path: root/camel/camel-mime-utils.c

/* -*- Mode: C; tab-width: 8; indent-tabs-mode: t; c-basic-offset: 8 -*- */
/*
 *  Copyright (C) 2000 Ximian Inc.
 *
 *  Authors: Michael Zucchi <notzed@ximian.com>
 *           Jeffrey Stedfast <fejj@ximian.com>
 *
 * This program is free software; you can redistribute it and/or
 * modify it under the terms of version 2 of the GNU General Public
 * License as published by the Free Software Foundation.
 *
 * 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, write to the
 * Free Software Foundation, Inc., 59 Temple Place - Suite 330,
 * Boston, MA 02111-1307, USA.
 */

/* dont touch this file without my permission - Michael */

#ifdef HAVE_CONFIG_H
#include <config.h>
#endif

#include <stdio.h>
#include <sys/types.h>
#include <sys/stat.h>
#include <sys/param.h>  /* for MAXHOSTNAMELEN */
#include <fcntl.h>
#include <stdlib.h>
#include <string.h>
#include <unistd.h>

#ifndef MAXHOSTNAMELEN
#define MAXHOSTNAMELEN 1024
#endif

#include <iconv.h>

#include <time.h>

#include <ctype.h>
#include <errno.h>
#include <regex.h>

#include <glib.h>
#include <gal/unicode/gunicode.h>
#include <gal/util/e-iconv.h>

#include "camel-mime-utils.h"
#include "camel-charset-map.h"

#ifdef ENABLE_THREADS
#include <pthread.h>
#endif

#ifndef CLEAN_DATE
#include "broken-date-parser.h"
#endif

#if 0
int strdup_count = 0;
int malloc_count = 0;
int free_count = 0;

#define g_strdup(x) (strdup_count++, g_strdup(x))
#define g_malloc(x) (malloc_count++, g_malloc(x))
#define g_free(x) (free_count++, g_free(x))
#endif

/* for all non-essential warnings ... */
#define w(x) 

#define d(x)
#define d2(x)

#define CAMEL_UUENCODE_CHAR(c)  ((c) ? (c) + ' ' : '`')
#define CAMEL_UUDECODE_CHAR(c)  (((c) - ' ') & 077)

static char *base64_alphabet =
"ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz0123456789+/";

static unsigned char tohex[16] = {
    '0', '1', '2', '3', '4', '5', '6', '7',
    '8', '9', 'A', 'B', 'C', 'D', 'E', 'F'
};

static unsigned short camel_mime_special_table[256];
static unsigned char camel_mime_base64_rank[256];

/* Flags bits set in the mime_special table, use the is_*() mactos to access them normally */
enum {
    IS_CTRL     = 1<<0,
    IS_LWSP     = 1<<1,
    IS_TSPECIAL = 1<<2,
    IS_SPECIAL  = 1<<3,
    IS_SPACE    = 1<<4,
    IS_DSPECIAL = 1<<5,
    IS_QPSAFE   = 1<<6,
    IS_ESAFE    = 1<<7, /* encoded word safe */
    IS_PSAFE    = 1<<8, /* encoded word in phrase safe */
};

#define is_ctrl(x) ((camel_mime_special_table[(unsigned char)(x)] & IS_CTRL) != 0)
#define is_lwsp(x) ((camel_mime_special_table[(unsigned char)(x)] & IS_LWSP) != 0)
#define is_tspecial(x) ((camel_mime_special_table[(unsigned char)(x)] & IS_TSPECIAL) != 0)
#define is_type(x, t) ((camel_mime_special_table[(unsigned char)(x)] & (t)) != 0)
#define is_ttoken(x) ((camel_mime_special_table[(unsigned char)(x)] & (IS_TSPECIAL|IS_LWSP|IS_CTRL)) == 0)
#define is_atom(x) ((camel_mime_special_table[(unsigned char)(x)] & (IS_SPECIAL|IS_SPACE|IS_CTRL)) == 0)
#define is_dtext(x) ((camel_mime_special_table[(unsigned char)(x)] & IS_DSPECIAL) == 0)
#define is_fieldname(x) ((camel_mime_special_table[(unsigned char)(x)] & (IS_CTRL|IS_SPACE)) == 0)
#define is_qpsafe(x) ((camel_mime_special_table[(unsigned char)(x)] & IS_QPSAFE) != 0)
#define is_especial(x) ((camel_mime_special_table[(unsigned char)(x)] & IS_ESPECIAL) != 0)
#define is_psafe(x) ((camel_mime_special_table[(unsigned char)(x)] & IS_PSAFE) != 0)

/* Used by table initialisation code for special characters */
#define CHARS_LWSP " \t\n\r"
#define CHARS_TSPECIAL "()<>@,;:\\\"/[]?="
#define CHARS_SPECIAL "()<>@,;:\\\".[]"
#define CHARS_CSPECIAL "()\\\r" /* not in comments */
#define CHARS_DSPECIAL "[]\\\r \t"  /* not in domains */
#define CHARS_ESPECIAL "()<>@,;:\"/[]?.=_" /* list of characters that must be encoded.
                          encoded word in text specials: rfc 2047 5(1)*/
#define CHARS_PSPECIAL "!*+-/" /* list of additional characters that can be left unencoded.
                  encoded word in phrase specials: rfc 2047 5(3) */

static void
header_remove_bits(unsigned short bit, unsigned char *vals)
{
    int i;

    for (i=0;vals[i];i++)
        camel_mime_special_table[vals[i]] &= ~ bit;
}

static void
header_init_bits(unsigned short bit, unsigned short bitcopy, int remove, unsigned char *vals)
{
    int i;
    int len = strlen(vals);

    if (!remove) {
        for (i=0;i<len;i++) {
            camel_mime_special_table[vals[i]] |= bit;
        }
        if (bitcopy) {
            for (i=0;i<256;i++) {
                if (camel_mime_special_table[i] & bitcopy)
                    camel_mime_special_table[i] |= bit;
            }
        }
    } else {
        for (i=0;i<256;i++)
            camel_mime_special_table[i] |= bit;
        for (i=0;i<len;i++) {
            camel_mime_special_table[vals[i]] &= ~bit;
        }
        if (bitcopy) {
            for (i=0;i<256;i++) {
                if (camel_mime_special_table[i] & bitcopy)
                    camel_mime_special_table[i] &= ~bit;
            }
        }
    }
}

static void
header_decode_init(void)
{
    int i;

    for (i=0;i<256;i++) {
        camel_mime_special_table[i] = 0;
        if (i<32)
            camel_mime_special_table[i] |= IS_CTRL;
        if ((i>=33 && i<=60) || (i>=62 && i<=126) || i==32 || i==9)
            camel_mime_special_table[i] |= (IS_QPSAFE|IS_ESAFE);
        if ((i>='0' && i<='9') || (i>='a' && i<='z') || (i>='A' && i<= 'Z'))
            camel_mime_special_table[i] |= IS_PSAFE;
    }
    camel_mime_special_table[127] |= IS_CTRL;
    camel_mime_special_table[' '] |= IS_SPACE;
    header_init_bits(IS_LWSP, 0, 0, CHARS_LWSP);
    header_init_bits(IS_TSPECIAL, IS_CTRL, 0, CHARS_TSPECIAL);
    header_init_bits(IS_SPECIAL, 0, 0, CHARS_SPECIAL);
    header_init_bits(IS_DSPECIAL, 0, FALSE, CHARS_DSPECIAL);
    header_remove_bits(IS_ESAFE, CHARS_ESPECIAL);
    header_init_bits(IS_PSAFE, 0, 0, CHARS_PSPECIAL);
}

static void
base64_init(void)
{
    int i;

    memset(camel_mime_base64_rank, 0xff, sizeof(camel_mime_base64_rank));
    for (i=0;i<64;i++) {
        camel_mime_base64_rank[(unsigned int)base64_alphabet[i]] = i;
    }
    camel_mime_base64_rank['='] = 0;
}

/* call this when finished encoding everything, to
   flush off the last little bit */
size_t
base64_encode_close(unsigned char *in, size_t inlen, gboolean break_lines, unsigned char *out, int *state, int *save)
{
    int c1, c2;
    unsigned char *outptr = out;

    if (inlen>0)
        outptr += base64_encode_step(in, inlen, break_lines, outptr, state, save);

    c1 = ((unsigned char *)save)[1];
    c2 = ((unsigned char *)save)[2];
    
    d(printf("mode = %d\nc1 = %c\nc2 = %c\n",
         (int)((char *)save)[0],
         (int)((char *)save)[1],
         (int)((char *)save)[2]));

    switch (((char *)save)[0]) {
    case 2:
        outptr[2] = base64_alphabet[ ( (c2 &0x0f) << 2 ) ];
        g_assert(outptr[2] != 0);
        goto skip;
    case 1:
        outptr[2] = '=';
    skip:
        outptr[0] = base64_alphabet[ c1 >> 2 ];
        outptr[1] = base64_alphabet[ c2 >> 4 | ( (c1&0x3) << 4 )];
        outptr[3] = '=';
        outptr += 4;
        break;
    }
    if (break_lines)
        *outptr++ = '\n';

    *save = 0;
    *state = 0;

    return outptr-out;
}

/*
  performs an 'encode step', only encodes blocks of 3 characters to the
  output at a time, saves left-over state in state and save (initialise to
  0 on first invocation).
*/
size_t
base64_encode_step(unsigned char *in, size_t len, gboolean break_lines, unsigned char *out, int *state, int *save)
{
    register unsigned char *inptr, *outptr;

    if (len<=0)
        return 0;

    inptr = in;
    outptr = out;

    d(printf("we have %d chars, and %d saved chars\n", len, ((char *)save)[0]));

    if (len + ((char *)save)[0] > 2) {
        unsigned char *inend = in+len-2;
        register int c1, c2, c3;
        register int already;

        already = *state;

        switch (((char *)save)[0]) {
        case 1: c1 = ((unsigned char *)save)[1]; goto skip1;
        case 2: c1 = ((unsigned char *)save)[1];
            c2 = ((unsigned char *)save)[2]; goto skip2;
        }
        
        /* yes, we jump into the loop, no i'm not going to change it, it's beautiful! */
        while (inptr < inend) {
            c1 = *inptr++;
        skip1:
            c2 = *inptr++;
        skip2:
            c3 = *inptr++;
            *outptr++ = base64_alphabet[ c1 >> 2 ];
            *outptr++ = base64_alphabet[ c2 >> 4 | ( (c1&0x3) << 4 ) ];
            *outptr++ = base64_alphabet[ ( (c2 &0x0f) << 2 ) | (c3 >> 6) ];
            *outptr++ = base64_alphabet[ c3 & 0x3f ];
            /* this is a bit ugly ... */
            if (break_lines && (++already)>=19) {
                *outptr++='\n';
                already = 0;
            }
        }

        ((char *)save)[0] = 0;
        len = 2-(inptr-inend);
        *state = already;
    }

    d(printf("state = %d, len = %d\n",
         (int)((char *)save)[0],
         len));

    if (len>0) {
        register char *saveout;

        /* points to the slot for the next char to save */
        saveout = & (((char *)save)[1]) + ((char *)save)[0];

        /* len can only be 0 1 or 2 */
        switch(len) {
        case 2: *saveout++ = *inptr++;
        case 1: *saveout++ = *inptr++;
        }
        ((char *)save)[0]+=len;
    }

    d(printf("mode = %d\nc1 = %c\nc2 = %c\n",
         (int)((char *)save)[0],
         (int)((char *)save)[1],
         (int)((char *)save)[2]));

    return outptr-out;
}


/**
 * base64_decode_step: decode a chunk of base64 encoded data
 * @in: input stream
 * @len: max length of data to decode
 * @out: output stream
 * @state: holds the number of bits that are stored in @save
 * @save: leftover bits that have not yet been decoded
 *
 * Decodes a chunk of base64 encoded data
 **/
size_t
base64_decode_step(unsigned char *in, size_t len, unsigned char *out, int *state, unsigned int *save)
{
    register unsigned char *inptr, *outptr;
    unsigned char *inend, c;
    register unsigned int v;
    int i;

    inend = in+len;
    outptr = out;

    /* convert 4 base64 bytes to 3 normal bytes */
    v=*save;
    i=*state;
    inptr = in;
    while (inptr<inend) {
        c = camel_mime_base64_rank[*inptr++];
        if (c != 0xff) {
            v = (v<<6) | c;
            i++;
            if (i==4) {
                *outptr++ = v>>16;
                *outptr++ = v>>8;
                *outptr++ = v;
                i=0;
            }
        }
    }

    *save = v;
    *state = i;

    /* quick scan back for '=' on the end somewhere */
    /* fortunately we can drop 1 output char for each trailing = (upto 2) */
    i=2;
    while (inptr>in && i) {
        inptr--;
        if (camel_mime_base64_rank[*inptr] != 0xff) {
            if (*inptr == '=')
                outptr--;
            i--;
        }
    }

