/*
 * This file is part of DisOrder
 * Copyright (C) 2005, 2007, 2008 Richard Kettlewell
 *
 * This program is free software: you can redistribute it and/or modify
 * it under the terms of the GNU General Public License as published by
 * the Free Software Foundation, either version 3 of the License, or
 * (at your option) any later version.
 * 
 * This program 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 General Public License for more details.
 * 
 * You should have received a copy of the GNU General Public License
 * along with this program.  If not, see <http://www.gnu.org/licenses/>.
 */
/** @file lib/mime.c
 * @brief Support for MIME and allied protocols
 */

#include "common.h"

#include <ctype.h>

#include "mem.h"
#include "mime.h"
#include "vector.h"
#include "hex.h"
#include "log.h"
#include "base64.h"
#include "kvp.h"

/** @brief Match whitespace characters */
static int whitespace(int c) {
  switch(c) {
  case ' ':
  case '\t':
  case '\r':
  case '\n':
    return 1;
  default:
    return 0;
  }
}

/** @brief Match RFC2045 tspecial characters */
int mime_tspecial(int c) {
  switch(c) {
  case '(':
  case ')':
  case '<':
  case '>':
  case '@':
  case ',':
  case ';':
  case ':':
  case '\\':
  case '"':
  case '/':
  case '[':
  case ']':
  case '?':
  case '=':
    return 1;
  default:
    return 0;
  }
}

/** @brief Match RFC2616 separator characters */
int mime_http_separator(int c) {
  switch(c) {
  case '(':
  case ')':
  case '<':
  case '>':
  case '@':
  case ',':
  case ';':
  case ':':
  case '\\':
  case '"':
  case '/':
  case '[':
  case ']':
  case '?':
  case '=':
  case '{':
  case '}':
  case ' ':
  case '\t':
    return 1;
  default:
    return 0;
  }
}

/** @brief Match CRLF */
static int iscrlf(const char *ptr) {
  return ptr[0] == '\r' && ptr[1] == '\n';
}

/** @brief Skip whitespace
 * @param s Pointer into string
 * @param rfc822_comments If true, skip RFC822 nested comments
 * @return Pointer into string after whitespace
 */
static const char *skipwhite(const char *s, int rfc822_comments) {
  int c, depth;
  
  for(;;) {
    switch(c = *s) {
    case ' ':
    case '\t':
    case '\r':
    case '\n':
      ++s;
      break;
    case '(':
      if(!rfc822_comments)
	return s;
      ++s;
      depth = 1;
      while(*s && depth) {
	c = *s++;
	switch(c) {
	case '(': ++depth; break;
	case ')': --depth; break;
	case '\\':
	  if(!*s)
	    return 0;
	  ++s;
	  break;
	}
      }
      if(depth)
	return 0;
      break;
    default:
      return s;
    }
  }
}

/** @brief Test for a word character
 * @param c Character to test
 * @param special mime_tspecial() (MIME/RFC2405) or mime_http_separator() (HTTP/RFC2616)
 * @return 1 if @p c is a word character, else 0
 */
static int iswordchar(int c, int (*special)(int)) {
  return !(c <= ' ' || c > '~' || special(c));
}

/** @brief Parse an RFC1521/RFC2616 word
 * @param s Pointer to start of word
 * @param valuep Where to store value
 * @param special mime_tspecial() (MIME/RFC2405) or mime_http_separator() (HTTP/RFC2616)
 * @return Pointer just after end of word or NULL if there's no word
 *
 * A word is a token or a quoted-string.
 */
const char *mime_parse_word(const char *s, char **valuep,
			    int (*special)(int)) {
  struct dynstr value[1];
  int c;

  dynstr_init(value);
  if(*s == '"') {
    ++s;
    while((c = *s++) != '"') {
      switch(c) {
      case '\\':
	if(!(c = *s++))
	  return 0;
      default:
	dynstr_append(value, c);
	break;
      }
    }
    if(!c)
      return 0;
  } else {
    if(!iswordchar((unsigned char)*s, special))
      return NULL;
    dynstr_init(value);
    while(iswordchar((unsigned char)*s, special))
      dynstr_append(value, *s++);
  }
  dynstr_terminate(value);
  *valuep = value->vec;
  return s;
}

/** @brief Parse an RFC1521/RFC2616 token
 * @param s Pointer to start of token
 * @param valuep Where to store value
 * @param special mime_tspecial() (MIME/RFC2405) or mime_http_separator() (HTTP/RFC2616)
 * @return Pointer just after end of token or NULL if there's no token
 */
static const char *parsetoken(const char *s, char **valuep,
			      int (*special)(int)) {
  if(*s == '"')
    return 0;
  return mime_parse_word(s, valuep, special);
}

