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

   1
   2 /*
   3  * mhparse.h -- definitions for parsing/building of MIME content
   4  *           -- (mhparse.c/mhbuildsbr.c)
   5  */
   6
   7 #define NPARTS  50
   8 #define NTYPES  20
   9 #define NPREFS  20
  10
  11 /*
  12  * Abstract type for header fields
  13  */
  14 typedef struct hfield *HF;
  15
  16 /*
  17  * Abstract types for MIME parsing/building
  18  */
  19 typedef struct cefile           *CE;
  20 typedef struct CTinfo           *CI;
  21 typedef struct Content          *CT;
  22 typedef struct Parameter        *PM;
  23
  24 /*
  25  * type for Init function (both type and transfer encoding)
  26  */
  27 typedef int (*InitFunc) (CT);
  28
  29 /*
  30  * types for various transfer encoding access functions
  31  */
  32 typedef int (*OpenCEFunc) (CT, char **);
  33 typedef void (*CloseCEFunc) (CT);
  34 typedef unsigned long (*SizeCEFunc) (CT);
  35
  36 /*
  37  * Structure for storing/encoding/decoding
  38  * a header field and its value.
  39  */
  40 struct hfield {
  41     char *name;         /* field name */
  42     char *value;        /* field body */
  43     int hf_encoding;    /* internal flag for transfer encoding to use */
  44     HF next;            /* link to next header field */
  45 };
  46
  47 /*
  48  * Structure for holding MIME parameter elements.
  49  */
  50 struct Parameter {
  51     char *pm_name;      /* Parameter name */
  52     char *pm_value;     /* Parameter value */
  53     char *pm_charset;   /* Parameter character set (optional) */
  54     char *pm_lang;      /* Parameter language tag (optional) */
  55     PM   pm_next;       /* Pointer to next element */
  56 };
  57
  58 /*
  59  * Structure for storing parsed elements
  60  * of the Content-Type component.
  61  */
  62 struct CTinfo {
  63     char *ci_type;              /* content type     */
  64     char *ci_subtype;           /* content subtype  */
  65     PM   ci_first_pm;           /* Pointer to first MIME parameter */
  66     PM   ci_last_pm;            /* Pointer to last MIME parameter */
  67     char *ci_comment;           /* RFC-822 comments */
  68     char *ci_magic;
  69 };
  70
  71 /*
  72  * Structure for storing decoded contents after
  73  * removing Content-Transfer-Encoding.
  74  */
  75 struct cefile {
  76     char *ce_file;      /* decoded content (file)   */
  77     FILE *ce_fp;        /* decoded content (stream) */
  78     int   ce_unlink;    /* remove file when done?   */
  79 };
  80
  81 /*
  82  * Primary structure for handling Content (Entity)
  83  *
  84  * Some more explanation of this:
  85  *
  86  * This structure recursively describes a complete MIME message.
  87  * At the top level, the c_first_hf list has a list of all message
  88  * headers.  If the content-type is multipart (c_type == CT_MULTIPART)
  89  * then c_ctparams will contain a pointer to a struct multipart.
  90  * A struct multipart contains (among other things) a linked list
  91  * of struct part elements, and THOSE contain a pointer to the sub-part's
  92  * Content structure.
