diplodocus.org Git - nmh/blob - h/mhparse.h

   1 /* mhparse.h -- definitions for parsing/building of MIME content
   2  *           -- (mhparse.c/mhbuildsbr.c)
   3  */
   4
   5 /*
   6  * Abstract type for header fields
   7  */
   8 typedef struct hfield *HF;
   9
  10 /*
  11  * Abstract types for MIME parsing/building
  12  */
  13 typedef struct cefile           *CE;
  14 typedef struct CTinfo           *CI;
  15 typedef struct Content          *CT;
  16 typedef struct Parameter        *PM;
  17
  18 /*
  19  * type for Init function (both type and transfer encoding)
  20  */
  21 typedef int (*InitFunc) (CT);
  22
  23 /*
  24  * types for various transfer encoding access functions
  25  */
  26 typedef int (*OpenCEFunc) (CT, char **);
  27 typedef void (*CloseCEFunc) (CT);
  28 typedef unsigned long (*SizeCEFunc) (CT);
  29
  30 /*
  31  * Structure for storing/encoding/decoding
  32  * a header field and its value.
  33  */
  34 struct hfield {
  35     char *name;         /* field name */
  36     char *value;        /* field body */
  37     int hf_encoding;    /* internal flag for transfer encoding to use */
  38     HF next;            /* link to next header field */
  39 };
  40
  41 /*
  42  * Structure for holding MIME parameter elements.
  43  */
  44 struct Parameter {
  45     char *pm_name;      /* Parameter name */
  46     char *pm_value;     /* Parameter value */
  47     char *pm_charset;   /* Parameter character set (optional) */
  48     char *pm_lang;      /* Parameter language tag (optional) */
  49     PM   pm_next;       /* Pointer to next element */
  50 };
  51
  52 /*
  53  * Structure for storing parsed elements
  54  * of the Content-Type component.
  55  */
  56 struct CTinfo {
  57     char *ci_type;              /* content type     */
  58     char *ci_subtype;           /* content subtype  */
  59     PM   ci_first_pm;           /* Pointer to first MIME parameter */
  60     PM   ci_last_pm;            /* Pointer to last MIME parameter */
  61     char *ci_comment;           /* RFC-822 comments */
  62     char *ci_magic;
  63 };
  64
  65 /*
  66  * Structure for storing decoded contents after
  67  * removing Content-Transfer-Encoding.
  68  */
  69 struct cefile {
  70     char *ce_file;      /* decoded content (file)   */
  71     FILE *ce_fp;        /* decoded content (stream) */
  72     int   ce_unlink;    /* remove file when done?   */
  73 };
  74
  75 /*
  76  * Primary structure for handling Content (Entity)
  77  *
  78  * Some more explanation of this:
  79  *
  80  * This structure recursively describes a complete MIME message.
  81  * At the top level, the c_first_hf list has a list of all message
  82  * headers.  If the content-type is multipart (c_type == CT_MULTIPART)
  83  * then c_ctparams will contain a pointer to a struct multipart.
  84  * A struct multipart contains (among other things) a linked list
  85  * of struct part elements, and THOSE contain a pointer to the sub-part's
  86  * Content structure.
  87  *
  88  * An extra note for message/external-body parts.  The enclosing
  89  * content structure is marked as a message/external-body; the c_ctparams
  90  * contains a pointer to a struct exbody, which contains a pointer to
  91  * (among other things) the "real" content (e.g., application/octet-stream).