/** @brief Parse a MIME content-type field
 * @param s Start of field
 * @param typep Where to store type
 * @param parametersp Where to store parameter list
 * @return 0 on success, non-0 on error
 *
 * See <a href="http://tools.ietf.org/html/rfc2045#section-5">RFC 2045 s5</a>.
 */
int mime_content_type(const char *s,
		      char **typep,
		      struct kvp **parametersp) {
  struct dynstr type, parametername;
  struct kvp *parameters = 0;
  char *parametervalue;

  dynstr_init(&type);
  if(!(s = skipwhite(s, 1)))
    return -1;
  if(!*s)
    return -1;
  while(*s && !mime_tspecial(*s) && !whitespace(*s))
    dynstr_append(&type, tolower((unsigned char)*s++));
  if(!(s = skipwhite(s, 1)))
    return -1;
  if(*s++ != '/')
    return -1;
  dynstr_append(&type, '/');
  if(!(s = skipwhite(s, 1)))
    return -1;
  while(*s && !mime_tspecial(*s) && !whitespace(*s))
    dynstr_append(&type, tolower((unsigned char)*s++));
  if(!(s = skipwhite(s, 1)))
    return -1;

  while(*s == ';') {
    dynstr_init(&parametername);
    ++s;
    if(!(s = skipwhite(s, 1)))
      return -1;
    if(!*s)
      return -1;
    while(*s && !mime_tspecial(*s) && !whitespace(*s))
      dynstr_append(&parametername, tolower((unsigned char)*s++));
    if(!(s = skipwhite(s, 1)))
      return -1;
    if(*s++ != '=')
      return -1;
    if(!(s = skipwhite(s, 1)))
      return -1;
    if(!(s = mime_parse_word(s, &parametervalue, mime_tspecial)))
      return -1;
    if(!(s = skipwhite(s, 1)))
      return -1;
    dynstr_terminate(&parametername);
    kvp_set(&parameters, parametername.vec, parametervalue);
  }
  dynstr_terminate(&type);
  *typep = type.vec;
  *parametersp = parameters;
  return 0;
}

/** @brief Parse a MIME message
 * @param s Start of message
 * @param callback Called for each header field
 * @param u Passed to callback
 * @return Pointer to decoded body (might be in original string), or NULL on error
 *
 * This does an RFC 822-style parse and honors Content-Transfer-Encoding as
 * described in <a href="http://tools.ietf.org/html/rfc2045#section-6">RFC 2045
 * s6</a>.  @p callback is called for each header field encountered, in order,
 * with ASCII characters in the header name forced to lower case.
 */
const char *mime_parse(const char *s,
		       int (*callback)(const char *name, const char *value,
				       void *u),
		       void *u) {
  struct dynstr name, value;
  char *cte = 0, *p;
  
  while(*s && !iscrlf(s)) {
    dynstr_init(&name);
    dynstr_init(&value);
    while(*s && !mime_tspecial(*s) && !whitespace(*s))
      dynstr_append(&name, tolower((unsigned char)*s++));
    if(!(s = skipwhite(s, 1)))
      return 0;
    if(*s != ':')
      return 0;
    ++s;
    while(*s && !(*s == '\n' && !(s[1] == ' ' || s[1] == '\t'))) {
      const int c = *s++;
      /* Strip leading whitespace */
      if(value.nvec || !(c == ' ' || c == '\t' || c == '\n' || c == '\r'))
	dynstr_append(&value, c);
    }
    /* Strip trailing whitespace */
    while(value.nvec > 0 && (value.vec[value.nvec - 1] == ' '
			     || value.vec[value.nvec - 1] == '\t'
			     || value.vec[value.nvec - 1] == '\n'
			     || value.vec[value.nvec - 1] == '\r'))
      --value.nvec;
    if(*s)
      ++s;
    dynstr_terminate(&name);
    dynstr_terminate(&value);
    if(!strcmp(name.vec, "content-transfer-encoding")) {
      cte = xstrdup(value.vec);
      for(p = cte; *p; p++)
	*p = tolower((unsigned char)*p);
    }
    if(callback(name.vec, value.vec, u))
      return 0;
  }
  if(*s)
    s += 2;
  if(cte) {
    if(!strcmp(cte, "base64"))
      return mime_base64(s, 0);
    if(!strcmp(cte, "quoted-printable"))
      return mime_qp(s);
    if(!strcmp(cte, "7bit") || !strcmp(cte, "8bit"))
      return s;
    disorder_error(0, "unknown content-transfer-encoding '%s'", cte);
    return 0;
  }
  return s;
}