  93  */
  94 struct Content {
  95     /* source (read) file */
  96     char *c_file;               /* read contents (file)              */
  97     FILE *c_fp;                 /* read contents (stream)            */
  98     int c_unlink;               /* remove file when done?            */
  99
 100     long c_begin;               /* where content body starts in file */
 101     long c_end;                 /* where content body ends in file   */
 102
 103     /* linked list of header fields */
 104     HF c_first_hf;              /* pointer to first header field     */
 105     HF c_last_hf;               /* pointer to last header field      */
 106
 107     /* copies of MIME related header fields */
 108     char *c_vrsn;               /* MIME-Version:                     */
 109     char *c_ctline;             /* Content-Type:                     */
 110     char *c_celine;             /* Content-Transfer-Encoding:        */
 111     char *c_id;                 /* Content-ID:                       */
 112     char *c_descr;              /* Content-Description:              */
 113     char *c_dispo;              /* Content-Disposition:              */
 114     char *c_dispo_type;         /* Type of Content-Disposition       */
 115     PM c_dispo_first;           /* Pointer to first disposition parm */
 116     PM c_dispo_last;            /* Pointer to last disposition parm  */
 117     char *c_partno;             /* within multipart content          */
 118
 119     /* Content-Type info */
 120     struct CTinfo c_ctinfo;     /* parsed elements of Content-Type   */
 121     int c_type;                 /* internal flag for content type    */
 122     int c_subtype;              /* internal flag for content subtype */
 123
 124     /* Content-Transfer-Encoding info (decoded contents) */
 125     struct cefile c_cefile;     /* structure holding decoded content */
 126     int c_encoding;             /* internal flag for encoding type   */
 127     int c_reqencoding;          /* Requested encoding (by mhbuild)   */
 128
 129     /* Content-MD5 info */
 130     int c_digested;             /* have we seen this header before?  */
 131     unsigned char c_digest[16]; /* decoded MD5 checksum              */
 132
 133     /* pointers to content-specific structures */
 134     void *c_ctparams;           /* content type specific data        */
 135     struct exbody *c_ctexbody;  /* data for type message/external    */
 136
 137     /* function pointers */
 138     InitFunc    c_ctinitfnx;    /* parse content body                */
 139     OpenCEFunc  c_ceopenfnx;    /* get a stream to decoded contents  */
 140     CloseCEFunc c_ceclosefnx;   /* release stream                    */
 141     SizeCEFunc  c_cesizefnx;    /* size of decoded contents          */
 142
 143     int c_umask;                /* associated umask                  */
 144     int c_rfc934;               /* rfc934 compatibility flag         */
 145
 146     char *c_showproc;           /* default, if not in profile        */
 147     char *c_termproc;           /* for charset madness...            */
 148     char *c_storeproc;          /* overrides profile entry, if any   */
 149
 150     char *c_storage;            /* write contents (file)             */
 151     char *c_folder;             /* write contents (folder)           */
 152 };
 153
 154 /*
 155  * Flags for Content-Type (Content->c_type)
 156  */
 157 #define CT_UNKNOWN      0x00
 158 #define CT_APPLICATION  0x01
 159 #define CT_AUDIO        0x02
 160 #define CT_IMAGE        0x03
 161 #define CT_MESSAGE      0x04
 162 #define CT_MULTIPART    0x05
 163 #define CT_TEXT         0x06
 164 #define CT_VIDEO        0x07
 165 #define CT_EXTENSION    0x08
 166
 167 /*
 168  * Flags for Content-Transfer-Encoding (Content->c_encoding)
 169  */
 170 #define CE_UNKNOWN      0x00
 171 #define CE_BASE64       0x01
 172 #define CE_QUOTED       0x02
 173 #define CE_8BIT         0x03
 174 #define CE_7BIT         0x04
 175 #define CE_BINARY       0x05
 176 #define CE_EXTENSION    0x06
 177 #define CE_EXTERNAL     0x07    /* for external-body */
 178
 179 /*
 180  * TEXT content
 181  */
 182
 183 /* Flags for subtypes of TEXT */
 184 #define TEXT_UNKNOWN    0x00
 185 #define TEXT_PLAIN      0x01
 186 #define TEXT_RICHTEXT   0x02
 187 #define TEXT_ENRICHED   0x03
 188
 189 /* Flags for character sets */
 190 #define CHARSET_SPECIFIED    0x00
 191 #define CHARSET_UNSPECIFIED 0x01  /* only needed when building drafts */
 192
 193 /* Structure for text content */
 194 struct text {
 195     int tx_charset;             /* flag for character set */
 196     int lf_line_endings;        /* Whether to use CR LF (0) or LF (1) line