  92  * The "real" content structure has the c_ctexbody pointer back to the
  93  * same struct exbody sees in the enclosing content structure (the struct
  94  * exbody contains parent pointers if you need to traverse up the content
  95  * structure).  Hopefully this makes it clearer:
  96  *
  97  * Enclosing content:
  98  *   Type: message/external-body
  99  *   c_ctparams: pointer to "struct exbody"
 100  *   c_ctexbody: NULL
 101  *
 102  * "Real" content:
 103  *   Type: application/octet-stream (or whatever)
 104  *   c_ctparams: NULL
 105  *   c_ctexbody: pointer to "struct exbody"
 106  *
 107  */
 108 struct Content {
 109     /* source (read) file */
 110     char *c_file;               /* read contents (file)              */
 111     FILE *c_fp;                 /* read contents (stream)            */
 112     int c_unlink;               /* remove file when done?            */
 113
 114     long c_begin;               /* where content body starts in file */
 115     long c_end;                 /* where content body ends in file   */
 116
 117     /* linked list of header fields */
 118     HF c_first_hf;              /* pointer to first header field     */
 119     HF c_last_hf;               /* pointer to last header field      */
 120
 121     /* copies of MIME related header fields */
 122     char *c_vrsn;               /* MIME-Version:                     */
 123     char *c_ctline;             /* Content-Type:                     */
 124     char *c_celine;             /* Content-Transfer-Encoding:        */
 125     char *c_id;                 /* Content-ID:                       */
 126     char *c_descr;              /* Content-Description:              */
 127     char *c_dispo;              /* Content-Disposition:              */
 128     char *c_dispo_type;         /* Type of Content-Disposition       */
 129     PM c_dispo_first;           /* Pointer to first disposition parm */
 130     PM c_dispo_last;            /* Pointer to last disposition parm  */
 131     char *c_partno;             /* within multipart content          */
 132
 133     /* Content-Type info */
 134     struct CTinfo c_ctinfo;     /* parsed elements of Content-Type   */
 135     int c_type;                 /* internal flag for content type    */
 136     int c_subtype;              /* internal flag for content subtype */
 137
 138     /* Content-Transfer-Encoding info (decoded contents) */
 139     struct cefile c_cefile;     /* structure holding decoded content */
 140     int c_encoding;             /* internal flag for encoding type   */
 141     int c_reqencoding;          /* Requested encoding (by mhbuild)   */
 142
 143     /* Content-MD5 info */
 144     int c_digested;             /* have we seen this header before?  */
 145     unsigned char c_digest[16]; /* decoded MD5 checksum              */
 146
 147     /* pointers to content-specific structures */
 148     void *c_ctparams;           /* content type specific data        */
 149     struct exbody *c_ctexbody;  /* data for type message/external    */
 150
 151     /* function pointers */
 152     InitFunc    c_ctinitfnx;    /* parse content body                */
 153     OpenCEFunc  c_ceopenfnx;    /* get a stream to decoded contents  */
 154     CloseCEFunc c_ceclosefnx;   /* release stream                    */
 155     SizeCEFunc  c_cesizefnx;    /* size of decoded contents          */
 156
 157     int c_umask;                /* associated umask                  */
 158     int c_rfc934;               /* RFC 934 compatibility flag        */
 159
 160     char *c_showproc;           /* default, if not in profile        */
 161     char *c_termproc;           /* for charset madness...            */
 162     char *c_storeproc;          /* overrides profile entry, if any   */
 163
 164     char *c_storage;            /* write contents (file)             */
 165     char *c_folder;             /* write contents (folder)           */
 166 };
 167
 168 /*
 169  * Flags for Content-Type (Content->c_type)
 170  */
 171 #define CT_UNKNOWN      0x00
 172 #define CT_APPLICATION  0x01
 173 #define CT_AUDIO        0x02
 174 #define CT_IMAGE        0x03
 175 #define CT_MESSAGE      0x04
 176 #define CT_MULTIPART    0x05
 177 #define CT_TEXT         0x06
 178 #define CT_VIDEO        0x07
 179 #define CT_EXTENSION    0x08
 180
 181 /*
 182  * Flags for Content-Transfer-Encoding (Content->c_encoding)
 183  */
 184 #define CE_UNKNOWN      0x00
 185 #define CE_BASE64       0x01
 186 #define CE_QUOTED       0x02
 187 #define CE_8BIT         0x03