/** @brief Match the boundary string */
static int isboundary(const char *ptr, const char *boundary, size_t bl) {
  return (ptr[0] == '-'
	  && ptr[1] == '-'
	  && !strncmp(ptr + 2, boundary, bl)
	  && (iscrlf(ptr + bl + 2)
	      || (ptr[bl + 2] == '-'
		  && ptr[bl + 3] == '-'
		  && (iscrlf(ptr + bl + 4) || *(ptr + bl + 4) == 0))));
}

/** @brief Match the final boundary string */
static int isfinal(const char *ptr, const char *boundary, size_t bl) {
  return (ptr[0] == '-'
	  && ptr[1] == '-'
	  && !strncmp(ptr + 2, boundary, bl)
	  && ptr[bl + 2] == '-'
	  && ptr[bl + 3] == '-'
	  && (iscrlf(ptr + bl + 4) || *(ptr + bl + 4) == 0));
}

/** @brief Parse a multipart MIME body
 * @param s Start of message
 * @param callback Callback for each part
 * @param boundary Boundary string
 * @param u Passed to callback
 * @return 0 on success, non-0 on error
 * 
 * See <a href="http://tools.ietf.org/html/rfc2046#section-5.1">RFC 2046
 * s5.1</a>.  @p callback is called for each part (not yet decoded in any way)
 * in succession; you should probably call mime_parse() for each part.
 */
int mime_multipart(const char *s,
		   int (*callback)(const char *s, void *u),
		   const char *boundary,
		   void *u) {
  size_t bl = strlen(boundary);
  const char *start, *e;
  int ret;

  /* We must start with a boundary string */
  if(!isboundary(s, boundary, bl)) {
    disorder_error(0, "mime_multipart: first line is not the boundary string");
    return -1;
  }
  /* Keep going until we hit a final boundary */
  while(!isfinal(s, boundary, bl)) {
    s = strstr(s, "\r\n") + 2;
    start = s;
    while(!isboundary(s, boundary, bl)) {
      if(!(e = strstr(s, "\r\n"))) {
	disorder_error(0, "mime_multipart: line does not end CRLF");
	return -1;
      }
      s = e + 2;
    }
    if((ret = callback(xstrndup(start,
				s == start ? 0 : s - start - 2),
		       u)))
      return ret;
  }
  return 0;
}

/** @brief Parse an RFC2388-style content-disposition field
 * @param s Start of field
 * @param dispositionp Where to store disposition
 * @param parameternamep Where to store parameter name
 * @param parametervaluep Wher to store parameter value
 * @return 0 on success, non-0 on error
 *
 * See <a href="http://tools.ietf.org/html/rfc2388#section-3">RFC 2388 s3</a>
 * and <a href="http://tools.ietf.org/html/rfc2183">RFC 2183</a>.
 */
int mime_rfc2388_content_disposition(const char *s,
				     char **dispositionp,
				     char **parameternamep,
				     char **parametervaluep) {
  struct dynstr disposition, parametername;

  dynstr_init(&disposition);
  if(!(s = skipwhite(s, 1)))
    return -1;
  if(!*s)
    return -1;
  while(*s && !mime_tspecial(*s) && !whitespace(*s))
    dynstr_append(&disposition, tolower((unsigned char)*s++));
  if(!(s = skipwhite(s, 1)))
    return -1;

  if(*s == ';') {
    dynstr_init(&parametername);
    ++s;
    if(!(s = skipwhite(s, 1)))
      return -1;
    if(!*s)
      return -1;
    while(*s && !mime_tspecial(*s) && !whitespace(*s))
      dynstr_append(&parametername, tolower((unsigned char)*s++));
    if(!(s = skipwhite(s, 1)))
      return -1;
    if(*s++ != '=')
      return -1;
    if(!(s = skipwhite(s, 1)))
      return -1;
    if(!(s = mime_parse_word(s, parametervaluep, mime_tspecial)))
      return -1;
    if(!(s = skipwhite(s, 1)))
      return -1;
    dynstr_terminate(&parametername);
    *parameternamep = parametername.vec;
  } else
    *parametervaluep = *parameternamep = 0;
  dynstr_terminate(&disposition);
  *dispositionp = disposition.vec;
  return 0;
}