    /* if i!= 0 then there is a truncation error! */
    return outptr-out;
}

char *
base64_encode_simple (const char *data, size_t len)
{
    unsigned char *out;
    int state = 0, outlen;
    unsigned int save = 0;
    
    out = g_malloc (len * 4 / 3 + 5);
    outlen = base64_encode_close ((unsigned char *)data, len, FALSE,
                      out, &state, &save);
    out[outlen] = '\0';
    return (char *)out;
}

size_t
base64_decode_simple (char *data, size_t len)
{
    int state = 0;
    unsigned int save = 0;

    return base64_decode_step ((unsigned char *)data, len,
                   (unsigned char *)data, &state, &save);
}

/**
 * uuencode_close: uuencode a chunk of data
 * @in: input stream
 * @len: input stream length
 * @out: output stream
 * @uubuf: temporary buffer of 60 bytes
 * @state: holds the number of bits that are stored in @save
 * @save: leftover bits that have not yet been encoded
 *
 * Returns the number of bytes encoded. Call this when finished
 * encoding data with uuencode_step to flush off the last little
 * bit.
 **/
size_t
uuencode_close (unsigned char *in, size_t len, unsigned char *out, unsigned char *uubuf, int *state, guint32 *save)
{
    register unsigned char *outptr, *bufptr;
    register guint32 saved;
    int uulen, i;
    
    outptr = out;
    
    if (len > 0)
        outptr += uuencode_step (in, len, out, uubuf, state, save);
    
    saved = *save;
    i = *state & 0xff;
    uulen = (*state >> 8) & 0xff;
    
    bufptr = uubuf + ((uulen / 3) * 4);
    
    if (i > 0) {
        while (i < 3) {
            saved <<= 8 | 0;
            i++;
        }
        
        if (i == 3) {
            /* convert 3 normal bytes into 4 uuencoded bytes */
            unsigned char b0, b1, b2;
            
            b0 = saved >> 16;
            b1 = saved >> 8 & 0xff;
            b2 = saved & 0xff;
            
            *bufptr++ = CAMEL_UUENCODE_CHAR ((b0 >> 2) & 0x3f);
            *bufptr++ = CAMEL_UUENCODE_CHAR (((b0 << 4) | ((b1 >> 4) & 0xf)) & 0x3f);
            *bufptr++ = CAMEL_UUENCODE_CHAR (((b1 << 2) | ((b2 >> 6) & 0x3)) & 0x3f);
            *bufptr++ = CAMEL_UUENCODE_CHAR (b2 & 0x3f);
            
            i = 0;
            saved = 0;
            uulen += 3;
        }
    }
    
    if (uulen > 0) {
        int cplen = ((uulen / 3) * 4);
        
        *outptr++ = CAMEL_UUENCODE_CHAR (uulen & 0xff);
        memcpy (outptr, uubuf, cplen);
        outptr += cplen;
        *outptr++ = '\n';
        uulen = 0;
    }
    
    *outptr++ = CAMEL_UUENCODE_CHAR (uulen & 0xff);
    *outptr++ = '\n';
    
    *save = 0;
    *state = 0;
    
    return outptr - out;
}


/**
 * uuencode_step: uuencode a chunk of data
 * @in: input stream
 * @len: input stream length
 * @out: output stream
 * @uubuf: temporary buffer of 60 bytes
 * @state: holds the number of bits that are stored in @save
 * @save: leftover bits that have not yet been encoded
 *
 * Returns the number of bytes encoded. Performs an 'encode step',
 * only encodes blocks of 45 characters to the output at a time, saves
 * left-over state in @uubuf, @state and @save (initialize to 0 on first
 * invocation).
 **/
size_t
uuencode_step (unsigned char *in, size_t len, unsigned char *out, unsigned char *uubuf, int *state, guint32 *save)
{
    register unsigned char *inptr, *outptr, *bufptr;
    unsigned char *inend;
    register guint32 saved;
    int uulen, i;
    
    saved = *save;
    i = *state & 0xff;
    uulen = (*state >> 8) & 0xff;
    
    inptr = in;
    inend = in + len;
    
    outptr = out;
    
    bufptr = uubuf + ((uulen / 3) * 4);
    
    while (inptr < inend) {
        while (uulen < 45 && inptr < inend) {
            while (i < 3 && inptr < inend) {
                saved = (saved << 8) | *inptr++;
                i++;
            }
            
            if (i == 3) {
                /* convert 3 normal bytes into 4 uuencoded bytes */
                unsigned char b0, b1, b2;
                
                b0 = saved >> 16;
                b1 = saved >> 8 & 0xff;
                b2 = saved & 0xff;
                
                *bufptr++ = CAMEL_UUENCODE_CHAR ((b0 >> 2) & 0x3f);
                *bufptr++ = CAMEL_UUENCODE_CHAR (((b0 << 4) | ((b1 >> 4) & 0xf)) & 0x3f);
                *bufptr++ = CAMEL_UUENCODE_CHAR (((b1 << 2) | ((b2 >> 6) & 0x3)) & 0x3f);
                *bufptr++ = CAMEL_UUENCODE_CHAR (b2 & 0x3f);
                
                i = 0;
                saved = 0;
                uulen += 3;
            }
        }
        
        if (uulen >= 45) {
            *outptr++ = CAMEL_UUENCODE_CHAR (uulen & 0xff);
            memcpy (outptr, uubuf, ((uulen / 3) * 4));
            outptr += ((uulen / 3) * 4);
            *outptr++ = '\n';
            uulen = 0;
            bufptr = uubuf;
        }
    }
    
    *save = saved;
    *state = ((uulen & 0xff) << 8) | (i & 0xff);
    
    return outptr - out;
}


/**
 * uudecode_step: uudecode a chunk of data
 * @in: input stream
 * @inlen: max length of data to decode ( normally strlen(in) ??)
 * @out: output stream
 * @state: holds the number of bits that are stored in @save
 * @save: leftover bits that have not yet been decoded
 *
 * Returns the number of bytes decoded. Performs a 'decode step' on
 * a chunk of uuencoded data. Assumes the "begin <mode> <file name>"
 * line has been stripped off.
 **/
size_t
uudecode_step (unsigned char *in, size_t len, unsigned char *out, int *state, guint32 *save)
{
    register unsigned char *inptr, *outptr;
    unsigned char *inend, ch;
    register guint32 saved;
    gboolean last_was_eoln;
    int uulen, i;
    
    if (*state & CAMEL_UUDECODE_STATE_END)
        return 0;
    
    saved = *save;
    i = *state & 0xff;
    uulen = (*state >> 8) & 0xff;
    if (uulen == 0)
        last_was_eoln = TRUE;
    else
        last_was_eoln = FALSE;
    
    inend = in + len;
    outptr = out;
    
    inptr = in;
    while (inptr < inend) {
        if (*inptr == '\n' || last_was_eoln) {
            if (last_was_eoln && *inptr != '\n') {
                uulen = CAMEL_UUDECODE_CHAR (*inptr);
                last_was_eoln = FALSE;
                if (uulen == 0) {
                    *state |= CAMEL_UUDECODE_STATE_END;
                    break;
                }
            } else {
                last_was_eoln = TRUE;
            }
            
            inptr++;
            continue;
        }
        
        ch = *inptr++;
        
        if (uulen > 0) {
            /* save the byte */
            saved = (saved << 8) | ch;
            i++;
            if (i == 4) {
                /* convert 4 uuencoded bytes to 3 normal bytes */
                unsigned char b0, b1, b2, b3;
                
                b0 = saved >> 24;
                b1 = saved >> 16 & 0xff;
                b2 = saved >> 8 & 0xff;
                b3 = saved & 0xff;
                
                if (uulen >= 3) {
                    *outptr++ = CAMEL_UUDECODE_CHAR (b0) << 2 | CAMEL_UUDECODE_CHAR (b1) >> 4;
                    *outptr++ = CAMEL_UUDECODE_CHAR (b1) << 4 | CAMEL_UUDECODE_CHAR (b2) >> 2;
                        *outptr++ = CAMEL_UUDECODE_CHAR (b2) << 6 | CAMEL_UUDECODE_CHAR (b3);
                } else {
                    if (uulen >= 1) {
                        *outptr++ = CAMEL_UUDECODE_CHAR (b0) << 2 | CAMEL_UUDECODE_CHAR (b1) >> 4;
                    }
                    if (uulen >= 2) {
                        *outptr++ = CAMEL_UUDECODE_CHAR (b1) << 4 | CAMEL_UUDECODE_CHAR (b2) >> 2;
                    }
                }
                
                i = 0;
                saved = 0;
                uulen -= 3;
            }
        } else {
            break;
        }
    }
    
    *save = saved;
    *state = (*state & CAMEL_UUDECODE_STATE_MASK) | ((uulen & 0xff) << 8) | (i & 0xff);
    
    return outptr - out;
}


/* complete qp encoding */
size_t
quoted_encode_close(unsigned char *in, size_t len, unsigned char *out, int *state, int *save)
{
    register unsigned char *outptr = out;
    int last;

    if (len>0)
        outptr += quoted_encode_step(in, len, outptr, state, save);

    last = *state;
    if (last != -1) {
        /* space/tab must be encoded if it's the last character on
           the line */
        if (is_qpsafe(last) && last!=' ' && last!=9) {
            *outptr++ = last;
        } else {
            *outptr++ = '=';
            *outptr++ = tohex[(last>>4) & 0xf];
            *outptr++ = tohex[last & 0xf];
        }
    }

    *save = 0;
    *state = -1;

    return outptr-out;
}

/* perform qp encoding, initialise state to -1 and save to 0 on first invocation */
size_t
quoted_encode_step (unsigned char *in, size_t len, unsigned char *out, int *statep, int *save)
{
    register guchar *inptr, *outptr, *inend;
    unsigned char c;
    register int sofar = *save;  /* keeps track of how many chars on a line */
    register int last = *statep; /* keeps track if last char to end was a space cr etc */
    
    inptr = in;
    inend = in + len;
    outptr = out;
    while (inptr < inend) {
        c = *inptr++;
        if (c == '\r') {
            if (last != -1) {
                *outptr++ = '=';
                *outptr++ = tohex[(last >> 4) & 0xf];
                *outptr++ = tohex[last & 0xf];
                sofar += 3;
            }
            last = c;
        } else if (c == '\n') {
            if (last != -1 && last != '\r') {
                *outptr++ = '=';
                *outptr++ = tohex[(last >> 4) & 0xf];
                *outptr++ = tohex[last & 0xf];
            }
            *outptr++ = '\n';
            sofar = 0;
            last = -1;
        } else {
            if (last != -1) {
                if (is_qpsafe(last)) {
                    *outptr++ = last;
                    sofar++;
                } else {
                    *outptr++ = '=';
                    *outptr++ = tohex[(last >> 4) & 0xf];
                    *outptr++ = tohex[last & 0xf];
                    sofar += 3;
                }
            }
            
            if (is_qpsafe(c)) {
                if (sofar > 74) {
                    *outptr++ = '=';
                    *outptr++ = '\n';
                    sofar = 0;
                }
                
                /* delay output of space char */
                if (c==' ' || c=='\t') {
                    last = c;
                } else {
                    *outptr++ = c;
                    sofar++;
                    last = -1;
                }
            } else {
                if (sofar > 72) {
                    *outptr++ = '=';
                    *outptr++ = '\n';
                    sofar = 3;
                } else
                    sofar += 3;
                
                *outptr++ = '=';
                *outptr++ = tohex[(c >> 4) & 0xf];
                *outptr++ = tohex[c & 0xf];
                last = -1;
            }
        }
    }
    *save = sofar;
    *statep = last;
    
    return (outptr - out);
}

/*
  FIXME: this does not strip trailing spaces from lines (as it should, rfc 2045, section 6.7)
  Should it also canonicalise the end of line to CR LF??