 188 #define CE_7BIT         0x04
 189 #define CE_BINARY       0x05
 190 #define CE_EXTENSION    0x06
 191 #define CE_EXTERNAL     0x07    /* for external-body */
 192
 193 /*
 194  * TEXT content
 195  */
 196
 197 /* Flags for subtypes of TEXT */
 198 #define TEXT_UNKNOWN    0x00
 199 #define TEXT_PLAIN      0x01
 200 #define TEXT_RICHTEXT   0x02
 201 #define TEXT_ENRICHED   0x03
 202
 203 /* Flags for character sets */
 204 #define CHARSET_SPECIFIED    0x00
 205 #define CHARSET_UNSPECIFIED 0x01  /* only needed when building drafts */
 206
 207 /* Structure for text content */
 208 struct text {
 209     int tx_charset;             /* flag for character set */
 210     int lf_line_endings;        /* Whether to use CR LF (0) or LF (1) line
 211                                    endings.  The meaning of 0 was selected so
 212                                    that CR LF is the default, in accordance
 213                                    with RFC 2046, Sec. 4.1.1, Par. 1. */
 214 };
 215
 216 /*
 217  * MULTIPART content
 218  */
 219
 220 /* Flags for subtypes of MULTIPART */
 221 #define MULTI_UNKNOWN   0x00
 222 #define MULTI_MIXED     0x01
 223 #define MULTI_ALTERNATE 0x02
 224 #define MULTI_DIGEST    0x03
 225 #define MULTI_PARALLEL  0x04
 226 #define MULTI_RELATED   0x05
 227
 228 /* Structure for subparts of a multipart content */
 229 struct part {
 230     CT mp_part;                 /* Content structure for subpart     */
 231     struct part *mp_next;       /* pointer to next subpart structure */
 232 };
 233
 234 /* Main structure for multipart content */
 235 struct multipart {
 236     char *mp_start;             /* boundary string separating parts   */
 237     char *mp_stop;              /* terminating boundary string        */
 238     char *mp_content_before;    /* any content before the first subpart */
 239     char *mp_content_after;     /* any content after the last subpart */
 240     struct part *mp_parts;      /* pointer to first subpart structure */
 241 };
 242
 243 /*
 244  * MESSAGE content
 245  */
 246
 247 /* Flags for subtypes of MESSAGE */
 248 #define MESSAGE_UNKNOWN  0x00
 249 #define MESSAGE_RFC822   0x01
 250 #define MESSAGE_PARTIAL  0x02
 251 #define MESSAGE_EXTERNAL 0x03
 252
 253 /* Structure for message/partial */
 254 struct partial {
 255     char *pm_partid;
 256     int pm_partno;
 257     int pm_maxno;
 258     int pm_marked;
 259     int pm_stored;
 260 };
 261
 262 /* Structure for message/external */
 263 struct exbody {
 264     CT eb_parent;       /* pointer to controlling content structure */
 265     CT eb_content;      /* pointer to internal content structure    */
 266     char *eb_partno;
 267     char *eb_access;
 268     int eb_flags;
 269     char *eb_name;
 270     char *eb_permission;
 271     char *eb_site;
 272     char *eb_dir;
 273     char *eb_mode;
 274     unsigned long eb_size;
 275     char *eb_server;
 276     char *eb_subject;
 277     char *eb_body;
 278     char *eb_url;
 279 };
 280
 281 /*
 282  * APPLICATION content
 283  */
 284
 285 /* Flags for subtype of APPLICATION */
 286 #define APPLICATION_UNKNOWN     0x00
 287 #define APPLICATION_OCTETS      0x01
 288 #define APPLICATION_POSTSCRIPT  0x02
 289
 290
 291 /*
 292  * Structures for mapping types to their internal flags
 293  */
 294 struct k2v {
 295     char *kv_key;
 296     int   kv_value;
 297 };
 298 extern struct k2v SubText[];
 299 extern struct k2v Charset[];
 300 extern struct k2v SubMultiPart[];
 301 extern struct k2v SubMessage[];
 302 extern struct k2v SubApplication[];
 303
 304 /*
 305  * Structures for mapping (content) types to
 306  * the functions to handle them.
 307  */
 308 struct str2init {
 309     char *si_key;
 310     int   si_val;
 311     InitFunc si_init;
 312 };
 313 extern struct str2init str2cts[];
 314 extern struct str2init str2ces[];
 315 extern struct str2init str2methods[];
 316
 317 /*
 318  * prototypes
 319  */
 320 CT parse_mime (char *);
 321
 322 /*
 323  * Translate a composition file into a MIME data structure.  Arguments are:
 324  *
 325  * infile       - Name of input filename
 326  * autobuild    - A flag to indicate if the composition file parser is
 327  *                being run in automatic mode or not.  In auto mode,
 328  *                if a MIME-Version header is encountered it is assumed
 329  *                that the composition file is already in MIME format
 330  *                and will not be processed further.  Otherwise, an
 331  *                error is generated.