 197                                    endings.  The meaning of 0 was selected so
 198                                    that CR LF is the default, in accordance
 199                                    with RFC 2046, Sec. 4.1.1, Par. 1. */
 200 };
 201
 202 /*
 203  * MULTIPART content
 204  */
 205
 206 /* Flags for subtypes of MULTIPART */
 207 #define MULTI_UNKNOWN   0x00
 208 #define MULTI_MIXED     0x01
 209 #define MULTI_ALTERNATE 0x02
 210 #define MULTI_DIGEST    0x03
 211 #define MULTI_PARALLEL  0x04
 212 #define MULTI_RELATED   0x05
 213
 214 /* Structure for subparts of a multipart content */
 215 struct part {
 216     CT mp_part;                 /* Content structure for subpart     */
 217     struct part *mp_next;       /* pointer to next subpart structure */
 218 };
 219
 220 /* Main structure for multipart content */
 221 struct multipart {
 222     char *mp_start;             /* boundary string separating parts   */
 223     char *mp_stop;              /* terminating boundary string        */
 224     char *mp_content_before;    /* any content before the first subpart */
 225     char *mp_content_after;     /* any content after the last subpart */
 226     struct part *mp_parts;      /* pointer to first subpart structure */
 227 };
 228
 229 /*
 230  * MESSAGE content
 231  */
 232
 233 /* Flags for subtypes of MESSAGE */
 234 #define MESSAGE_UNKNOWN  0x00
 235 #define MESSAGE_RFC822   0x01
 236 #define MESSAGE_PARTIAL  0x02
 237 #define MESSAGE_EXTERNAL 0x03
 238
 239 /* Structure for message/partial */
 240 struct partial {
 241     char *pm_partid;
 242     int pm_partno;
 243     int pm_maxno;
 244     int pm_marked;
 245     int pm_stored;
 246 };
 247
 248 /* Structure for message/external */
 249 struct exbody {
 250     CT eb_parent;       /* pointer to controlling content structure */
 251     CT eb_content;      /* pointer to internal content structure    */
 252     char *eb_partno;
 253     char *eb_access;
 254     int eb_flags;
 255     char *eb_name;
 256     char *eb_permission;
 257     char *eb_site;
 258     char *eb_dir;
 259     char *eb_mode;
 260     unsigned long eb_size;
 261     char *eb_server;
 262     char *eb_subject;
 263     char *eb_body;
 264     char *eb_url;
 265 };
 266
 267 /*
 268  * APPLICATION content
 269  */
 270
 271 /* Flags for subtype of APPLICATION */
 272 #define APPLICATION_UNKNOWN     0x00
 273 #define APPLICATION_OCTETS      0x01
 274 #define APPLICATION_POSTSCRIPT  0x02
 275
 276
 277 /*
 278  * Structures for mapping types to their internal flags
 279  */
 280 struct k2v {
 281     char *kv_key;
 282     int   kv_value;
 283 };
 284 extern struct k2v SubText[];
 285 extern struct k2v Charset[];
 286 extern struct k2v SubMultiPart[];
 287 extern struct k2v SubMessage[];
 288 extern struct k2v SubApplication[];
 289
 290 /*
 291  * Structures for mapping (content) types to
 292  * the functions to handle them.
 293  */
 294 struct str2init {
 295     char *si_key;
 296     int   si_val;
 297     InitFunc si_init;
 298 };
 299 extern struct str2init str2cts[];
 300 extern struct str2init str2ces[];
 301 extern struct str2init str2methods[];
 302
 303 /*
 304  * prototypes
 305  */
 306 CT parse_mime (char *);
 307
 308 /*
 309  * Translate a composition file into a MIME data structure.  Arguments are:
 310  *
 311  * infile       - Name of input filename
 312  * autobuild    - A flag to indicate if the composition file parser is
 313  *                being run in automatic mode or not.  In auto mode,
 314  *                if a MIME-Version header is encountered it is assumed
 315  *                that the composition file is already in MIME format
 316  *                and will not be processed further.  Otherwise, an
 317  *                error is generated.
 318  * dist         - A flag to indicate if we are being run by "dist".  In
 319  *                that case, add no MIME headers to the message.  Existing
 320  *                headers will still be encoded by RFC 2047.
 321  * directives   - A flag to control whether or not build directives are
 322  *                processed by default.
 323  * encoding     - The default encoding to use when doing RFC 2047 header
 324  *                encoding.  Must be one of CE_UNKNOWN, CE_BASE64, or
 325  *                CE_QUOTED.
 326  * maxunencoded - The maximum line length before the default encoding for
 327  *                text parts is quoted-printable.
 328  * verbose      - If 1, output verbose information during message composition
 329  *
 330  * Returns a CT structure describing the resulting MIME message.  If the
 331  * -auto flag is set and a MIME-Version header is encountered, the return
 332  * value is NULL.