  Note: Trailing rubbish (at the end of input), like = or =x or =\r will be lost.
*/ 

size_t
quoted_decode_step(unsigned char *in, size_t len, unsigned char *out, int *savestate, int *saveme)
{
    register unsigned char *inptr, *outptr;
    unsigned char *inend, c;
    int state, save;

    inend = in+len;
    outptr = out;

    d(printf("quoted-printable, decoding text '%.*s'\n", len, in));

    state = *savestate;
    save = *saveme;
    inptr = in;
    while (inptr<inend) {
        switch (state) {
        case 0:
            while (inptr<inend) {
                c = *inptr++;
                if (c=='=') { 
                    state = 1;
                    break;
                }
#ifdef CANONICALISE_EOL
                /*else if (c=='\r') {
                    state = 3;
                } else if (c=='\n') {
                    *outptr++ = '\r';
                    *outptr++ = c;
                    } */
#endif
                else {
                    *outptr++ = c;
                }
            }
            break;
        case 1:
            c = *inptr++;
            if (c=='\n') {
                /* soft break ... unix end of line */
                state = 0;
            } else {
                save = c;
                state = 2;
            }
            break;
        case 2:
            c = *inptr++;
            if (isxdigit(c) && isxdigit(save)) {
                c = toupper(c);
                save = toupper(save);
                *outptr++ = (((save>='A'?save-'A'+10:save-'0')&0x0f) << 4)
                    | ((c>='A'?c-'A'+10:c-'0')&0x0f);
            } else if (c=='\n' && save == '\r') {
                /* soft break ... canonical end of line */
            } else {
                /* just output the data */
                *outptr++ = '=';
                *outptr++ = save;
                *outptr++ = c;
            }
            state = 0;
            break;
#ifdef CANONICALISE_EOL
        case 3:
            /* convert \r -> to \r\n, leaves \r\n alone */
            c = *inptr++;
            if (c=='\n') {
                *outptr++ = '\r';
                *outptr++ = c;
            } else {
                *outptr++ = '\r';
                *outptr++ = '\n';
                *outptr++ = c;
            }
            state = 0;
            break;
#endif
        }
    }

    *savestate = state;
    *saveme = save;

    return outptr-out;
}

/*
  this is for the "Q" encoding of international words,
  which is slightly different than plain quoted-printable (mainly by allowing 0x20 <> _)
*/
static int
quoted_decode(const unsigned char *in, size_t len, unsigned char *out)
{
    register const unsigned char *inptr;
    register unsigned char *outptr;
    unsigned const char *inend;
    unsigned char c, c1;
    int ret = 0;

    inend = in+len;
    outptr = out;

    d(printf("decoding text '%.*s'\n", len, in));

    inptr = in;
    while (inptr<inend) {
        c = *inptr++;
        if (c=='=') {
            /* silently ignore truncated data? */
            if (inend-in>=2) {
                c = toupper(*inptr++);
                c1 = toupper(*inptr++);
                *outptr++ = (((c>='A'?c-'A'+10:c-'0')&0x0f) << 4)
                    | ((c1>='A'?c1-'A'+10:c1-'0')&0x0f);
            } else {
                ret = -1;
                break;
            }
        } else if (c=='_') {
            *outptr++ = 0x20;
        } else if (c==' ' || c==0x09) {
            /* FIXME: this is an error! ignore for now ... */
            ret = -1;
            break;
        } else {
            *outptr++ = c;
        }
    }
    if (ret==0) {
        return outptr-out;
    }
    return -1;
}

/* rfc2047 version of quoted-printable */
/* safemask is the mask to apply to the camel_mime_special_table to determine what
   characters can safely be included without encoding */
static int
quoted_encode (const unsigned char *in, size_t len, unsigned char *out, unsigned short safemask)
{
    register const unsigned char *inptr, *inend;
    unsigned char *outptr;
    unsigned char c;

    inptr = in;
    inend = in + len;
    outptr = out;
    while (inptr < inend) {
        c = *inptr++;
        if (c==' ') {
            *outptr++ = '_';
        } else if (camel_mime_special_table[c] & safemask) {
            *outptr++ = c;
        } else {
            *outptr++ = '=';
            *outptr++ = tohex[(c >> 4) & 0xf];
            *outptr++ = tohex[c & 0xf];
        }
    }

    d(printf("encoding '%.*s' = '%.*s'\n", len, in, outptr-out, out));

    return (outptr - out);
}


static void
header_decode_lwsp(const char **in)
{
    const char *inptr = *in;
    char c;

    d2(printf("is ws: '%s'\n", *in));

    while (is_lwsp(*inptr) || (*inptr =='(' && *inptr != '\0')) {
        while (is_lwsp(*inptr) && inptr != '\0') {
            d2(printf("(%c)", *inptr));
            inptr++;
        }
        d2(printf("\n"));

        /* check for comments */
        if (*inptr == '(') {
            int depth = 1;
            inptr++;
            while (depth && (c=*inptr) && *inptr != '\0') {
                if (c=='\\' && inptr[1]) {
                    inptr++;
                } else if (c=='(') {
                    depth++;
                } else if (c==')') {
                    depth--;
                }
                inptr++;
            }
        }
    }
    *in = inptr;
}

/* decode rfc 2047 encoded string segment */
static char *
rfc2047_decode_word(const char *in, size_t len)
{
    const char *inptr = in+2;
    const char *inend = in+len-2;
    const char *inbuf;
    const char *charset;
    char *encname, *p;
    int tmplen;
    size_t ret;
    char *decword = NULL;
    char *decoded = NULL;
    char *outbase = NULL;
    char *outbuf;
    size_t inlen, outlen;
    gboolean retried = FALSE;
    iconv_t ic;
    
    d(printf("rfc2047: decoding '%.*s'\n", len, in));

    /* quick check to see if this could possibly be a real encoded word */
    if (len < 8 || !(in[0] == '=' && in[1] == '?' && in[len-1] == '=' && in[len-2] == '?')) {
        d(printf("invalid\n"));
        return NULL;
    }
    
    /* skip past the charset to the encoding type */
    inptr = memchr (inptr, '?', inend-inptr);
    if (inptr != NULL && inptr < inend + 2 && inptr[2] == '?') {
        d(printf("found ?, encoding is '%c'\n", inptr[0]));
        inptr++;
        tmplen = inend-inptr-2;
        decword = alloca(tmplen); /* this will always be more-than-enough room */
        switch(toupper(inptr[0])) {
        case 'Q':
            inlen = quoted_decode(inptr+2, tmplen, decword);
            break;
        case 'B': {
            int state = 0;
            unsigned int save = 0;
            
            inlen = base64_decode_step((char *)inptr+2, tmplen, decword, &state, &save);
            /* if state != 0 then error? */
            break;
        }
        default:
            /* uhhh, unknown encoding type - probably an invalid encoded word string */
            return NULL;
        }
        d(printf("The encoded length = %d\n", inlen));
        if (inlen > 0) {
            /* yuck, all this snot is to setup iconv! */
            tmplen = inptr - in - 3;
            encname = alloca (tmplen + 1);
            memcpy (encname, in + 2, tmplen);
            encname[tmplen] = '\0';
            
            /* rfc2231 updates rfc2047 encoded words...
             * The ABNF given in RFC 2047 for encoded-words is:
             *   encoded-word := "=?" charset "?" encoding "?" encoded-text "?="
             * This specification changes this ABNF to:
             *   encoded-word := "=?" charset ["*" language] "?" encoding "?" encoded-text "?="
             */
            
            /* trim off the 'language' part if it's there... */
            p = strchr (encname, '*');
            if (p)
                *p = '\0';
            
            charset = e_iconv_charset_name (encname);
            
            inbuf = decword;
            
            outlen = inlen * 6 + 16;
            outbase = alloca (outlen);
            outbuf = outbase;
            
        retry:
            ic = e_iconv_open ("UTF-8", charset);
            if (ic != (iconv_t) -1) {
                ret = e_iconv (ic, &inbuf, &inlen, &outbuf, &outlen);
                if (ret != (size_t) -1) {
                    e_iconv (ic, NULL, 0, &outbuf, &outlen);
                    *outbuf = 0;
                    decoded = g_strdup (outbase);
                }
                e_iconv_close (ic);
            } else {
                w(g_warning ("Cannot decode charset, header display may be corrupt: %s: %s",
                         charset, g_strerror (errno)));
                
                if (!retried) {
                    charset = e_iconv_locale_charset ();
                    if (!charset)
                        charset = "iso-8859-1";
                    
                    retried = TRUE;
                    goto retry;
                }
                
                /* we return the encoded word here because we've got to return valid utf8 */
                decoded = g_strndup (in, inlen);
            }
        }
    }
    
    d(printf("decoded '%s'\n", decoded));
    
    return decoded;
}

/* grrr, glib should have this ! */
static GString *
g_string_append_len(GString *st, const char *s, size_t l)
{
    char *tmp;

    tmp = alloca(l+1);
    tmp[l]=0;
    memcpy(tmp, s, l);
    return g_string_append(st, tmp);
}

/* ok, a lot of mailers are BROKEN, and send iso-latin1 encoded
   headers, when they should just be sticking to US-ASCII
   according to the rfc's.  Anyway, since the conversion to utf-8
   is trivial, just do it here without iconv */
static GString *
append_latin1 (GString *out, const char *in, size_t len)
{
    unsigned int c;
    
    while (len) {
        c = (unsigned int)*in++;
        len--;
        if (c & 0x80) {
            out = g_string_append_c (out, 0xc0 | ((c >> 6) & 0x3));  /* 110000xx */
            out = g_string_append_c (out, 0x80 | (c & 0x3f));        /* 10xxxxxx */
        } else {
            out = g_string_append_c (out, c);
        }
    }
    return out;
}

static int
append_8bit (GString *out, const char *inbuf, size_t inlen, const char *charset)
{
    char *outbase, *outbuf;
    size_t outlen;
    iconv_t ic;
    
    ic = e_iconv_open ("UTF-8", charset);
    if (ic == (iconv_t) -1)
        return FALSE;

    outlen = inlen * 6 + 16;
    outbuf = outbase = g_malloc(outlen);
    
    if (e_iconv(ic, &inbuf, &inlen, &outbuf, &outlen) == (size_t) -1) {
        w(g_warning("Conversion to '%s' failed: %s", charset, strerror(errno)));
        g_free(outbase);
        e_iconv_close(ic);
        return FALSE;
    }

    *outbuf = 0;
    g_string_append(out, outbase);
    g_free(outbase);
    e_iconv_close(ic);

    return TRUE;
    
}

/* decodes a simple text, rfc822 + rfc2047 */
static char *
header_decode_text (const char *in, size_t inlen, const char *default_charset)
{
    GString *out;
    const char *inptr, *inend, *start, *chunk, *locale_charset;
    char *dword = NULL;

    locale_charset = e_iconv_locale_charset();

    out = g_string_new("");
    inptr = in;
    inend = inptr + inlen;
    chunk = NULL;

    while (inptr < inend) {
        start = inptr;
        while (inptr < inend && is_lwsp(*inptr))
            inptr++;

        if (inptr == inend) {
            g_string_append_len(out, start, inptr-start);
            break;
        } else if (dword == NULL) {
            g_string_append_len(out, start, inptr-start);
        } else {
            chunk = start;
        }

        start = inptr;
        while (inptr < inend && !is_lwsp(*inptr))
            inptr++;

        dword = rfc2047_decode_word(start, inptr-start);
        if (dword) {
            g_string_append(out, dword);
            g_free(dword);
        } else {
            if (!chunk)
                chunk = start;
            
            if ((default_charset == NULL || !append_8bit (out, chunk, inptr-chunk, default_charset))
                && (locale_charset == NULL || !append_8bit(out, chunk, inptr-chunk, locale_charset)))
                append_latin1(out, chunk, inptr-chunk);
        }
        
        chunk = NULL;
    }

    dword = out->str;
    g_string_free (out, FALSE);
    
    return dword;
}

char *
header_decode_string (const char *in, const char *default_charset)
{
    if (in == NULL)
        return NULL;
    return header_decode_text (in, strlen (in), default_charset);
}

/* how long a sequence of pre-encoded words should be less than, to attempt to 
   fit into a properly folded word.  Only a guide. */
#define CAMEL_FOLD_PREENCODED (24)

/* FIXME: needs a way to cache iconv opens for different charsets? */
static void
rfc2047_encode_word(GString *outstring, const char *in, size_t len, const char *type, unsigned short safemask)
{
    iconv_t ic = (iconv_t *)-1;
    char *buffer, *out, *ascii;
    size_t inlen, outlen, enclen, bufflen;
    const char *inptr, *p;
    int first = 1;

    d(printf("Converting [%d] '%.*s' to %s\n", len, len, in, type));

    /* convert utf8->encoding */
    bufflen = len * 6 + 16;
    buffer = alloca (bufflen);
    inlen = len;
    inptr = in;
    
    ascii = alloca (bufflen);
    
    if (strcasecmp (type, "UTF-8") != 0)
        ic = e_iconv_open (type, "UTF-8");
    
    while (inlen) {
        size_t convlen, proclen;
        int i;
        
