msg_mime.c

Sofia SIP is an open-source SIP User-Agent library, compliant with the IETF RFC3261 specification.

/*
 * This file is part of the Sofia-SIP package
 *
 * Copyright (C) 2005 Nokia Corporation.
 *
 * Contact: Pekka Pessi <[e-mail]>
 *
 * This library is free software; you can redistribute it and/or
 * modify it under the terms of the GNU Lesser General Public License
 * as published by the Free Software Foundation; either version 2.1 of
 * the License, or (at your option) any later version.
 *
 * This library is distributed in the hope that it will be useful, but
 * WITHOUT ANY WARRANTY; without even the implied warranty of
 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
 * Lesser General Public License for more details.
 *
 * You should have received a copy of the GNU Lesser General Public
 * License along with this library; if not, write to the Free Software
 * Foundation, Inc., 51 Franklin St, Fifth Floor, Boston, MA
 * 02110-1301 USA
 *
 */

/**@ingroup msg_mime
 * @CFILE msg_mime.c
 *
 * MIME-related headers and MIME multipart bodies for SIP/HTTP/RTSP.
 *
 * @author Pekka Pessi <[e-mail]>
 *
 * @date Created: Tue Jun 13 02:57:51 2000 ppessi
 *
 *
 */

#include "config.h"

#define _GNU_SOURCE 1

#include <stddef.h>
#include <stdlib.h>
#include <string.h>
#include <limits.h>
#include <errno.h>
#include <assert.h>

#include <sofia-sip/su_alloc.h>

#include "msg_internal.h"
#include "sofia-sip/msg.h"
#include "sofia-sip/msg_mime.h"

#include <sofia-sip/su_uniqueid.h>
#include <sofia-sip/su_errno.h>

#if !HAVE_MEMMEM
void *memmem(const void *haystack, size_t haystacklen,
	     const void *needle, size_t needlelen);
#endif

size_t memspn(const void *mem, size_t memlen,
	      const void *accept, size_t acceptlen);
size_t memcspn(const void *mem, size_t memlen,
	       const void *reject, size_t rejectlen);

/** Protocol version of MIME */
char const msg_mime_version_1_0[] = "MIME/1.0";

#include <sofia-sip/msg_parser.h>
#include <sofia-sip/msg_mime_protos.h>

/** Define a header class for headers without any extra data to copy */
#define MSG_HEADER_CLASS_G(c, l, s, kind) \
  MSG_HEADER_CLASS(msg_, c, l, s, g_common, kind, msg_generic, msg_generic)

#define msg_generic_update NULL

/** Define a header class for a msg_list_t kind of header */
#define MSG_HEADER_CLASS_LIST(c, l, s, kind) \
  MSG_HEADER_CLASS(msg_, c, l, s, k_items, kind, msg_list, msg_list)

#define msg_list_update NULL

/* ====================================================================== */

/** Calculate length of line ending (0, 1 or 2). @internal */
#define CRLF_TEST(b) ((b)[0] == '\r' ? ((b)[1] == '\n') + 1 : (b)[0] =='\n')

