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  * An extra note for message/external-body parts.  The enclosing
  95  * content structure is marked as a message/external-body; the c_ctparams
  96  * contains a pointer to a struct exbody, which contains a pointer to
  97  * (among other things) the "real" content (e.g., application/octet-stream).
  98  * The "real" content structure has the c_ctexbody pointer back to the
  99  * same struct exbody sees in the enclosing content structure (the struct
 100  * exbody contains parent pointers if you need to traverse up the content
 101  * structure).  Hopefully this makes it clearer:
 102  *
 103  * Enclosing content:
 104  *   Type: message/external-body
 105  *   c_ctparams: pointer to "struct exbody"
 106  *   c_ctexbody: NULL
 107  *
 108  * "Real" content:
 109  *   Type: application/octet-stream (or whatever)
 110  *   c_ctparams: NULL
 111  *   c_ctexbody: pointer to "struct exbody"
 112  *
 113  */
 114 struct Content {
 115     /* source (read) file */
 116     char *c_file;               /* read contents (file)              */
 117     FILE *c_fp;                 /* read contents (stream)            */
 118     int c_unlink;               /* remove file when done?            */
 119
 120     long c_begin;               /* where content body starts in file */
 121     long c_end;                 /* where content body ends in file   */
 122
 123     /* linked list of header fields */
 124     HF c_first_hf;              /* pointer to first header field     */
 125     HF c_last_hf;               /* pointer to last header field      */
 126
 127     /* copies of MIME related header fields */
 128     char *c_vrsn;               /* MIME-Version:                     */
 129     char *c_ctline;             /* Content-Type:                     */
 130     char *c_celine;             /* Content-Transfer-Encoding:        */
 131     char *c_id;                 /* Content-ID:                       */
 132     char *c_descr;              /* Content-Description:              */
 133     char *c_dispo;              /* Content-Disposition:              */
 134     char *c_dispo_type;         /* Type of Content-Disposition       */
 135     PM c_dispo_first;           /* Pointer to first disposition parm */
 136     PM c_dispo_last;            /* Pointer to last disposition parm  */
 137     char *c_partno;             /* within multipart content          */
 138
 139     /* Content-Type info */
 140     struct CTinfo c_ctinfo;     /* parsed elements of Content-Type   */
 141     int c_type;                 /* internal flag for content type    */
 142     int c_subtype;              /* internal flag for content subtype */
 143
 144     /* Content-Transfer-Encoding info (decoded contents) */
 145     struct cefile c_cefile;     /* structure holding decoded content */
 146     int c_encoding;             /* internal flag for encoding type   */
 147     int c_reqencoding;          /* Requested encoding (by mhbuild)   */
 148
 149     /* Content-MD5 info */
 150     int c_digested;             /* have we seen this header before?  */
 151     unsigned char c_digest[16]; /* decoded MD5 checksum              */
 152
 153     /* pointers to content-specific structures */
 154     void *c_ctparams;           /* content type specific data        */
 155     struct exbody *c_ctexbody;  /* data for type message/external    */
 156
 157     /* function pointers */
 158     InitFunc    c_ctinitfnx;    /* parse content body                */
 159     OpenCEFunc  c_ceopenfnx;    /* get a stream to decoded contents  */
 160     CloseCEFunc c_ceclosefnx;   /* release stream                    */
 161     SizeCEFunc  c_cesizefnx;    /* size of decoded contents          */
 162
 163     int c_umask;                /* associated umask                  */
 164     int c_rfc934;               /* rfc934 compatibility flag         */
 165
 166     char *c_showproc;           /* default, if not in profile        */
 167     char *c_termproc;           /* for charset madness...            */
 168     char *c_storeproc;          /* overrides profile entry, if any   */
 169
 170     char *c_storage;            /* write contents (file)             */
 171     char *c_folder;             /* write contents (folder)           */