        /* break up words into smaller bits, what we really want is encoded + overhead < 75,
           but we'll just guess what that means in terms of input chars, and assume its good enough */

        out = buffer;
        outlen = bufflen;

        if (ic == (iconv_t) -1) {
            /* native encoding case, the easy one (?) */
            /* we work out how much we can convert, and still be in length */
            /* proclen will be the result of input characters that we can convert, to the nearest
               (approximated) valid utf8 char */
            convlen = 0;
            proclen = 0;
            p = inptr;
            i = 0;
            while (p < (in+len) && convlen < (75 - strlen("=?utf-8?q\?\?="))) {
                unsigned char c = *p++;

                if (c >= 0xc0)
                    proclen = i;
                i++;
                if (c < 0x80)
                    proclen = i;
                if (camel_mime_special_table[c] & safemask)
                    convlen += 1;
                else
                    convlen += 3;
            }
            /* well, we probably have broken utf8, just copy it anyway what the heck */
            if (proclen == 0) {
                w(g_warning("Appear to have truncated utf8 sequence"));
                proclen = inlen;
            }
            memcpy(out, inptr, proclen);
            inptr += proclen;
            inlen -= proclen;
            out += proclen;
        } else {
            /* well we could do similar, but we can't (without undue effort), we'll just break it up into
               hopefully-small-enough chunks, and leave it at that */
            convlen = MIN(inlen, CAMEL_FOLD_PREENCODED);
            p = inptr;
            if (e_iconv (ic, &inptr, &convlen, &out, &outlen) == (size_t) -1) {
                w(g_warning("Conversion problem: conversion truncated: %s", strerror(errno)));
                /* blah, we include it anyway, better than infinite loop ... */
                inptr = p + convlen;
            } else {
                /* make sure we flush out any shift state */
                e_iconv(ic, NULL, 0, &out, &outlen);
            }
            inlen -= (inptr - p);
        }
        
        enclen = out-buffer;
        
        if (enclen) {
            /* create token */
            out = ascii;
            if (first)
                first = 0;
            else
                *out++ = ' ';
            out += sprintf (out, "=?%s?Q?", type);
            out += quoted_encode (buffer, enclen, out, safemask);
            sprintf (out, "?=");
            
            d(printf("converted part = %s\n", ascii));
            
            g_string_append (outstring, ascii);
        }
    }

    if (ic != (iconv_t) -1)
        e_iconv_close(ic);
}


/* TODO: Should this worry about quotes?? */
char *
header_encode_string (const unsigned char *in)
{
    const unsigned char *inptr = in, *start, *word;
    gboolean last_was_encoded = FALSE;
    gboolean last_was_space = FALSE;
    int encoding;
    GString *out;
    char *outstr;

    g_return_val_if_fail (g_utf8_validate (in, -1, NULL), NULL);
    
    if (in == NULL)
        return NULL;
    
    /* do a quick us-ascii check (the common case?) */
    while (*inptr) {
        if (*inptr > 127)
            break;
        inptr++;
    }
    if (*inptr == '\0')
        return g_strdup (in);
    
    /* This gets each word out of the input, and checks to see what charset
       can be used to encode it. */
    /* TODO: Work out when to merge subsequent words, or across word-parts */
    out = g_string_new ("");
    inptr = in;
    encoding = 0;
    word = NULL;
    start = inptr;
    while (inptr && *inptr) {
        gunichar c;
        const char *newinptr;
        
        newinptr = g_utf8_next_char (inptr);
        c = g_utf8_get_char (inptr);
        if (newinptr == NULL || !g_unichar_validate (c)) {
            w(g_warning ("Invalid UTF-8 sequence encountered (pos %d, char '%c'): %s",
                     (inptr-in), inptr[0], in));
            inptr++;
            continue;
        }

        
        if (g_unichar_isspace (c) && !last_was_space) {
            /* we've reached the end of a 'word' */
            if (word && !(last_was_encoded && encoding)) {
                g_string_append_len (out, start, word - start);
                start = word;
            }
            
            switch (encoding) {
            case 0:
                out = g_string_append_len (out, word, inptr - start);
                last_was_encoded = FALSE;
                break;
            case 1:
                if (last_was_encoded)
                    g_string_append_c (out, ' ');
                
                rfc2047_encode_word (out, start, inptr - start, "ISO-8859-1", IS_ESAFE);
                last_was_encoded = TRUE;
                break;
            case 2:
                if (last_was_encoded)
                    g_string_append_c (out, ' ');
                
                rfc2047_encode_word (out, start, inptr - start,
                             camel_charset_best (start, inptr - start), IS_ESAFE);
                last_was_encoded = TRUE;
                break;
            }
            
            last_was_space = TRUE;
            start = inptr;
            word = NULL;
            encoding = 0;
        } else if (c > 127 && c < 256) {
            encoding = MAX (encoding, 1);
            last_was_space = FALSE;
        } else if (c >= 256) {
            encoding = MAX (encoding, 2);
            last_was_space = FALSE;
        } else if (!g_unichar_isspace (c)) {
            last_was_space = FALSE;
        }
        
        if (!g_unichar_isspace (c) && !word)
            word = inptr;
        
        inptr = newinptr;
    }
    
    if (inptr - start) {
        if (word && !(last_was_encoded && encoding)) {
            g_string_append_len (out, start, word - start);
            start = word;
        }
        
        switch (encoding) {
        case 0:
            out = g_string_append_len (out, start, inptr - start);
            break;
        case 1:
            if (last_was_encoded)
                g_string_append_c (out, ' ');
            
            rfc2047_encode_word (out, start, inptr - start, "ISO-8859-1", IS_ESAFE);
            break;
        case 2:
            if (last_was_encoded)
                g_string_append_c (out, ' ');
            
            rfc2047_encode_word (out, start, inptr - start,
                         camel_charset_best (start, inptr - start - 1), IS_ESAFE);
            break;
        }
    }
    
    outstr = out->str;
    g_string_free (out, FALSE);
    
    return outstr;
}

/* apply quoted-string rules to a string */
static void
quote_word(GString *out, gboolean do_quotes, const char *start, size_t len)
{
    int i, c;

    /* TODO: What about folding on long lines? */
    if (do_quotes)
        g_string_append_c(out, '"');
    for (i=0;i<len;i++) {
        c = *start++;
        if (c == '\"' || c=='\\' || c=='\r')
            g_string_append_c(out, '\\');
        g_string_append_c(out, c);
    }
    if (do_quotes)
        g_string_append_c(out, '"');
}

/* incrementing possibility for the word type */
enum _phrase_word_t {
    WORD_ATOM,
    WORD_QSTRING,
    WORD_2047
};

struct _phrase_word {
    const unsigned char *start, *end;
    enum _phrase_word_t type;
    int encoding;
};

static gboolean
word_types_compatable (enum _phrase_word_t type1, enum _phrase_word_t type2)
{
    switch (type1) {
    case WORD_ATOM:
        return type2 == WORD_QSTRING;
    case WORD_QSTRING:
        return type2 != WORD_2047;
    case WORD_2047:
        return type2 == WORD_2047;
    default:
        return FALSE;
    }
}

/* split the input into words with info about each word
 * merge common word types clean up */
static GList *
header_encode_phrase_get_words (const unsigned char *in)
{
    const unsigned char *inptr = in, *start, *last;
    struct _phrase_word *word;
    enum _phrase_word_t type;
    int encoding, count = 0;
    GList *words = NULL;
    
    /* break the input into words */
    type = WORD_ATOM;
    last = inptr;
    start = inptr;
    encoding = 0;
    while (inptr && *inptr) {
        gunichar c;
        const char *newinptr;
        
        newinptr = g_utf8_next_char (inptr);
        c = g_utf8_get_char (inptr);
        
        if (!g_unichar_validate (c)) {
            w(g_warning ("Invalid UTF-8 sequence encountered (pos %d, char '%c'): %s",
                     (inptr - in), inptr[0], in));
            inptr++;
            continue;
        }
        
        inptr = newinptr;
        if (g_unichar_isspace (c)) {
            if (count > 0) {
                word = g_new0 (struct _phrase_word, 1);
                word->start = start;
                word->end = last;
                word->type = type;
                word->encoding = encoding;
                words = g_list_append (words, word);
                count = 0;
            }
            
            start = inptr;
            type = WORD_ATOM;
            encoding = 0;
        } else {
            count++;
            if (c < 128) {
                if (!is_atom (c))
                    type = MAX (type, WORD_QSTRING);
            } else if (c > 127 && c < 256) {
                type = WORD_2047;
                encoding = MAX (encoding, 1);
            } else if (c >= 256) {
                type = WORD_2047;
                encoding = MAX (encoding, 2);
            }
        }
        
        last = inptr;
    }
    
    if (count > 0) {
        word = g_new0 (struct _phrase_word, 1);
        word->start = start;
        word->end = last;
        word->type = type;
        word->encoding = encoding;
        words = g_list_append (words, word);
    }
    
    return words;
}

static gboolean
header_encode_phrase_merge_words (GList **wordsp)
{
    GList *wordl, *nextl, *words = *wordsp;
    struct _phrase_word *word, *next;
    gboolean merged = FALSE;
    
    /* scan the list, checking for words of similar types that can be merged */
    wordl = words;
    while (wordl) {
        word = wordl->data;
        nextl = g_list_next (wordl);
        
        while (nextl) {
            next = nextl->data;
            /* merge nodes of the same type AND we are not creating too long a string */
            if (word_types_compatable (word->type, next->type)) {
                if (next->end - word->start < CAMEL_FOLD_PREENCODED) {
                    /* the resulting word type is the MAX of the 2 types */
                    word->type = MAX(word->type, next->type);
                    
                    word->end = next->end;
                    words = g_list_remove_link (words, nextl);
                    g_free (next);
                    nextl = g_list_next (wordl);
                    
                    merged = TRUE;
                } else {
                    /* if it is going to be too long, make sure we include the
                       separating whitespace */
                    word->end = next->start;
                    break;
                }
            } else {
                break;
            }
        }
        
        wordl = g_list_next (wordl);
    }
    
    *wordsp = words;
    
    return merged;
}

/* encodes a phrase sequence (different quoting/encoding rules to strings) */
char *
header_encode_phrase (const unsigned char *in)
{
    struct _phrase_word *word = NULL, *last_word = NULL;
    GList *words, *wordl;
    GString *out;
    char *outstr;
    
    if (in == NULL)
        return NULL;
    
    words = header_encode_phrase_get_words (in);
    if (!words)
        return NULL;
    
    while (header_encode_phrase_merge_words (&words));
    
    out = g_string_new ("");
    
    /* output words now with spaces between them */
    wordl = words;
    while (wordl) {
        const char *start;
        size_t len;
        
        word = wordl->data;
        
        /* append correct number of spaces between words */
        if (last_word && !(last_word->type == WORD_2047 && word->type == WORD_2047)) {
            /* one or both of the words are not encoded so we write the spaces out untouched */
            len = word->start - last_word->end;
            out = g_string_append_len (out, last_word->end, len);
        }
        
        switch (word->type) {
        case WORD_ATOM:
            out = g_string_append_len (out, word->start, word->end - word->start);
            break;
        case WORD_QSTRING:
            quote_word (out, TRUE, word->start, word->end - word->start);
            break;
        case WORD_2047:
            if (last_word && last_word->type == WORD_2047) {
                /* include the whitespace chars between these 2 words in the
                                   resulting rfc2047 encoded word. */
                len = word->end - last_word->end;
                start = last_word->end;
                
                /* encoded words need to be separated by linear whitespace */
                g_string_append_c (out, ' ');
            } else {
                len = word->end - word->start;
                start = word->start;
            }
            
            if (word->encoding == 1)
                rfc2047_encode_word (out, start, len, "ISO-8859-1", IS_PSAFE);
            else
                rfc2047_encode_word (out, start, len,
                             camel_charset_best (start, len), IS_PSAFE);
            break;
        }
        
        g_free (last_word);
        wordl = g_list_next (wordl);
        
        last_word = word;
    }
    
    /* and we no longer need the list */
    g_free (word);
    g_list_free (words);
    
    outstr = out->str;
    g_string_free (out, FALSE);
    
    return outstr;
}


/* these are all internal parser functions */

static char *
decode_token (const char **in)
{
    const char *inptr = *in;
    const char *start;
    
    header_decode_lwsp (&inptr);
    start = inptr;
    while (is_ttoken (*inptr))
        inptr++;
    if (inptr > start) {
        *in = inptr;
        return g_strndup (start, inptr - start);
    } else {
        return NULL;
    }
}

char *
header_token_decode(const char *in)
{
    if (in == NULL)
        return NULL;

    return decode_token(&in);
}

/*
   <"> * ( <any char except <"> \, cr  /  \ <any char> ) <">
*/
static char *
header_decode_quoted_string(const char **in)
{
    const char *inptr = *in;
    char *out = NULL, *outptr;
    size_t outlen;
    int c;

    header_decode_lwsp(&inptr);
    if (*inptr == '"') {
        const char *intmp;
        int skip = 0;

        /* first, calc length */
        inptr++;
        intmp = inptr;
        while ( (c = *intmp++) && c!= '"') {
            if (c=='\\' && *intmp) {
                intmp++;
                skip++;
            }
        }
        outlen = intmp-inptr-skip;
        out = outptr = g_malloc(outlen+1);
        while ( (c = *inptr++) && c!= '"') {
            if (c=='\\' && *inptr) {
                c = *inptr++;
            }
            *outptr++ = c;
        }
        *outptr = '\0';
    }
    *in = inptr;
    return out;
}

static char *
header_decode_atom(const char **in)
{
    const char *inptr = *in, *start;

    header_decode_lwsp(&inptr);
    start = inptr;
    while (is_atom(*inptr))
        inptr++;
    *in = inptr;
    if (inptr > start)
        return g_strndup(start, inptr-start);
    else
        return NULL;
}

static char *
header_decode_word(const char **in)
{
    const char *inptr = *in;

    header_decode_lwsp(&inptr);
    if (*inptr == '"') {
        *in = inptr;
        return header_decode_quoted_string(in);
    } else {
        *in = inptr;
        return header_decode_atom(in);
    }
}

static char *
header_decode_value(const char **in)
{
    const char *inptr = *in;

    header_decode_lwsp(&inptr);
    if (*inptr == '"') {
        d(printf("decoding quoted string\n"));
        return header_decode_quoted_string(in);
    } else if (is_ttoken(*inptr)) {
        d(printf("decoding token\n"));
        /* this may not have the right specials for all params? */
        return decode_token(in);
    }
    return NULL;
}

