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