 172 };
 173
 174 /*
 175  * Flags for Content-Type (Content->c_type)
 176  */
 177 #define CT_UNKNOWN      0x00
 178 #define CT_APPLICATION  0x01
 179 #define CT_AUDIO        0x02
 180 #define CT_IMAGE        0x03
 181 #define CT_MESSAGE      0x04
 182 #define CT_MULTIPART    0x05
 183 #define CT_TEXT         0x06
 184 #define CT_VIDEO        0x07
 185 #define CT_EXTENSION    0x08
 186
 187 /*
 188  * Flags for Content-Transfer-Encoding (Content->c_encoding)
 189  */
 190 #define CE_UNKNOWN      0x00
 191 #define CE_BASE64       0x01
 192 #define CE_QUOTED       0x02
 193 #define CE_8BIT         0x03
 194 #define CE_7BIT         0x04
 195 #define CE_BINARY       0x05
 196 #define CE_EXTENSION    0x06
 197 #define CE_EXTERNAL     0x07    /* for external-body */
 198
 199 /*
 200  * TEXT content
 201  */
 202
 203 /* Flags for subtypes of TEXT */
 204 #define TEXT_UNKNOWN    0x00
 205 #define TEXT_PLAIN      0x01
 206 #define TEXT_RICHTEXT   0x02
 207 #define TEXT_ENRICHED   0x03
 208
 209 /* Flags for character sets */
 210 #define CHARSET_SPECIFIED    0x00
 211 #define CHARSET_UNSPECIFIED 0x01  /* only needed when building drafts */
 212
 213 /* Structure for text content */
 214 struct text {
 215     int tx_charset;             /* flag for character set */
 216     int lf_line_endings;        /* Whether to use CR LF (0) or LF (1) line
 217                                    endings.  The meaning of 0 was selected so
 218                                    that CR LF is the default, in accordance
 219                                    with RFC 2046, Sec. 4.1.1, Par. 1. */
 220 };
 221
 222 /*
 223  * MULTIPART content
 224  */
 225
 226 /* Flags for subtypes of MULTIPART */
 227 #define MULTI_UNKNOWN   0x00
 228 #define MULTI_MIXED     0x01
 229 #define MULTI_ALTERNATE 0x02
 230 #define MULTI_DIGEST    0x03
 231 #define MULTI_PARALLEL  0x04
 232 #define MULTI_RELATED   0x05
 233
 234 /* Structure for subparts of a multipart content */
 235 struct part {
 236     CT mp_part;                 /* Content structure for subpart     */
 237     struct part *mp_next;       /* pointer to next subpart structure */
 238 };
 239
 240 /* Main structure for multipart content */
 241 struct multipart {
 242     char *mp_start;             /* boundary string separating parts   */
 243     char *mp_stop;              /* terminating boundary string        */
 244     char *mp_content_before;    /* any content before the first subpart */
 245     char *mp_content_after;     /* any content after the last subpart */
 246     struct part *mp_parts;      /* pointer to first subpart structure */
 247 };
 248
 249 /*
 250  * MESSAGE content
 251  */
 252
 253 /* Flags for subtypes of MESSAGE */
 254 #define MESSAGE_UNKNOWN  0x00
 255 #define MESSAGE_RFC822   0x01
 256 #define MESSAGE_PARTIAL  0x02
 257 #define MESSAGE_EXTERNAL 0x03
 258
 259 /* Structure for message/partial */
 260 struct partial {
 261     char *pm_partid;
 262     int pm_partno;
 263     int pm_maxno;
 264     int pm_marked;
 265     int pm_stored;
 266 };
 267
 268 /* Structure for message/external */
 269 struct exbody {
 270     CT eb_parent;       /* pointer to controlling content structure */
 271     CT eb_content;      /* pointer to internal content structure    */
 272     char *eb_partno;
 273     char *eb_access;
 274     int eb_flags;
 275     char *eb_name;
 276     char *eb_permission;
 277     char *eb_site;
 278     char *eb_dir;
 279     char *eb_mode;
 280     unsigned long eb_size;
 281     char *eb_server;
 282     char *eb_subject;
 283     char *eb_body;
 284     char *eb_url;
 285 };
 286
 287 /*
 288  * APPLICATION content
 289  */
 290
 291 /* Flags for subtype of APPLICATION */
 292 #define APPLICATION_UNKNOWN     0x00
 293 #define APPLICATION_OCTETS      0x01
 294 #define APPLICATION_POSTSCRIPT  0x02
 295
 296
 297 /*
 298  * Structures for mapping types to their internal flags
 299  */
 300 struct k2v {
 301     char *kv_key;
 302     int   kv_value;
 303 };
 304 extern struct k2v SubText[];
 305 extern struct k2v Charset[];
 306 extern struct k2v SubMultiPart[];
 307 extern struct k2v SubMessage[];
 308 extern struct k2v SubApplication[];
 309
 310 /*
 311  * Structures for mapping (content) types to
 312  * the functions to handle them.