 333  */
 334 CT build_mime (char *infile, int autobuild, int dist, int directives,
 335                int encoding, size_t maxunencoded, int verbose);
 336
 337 int add_header (CT, char *, char *);
 338 int get_ctinfo (char *, CT, int);
 339 int params_external (CT, int);
 340 int open7Bit (CT, char **);
 341 void close_encoding (CT);
 342 void free_content (CT);
 343 char *ct_type_str (int);
 344 char *ct_subtype_str (int, int);
 345 int ct_str_type (const char *);
 346 int ct_str_subtype (int, const char *);
 347 const struct str2init *get_ct_init (int);
 348 const char *ce_str (int);
 349 const struct str2init *get_ce_method (const char *);
 350 char *content_charset (CT);
 351 int convert_charset (CT, char *, int *);
 352 void reverse_alternative_parts (CT);
 353
 354 /*
 355  * Given a content structure, return true if the content has a disposition
 356  * of "inline".
 357  *
 358  * Arguments are:
 359  *
 360  * ct           - Content structure to examine
 361  */
 362 int is_inline(CT ct);
 363
 364 /*
 365  * Given a list of messages, display information about them on standard
 366  * output.
 367  *
 368  * Arguments are:
 369  *
 370  * cts          - An array of CT elements of messages that need to be
 371  *                displayed.  Array is terminated by a NULL.
 372  * headsw       - If 1, display a column header.
 373  * sizesw       - If 1, display the size of the part.
 374  * verbosw      - If 1, display verbose information
 375  * debugsw      - If 1, turn on debugging for the output.
 376  * disposw      - If 1, display MIME part disposition information.
 377  *
 378  */
 379 void list_all_messages(CT *cts, int headsw, int sizesw, int verbosw,
 380                        int debugsw, int disposw);
 381
 382 /*
 383  * List the content information of a single MIME part on stdout.
 384  *
 385  * Arguments are:
 386  *
 387  * ct           - MIME Content structure to display.
 388  * toplevel     - If set, we're at the top level of a message
 389  * realsize     - If set, determine the real size of the content
 390  * verbose      - If set, output verbose information
 391  * debug        - If set, turn on debugging for the output
 392  * dispo        - If set, display MIME part disposition information.
 393  *
 394  * Returns OK on success, NOTOK otherwise.
 395  */
 396 int list_content(CT ct, int toplevel, int realsize, int verbose, int debug,
 397                  int dispo);
 398
 399 /*
 400  * Display content-appropriate information on MIME parts, decending recursively
 401  * into multipart content if appropriate.  Uses list_content() for displaying
 402  * generic information.
 403  *
 404  * Arguments and return value are the same as list_content().
 405  */
 406 int list_switch(CT ct, int toplevel, int realsize, int verbose, int debug,
 407                 int dispo);
 408
 409 /*
 410  * Given a linked list of parameters, build an output string for them.  This
 411  * string is designed to be concatenated on an already-built header.
 412  *
 413  * Arguments are:
 414  *
 415  * initialwidth - Current width of the header.  Used to compute when to wrap
 416  *                parameters on the first line.  The following lines will
 417  *                be prefixed by a tab (\t) character.
 418  * params       - Pointer to head of linked list of parameters.
 419  * offsetout    - The final line offset after all the parameters have been
 420  *                output.  May be NULL.
 421  * external     - If set, outputting an external-body type and will not
 422  *                output a "body" parameter.
 423
 424  * Returns a pointer to the resulting parameter string.  This string must
 425  * be free()'d by the caller.  Returns NULL on error.
 426  */
 427 char *output_params(size_t initialwidth, PM params, int *offsetout,
 428                     int external);
 429
 430 /*
 431  * Encode a parameter value using RFC 2231 encode.
 432  *
 433  * Arguments are:
 434  *
 435  * pm           - PM containing the parameter value and related info.
 436  * output       - Output buffer.
 437  * len          - Size, in octets, of output buffer.
 438  * valuelen     - Number of characters in the value
 439  * valueoff     - Offset into value field (pm->pm_value).
 440  * index        - If 0, output character set and language tag.
 441  */
 442 size_t encode_param(PM pm, char *output, size_t len, size_t valuelen,
 443                     size_t valueoff, int index);
 444
 445 /*
 446  * Add a parameter to the parameter linked list.
 447  *
 448  * Arguments are:
 449  *
 450  * first        - Pointer to head of linked list
 451  * last         - Pointer to tail of linked list
 452  * name         - Name of parameter
 453  * value        - Value of parameter
 454  * nocopy       - If set, will use the pointer values directly for "name"
 455  *                and "value" instead of making their own copy.  These
 456  *                pointers will be free()'d later by the MIME routines, so
 457  *                they should not be used after calling this function!