/**@ingroup msg_mime
 * @defgroup msg_multipart MIME Multipart Body
 *
 * Representing MIME multipart bodies and their manipulation.
 *
 * The #msg_multipart_t is an object for storing MIME multipart message
 * bodies. It includes message components used for framing and identifying
 * message parts. Its syntax is defined in  @RFC2046 as follows:
 *
 * @code
 *
 *   multipart-body := [preamble CRLF]
 *                     dash-boundary transport-padding CRLF
 *                     body-part *encapsulation
 *                     close-delimiter transport-padding
 *                     [CRLF epilogue]
 *
 *   preamble := discard-text
 *
 *   discard-text := *(*text CRLF)
 *                   ; May be ignored or discarded.
 *
 *   dash-boundary := "--" boundary
 *                    ; boundary taken from the value of boundary parameter
 *                    ; of the Content-Type field.
 *
 *   boundary := 0*69<bchars> bcharsnospace
 *
 *   bchars := bcharsnospace / " "
 *
 *   bcharsnospace := DIGIT / ALPHA / "'" / "(" / ")" /
 *                    "+" / "_" / "," / "-" / "." /
 *                    "/" / ":" / "=" / "?"
 *
 *   transport-padding := *LWSP-char
 *                        ; Composers MUST NOT generate non-zero length
 *                        ; transport padding, but receivers MUST be able to
 *                        ; handle padding added by message transports.
 *
 *   body-part := <"message" as defined in @RFC822, with all header fields
 *                 optional, not starting with the specified dash-boundary,
 *                 and with the delimiter not occurring anywhere in the body
 *                 part. Note that the semantics of a part differ from the
 *                 semantics of a message, as described in the text.>
 *
 *   encapsulation := delimiter transport-padding CRLF 
 *                    body-part
 *
 *   close-delimiter := delimiter "--"
 *
 *   delimiter := CRLF dash-boundary
 *
 *   epilogue := discard-text
 *
 * @endcode
 *
 * @par Parsing a Multipart Message
 * 
 * When a message body contains a multipart entity (in other words, it has a
 * MIME media type of "multipart"), the application can split the multipart
 * entity into body parts
 *
 * The parsing is relatively simple, the application just gives a memory
 * home object, a Content-Type header object and message body object as an 
 * argument to msg_multipart_parse() function:
 * @code
 *    if (sip->sip_content_type && 
 *        strncasecmp(sip->sip_content_type, "multipart/", 10) == 0) {
 *      msg_multipart_t *mp;
 *
 *      if (sip->sip_multipart)
 *        mp = sip->sip_multipart;
 *      else
 *        mp = msg_multipart_parse(msg_home(msg), 
 *                                 sip->sip_content_type,
 *                                 (sip_payload_t *)sip->sip_payload);
 * 
 *      if (mp)
 *        ... processing multipart ...
 *      else
 *        ... error handling ...
 *    }
 * @endcode
 *
 * The resulting list of msg_multipart_t structures contain the parts of the
 * multipart entity, each part represented by a separate #msg_multipart_t
 * structure. Please note that in order to make error recovery possible, the
 * parsing is not recursive - if multipart contains another multipart, the
 * application is responsible for scanning for it and parsing it.
 *
 * @par Constructing a Multipart Message
 *
 * Constructing a multipart body is a bit more hairy. The application needs
 * a message object (#msg_t), which is used to buffer the encoding of
 * multipart components.
 *
 * As an example, let us create a "multipart/mixed" multipart entity with a
 * HTML and GIF contents, and convert it into a #sip_payload_t structure:
 * @code
 *   msg_t *msg = msg_create(sip_default_mclass, 0);
 *   su_home_t *home = msg_home(msg);
 *   sip_t *sip = sip_object(msg);
 *   sip_content_type_t *c;
 *   msg_multipart_t *mp = NULL;
 *   msg_header_t *h = NULL;
 *   char *b;
 *   size_t len, offset;
 * 
 *   mp = msg_multipart_create(home, "text/html;level=3", html, strlen(html));
 *   mp->mp_next = msg_multipart_create(home, "image/gif", gif, giflen);
 *
 *   c = sip_content_type_make(home, "multipart/mixed");
 * 
 *   // Add delimiters to multipart, and boundary parameter to content-type
 *   if (msg_multipart_complete(home, c, mp) < 0)
 *     return -1;		// Error
 *
 *   // Combine multipart components into the chain
 *   h = NULL;
 *   if (msg_multipart_serialize(&h, mp) < 0)
 *     return -1;		// Error
 *
 *   // Encode all multipart components 
 *   len = msg_multipart_prepare(msg, mp, 0);
 *   if (len < 0)
 *     return -1;		// Error
 *
 *   pl = sip_payload_create(home, NULL, len);
 *
 *   // Copy each element from multipart to pl_data 
 *   b = pl->pl_data;
 *   for (offset = 0, h = mp; offset < len; h = h->sh_succ) {
 *     memcpy(b + offset, h->sh_data, h->sh_len);
 *     offset += h->sh_len;
 *   }
 * @endcode
 *
 */

/**Create a part for MIME multipart entity.
 *
 * The function msg_multipart_create() allocates a new #msg_multipart_t
 * object from memory home @a home. If @a content_type is non-NULL, it makes
 * a #msg_content_type_t header object and adds the header to the
 * #msg_multipart_t object. If @a dlen is nonzero, it allocates a
 * msg_payload_t structure of @a dlen bytes for the payload of the newly
 * created #msg_multipart_t object. If @a data is non-NULL, it copies the @a
 * dlen bytes of of data to the payload of the newly created
 * #msg_multipart_t object.
 *
 * @return A pointer to the newly created #msg_multipart_t object, or NULL
 * upon an error.
 */
msg_multipart_t *msg_multipart_create(su_home_t *home,
				      char const *content_type,
				      void const *data,
				      isize_t dlen)
{
  msg_multipart_t *mp;

  mp = (msg_multipart_t *)msg_header_alloc(home, msg_multipart_class, 0);