 332  * dist         - A flag to indicate if we are being run by "dist".  In
 333  *                that case, add no MIME headers to the message.  Existing
 334  *                headers will still be encoded by RFC 2047.
 335  * directives   - A flag to control whether or not build directives are
 336  *                processed by default.
 337  * encoding     - The default encoding to use when doing RFC 2047 header
 338  *                encoding.  Must be one of CE_UNKNOWN, CE_BASE64, or
 339  *                CE_QUOTED.
 340  * maxunencoded - The maximum line length before the default encoding for
 341  *                text parts is quoted-printable.
 342  * verbose      - If 1, output verbose information during message composition
 343  *
 344  * Returns a CT structure describing the resulting MIME message.  If the
 345  * -auto flag is set and a MIME-Version header is encountered, the return
 346  * value is NULL.
 347  */
 348 CT build_mime (char *infile, int autobuild, int dist, int directives,
 349                int encoding, size_t maxunencoded, int verbose);
 350
 351 int add_header (CT, char *, char *);
 352 int get_ctinfo (char *, CT, int);
 353 int params_external (CT, int);
 354 int open7Bit (CT, char **);
 355 void close_encoding (CT);
 356 char *ct_type_str (int) CONST;
 357 char *ct_subtype_str (int, int) CONST;
 358 int ct_str_type (const char *) PURE;
 359 int ct_str_subtype (int, const char *) PURE;
 360 const struct str2init *get_ct_init (int) PURE;
 361 const char *ce_str (int) CONST;
 362 const struct str2init *get_ce_method (const char *) PURE;
 363 char *content_charset (CT);
 364 void reverse_alternative_parts (CT);
 365
 366 /*
 367  * Given a list of messages, display information about them on standard
 368  * output.
 369  *
 370  * Arguments are:
 371  *
 372  * cts          - An array of CT elements of messages that need to be
 373  *                displayed.  Array is terminated by a NULL.
 374  * headsw       - If 1, display a column header.
 375  * sizesw       - If 1, display the size of the part.
 376  * verbosw      - If 1, display verbose information
 377  * debugsw      - If 1, turn on debugging for the output.
 378  * disposw      - If 1, display MIME part disposition information.
 379  *
 380  */
 381 void list_all_messages(CT *cts, int headsw, int sizesw, int verbosw,
 382                        int debugsw, int disposw);
 383
 384 /*
 385  * List the content information of a single MIME part on stdout.
 386  *
 387  * Arguments are:
 388  *
 389  * ct           - MIME Content structure to display.
 390  * toplevel     - If set, we're at the top level of a message
 391  * realsize     - If set, determine the real size of the content
 392  * verbose      - If set, output verbose information
 393  * debug        - If set, turn on debugging for the output
 394  * dispo        - If set, display MIME part disposition information.
 395  *
 396  * Returns OK on success, NOTOK otherwise.
 397  */
 398 int list_content(CT ct, int toplevel, int realsize, int verbose, int debug,
 399                  int dispo);
 400
 401 /*
 402  * Display content-appropriate information on MIME parts, descending recursively
 403  * into multipart content if appropriate.  Uses list_content() for displaying
 404  * generic information.
 405  *
 406  * Arguments and return value are the same as list_content().
 407  */
 408 int list_switch(CT ct, int toplevel, int realsize, int verbose, int debug,
 409                 int dispo);
 410
 411 /*
 412  * Given a linked list of parameters, build an output string for them.  This
 413  * string is designed to be concatenated on an already-built header.
 414  *
 415  * Arguments are:
 416  *
 417  * initialwidth - Current width of the header.  Used to compute when to wrap
 418  *                parameters on the first line.  The following lines will
 419  *                be prefixed by a tab (\t) character.
 420  * params       - Pointer to head of linked list of parameters.
 421  * offsetout    - The final line offset after all the parameters have been
 422  *                output.  May be NULL.
 423  * external     - If set, outputting an external-body type and will not
 424  *                output a "body" parameter.
 425
 426  * Returns a pointer to the resulting parameter string.  This string must
 427  * be free()'d by the caller.  Returns NULL on error.