/* should this return -1 for no int? */
int
header_decode_int(const char **in)
{
    const char *inptr = *in;
    int c, v=0;

    header_decode_lwsp(&inptr);
    while ( (c=*inptr++ & 0xff)
        && isdigit(c) ) {
        v = v*10+(c-'0');
    }
    *in = inptr-1;
    return v;
}

#define HEXVAL(c) (isdigit (c) ? (c) - '0' : tolower (c) - 'a' + 10)

static char *
hex_decode (const char *in, size_t len)
{
    const unsigned char *inend = in + len;
    unsigned char *inptr, *outptr;
    char *outbuf;
    
    outptr = outbuf = g_malloc (len + 1);
    
    inptr = (unsigned char *) in;
    while (inptr < inend) {
        if (*inptr == '%') {
            if (isxdigit (inptr[1]) && isxdigit (inptr[2])) {
                *outptr++ = HEXVAL (inptr[1]) * 16 + HEXVAL (inptr[2]);
                inptr += 3;
            } else
                *outptr++ = *inptr++;
        } else
            *outptr++ = *inptr++;
    }
    
    *outptr = '\0';
    
    return outbuf;
}

/* an rfc2184 encoded string looks something like:
 * us-ascii'en'This%20is%20even%20more%20
 */

static char *
rfc2184_decode (const char *in, size_t len)
{
    const char *inptr = in;
    const char *inend = in + len;
    const char *charset;
    char *decoded = NULL;
    char *encoding;
    
    inptr = memchr (inptr, '\'', len);
    if (!inptr)
        return NULL;
    
    encoding = g_strndup (in, inptr - in);
    charset = e_iconv_charset_name (encoding);
    g_free (encoding);
    
    inptr = memchr (inptr + 1, '\'', inend - inptr - 1);
    if (!inptr)
        return NULL;
    
    inptr++;
    if (inptr < inend) {
        char *decword, *outbase, *outbuf;
        const char *inbuf;
        size_t inlen, outlen;
        iconv_t ic;
        
        inbuf = decword = hex_decode (inptr, inend - inptr);
        inlen = strlen (inbuf);
        
        ic = e_iconv_open ("UTF-8", charset);
        if (ic != (iconv_t) -1) {
            size_t ret;
            
            outlen = inlen * 6 + 16;
            outbuf = outbase = g_malloc (outlen);
            
            ret = e_iconv (ic, &inbuf, &inlen, &outbuf, &outlen);
            if (ret != (size_t) -1) {
                e_iconv (ic, NULL, 0, &outbuf, &outlen);
                *outbuf = '\0';
                g_free (decoded);
                decoded = outbase;
            }
            
            e_iconv_close (ic);
        } else {
            decoded = decword;
        }
    }
    
    return decoded;
}

/* This function is basically the same as decode_token()
 * except that it will not accept *'s which have a special
 * meaning for rfc2184 params */
static char *
decode_param_token (const char **in)
{
    const char *inptr = *in;
    const char *start;
    
    header_decode_lwsp (&inptr);
    start = inptr;
    while (is_ttoken (*inptr) && *inptr != '*')
        inptr++;
    if (inptr > start) {
        *in = inptr;
        return g_strndup (start, inptr - start);
    } else {
        return NULL;
    }
}

static gboolean
header_decode_rfc2184_param (const char **in, char **paramp, gboolean *value_is_encoded, int *part)
{
    gboolean is_rfc2184 = FALSE;
    const char *inptr = *in;
    char *param;
    
    *value_is_encoded = FALSE;
    *part = -1;
    
    param = decode_param_token (&inptr);
    header_decode_lwsp (&inptr);
    
    if (*inptr == '*') {
        is_rfc2184 = TRUE;
        inptr++;
        header_decode_lwsp (&inptr);
        if (*inptr == '=') {
            /* form := param*=value */
            if (value_is_encoded)
                *value_is_encoded = TRUE;
        } else {
            /* form := param*#=value or param*#*=value */
            *part = header_decode_int (&inptr);
            header_decode_lwsp (&inptr);
            if (*inptr == '*') {
                /* form := param*#*=value */
                if (value_is_encoded)
                    *value_is_encoded = TRUE;
                inptr++;
                header_decode_lwsp (&inptr);
            }
        }
    }
    
    if (paramp)
        *paramp = param;
    
    if (param)
        *in = inptr;
    
    return is_rfc2184;
}

static int
header_decode_param (const char **in, char **paramp, char **valuep, int *is_rfc2184_param, int *rfc2184_part)
{
    gboolean is_rfc2184_encoded = FALSE;
    gboolean is_rfc2184 = FALSE;
    const char *inptr = *in;
    char *param, *value = NULL;
    
    *is_rfc2184_param = FALSE;
    *rfc2184_part = -1;
    
    is_rfc2184 = header_decode_rfc2184_param (&inptr, &param, &is_rfc2184_encoded, rfc2184_part);
    
    if (*inptr == '=') {
        inptr++;
        value = header_decode_value (&inptr);
        
        if (is_rfc2184) {
            /* We have ourselves an rfc2184 parameter */
            
            if (*rfc2184_part == -1) {
                /* rfc2184 allows the value to be broken into
                 * multiple parts - this isn't one of them so
                 * it is safe to decode it.
                 */
                char *val;
                
                val = rfc2184_decode (value, strlen (value));
                if (val) {
                    g_free (value);
                    value = val;
                }
            } else {
                /* Since we are expecting to find the rest of
                 * this paramter value later, let our caller know.
                 */
                *is_rfc2184_param = TRUE;
            }
        } else if (value && !strncmp (value, "=?", 2)) {
            /* We have a broken param value that is rfc2047 encoded.
             * Since both Outlook and Netscape/Mozilla do this, we
             * should handle this case.
             */
            char *val;
            
            val = rfc2047_decode_word (value, strlen (value));
            if (val) {
                g_free (value);
                value = val;
            }
        }
    }
    
    if (value && !g_utf8_validate (value, -1, NULL)) {
        /* The (broken) mailer sent us an unencoded 8bit value
         * attempt to save it by assuming it's in the user's
         * locale and converting to utf8 */
        char *outbase, *outbuf, *p;
        const char *charset, *inbuf;
        size_t inlen, outlen;
        iconv_t ic;
        
        inbuf = value;
        inlen = strlen (inbuf);
        
        charset = e_iconv_locale_charset ();
        ic = e_iconv_open ("UTF-8", charset ? charset : "ISO-8859-1");
        if (ic != (iconv_t) -1) {
            size_t ret;
            
            outlen = inlen * 6 + 16;
            outbuf = outbase = g_malloc (outlen);
            
            ret = e_iconv (ic, &inbuf, &inlen, &outbuf, &outlen);
            if (ret != (size_t) -1) {
                e_iconv (ic, NULL, 0, &outbuf, &outlen);
                *outbuf = '\0';
            }
            
            e_iconv_close (ic);
            
            g_free (value);
            value = outbase;
        } else {
            /* Okay, so now what? I guess we convert invalid chars to _'s? */
            for (p = value; *p; p++)
                if (!isascii ((unsigned) *p))
                    *p = '_';
        }
    }
    
    if (param && value) {
        *paramp = param;
        *valuep = value;
        *in = inptr;
        return 0;
    } else {
        g_free (param);
        g_free (value);
        return 1;
    }
}

char *
header_param (struct _header_param *p, const char *name)
{
    while (p && strcasecmp (p->name, name) != 0)
        p = p->next;
    if (p)
        return p->value;
    return NULL;
}

struct _header_param *
header_set_param (struct _header_param **l, const char *name, const char *value)
{
    struct _header_param *p = (struct _header_param *)l, *pn;
    
    if (name == NULL)
        return NULL;
    
    while (p->next) {
        pn = p->next;
        if (!strcasecmp (pn->name, name)) {
            g_free (pn->value);
            if (value) {
                pn->value = g_strdup (value);
                return pn;
            } else {
                p->next = pn->next;
                g_free (pn->name);
                g_free (pn);
                return NULL;
            }
        }
        p = pn;
    }

    if (value == NULL)
        return NULL;

    pn = g_malloc (sizeof (*pn));
    pn->next = 0;
    pn->name = g_strdup (name);
    pn->value = g_strdup (value);
    p->next = pn;

    return pn;
}

const char *
header_content_type_param (struct _header_content_type *t, const char *name)
{
    if (t==NULL)
        return NULL;
    return header_param (t->params, name);
}

void
header_content_type_set_param (struct _header_content_type *t, const char *name, const char *value)
{
    header_set_param (&t->params, name, value);
}

/**
 * header_content_type_is:
 * @ct: A content type specifier, or #NULL.
 * @type: A type to check against.
 * @subtype: A subtype to check against, or "*" to match any subtype.
 * 
 * Returns #TRUE if the content type @ct is of type @type/@subtype.
 * The subtype of "*" will match any subtype.  If @ct is #NULL, then
 * it will match the type "text/plain".
 * 
 * Return value: #TRUE or #FALSE depending on the matching of the type.
 **/
int
header_content_type_is(struct _header_content_type *ct, const char *type, const char *subtype)
{
    /* no type == text/plain or text/"*" */
    if (ct==NULL || (ct->type == NULL && ct->subtype == NULL)) {
        return (!strcasecmp(type, "text")
            && (!strcasecmp(subtype, "plain")
                || !strcasecmp(subtype, "*")));
    }

    return (ct->type != NULL
        && (!strcasecmp(ct->type, type)
            && ((ct->subtype != NULL
             && !strcasecmp(ct->subtype, subtype))
            || !strcasecmp("*", subtype))));
}

void
header_param_list_free(struct _header_param *p)
{
    struct _header_param *n;

    while (p) {
        n = p->next;
        g_free(p->name);
        g_free(p->value);
        g_free(p);
        p = n;
    }
}

struct _header_content_type *
header_content_type_new(const char *type, const char *subtype)
{
    struct _header_content_type *t = g_malloc(sizeof(*t));

    t->type = g_strdup(type);
    t->subtype = g_strdup(subtype);
    t->params = NULL;
    t->refcount = 1;
    return t;
}

void
header_content_type_ref(struct _header_content_type *ct)
{
    if (ct)
        ct->refcount++;
}


void
header_content_type_unref(struct _header_content_type *ct)
{
    if (ct) {
        if (ct->refcount <= 1) {
            header_param_list_free(ct->params);
            g_free(ct->type);
            g_free(ct->subtype);
            g_free(ct);
        } else {
            ct->refcount--;
        }
    }
}

/* for decoding email addresses, canonically */
static char *
header_decode_domain(const char **in)
{
    const char *inptr = *in, *start;
    int go = TRUE;
    char *ret;
    GString *domain = g_string_new("");

                /* domain ref | domain literal */
    header_decode_lwsp(&inptr);
    while (go) {
        if (*inptr == '[') { /* domain literal */
            domain = g_string_append_c(domain, '[');
            inptr++;
            header_decode_lwsp(&inptr);
            start = inptr;
            while (is_dtext(*inptr)) {
                domain = g_string_append_c(domain, *inptr);
                inptr++;
            }
            if (*inptr == ']') {
                domain = g_string_append_c(domain, ']');
                inptr++;
            } else {
                w(g_warning("closing ']' not found in domain: %s", *in));
            }
        } else {
            char *a = header_decode_atom(&inptr);
            if (a) {
                domain = g_string_append(domain, a);
                g_free(a);
            } else {
                w(g_warning("missing atom from domain-ref"));
                break;
            }
        }
        header_decode_lwsp(&inptr);
        if (*inptr == '.') { /* next sub-domain? */
            domain = g_string_append_c(domain, '.');
            inptr++;
            header_decode_lwsp(&inptr);
        } else
            go = FALSE;
    }

    *in = inptr;

    ret = domain->str;
    g_string_free(domain, FALSE);
    return ret;
}

static char *
header_decode_addrspec(const char **in)
{
    const char *inptr = *in;
    char *word;
    GString *addr = g_string_new("");

    header_decode_lwsp(&inptr);