  if (mp) {
    if (content_type)
      mp->mp_content_type = msg_content_type_make(home, content_type);
    if (dlen)
      mp->mp_payload = msg_payload_create(home, data, dlen);

    if ((!mp->mp_content_type && content_type) ||
	(!mp->mp_payload && dlen)) {
      su_free(home, mp->mp_content_type);
      su_free(home, mp->mp_payload);
      su_free(home, mp);
      mp = NULL;
    }
  }

  return mp;
}

/** Convert boundary parameter to a search string. */
static char *
msg_multipart_boundary(su_home_t *home, char const *b)
{
  char *boundary;

  if (!b || !(boundary = su_alloc(home, 2 + 2 + strlen(b) + 2 + 1)))
    return NULL;

  strcpy(boundary, CR LF "--");

  if (b[0] == '"') /* " See http://bugzilla.gnome.org/show_bug.cgi?id=134216 */

    msg_unquote(boundary + 4, b);
  else
    strcpy(boundary + 4, b);


  strcat(boundary + 4, CR LF);

  return boundary;
}


/** Boundary chars. */
static char const bchars[] = 
"'()+_,-./:=?"
"0123456789"
"ABCDEFGHIJKLMNOPQRSTUVWXYZ"
"abcdefghijklmnopqrstuvwxyz"
" ";

#define bchars_len (sizeof(bchars) - 1)

/** Search for a suitable boundary from MIME. */
static char *
msg_multipart_search_boundary(su_home_t *home, char const *p, size_t len)
{
  size_t m;
  unsigned crlf;
  char const *end = p + len;
  char *boundary;

  if (len < 2)
    return NULL;

  /* Boundary looks like LF -- string SP* [CR] LF */
  if (memcmp("--", p, 2) == 0) {
    /* We can be at boundary beginning, there is no CR LF */
    m = 2 + memspn(p + 2, len - 2, bchars, bchars_len);
    if (m + 2 >= len)
      return NULL;
    crlf = p[m] == '\r' ? 1 + (p[m + 1] == '\n') : (p[m] == '\n');
    while (p[m - 1] == ' ' || p[m - 1] == '\t')
      m--;
    if (m > 2 && crlf) {
      boundary = su_alloc(home, 2 + m + 2 + 1);
      if (boundary) {
	memcpy(boundary, CR LF, 2);
	memcpy(boundary + 2, p, m);
	strcpy(boundary + m + 2, CR LF);
      }
      return boundary;
    }
  } 

  /* Look for LF -- */
  for (;(p = memmem(p, end - p, LF "--", 3)); p += 3) {
    len = end - p;
    m = 3 + memspn(p + 3, len - 3, bchars, bchars_len);
    if (m + 2 >= len)
      return NULL;
    crlf = p[m] == '\r' ? 1 + (p[m + 1] == '\n') : (p[m] == '\n');
    while (p[m - 1] == ' ' || p[m - 1] == '\t')
      m--;
    m--;
    if (m > 2 && crlf) {
      boundary = su_alloc(home, 2 + m + 2 + 1);
      if (boundary) {
	memcpy(boundary, CR LF, 2);
	memcpy(boundary + 2, p + 1, m);
	strcpy(boundary + 2 + m, CR LF);
      }
      return boundary;
    }
  }

  return NULL;
}

/** Parse a MIME multipart.
 *
 * The function msg_multipart_parse() parses a MIME multipart message. The
 * common syntax of multiparts is described in @RFC2046 (section 7).
 *
 * @param[in,out] home home for allocating structures 
 * @param[in]     c    content-type header for multipart 
 * @param[in]     pl   payload structure for multipart 
 *
 * After parsing, the @a pl will contain the plain-text preamble (if any).
 *
 * @note If no @b Content-Type header is given, the msg_multipart_parse()
 * tries to look for a suitable boundary. Currently, it takes first
 * boundary-looking string and uses that, so it can be fooled with, for
 * instance, signature @c "--Pekka".
 */
msg_multipart_t *msg_multipart_parse(su_home_t *home,
				     msg_content_type_t const *c,
				     msg_payload_t *pl)
{
  msg_multipart_t *mp = NULL, *all = NULL, **mmp = &all;
  /* Dummy msg object */
  msg_t msg[1] = {{{ SU_HOME_INIT(msg) }}};
  size_t len, m, blen;
  char *boundary, *p, *next, save;
  char const *b, *end;
  msg_param_t param;

  p = pl->pl_data; len = pl->pl_len; end = p + len;

  su_home_init(msg_home(msg));
  msg->m_class = msg_multipart_mclass;
  msg->m_tail = &msg->m_chain;