 428  */
 429 char *output_params(size_t initialwidth, PM params, int *offsetout,
 430                     int external);
 431
 432 /*
 433  * Encode a parameter value using RFC 2231 encode.
 434  *
 435  * Arguments are:
 436  *
 437  * pm           - PM containing the parameter value and related info.
 438  * output       - Output buffer.
 439  * len          - Size, in octets, of output buffer.
 440  * valuelen     - Number of characters in the value
 441  * valueoff     - Offset into value field (pm->pm_value).
 442  * index        - If 0, output character set and language tag.
 443  */
 444 size_t encode_param(PM pm, char *output, size_t len, size_t valuelen,
 445                     size_t valueoff, int index);
 446
 447 /*
 448  * Add a parameter to the parameter linked list.
 449  *
 450  * Arguments are:
 451  *
 452  * first        - Pointer to head of linked list
 453  * last         - Pointer to tail of linked list
 454  * name         - Name of parameter
 455  * value        - Value of parameter
 456  * nocopy       - If set, will use the pointer values directly for "name"
 457  *                and "value" instead of making their own copy.  These
 458  *                pointers will be free()'d later by the MIME routines, so
 459  *                they should not be used after calling this function!
 460  *
 461  * Returns allocated parameter element
 462  */
 463 PM add_param(PM *first, PM *last, char *name, char *value, int nocopy);
 464
 465 /*
 466  * Replace (or add) a parameter to the parameter linked list.
 467  *
 468  * If the named parameter already exists on the parameter linked list,
 469  * replace the value with the new one.  Otherwise add it to the linked
 470  * list.  All parameters are identical to add_param().
 471  */
 472 PM replace_param(PM *first, PM *last, char *name, char *value, int nocopy);
 473
 474 /*
 475  * Retrieve a parameter value from a parameter linked list.  Convert to the
 476  * local character set if required.
 477  *
 478  * Arguments are:
 479  *
 480  * first        - Pointer to head of parameter linked list.
 481  * name         - Name of parameter.
 482  * replace      - If characters in the parameter list cannot be converted to
 483  *                the local character set, replace with this character.
 484  * fetchonly    - If true, return pointer to original value, no conversion
 485  *                performed.
 486  *
 487  * Returns parameter value if found, NULL otherwise.  Memory must be free()'d
 488  * unless fetchonly is set.
 489  */
 490
 491 char *get_param(PM first, const char *name, char replace, int fetchonly);
 492
 493 /*
 494  * Fetch a parameter value from a parameter structure, converting it to
 495  * the local character set.
 496  *
 497  * Arguments are:
 498  *
 499  * pm           - Pointer to parameter structure
 500  * replace      - If characters in the parameter list cannot be converted to
 501  *                the local character set, replace with this character.
 502  *
 503  * Returns a pointer to the parameter value.  Memory is stored in an
 504  * internal buffer, so the returned value is only valid until the next
 505  * call to get_param_value() or get_param() (get_param() uses get_param_value()
 506  * internally).
 507  */
 508 char *get_param_value(PM pm, char replace);
 509
 510 extern int checksw;     /* Add Content-MD5 field */
 511
 512 /*
 513  * mhstore
 514  * Put it here because it uses the CT typedef.
 515  */
 516 typedef struct mhstoreinfo *mhstoreinfo_t;
 517 mhstoreinfo_t mhstoreinfo_create(CT *, char *, const char *, int, int);
 518 int mhstoreinfo_files_not_clobbered(const mhstoreinfo_t) PURE;
 519 void mhstoreinfo_free(mhstoreinfo_t);
 520 void store_all_messages (mhstoreinfo_t);
 521
 522 extern bool skip_mp_cte_check;
 523 extern bool suppress_bogus_mp_content_warning;
 524 extern bool bogus_mp_content;
 525 extern bool suppress_extraneous_trailing_semicolon_warning;
 526
 527 extern bool suppress_multiple_mime_version_warning;
 528
 529 #define NPREFS  20
 530 typedef struct {
 531     char *type;
 532     char *subtype;
 533 } mime_type_subtype;
 534 extern mime_type_subtype mime_preference[NPREFS];
 535 extern int npreferred;