 458  *
 459  * Returns allocated parameter element
 460  */
 461 PM add_param(PM *first, PM *last, char *name, char *value, int nocopy);
 462
 463 /*
 464  * Replace (or add) a parameter to the parameter linked list.
 465  *
 466  * If the named parameter already exists on the parameter linked list,
 467  * replace the value with the new one.  Otherwise add it to the linked
 468  * list.  All parameters are identical to add_param().
 469  */
 470 PM replace_param(PM *first, PM *last, char *name, char *value, int nocopy);
 471
 472 /*
 473  * Retrieve a parameter value from a parameter linked list.  Convert to the
 474  * local character set if required.
 475  *
 476  * Arguments are:
 477  *
 478  * first        - Pointer to head of parameter linked list.
 479  * name         - Name of parameter.
 480  * replace      - If characters in the parameter list cannot be converted to
 481  *                the local character set, replace with this character.
 482  * fetchonly    - If true, return pointer to original value, no conversion
 483  *                performed.
 484  *
 485  * Returns parameter value if found, NULL otherwise.  Memory must be free()'d
 486  * unless fetchonly is set.
 487  */
 488
 489 char *get_param(PM first, const char *name, char replace, int fetchonly);
 490
 491 /*
 492  * Fetch a parameter value from a parameter structure, converting it to
 493  * the local character set.
 494  *
 495  * Arguments are:
 496  *
 497  * pm           - Pointer to parameter structure
 498  * replace      - If characters in the parameter list cannot be converted to
 499  *                the local character set, replace with this character.
 500  *
 501  * Returns a pointer to the parameter value.  Memory is stored in an
 502  * internal buffer, so the returned value is only valid until the next
 503  * call to get_param_value() or get_param() (get_param() uses get_param_value()
 504  * internally).
 505  */
 506 char *get_param_value(PM pm, char replace);
 507
 508 /*
 509  * Display MIME message(s) on standard out.
 510  *
 511  * Arguments are:
 512  *
 513  * cts          - NULL terminated array of CT structures for messages
 514  *                to display
 515  * concat       - If true, concatenate all MIME parts.  If false, show each
 516  *                MIME part under a separate pager.
 517  * textonly     - If true, only display "text" MIME parts
 518  * inlineonly   - If true, only display MIME parts that are marked with
 519  *                a disposition of "inline" (includes parts that lack a
 520  *                Content-Disposition header).
 521  * markerform   - The name of a file containing mh-format(5) code used to
 522  *                display markers about non-displayed MIME parts.
 523  */
 524 void show_all_messages(CT *cts, int concat, int textonly, int inlineonly);
 525
 526 /*
 527  * Display (or store) a single MIME part using the specified command
 528  *
 529  * Arguments are:
 530  *
 531  * ct           - The Content structure of the MIME part we wish to display
 532  * alternate    - Set this to true if this is one part of a MIME
 533  *                multipart/alternative part.  Will suppress some errors and
 534  *                will cause the function to return DONE instead of OK on
 535  *                success.
 536  * cp           - The command string to execute.  Will be run through the
 537  *                parser for %-escapes as described in mhshow(1).
 538  * cracked      - If set, chdir() to this directory before executing the
 539  *                command in "cp".  Only used by mhstore(1).
 540  * fmt          - A series of mh-format(5) instructions to execute if the
 541  *                command string indicates a marker is desired.  Can be NULL.
 542  *
 543  * Returns NOTOK if we could not display the part, DONE if alternate was
 544  * set and we could display the part, and OK if alternate was not set and
 545  * we could display the part.
 546  */
 547 struct format;
 548 int show_content_aux(CT ct, int alternate, char *cp, char *cracked,
 549                      struct format *fmt);
 550
 551 extern int checksw;     /* Add Content-MD5 field */
 552
 553 /*
 554  * mhstore
 555  * Put it here because it uses the CT typedef.
 556  */
 557 typedef struct mhstoreinfo *mhstoreinfo_t;
 558 mhstoreinfo_t mhstoreinfo_create(CT *, char *, const char *, int, int);
 559 int mhstoreinfo_files_not_clobbered(const mhstoreinfo_t);
 560 void mhstoreinfo_free(mhstoreinfo_t);
 561 void store_all_messages (mhstoreinfo_t);