 313  */
 314 struct str2init {
 315     char *si_key;
 316     int   si_val;
 317     InitFunc si_init;
 318 };
 319 extern struct str2init str2cts[];
 320 extern struct str2init str2ces[];
 321 extern struct str2init str2methods[];
 322
 323 /*
 324  * prototypes
 325  */
 326 CT parse_mime (char *);
 327
 328 /*
 329  * Translate a composition file into a MIME data structure.  Arguments are:
 330  *
 331  * infile       - Name of input filename
 332  * autobuild    - A flag to indicate if the composition file parser is
 333  *                being run in automatic mode or not.  In auto mode,
 334  *                if a MIME-Version header is encountered it is assumed
 335  *                that the composition file is already in MIME format
 336  *                and will not be processed further.  Otherwise, an
 337  *                error is generated.
 338  * dist         - A flag to indicate if we are being run by "dist".  In
 339  *                that case, add no MIME headers to the message.  Existing
 340  *                headers will still be encoded by RFC 2047.
 341  * directives   - A flag to control whether or not build directives are
 342  *                processed by default.
 343  * encoding     - The default encoding to use when doing RFC 2047 header
 344  *                encoding.  Must be one of CE_UNKNOWN, CE_BASE64, or
 345  *                CE_QUOTED.
 346  * maxunencoded - The maximum line length before the default encoding for
 347  *                text parts is quoted-printable.
 348  * verbose      - If 1, output verbose information during message composition
 349  *
 350  * Returns a CT structure describing the resulting MIME message.  If the
 351  * -auto flag is set and a MIME-Version header is encountered, the return
 352  * value is NULL.
 353  */
 354 CT build_mime (char *infile, int autobuild, int dist, int directives,
 355                int encoding, size_t maxunencoded, int verbose);
 356
 357 int add_header (CT, char *, char *);
 358 int get_ctinfo (char *, CT, int);
 359 int params_external (CT, int);
 360 int open7Bit (CT, char **);
 361 void close_encoding (CT);
 362 void free_content (CT);
 363 char *ct_type_str (int);
 364 char *ct_subtype_str (int, int);
 365 int ct_str_type (const char *);
 366 int ct_str_subtype (int, const char *);
 367 const struct str2init *get_ct_init (int);
 368 const char *ce_str (int);
 369 const struct str2init *get_ce_method (const char *);
 370 char *content_charset (CT);
 371 int convert_charset (CT, char *, int *);
 372 void reverse_alternative_parts (CT);
 373
 374 /*
 375  * Given a content structure, return true if the content has a disposition
 376  * of "inline".
 377  *
 378  * Arguments are:
 379  *
 380  * ct           - Content structure to examine
 381  */
 382 int is_inline(CT ct);
 383
 384 /*
 385  * Given a list of messages, display information about them on standard
 386  * output.