/** @brief Convert MIME quoted-printable
 * @param s Quoted-printable data
 * @return Decoded data
 *
 * See <a href="http://tools.ietf.org/html/rfc2045#section-6.7">RFC 2045
 * s6.7</a>.
 */
char *mime_qp(const char *s) {
  struct dynstr d;
  int c, a, b;
  const char *t;

  dynstr_init(&d);
  while((c = *s++)) {
    switch(c) {
    case '=':
      if((a = unhexdigitq(s[0])) != -1
	 && (b = unhexdigitq(s[1])) != -1) {
	dynstr_append(&d, a * 16 + b);
	s += 2;
      } else {
	t = s;
	while(*t == ' ' || *t == '\t') ++t;
	if(iscrlf(t)) {
	  /* soft line break */
	  s = t + 2;
	} else
	  return 0;
      }
      break;
    case ' ':
    case '\t':
      t = s;
      while(*t == ' ' || *t == '\t') ++t;
      if(iscrlf(t))
	/* trailing space is always eliminated */
	s = t;
      else
	dynstr_append(&d, c);
      break;
    default:
      dynstr_append(&d, c);
      break;
    }
  }
  dynstr_terminate(&d);
  return d.vec;
}

/** @brief Match cookie separator characters
 *
 * This is a subset of the RFC2616 specials, and technically is in breach of
 * the specification.  However rejecting (in particular) slashes is
 * unreasonably strict and has broken at least one (admittedly somewhat
 * obscure) browser, so we're more forgiving.
 */
static int cookie_separator(int c) {
  switch(c) {
  case '(':
  case ')':
  case ',':
  case ';':
  case '=':
  case ' ':
  case '"':
  case '\t':
    return 1;

  default:
    return 0;
  }
}

/** @brief Match cookie value separator characters
 *
 * Same as cookie_separator() but allows for @c = in cookie values.
 */
static int cookie_value_separator(int c) {
  switch(c) {
  case '(':
  case ')':
  case ',':
  case ';':
  case ' ':
  case '"':
  case '\t':
    return 1;

  default:
    return 0;
  }
}

/** @brief Parse a RFC2109 Cookie: header
 * @param s Header field value
 * @param cd Where to store result
 * @return 0 on success, non-0 on error
 *
 * See <a href="http://tools.ietf.org/html/rfc2109">RFC 2109</a>.
 */
int parse_cookie(const char *s,
		 struct cookiedata *cd) {
  char *n = 0, *v = 0;

  memset(cd, 0, sizeof *cd);
  s = skipwhite(s, 0);
  while(*s) {
    /* Skip separators */
    if(*s == ';' || *s == ',') {
      ++s;
      s = skipwhite(s, 0);
      continue;
    }
    if(!(s = parsetoken(s, &n, cookie_separator))) {
      disorder_error(0, "parse_cookie: cannot parse attribute name");
      return -1;
    }
    s = skipwhite(s, 0);
    if(*s++ != '=') {
      disorder_error(0, "parse_cookie: did not find expected '='");
      return -1;
    }
    s = skipwhite(s, 0);
    if(!(s = mime_parse_word(s, &v, cookie_value_separator))) {
      disorder_error(0, "parse_cookie: cannot parse value for '%s'", n);
      return -1;
    }
    if(n[0] == '$') {
      /* Some bit of meta-information */
      if(!strcmp(n, "$Version"))
	cd->version = v;
      else if(!strcmp(n, "$Path")) {
	if(cd->ncookies > 0 && cd->cookies[cd->ncookies-1].path == 0)
	  cd->cookies[cd->ncookies-1].path = v;
	else {
	  disorder_error(0, "redundant $Path in Cookie: header");
	  return -1;
	}
      } else if(!strcmp(n, "$Domain")) {
	if(cd->ncookies > 0 && cd->cookies[cd->ncookies-1].domain == 0)
	  cd->cookies[cd->ncookies-1].domain = v;
	else {
	  disorder_error(0, "redundant $Domain in Cookie: header");
	  return -1;
	}
      }
    } else {
      /* It's a new cookie */
      cd->cookies = xrealloc(cd->cookies,
			     (cd->ncookies + 1) * sizeof (struct cookie));
      cd->cookies[cd->ncookies].name = n;
      cd->cookies[cd->ncookies].value = v;
      cd->cookies[cd->ncookies].path = 0;
      cd->cookies[cd->ncookies].domain = 0;
      ++cd->ncookies;
    }
    s = skipwhite(s, 0);
    if(*s && (*s != ',' && *s != ';')) {
      disorder_error(0, "missing separator in Cookie: header");
      return -1;
    }
  }
  return 0;
}