    /* addr-spec */
    word = header_decode_word(&inptr);
    if (word) {
        addr = g_string_append(addr, word);
        header_decode_lwsp(&inptr);
        g_free(word);
        while (*inptr == '.' && word) {
            inptr++;
            addr = g_string_append_c(addr, '.');
            word = header_decode_word(&inptr);
            if (word) {
                addr = g_string_append(addr, word);
                header_decode_lwsp(&inptr);
                g_free(word);
            } else {
                w(g_warning("Invalid address spec: %s", *in));
            }
        }
        if (*inptr == '@') {
            inptr++;
            addr = g_string_append_c(addr, '@');
            word = header_decode_domain(&inptr);
            if (word) {
                addr = g_string_append(addr, word);
                g_free(word);
            } else {
                w(g_warning("Invalid address, missing domain: %s", *in));
            }
        } else {
            w(g_warning("Invalid addr-spec, missing @: %s", *in));
        }
    } else {
        w(g_warning("invalid addr-spec, no local part"));
    }

    /* FIXME: return null on error? */

    *in = inptr;
    word = addr->str;
    g_string_free(addr, FALSE);
    return word;
}

/*
  address:
   word *('.' word) @ domain |
   *(word) '<' [ *('@' domain ) ':' ] word *( '.' word) @ domain |

   1*word ':' [ word ... etc (mailbox, as above) ] ';'
 */

/* mailbox:
   word *( '.' word ) '@' domain
   *(word) '<' [ *('@' domain ) ':' ] word *( '.' word) @ domain
   */

static struct _header_address *
header_decode_mailbox(const char **in)
{
    const char *inptr = *in;
    char *pre;
    int closeme = FALSE;
    GString *addr;
    GString *name = NULL;
    struct _header_address *address = NULL;
    const char *comment = NULL;

    addr = g_string_new("");

    /* for each address */
    pre = header_decode_word(&inptr);
    header_decode_lwsp(&inptr);
    if (!(*inptr == '.' || *inptr == '@' || *inptr==',' || *inptr=='\0')) {
        /* ',' and '\0' required incase it is a simple address, no @ domain part (buggy writer) */
        name = g_string_new ("");
        while (pre) {
            char *text, *last;

            /* perform internationalised decoding, and append */
            text = header_decode_string (pre, NULL);
            g_string_append (name, text);
            last = pre;
            g_free(text);

            pre = header_decode_word(&inptr);
            if (pre) {
                size_t l = strlen (last);
                size_t p = strlen (pre);
                
                /* dont append ' ' between sucsessive encoded words */
                if ((l>6 && last[l-2] == '?' && last[l-1] == '=')
                    && (p>6 && pre[0] == '=' && pre[1] == '?')) {
                    /* dont append ' ' */
                } else {
                    name = g_string_append_c(name, ' ');
                }
            } else {
                /* Fix for stupidly-broken-mailers that like to put '.''s in names unquoted */
                /* see bug #8147 */
                if (*inptr && *inptr != '<') {
                    w(g_warning("Working around stupid mailer bug #5: unescaped characters in names"));
                    name = g_string_append_c(name, *inptr++);
                    pre = header_decode_word(&inptr);
                }
            }
            g_free(last);
        }
        header_decode_lwsp(&inptr);
        if (*inptr == '<') {
            closeme = TRUE;
        try_address_again:
            inptr++;
            header_decode_lwsp(&inptr);
            if (*inptr == '@') {
                while (*inptr == '@') {
                    inptr++;
                    header_decode_domain(&inptr);
                    header_decode_lwsp(&inptr);
                    if (*inptr == ',') {
                        inptr++;
                        header_decode_lwsp(&inptr);
                    }
                }
                if (*inptr == ':') {
                    inptr++;
                } else {
                    w(g_warning("broken route-address, missing ':': %s", *in));
                }
            }
            pre = header_decode_word(&inptr);
            header_decode_lwsp(&inptr);
        } else {
            w(g_warning("broken address? %s", *in));
        }
    }

    if (pre) {
        addr = g_string_append(addr, pre);
    } else {
        w(g_warning("No local-part for email address: %s", *in));
    }

    /* should be at word '.' localpart */
    while (*inptr == '.' && pre) {
        inptr++;
        g_free(pre);
        pre = header_decode_word(&inptr);
        addr = g_string_append_c(addr, '.');
        if (pre)
            addr = g_string_append(addr, pre);
        comment = inptr;
        header_decode_lwsp(&inptr);
    }
    g_free(pre);

    /* now at '@' domain part */
    if (*inptr == '@') {
        char *dom;

        inptr++;
        addr = g_string_append_c(addr, '@');
        comment = inptr;
        dom = header_decode_domain(&inptr);
        addr = g_string_append(addr, dom);
        g_free(dom);
    } else {
        /* If we get a <, the address was probably a name part, lets try again shall we? */
        /* Another fix for seriously-broken-mailers */
        if (*inptr && *inptr != ',') {
            char *text;

            w(g_warning("We didn't get an '@' where we expected in '%s', trying again", *in));
            w(g_warning("Name is '%s', Addr is '%s' we're at '%s'\n", name?name->str:"<UNSET>", addr->str, inptr));

            /* need to keep *inptr, as try_address_again will drop the current character */
            if (*inptr == '<')
                closeme = TRUE;
            else
                g_string_append_c(addr, *inptr);

            /* check for address is encoded word ... */
            text = header_decode_string(addr->str, NULL);
            if (name == NULL) {
                name = addr;
                addr = g_string_new("");
                if (text) {
                    g_string_truncate(name, 0);
                    g_string_append(name, text);
                }
            } else {
                g_string_append(name, text?text:addr->str);
                g_string_truncate(addr, 0);
            }
            g_free(text);

            /* or maybe that we've added up a bunch of broken bits to make an encoded word */
            text = header_decode_string(name->str, NULL);
            if (text) {
                g_string_truncate(name, 0);
                g_string_append(name, text);
                g_free(text);
            }

            goto try_address_again;
        }
        w(g_warning("invalid address, no '@' domain part at %c: %s", *inptr, *in));
    }

    if (closeme) {
        header_decode_lwsp(&inptr);
        if (*inptr == '>') {
            inptr++;
        } else {
            w(g_warning("invalid route address, no closing '>': %s", *in));
        } 
    } else if (name == NULL && comment != NULL && inptr>comment) { /* check for comment after address */
        char *text, *tmp;
        const char *comstart, *comend;

        /* this is a bit messy, we go from the last known position, because
           decode_domain/etc skip over any comments on the way */
        /* FIXME: This wont detect comments inside the domain itself,
           but nobody seems to use that feature anyway ... */

        d(printf("checking for comment from '%s'\n", comment));

        comstart = strchr(comment, '(');
        if (comstart) {
            comstart++;
            header_decode_lwsp(&inptr);
            comend = inptr-1;
            while (comend > comstart && comend[0] != ')')
                comend--;
            
            if (comend > comstart) {
                d(printf("  looking at subset '%.*s'\n", comend-comstart, comstart));
                tmp = g_strndup (comstart, comend-comstart);
                text = header_decode_string (tmp, NULL);
                name = g_string_new (text);
                g_free (tmp);
                g_free (text);
            }
        }
    }
    
    *in = inptr;
    
    if (addr->len > 0) {
        address = header_address_new_name(name ? name->str : "", addr->str);
    }

    g_string_free(addr, TRUE);
    if (name)
        g_string_free(name, TRUE);

    d(printf("got mailbox: %s\n", addr->str));
    return address;
}

static struct _header_address *
header_decode_address(const char **in)
{
    const char *inptr = *in;
    char *pre;
    GString *group = g_string_new("");
    struct _header_address *addr = NULL, *member;

    /* pre-scan, trying to work out format, discard results */
    header_decode_lwsp(&inptr);
    while ( (pre = header_decode_word(&inptr)) ) {
        group = g_string_append(group, pre);
        group = g_string_append(group, " ");
        g_free(pre);
    }
    header_decode_lwsp(&inptr);
    if (*inptr == ':') {
        d(printf("group detected: %s\n", group->str));
        addr = header_address_new_group(group->str);
        /* that was a group spec, scan mailbox's */
        inptr++;
        /* FIXME: check rfc 2047 encodings of words, here or above in the loop */
        header_decode_lwsp(&inptr);
        if (*inptr != ';') {
            int go = TRUE;
            do {
                member = header_decode_mailbox(&inptr);
                if (member)
                    header_address_add_member(addr, member);
                header_decode_lwsp(&inptr);
                if (*inptr == ',')
                    inptr++;
                else
                    go = FALSE;
            } while (go);
            if (*inptr == ';') {
                inptr++;
            } else {
                w(g_warning("Invalid group spec, missing closing ';': %s", *in));
            }
        } else {
            inptr++;
        }
        *in = inptr;
    } else {
        addr = header_decode_mailbox(in);
    }

    g_string_free(group, TRUE);

    return addr;
}

static char *
header_msgid_decode_internal(const char **in)
{
    const char *inptr = *in;
    char *msgid = NULL;

    d(printf("decoding Message-ID: '%s'\n", *in));

    header_decode_lwsp(&inptr);
    if (*inptr == '<') {
        inptr++;
        header_decode_lwsp(&inptr);
        msgid = header_decode_addrspec(&inptr);
        if (msgid) {
            header_decode_lwsp(&inptr);
            if (*inptr == '>') {
                inptr++;
            } else {
                w(g_warning("Missing closing '>' on message id: %s", *in));
            }
        } else {
            w(g_warning("Cannot find message id in: %s", *in));
        }
    } else {
        w(g_warning("missing opening '<' on message id: %s", *in));
    }
    *in = inptr;

    return msgid;
}

char *
header_msgid_decode(const char *in)
{
    if (in == NULL)
        return NULL;

    return header_msgid_decode_internal(&in);
}

void
header_references_list_append_asis(struct _header_references **list, char *ref)
{
    struct _header_references *w = (struct _header_references *)list, *n;
    while (w->next)
        w = w->next;
    n = g_malloc(sizeof(*n));
    n->id = ref;
    n->next = 0;
    w->next = n;
}

int
header_references_list_size(struct _header_references **list)
{
    int count = 0;
    struct _header_references *w = *list;
    while (w) {
        count++;
        w = w->next;
    }
    return count;
}

void
header_references_list_clear(struct _header_references **list)
{
    struct _header_references *w = *list, *n;
    while (w) {
        n = w->next;
        g_free(w->id);
        g_free(w);
        w = n;
    }
    *list = NULL;
}

/* generate a list of references, from most recent up */
struct _header_references *
header_references_decode(const char *in)
{
    const char *inptr = in;
    struct _header_references *head = NULL, *node;
    char *id, *word;

    if (in == NULL || in[0] == '\0')
        return NULL;

    while (*inptr) {
        header_decode_lwsp(&inptr);
        if (*inptr == '<') {
            id = header_msgid_decode_internal(&inptr);
            if (id) {
                node = g_malloc(sizeof(*node));
                node->next = head;
                head = node;
                node->id = id;
            }
        } else {
            word = header_decode_word(&inptr);
            if (word)
                g_free (word);
            else if (*inptr != '\0')
                inptr++; /* Stupid mailer tricks */
        }
    }

    return head;
}

struct _header_references *
header_references_dup(const struct _header_references *list)
{
    struct _header_references *new = NULL, *tmp;

    while (list) {
        tmp = g_new(struct _header_references, 1);
        tmp->next = new;
        tmp->id = g_strdup(list->id);
        new = tmp;
        list = list->next;
    }
    return new;
}

struct _header_address *
header_mailbox_decode(const char *in)
{
    if (in == NULL)
        return NULL;

    return header_decode_mailbox(&in);
}

struct _header_address *
header_address_decode(const char *in)
{
    const char *inptr = in, *last;
    struct _header_address *list = NULL, *addr;

    d(printf("decoding To: '%s'\n", in));

    if (in == NULL)
        return NULL;

    header_decode_lwsp(&inptr);
    if (*inptr == 0)
        return NULL;

    do {
        last = inptr;
        addr = header_decode_address(&inptr);
        if (addr)
            header_address_list_append(&list, addr);
        header_decode_lwsp(&inptr);
        if (*inptr == ',')
            inptr++;
        else
            break;
    } while (inptr != last);

    if (*inptr) {
        w(g_warning("Invalid input detected at %c (%d): %s\n or at: %s", *inptr, inptr-in, in, inptr));
    }

    if (inptr == last) {
        w(g_warning("detected invalid input loop at : %s", last));
    }

    return list;
}

void
header_mime_decode(const char *in, int *maj, int *min)
{
    const char *inptr = in;
    int major=-1, minor=-1;

    d(printf("decoding MIME-Version: '%s'\n", in));

    if (in != NULL) {
        header_decode_lwsp(&inptr);
        if (isdigit(*inptr)) {
            major = header_decode_int(&inptr);
            header_decode_lwsp(&inptr);
            if (*inptr == '.') {
                inptr++;
                header_decode_lwsp(&inptr);
                if (isdigit(*inptr))
                    minor = header_decode_int(&inptr);
            }
        }
    }

    if (maj)
        *maj = major;
    if (min)
        *min = minor;

    d(printf("major = %d, minor = %d\n", major, minor));
}

static struct _header_param *
header_decode_param_list (const char **in)
{
    const char *inptr = *in;
    struct _header_param *head = NULL, *tail = NULL;
    gboolean last_was_rfc2184 = FALSE;
    gboolean is_rfc2184 = FALSE;
    
    header_decode_lwsp (&inptr);
    
    while (*inptr == ';') {
        struct _header_param *param;
        char *name, *value;
        int rfc2184_part;
        
        inptr++;
        /* invalid format? */
        if (header_decode_param (&inptr, &name, &value, &is_rfc2184, &rfc2184_part) != 0)
            break;
        
        if (is_rfc2184 && tail && !strcasecmp (name, tail->name)) {
            /* rfc2184 allows a parameter to be broken into multiple parts
             * and it looks like we've found one. Append this value to the
             * last value.
             */
            /* FIXME: we should be ordering these based on rfc2184_part id */
            GString *gvalue;
            
            gvalue = g_string_new (tail->value);
            g_string_append (gvalue, value);
            g_free (tail->value);
            g_free (value);
            g_free (name);
            
            tail->value = gvalue->str;
            g_string_free (gvalue, FALSE);
        } else {
            if (last_was_rfc2184) {
                /* We've finished gathering the values for the last param
                 * so it is now safe to decode it.
                 */
                char *val;
                
                val = rfc2184_decode (tail->value, strlen (tail->value));
                if (val) {
                    g_free (tail->value);
                    tail->value = val;
                }
            }
            
            param = g_malloc (sizeof (struct _header_param));
            param->name = name;
            param->value = value;
            param->next = NULL;
            if (head == NULL)
                head = param;
            if (tail)
                tail->next = param;
            tail = param;
        }
        
        last_was_rfc2184 = is_rfc2184;
        
        header_decode_lwsp (&inptr);
    }
    
    if (last_was_rfc2184) {
        /* We've finished gathering the values for the last param
         * so it is now safe to decode it.
         */
        char *val;
        
        val = rfc2184_decode (tail->value, strlen (tail->value));
        if (val) {
            g_free (tail->value);
            tail->value = val;
        }
    }
    
