if (strucmp (typespec, t2e->from.mime_type) == 0) {
 352             while((try_extension = strtok(NULL, " \t\n\r")) != NULL){
 353                 if (strlen (try_extension) <= MT_MAX_FILE_EXTENSION) {
 354                     strncpy (t2e->to.ext, try_extension, 32);
 355                     /*
 356                      * not sure of the 32, so don't write to byte 32
 357                      * on purpose with
 358                      *     t2e->to.ext[31] = '\0';
 359                      * in case that breaks something
 360                      */
 361
 362                     return (1);
 363                 }
 364             }
 365         }
 366     }
 367
 368     dprint((5, "traverse: search failed.\n"));
 369     return 0;
 370 }
 371
 372
 373 /*
 374  * Translate a character string representing a content type into a short
 375  * integer number, according to the coding described in c-client/mail.h
 376  * List of content types taken from rfc1521, September 1993.
 377  */
 378 int
 379 mt_translate_type(char *type)
 380 {
 381     int i;
 382
 383     for (i=0;(i<=TYPEMAX) && body_types[i] && strucmp(type,body_types[i]);i++)
 384       ;
 385
 386     if (i > TYPEMAX)
 387       i = TYPEOTHER;
 388     else if (!body_types[i])    /* if empty slot, assign it to this type */
 389       body_types[i] = cpystr (type);
 390
 391     return(i);
 392 }