/** @brief Find a named cookie
 * @param cd Parse cookie data
 * @param name Name of cookie
 * @return Cookie structure or NULL if not found
 */
const struct cookie *find_cookie(const struct cookiedata *cd,
				 const char *name) {
  int n;

  for(n = 0; n < cd->ncookies; ++n)
    if(!strcmp(cd->cookies[n].name, name))
      return &cd->cookies[n];
  return 0;
}

/** @brief RFC822 quoting
 * @param s String to quote
 * @param force If non-0, always quote
 * @return Possibly quoted string
 */
char *quote822(const char *s, int force) {
  const char *t;
  struct dynstr d[1];
  int c;

  if(!force) {
    /* See if we need to quote */
    for(t = s; (c = (unsigned char)*t); ++t) {
      if(mime_tspecial(c) || mime_http_separator(c) || whitespace(c))
	break;
    }
    if(*t)
      force = 1;
  }

  if(!force)
    return xstrdup(s);

  dynstr_init(d);
  dynstr_append(d, '"');
  for(t = s; (c = (unsigned char)*t); ++t) {
    if(c == '"' || c == '\\')
      dynstr_append(d, '\\');
    dynstr_append(d, c);
  }
  dynstr_append(d, '"');
  dynstr_terminate(d);
  return d->vec;
}

/** @brief Return true if @p ptr points at trailing space */
static int is_trailing_space(const char *ptr) {
  if(*ptr == ' ' || *ptr == '\t') {
    while(*ptr == ' ' || *ptr == '\t')
      ++ptr;
    return *ptr == '\n' || *ptr == 0;
  } else
    return 0;
}

/** @brief Encoding text as quoted-printable
 * @param text String to encode
 * @return Encoded string
 *
 * See <a href="http://tools.ietf.org/html/rfc2045#section-6.7">RFC2045
 * s6.7</a>.
 */
char *mime_to_qp(const char *text) {
  struct dynstr d[1];
  int linelength = 0;			/* length of current line */
  char buffer[10];

  dynstr_init(d);
  /* The rules are:
   * 1. Anything except newline can be replaced with =%02X
   * 2. Newline, 33-60 and 62-126 stand for themselves (i.e. not '=')
   * 3. Non-trailing space/tab stand for themselves.
   * 4. Output lines are limited to 76 chars, with =<newline> being used
   *    as a soft line break
   * 5. Newlines aren't counted towards the 76 char limit.
   */
  while(*text) {
    const int c = (unsigned char)*text;
    if(c == '\n') {
      /* Newline stands as itself */
      dynstr_append(d, '\n');
      linelength = 0;
    } else if((c >= 33 && c <= 126 && c != '=')
	      || ((c == ' ' || c == '\t')
		  && !is_trailing_space(text))) {
      /* Things that can stand for themselves  */
      dynstr_append(d, c);
      ++linelength;
    } else {
      /* Anything else that needs encoding */
      snprintf(buffer, sizeof buffer, "=%02X", c);
      dynstr_append_string(d, buffer);
      linelength += 3;
    }
    ++text;
    if(linelength > 73 && *text && *text != '\n') {
      /* Next character might overflow 76 character limit if encoded, so we
       * insert a soft break */
      dynstr_append_string(d, "=\n");
      linelength = 0;
    }
  }
  /* Ensure there is a final newline */
  if(linelength)
    dynstr_append(d, '\n');
  /* That's all */
  dynstr_terminate(d);
  return d->vec;
}

/** @brief Encode text
 * @param text Underlying UTF-8 text
 * @param charsetp Where to store charset string
 * @param encodingp Where to store encoding string
 * @return Encoded text (might be @p text)
 */
const char *mime_encode_text(const char *text,
			     const char **charsetp,
			     const char **encodingp) {
  const char *ptr;

  /* See if there are in fact any non-ASCII characters */
  for(ptr = text; *ptr; ++ptr)
    if((unsigned char)*ptr >= 128)
      break;
  if(!*ptr) {
    /* Plain old ASCII, no encoding required */
    *charsetp = "us-ascii";
    *encodingp = "7bit";
    return text;
  }
  *charsetp = "utf-8";
  *encodingp = "quoted-printable";
  return mime_to_qp(text);
}

/*
Local Variables:
c-basic-offset:2
comment-column:40
fill-column:79
End:
*/