    *in = inptr;
    
    return head;
}

struct _header_param *
header_param_list_decode(const char *in)
{
    if (in == NULL)
        return NULL;

    return header_decode_param_list(&in);
}


static char *
header_encode_param (const unsigned char *in, gboolean *encoded)
{
    const unsigned char *inptr = in;
    char *outstr, *charset;
    int encoding;
    GString *out;
    
    *encoded = FALSE;
    
    g_return_val_if_fail (in != NULL, NULL);
    g_return_val_if_fail (g_utf8_validate (in, -1, NULL), NULL);
    
    /* do a quick us-ascii check (the common case?) */
    while (*inptr) {
        if (*inptr > 127)
            break;
        inptr++;
    }
    
    if (*inptr == '\0')
        return g_strdup (in);
    
    out = g_string_new ("");
    inptr = in;
    encoding = 0;
    while (inptr && *inptr) {
        gunichar c;
        const char *newinptr;
        
        newinptr = g_utf8_next_char (inptr);
        c = g_utf8_get_char (inptr);
        if (newinptr == NULL || !g_unichar_validate (c)) {
            w(g_warning ("Invalid UTF-8 sequence encountered (pos %d, char '%c'): %s",
                     (inptr-in), inptr[0], in));
            inptr++;
            continue;
        }
        
        /* FIXME: make sure that '\'', '*', and ';' are also encoded */
        
        if (c > 127 && c < 256) {
            encoding = MAX (encoding, 1);
            g_string_sprintfa (out, "%%%c%c", tohex[(c >> 4) & 0xf], tohex[c & 0xf]);
        } else if (c >= 256) {
            encoding = MAX (encoding, 2);
            g_string_sprintfa (out, "%%%c%c", tohex[(c >> 4) & 0xf], tohex[c & 0xf]);
        } else if (is_lwsp (c) || !(camel_mime_special_table[c] & IS_ESAFE)) {
            g_string_sprintfa (out, "%%%c%c", tohex[(c >> 4) & 0xf], tohex[c & 0xf]);
        } else {
            g_string_append_c (out, c);
        }
        
        inptr = newinptr;
    }
    
    /* FIXME: set the 'language' as well, assuming we can get that info...? */
    switch (encoding) {
    default:
        g_string_prepend (out, "iso-8859-1''");
        break;
    case 2:
        charset = g_strdup_printf ("%s''", camel_charset_best (in, inptr - in));
        g_string_prepend (out, charset);
        g_free (charset);
        break;
    }
    
    outstr = out->str;
    g_string_free (out, FALSE);
    *encoded = TRUE;
    
    return outstr;
}

void
header_param_list_format_append (GString *out, struct _header_param *p)
{
    int used = out->len;
    
    while (p) {
        gboolean encoded = FALSE;
        gboolean quote = FALSE;
        int here = out->len;
        size_t nlen, vlen;
        char *value;
        
        if (!p->value) {
            p = p->next;
            continue;
        }
        
        value = header_encode_param (p->value, &encoded);
        if (!value) {
            w(g_warning ("appending parameter %s=%s violates rfc2184", p->name, p->value));
            value = g_strdup (p->value);
        }
        
        if (!encoded) {
            char *ch;
            
            for (ch = value; *ch; ch++) {
                if (is_tspecial (*ch) || is_lwsp (*ch))
                    break;
            }
            
            quote = ch && *ch;
        }
        
        nlen = strlen (p->name);
        vlen = strlen (value);
        
        if (used + nlen + vlen > CAMEL_FOLD_SIZE - 8) {
            out = g_string_append (out, ";\n\t");
            here = out->len;
            used = 0;
        } else
            out = g_string_append (out, "; ");
        
        if (nlen + vlen > CAMEL_FOLD_SIZE - 8) {
            /* we need to do special rfc2184 parameter wrapping */
            int maxlen = CAMEL_FOLD_SIZE - (nlen + 8);
            char *inptr, *inend;
            int i = 0;
            
            inptr = value;
            inend = value + vlen;
            
            while (inptr < inend) {
                char *ptr = inptr + MIN (inend - inptr, maxlen);
                
                if (encoded && ptr < inend) {
                    /* be careful not to break an encoded char (ie %20) */
                    char *q = ptr;
                    int j = 2;
                    
                    for ( ; j > 0 && q > inptr && *q != '%'; j--, q--);
                    if (*q == '%')
                        ptr = q;
                }
                
                if (i != 0) {
                    g_string_append (out, ";\n\t");
                    here = out->len;
                    used = 0;
                }
                
                g_string_sprintfa (out, "%s*%d%s=", p->name, i++, encoded ? "*" : "");
                if (encoded || !quote)
                    g_string_append_len (out, inptr, ptr - inptr);
                else
                    quote_word (out, TRUE, inptr, ptr - inptr);
                
                d(printf ("wrote: %s\n", out->str + here));
                
                used += (out->len - here);
                
                inptr = ptr;
            }
        } else {
            g_string_sprintfa (out, "%s%s=", p->name, encoded ? "*" : "");
            
            if (encoded || !quote)
                g_string_append (out, value);
            else
                quote_word (out, TRUE, value, vlen);
            
            used += (out->len - here);
        }
        
        g_free (value);
        
        p = p->next;
    }
}

char *
header_param_list_format(struct _header_param *p)
{
    GString *out = g_string_new("");
    char *ret;

    header_param_list_format_append(out, p);
    ret = out->str;
    g_string_free(out, FALSE);
    return ret;
}

struct _header_content_type *
header_content_type_decode(const char *in)
{
    const char *inptr = in;
    char *type, *subtype = NULL;
    struct _header_content_type *t = NULL;

    if (in==NULL)
        return NULL;

    type = decode_token(&inptr);
    header_decode_lwsp(&inptr);
    if (type) {
        if  (*inptr == '/') {
            inptr++;
            subtype = decode_token(&inptr);
        }
        if (subtype == NULL && (!strcasecmp(type, "text"))) {
            w(g_warning("text type with no subtype, resorting to text/plain: %s", in));
            subtype = g_strdup("plain");
        }
        if (subtype == NULL) {
            w(g_warning("MIME type with no subtype: %s", in));
        }

        t = header_content_type_new(type, subtype);
        t->params = header_decode_param_list(&inptr);
        g_free(type);
        g_free(subtype);
    } else {
        g_free(type);
        d(printf("cannot find MIME type in header (2) '%s'", in));
    }
    return t;
}

void
header_content_type_dump(struct _header_content_type *ct)
{
    struct _header_param *p;

    printf("Content-Type: ");
    if (ct==NULL) {
        printf("<NULL>\n");
        return;
    }
    printf("%s / %s", ct->type, ct->subtype);
    p = ct->params;
    if (p) {
        while (p) {
            printf(";\n\t%s=\"%s\"", p->name, p->value);
            p = p->next;
        }
    }
    printf("\n");
}

char *
header_content_type_format(struct _header_content_type *ct)
{
    GString *out;
    char *ret;

    if (ct==NULL)
        return NULL;

    out = g_string_new("");
    if (ct->type == NULL) {
        g_string_sprintfa(out, "text/plain");
        w(g_warning("Content-Type with no main type"));
    } else if (ct->subtype == NULL) {
        w(g_warning("Content-Type with no sub type: %s", ct->type));
        if (!strcasecmp(ct->type, "multipart"))
            g_string_sprintfa(out, "%s/mixed", ct->type);
        else
            g_string_sprintfa(out, "%s", ct->type);
    } else {
        g_string_sprintfa(out, "%s/%s", ct->type, ct->subtype);
    }
    header_param_list_format_append(out, ct->params);

    ret = out->str;
    g_string_free(out, FALSE);
    return ret;
}

char *
header_content_type_simple (struct _header_content_type *ct)
{
    if (ct->type == NULL) {
        w(g_warning ("Content-Type with no main type"));
        return g_strdup ("text/plain");
    } else if (ct->subtype == NULL) {
        w(g_warning ("Content-Type with no sub type: %s", ct->type));
        if (!strcasecmp (ct->type, "multipart"))
            return g_strdup_printf ("%s/mixed", ct->type);
        else
            return g_strdup (ct->type);
    } else
        return g_strdup_printf ("%s/%s", ct->type, ct->subtype);
}

char *
header_content_encoding_decode(const char *in)
{
    if (in)
        return decode_token(&in);
    return NULL;
}

CamelMimeDisposition *
header_disposition_decode(const char *in)
{
    CamelMimeDisposition *d = NULL;
    const char *inptr = in;

    if (in == NULL)
        return NULL;

    d = g_malloc(sizeof(*d));
    d->refcount = 1;
    d->disposition = decode_token(&inptr);
    if (d->disposition == NULL)
        w(g_warning("Empty disposition type"));
    d->params = header_decode_param_list(&inptr);
    return d;
}

void
header_disposition_ref(CamelMimeDisposition *d)
{
    if (d)
        d->refcount++;
}

void
header_disposition_unref(CamelMimeDisposition *d)
{
    if (d) {
        if (d->refcount<=1) {
            header_param_list_free(d->params);
            g_free(d->disposition);
            g_free(d);
        } else {
            d->refcount--;
        }
    }
}

char *
header_disposition_format(CamelMimeDisposition *d)
{
    GString *out;
    char *ret;

    if (d==NULL)
        return NULL;

    out = g_string_new("");
    if (d->disposition)
        out = g_string_append(out, d->disposition);
    else
        out = g_string_append(out, "attachment");
    header_param_list_format_append(out, d->params);

    ret = out->str;
    g_string_free(out, FALSE);
    return ret;
}

/* hrm, is there a library for this shit? */
static struct {
    char *name;
    int offset;
} tz_offsets [] = {
    { "UT", 0 },
    { "GMT", 0 },
    { "EST", -500 },    /* these are all US timezones.  bloody yanks */
    { "EDT", -400 },
    { "CST", -600 },
    { "CDT", -500 },
    { "MST", -700 },
    { "MDT", -600 },
    { "PST", -800 },
    { "PDT", -700 },
    { "Z", 0 },
    { "A", -100 },
    { "M", -1200 },
    { "N", 100 },
    { "Y", 1200 },
};

static char *tz_months [] = {
    "Jan", "Feb", "Mar", "Apr", "May", "Jun",
    "Jul", "Aug", "Sep", "Oct", "Nov", "Dec"
};

char *
header_format_date(time_t time, int offset)
{
    struct tm tm;

    d(printf("offset = %d\n", offset));

    d(printf("converting date %s", ctime(&time)));

    time += ((offset / 100) * (60*60)) + (offset % 100)*60;

    d(printf("converting date %s", ctime(&time)));

    memcpy(&tm, gmtime(&time), sizeof(tm));

    return g_strdup_printf("%02d %s %04d %02d:%02d:%02d %+05d",
                   tm.tm_mday, tz_months[tm.tm_mon],
                   tm.tm_year + 1900,
                   tm.tm_hour, tm.tm_min, tm.tm_sec,
                   offset);
}