  /* Get boundary from Content-Type */
  if (c && (param = msg_header_find_param(c->c_common, "boundary=")))
    boundary = msg_multipart_boundary(msg_home(msg), param);
  else
    boundary = msg_multipart_search_boundary(msg_home(msg), p, len);

  if (!boundary)
    return NULL;

  m = strlen(boundary) - 2, blen = m - 1;

  /* Find first delimiter */
  if (memcmp(boundary + 2, p, m - 2) == 0)
    b = p, p = p + m - 2, len -= m - 2;
  else if ((p = memmem(p, len, boundary + 1, m - 1))) {
    if (p != pl->pl_data && p[-1] == '\r')
      b = --p, p = p + m, len -= m;
    else
      b = p, p = p + m - 1, len -= m - 1;
  }
  else {
    su_home_deinit(msg_home(msg));
    return NULL;
  }

  /* Split multipart into parts */
  for (;;) {
    while (p[0] == ' ') 
      p++;

    p += p[0] == '\r' ? 1 + (p[1] == '\n') : (p[0] == '\n');

    len = end - p;

    if (len < blen)
      break;

    next = memmem(p, len, boundary + 1, m = blen);

    if (!next)
      break;			/* error */

    if (next != p && next[-1] == '\r')
      next--, m++;

    mp = (msg_multipart_t *)msg_header_alloc(msg_home(msg), msg_multipart_class, 0);
    if (mp == NULL)
      break;			/* error */
    *mmp = mp; mmp = &mp->mp_next;

    /* Put delimiter transport-padding CRLF here */
    mp->mp_common->h_data = b;
    mp->mp_common->h_len = p - b;

    /* .. and body-part here */
    mp->mp_data = p;
    mp->mp_len = next - p;

    if (next[m] == '-' && next[m + 1] == '-') {
      /* We found close-delimiter */
      assert(mp);
      if (!mp)
	break;			/* error */
      mp->mp_close_delim = (msg_payload_t *)
	msg_header_alloc(msg_home(msg), msg_payload_class, 0);
      if (!mp->mp_close_delim)
	break;			/* error */
      /* Include also transport-padding and epilogue in the close-delimiter */
      mp->mp_close_delim->pl_data = next;
      mp->mp_close_delim->pl_len = p + len - next;
      break;
    }

    b = next; p = next + m; 
  }

  if (!mp || !mp->mp_close_delim) {
    su_home_deinit(msg_home(msg));
    /* Delimiter error */
    return NULL;
  }

  /* Parse each part */
  for (mp = all; mp; mp = mp->mp_next) {
    msg->m_object = (msg_pub_t *)mp; p = mp->mp_data; next = p + mp->mp_len;

    if (msg->m_tail)
      mp->mp_common->h_prev = msg->m_tail,
	*msg->m_tail = (msg_header_t *)mp;

    msg->m_chain = (msg_header_t *)mp;
    msg->m_tail = &mp->mp_common->h_succ;

    save = *next; *next = '\0';	/* NUL-terminate this part */

    for (len = next - p; len > 0; len -= m, p += m) {
      if (IS_CRLF(p[0])) {
	m = msg_extract_separator(msg, (msg_pub_t*)mp, p, len, 1);
	assert(m > 0);

	p += m; len -= m;

	if (len > 0) {
	  m = msg_extract_payload(msg, (msg_pub_t*)mp, NULL, len, p, len, 1);
	  assert(m > 0);
	  assert(len == m);
	}
	break;
      }

      m = msg_extract_header(msg, (msg_pub_t*)mp, p, len, 1);

      if (m <= 0) {
	assert(m > 0);
	/* Xyzzy */
      }
    }

    *next = save; /* XXX - Should we leave the payload NUL-terminated? */
  }

  /* Postprocess */
  blen = strlen(boundary);

  for (mp = all; mp; mp = mp->mp_next) {
    mp->mp_data = boundary; 
    mp->mp_len = blen;

    assert(mp->mp_payload || mp->mp_separator);

    if (mp->mp_close_delim) {
      msg_header_t **tail;

      if (mp->mp_payload)
	tail = &mp->mp_payload->pl_common->h_succ;
      else
	tail = &mp->mp_separator->sep_common->h_succ;

      assert(msg->m_chain == (msg_header_t *)mp);
      assert(*tail == NULL);

      mp->mp_close_delim->pl_common->h_prev = tail;
      *tail = (msg_header_t *)mp->mp_close_delim;
    }
  }