 387  *
 388  * Arguments are:
 389  *
 390  * cts          - An array of CT elements of messages that need to be
 391  *                displayed.  Array is terminated by a NULL.
 392  * headsw       - If 1, display a column header.
 393  * sizesw       - If 1, display the size of the part.
 394  * verbosw      - If 1, display verbose information
 395  * debugsw      - If 1, turn on debugging for the output.
 396  * disposw      - If 1, display MIME part disposition information.
 397  *
 398  */
 399 void list_all_messages(CT *cts, int headsw, int sizesw, int verbosw,
 400                        int debugsw, int disposw);
 401
 402 /*
 403  * List the content information of a single MIME part on stdout.
 404  *
 405  * Arguments are:
 406  *
 407  * ct           - MIME Content structure to display.
 408  * toplevel     - If set, we're at the top level of a message
 409  * realsize     - If set, determine the real size of the content
 410  * verbose      - If set, output verbose information
 411  * debug        - If set, turn on debugging for the output
 412  * dispo        - If set, display MIME part disposition information.
 413  *
 414  * Returns OK on success, NOTOK otherwise.
 415  */
 416 int list_content(CT ct, int toplevel, int realsize, int verbose, int debug,
 417                  int dispo);
 418
 419 /*
 420  * Display content-appropriate information on MIME parts, decending recursively
 421  * into multipart content if appropriate.  Uses list_content() for displaying
 422  * generic information.
 423  *
 424  * Arguments and return value are the same as list_content().
 425  */
 426 int list_switch(CT ct, int toplevel, int realsize, int verbose, int debug,
 427                 int dispo);
 428
 429 /*
 430  * Given a linked list of parameters, build an output string for them.  This
 431  * string is designed to be concatenated on an already-built header.
 432  *
 433  * Arguments are:
 434  *
 435  * initialwidth - Current width of the header.  Used to compute when to wrap
 436  *                parameters on the first line.  The following lines will
 437  *                be prefixed by a tab (\t) character.
 438  * params       - Pointer to head of linked list of parameters.
 439  * offsetout    - The final line offset after all the parameters have been
 440  *                output.  May be NULL.
 441  * external     - If set, outputting an external-body type and will not
 442  *                output a "body" parameter.
 443
 444  * Returns a pointer to the resulting parameter string.  This string must
 445  * be free()'d by the caller.  Returns NULL on error.
 446  */
 447 char *output_params(size_t initialwidth, PM params, int *offsetout,
 448                     int external);
 449
 450 /*
 451  * Encode a parameter value using RFC 2231 encode.
 452  *
 453  * Arguments are:
 454  *
 455  * pm           - PM containing the parameter value and related info.
 456  * output       - Output buffer.
 457  * len          - Size, in octets, of output buffer.
 458  * valuelen     - Number of characters in the value
 459  * valueoff     - Offset into value field (pm->pm_value).
 460  * index        - If 0, output character set and language tag.
 461  */
 462 size_t encode_param(PM pm, char *output, size_t len, size_t valuelen,
 463                     size_t valueoff, int index);
 464
 465 /*
 466  * Add a parameter to the parameter linked list.
 467  *
 468  * Arguments are:
 469  *
 470  * first        - Pointer to head of linked list
 471  * last         - Pointer to tail of linked list
 472  * name         - Name of parameter
 473  * value        - Value of parameter
 474  * nocopy       - If set, will use the pointer values directly for "name"
 475  *                and "value" instead of making their own copy.  These
 476  *                pointers will be free()'d later by the MIME routines, so
 477  *                they should not be used after calling this function!
 478  *
 479  * Returns allocated parameter element
 480  */
 481 PM add_param(PM *first, PM *last, char *name, char *value, int nocopy);
 482
 483 /*
 484  * Replace (or add) a parameter to the parameter linked list.
 485  *
 486  * If the named parameter already exists on the parameter linked list,
 487  * replace the value with the new one.  Otherwise add it to the linked
 488  * list.  All parameters are identical to add_param().
 489  */
 490 PM replace_param(PM *first, PM *last, char *name, char *value, int nocopy);
 491
 492 /*
 493  * Retrieve a parameter value from a parameter linked list.  Convert to the
 494  * local character set if required.
 495  *
 496  * Arguments are:
 497  *
 498  * first        - Pointer to head of parameter linked list.
 499  * name         - Name of parameter.
 500  * replace      - If characters in the parameter list cannot be converted to
 501  *                the local character set, replace with this character.
 502  * fetchonly    - If true, return pointer to original value, no conversion
 503  *                performed.
 504  *
 505  * Returns parameter value if found, NULL otherwise.  Memory must be free()'d
 506  * unless fetchonly is set.
 507  */
 508
 509 char *get_param(PM first, const char *name, char replace, int fetchonly);
 510
 511 /*
 512  * Fetch a parameter value from a parameter structure, converting it to
 513  * the local character set.
 514  *
 515  * Arguments are:
 516  *
 517  * pm           - Pointer to parameter structure
 518  * replace      - If characters in the parameter list cannot be converted to
 519  *                the local character set, replace with this character.
 520  *
 521  * Returns a pointer to the parameter value.  Memory is stored in an
 522  * internal buffer, so the returned value is only valid until the next
 523  * call to get_param_value() or get_param() (get_param() uses get_param_value()
 524  * internally).
 525  */
 526 char *get_param_value(PM pm, char replace);
 527
 528 /*
 529  * Display MIME message(s) on standard out.
 530  *
 531  * Arguments are:
 532  *
 533  * cts          - NULL terminated array of CT structures for messages
 534  *                to display
 535  * concat       - If true, concatenate all MIME parts.  If false, show each
 536  *                MIME part under a separate pager.
 537  * textonly     - If true, only display "text" MIME parts
 538  * inlineonly   - If true, only display MIME parts that are marked with
 539  *                a disposition of "inline" (includes parts that lack a
 540  *                Content-Disposition header).
 541  * markerform   - The name of a file containing mh-format(5) code used to
 542  *                display markers about non-displayed MIME parts.
 543  */
 544 void show_all_messages(CT *cts, int concat, int textonly, int inlineonly);
 545
 546 /*
 547  * Display (or store) a single MIME part using the specified command
 548  *
 549  * Arguments are:
 550  *
 551  * ct           - The Content structure of the MIME part we wish to display
 552  * alternate    - Set this to true if this is one part of a MIME
 553  *                multipart/alternative part.  Will suppress some errors and
 554  *                will cause the function to return DONE instead of OK on
 555  *                success.
 556  * cp           - The command string to execute.  Will be run through the
 557  *                parser for %-escapes as described in mhshow(1).
 558  * cracked      - If set, chdir() to this directory before executing the
 559  *                command in "cp".  Only used by mhstore(1).
 560  * fmt          - A series of mh-format(5) instructions to execute if the
 561  *                command string indicates a marker is desired.  Can be NULL.
 562  *
 563  * Returns NOTOK if we could not display the part, DONE if alternate was
 564  * set and we could display the part, and OK if alternate was not set and
 565  * we could display the part.
 566  */
 567 struct format;
 568 int show_content_aux(CT ct, int alternate, char *cp, char *cracked,
 569                      struct format *fmt);
 570
 571 extern int checksw;     /* Add Content-MD5 field */
 572
 573 /*
 574  * mhstore
 575  * Put it here because it uses the CT typedef.
 576  */
 577 typedef struct mhstoreinfo *mhstoreinfo_t;
 578 mhstoreinfo_t mhstoreinfo_create(CT *, char *, const char *, int, int);
 579 int mhstoreinfo_files_not_clobbered(const mhstoreinfo_t);
 580 void mhstoreinfo_free(mhstoreinfo_t);
 581 void store_all_messages (mhstoreinfo_t);