/* convert a date to time_t representation */
/* this is an awful mess oh well */
time_t
header_decode_date(const char *in, int *saveoffset)
{
    const char *inptr = in;
    char *monthname;
    int year, offset = 0;
    struct tm tm;
    int i;
    time_t t;

    if (in == NULL) {
        if (saveoffset)
            *saveoffset = 0;
        return 0;
    }

    d(printf ("\ndecoding date '%s'\n", inptr));

    memset (&tm, 0, sizeof(tm));

    header_decode_lwsp (&inptr);
    if (!isdigit (*inptr)) {
        char *day = decode_token (&inptr);
        /* we dont really care about the day, it's only for display */
        if (day) {
            d(printf ("got day: %s\n", day));
            g_free (day);
            header_decode_lwsp (&inptr);
            if (*inptr == ',') {
                inptr++;
            } else {
#ifndef CLEAN_DATE
                return parse_broken_date (in, saveoffset);
#else
                if (saveoffset)
                    *saveoffset = 0;
                return 0;
#endif /* ! CLEAN_DATE */
            }
        }
    }
    tm.tm_mday = header_decode_int(&inptr);
    monthname = decode_token(&inptr);
    if (monthname) {
        for (i=0;i<sizeof(tz_months)/sizeof(tz_months[0]);i++) {
            if (!strcasecmp(tz_months[i], monthname)) {
                tm.tm_mon = i;
                break;
            }
        }
        g_free(monthname);
    }
    year = header_decode_int(&inptr);
    if (year < 69 || (year >= 100 && year < 1900)) {
        tm.tm_year = 100 + year;
    } else if (year < 100) {
        tm.tm_year = year;
    } else {
        tm.tm_year = year - 1900;
    }
    /* get the time ... yurck */
    tm.tm_hour = header_decode_int(&inptr);
    header_decode_lwsp(&inptr);
    if (*inptr == ':')
        inptr++;
    tm.tm_min = header_decode_int(&inptr);
    header_decode_lwsp(&inptr);
    if (*inptr == ':')
        inptr++;
    tm.tm_sec = header_decode_int(&inptr);
    header_decode_lwsp(&inptr);
    if (*inptr == '+'
        || *inptr == '-') {
        offset = (*inptr++)=='-'?-1:1;
        offset = offset * header_decode_int(&inptr);
        d(printf("abs signed offset = %d\n", offset));
    } else if (isdigit(*inptr)) {
        offset = header_decode_int(&inptr);
        d(printf("abs offset = %d\n", offset));
    } else {
        char *tz = decode_token(&inptr);

        if (tz) {
            for (i=0;i<sizeof(tz_offsets)/sizeof(tz_offsets[0]);i++) {
                if (!strcasecmp(tz_offsets[i].name, tz)) {
                    offset = tz_offsets[i].offset;
                    break;
                }
            }
            g_free(tz);
        }
        /* some broken mailers seem to put in things like GMT+1030 instead of just +1030 */
        header_decode_lwsp(&inptr);
        if (*inptr == '+' || *inptr == '-') {
            int sign = (*inptr++)=='-'?-1:1;
            offset = offset + (header_decode_int(&inptr)*sign);
        }
        d(printf("named offset = %d\n", offset));
    }

    t = mktime(&tm);
#if defined(HAVE_TIMEZONE)
    t -= timezone;
#elif defined(HAVE_TM_GMTOFF)
    t += tm.tm_gmtoff;
#else
#error Neither HAVE_TIMEZONE nor HAVE_TM_GMTOFF defined. Rerun autoheader, autoconf, etc.
#endif

    /* t is now GMT of the time we want, but not offset by the timezone ... */

    d(printf(" gmt normalized? = %s\n", ctime(&t)));

    /* this should convert the time to the GMT equiv time */
    t -= ( (offset/100) * 60*60) + (offset % 100)*60;

    d(printf(" gmt normalized for timezone? = %s\n", ctime(&t)));

    d({
        char *tmp;
        tmp = header_format_date(t, offset);
        printf(" encoded again: %s\n", tmp);
        g_free(tmp);
    });

    if (saveoffset)
        *saveoffset = offset;

    return t;
}

char *
header_location_decode(const char *in)
{
    const char *p;

    /* Sigh. RFC2557 says:
     *   content-location =   "Content-Location:" [CFWS] URI [CFWS]
     *      where URI is restricted to the syntax for URLs as
     *      defined in Uniform Resource Locators [URL] until
     *      IETF specifies other kinds of URIs.
     *
     * But Netscape puts quotes around the URI when sending web
     * pages.
     */

    header_decode_lwsp(&in);
    if (*in == '"')
        return header_decode_quoted_string(&in);
    else {
        for (p = in; *p && !is_lwsp(*p); p++)
            ;
        return g_strndup(in, p - in);
    }
}


/* extra rfc checks */
#define CHECKS

#ifdef CHECKS
static void
check_header(struct _header_raw *h)
{
    unsigned char *p;

    p = h->value;
    while (p && *p) {
        if (!isascii(*p)) {
            w(g_warning("Appending header violates rfc: %s: %s", h->name, h->value));
            return;
        }
        p++;
    }
}
#endif

void
header_raw_append_parse(struct _header_raw **list, const char *header, int offset)
{
    register const char *in;
    size_t fieldlen;
    char *name;

    in = header;
    while (is_fieldname(*in) || *in==':')
        in++;
    fieldlen = in-header-1;
    while (is_lwsp(*in))
        in++;
    if (fieldlen == 0 || header[fieldlen] != ':') {
        printf("Invalid header line: '%s'\n", header);
        return;
    }
    name = alloca(fieldlen+1);
    memcpy(name, header, fieldlen);
    name[fieldlen] = 0;

    header_raw_append(list, name, in, offset);
}

void
header_raw_append(struct _header_raw **list, const char *name, const char *value, int offset)
{
    struct _header_raw *l, *n;

    d(printf("Header: %s: %s\n", name, value));

    n = g_malloc(sizeof(*n));
    n->next = NULL;
    n->name = g_strdup(name);
    n->value = g_strdup(value);
    n->offset = offset;
#ifdef CHECKS
    check_header(n);
#endif
    l = (struct _header_raw *)list;
    while (l->next) {
        l = l->next;
    }
    l->next = n;

    /* debug */
#if 0
    if (!strcasecmp(name, "To")) {
        printf("- Decoding To\n");
        header_to_decode(value);
    } else if (!strcasecmp(name, "Content-type")) {
        printf("- Decoding content-type\n");
        header_content_type_dump(header_content_type_decode(value));        
    } else if (!strcasecmp(name, "MIME-Version")) {
        printf("- Decoding mime version\n");
        header_mime_decode(value);
    }
#endif
}

static struct _header_raw *
header_raw_find_node(struct _header_raw **list, const char *name)
{
    struct _header_raw *l;

    l = *list;
    while (l) {
        if (!strcasecmp(l->name, name))
            break;
        l = l->next;
    }
    return l;
}

const char *
header_raw_find(struct _header_raw **list, const char *name, int *offset)
{
    struct _header_raw *l;

    l = header_raw_find_node(list, name);
    if (l) {
        if (offset)
            *offset = l->offset;
        return l->value;
    } else
        return NULL;
}

const char *
header_raw_find_next(struct _header_raw **list, const char *name, int *offset, const char *last)
{
    struct _header_raw *l;

    if (last == NULL || name == NULL)
        return NULL;

    l = *list;
    while (l && l->value != last)
        l = l->next;
    return header_raw_find(&l, name, offset);
}

static void
header_raw_free(struct _header_raw *l)
{
    g_free(l->name);
    g_free(l->value);
    g_free(l);
}

void
header_raw_remove(struct _header_raw **list, const char *name)
{
    struct _header_raw *l, *p;

    /* the next pointer is at the head of the structure, so this is safe */
    p = (struct _header_raw *)list;
    l = *list;
    while (l) {
        if (!strcasecmp(l->name, name)) {
            p->next = l->next;
            header_raw_free(l);
            l = p->next;
        } else {
            p = l;
            l = l->next;
        }
    }
}

void
header_raw_replace(struct _header_raw **list, const char *name, const char *value, int offset)
{
    header_raw_remove(list, name);
    header_raw_append(list, name, value, offset);
}

void
header_raw_clear(struct _header_raw **list)
{
    struct _header_raw *l, *n;
    l = *list;
    while (l) {
        n = l->next;
        header_raw_free(l);
        l = n;
    }
    *list = NULL;
}

char *
header_msgid_generate (void)
{
    char host[MAXHOSTNAMELEN];
#ifdef ENABLE_THREADS
    static pthread_mutex_t count_lock = PTHREAD_MUTEX_INITIALIZER;
#define COUNT_LOCK() pthread_mutex_lock (&count_lock)
#define COUNT_UNLOCK() pthread_mutex_unlock (&count_lock)
#else
#define COUNT_LOCK()
#define COUNT_UNLOCK()
#endif /* ENABLE_THREADS */
    static int count = 0;
    int hrv;
    char *ret;
    
    hrv = gethostname (host, sizeof (host));
    
    COUNT_LOCK ();
    ret = g_strdup_printf ("%d.%d.%d.camel@%s", (gint) time (NULL), getpid (), count++,
                   (hrv == 0 && host && *host) ? host : "unknown.host");
    COUNT_UNLOCK ();
    
    return ret;
}


static struct {
    char *name;
    char *pattern;
    regex_t regex;
} mail_list_magic[] = {
    /* Sender: owner-gnome-hackers@gnome.org */
    /* Sender: owner-gnome-hacekrs */
    { "Sender", " *owner-([^@]+)@?([^ @\n\t\r>]*)" },
    /* Sender: gnome-hackers-owner@gnome.org */
    /* Sender: gnome-hackers-owner */
    { "Sender", " *([^@]+)-owner@?([^ @\n\t\r>]*)" },
    /* Sender: owner-gnome-hackers@gnome.org */
    /* Sender: <owner-gnome-hackers@gnome.org> */
    /* Sender: owner-gnome-hackers */
    /* Sender: <owner-gnome-hackers> */
    { "Return-Path", " <?owner-([^@>]+)@?([^ \n\t\r>]*)" },
    /* X-BeenThere: gnome-hackers@gnome.org */
    /* X-BeenThere: gnome-hackers */
    { "X-BeenThere", " *([^@]+)@?([^ \n\t\r>]*)" },
    /* Delivered-To: mailing list gnome-hackers@gnome.org */
    /* Delivered-To: mailing list gnome-hackers */
    { "Delivered-To", " *mailing list ([^@]+)@?([^ \n\t\r>]*)" },
    /* X-Mailing-List: <gnome-hackers@gnome.org> arcive/latest/100 */
    /* X-Mailing-List: gnome-hackers@gnome.org */
    /* X-Mailing-List: gnome-hackers */
    /* X-Mailing-List: <gnome-hackers> */
    { "X-Mailing-List", " <?([^@>]+)@?([^ \n\t\r>]*)" },
    /* X-Loop: gnome-hackers@gnome.org */
    { "X-Loop", " *([^@]+)@?([^ \n\t\r>]*)" },
    /* List-Id: GNOME stuff <gnome-hackers.gnome.org> */
    /* List-Id: <gnome-hackers.gnome.org> */
    /* List-Id: <gnome-hackers> */
    /* This old one wasn't very useful: { "List-Id", " *([^<]+)" },*/
    { "List-Id", "[^<]*<([^\\.>]+)\\.?([^ \n\t\r>]*)" },
    /* List-Post: <mailto:gnome-hackers@gnome.org> */
    /* List-Post: <mailto:gnome-hackers> */
    { "List-Post", " *<mailto:([^@>]+)@?([^ \n\t\r>]*)" },
    /* Mailing-List: list gnome-hackers@gnome.org; contact gnome-hackers-owner@gnome.org */
    { "Mailing-List", " *list ([^@]+)@?([^ \n\t\r>;])*" },
    /* Originator: gnome-hackers@gnome.org */
    { "Originator", " *([^@]+)@?([^ \n\t\r>]*)" },
    /* X-List: gnome-hackers */
    /* X-List: gnome-hackers@gnome.org */
    { "X-List", " *([^@]+)@?([^ \n\t\r>]*)" },  
};

char *
header_raw_check_mailing_list(struct _header_raw **list)
{
    const char *v;
    regmatch_t match[3];
    int i;
    
    for (i = 0; i < sizeof (mail_list_magic) / sizeof (mail_list_magic[0]); i++) {
        v = header_raw_find (list, mail_list_magic[i].name, NULL);
        if (v != NULL && regexec (&mail_list_magic[i].regex, v, 3, match, 0) == 0 && match[1].rm_so != -1) {
            char *list;
            int len1, len2;

            len1 = match[1].rm_eo - match[1].rm_so;
            len2 = match[2].rm_eo - match[2].rm_so;

            list = g_malloc(len1+len2+2);
            memcpy(list, v + match[1].rm_so, len1);
            if (len2) {
                list[len1] = '@';
                memcpy(list+len1+1, v+match[2].rm_so, len2);
                list[len1+len2+1]=0;
            } else {
                list[len1] = 0;
            }

            return list;
        }
    }

    return NULL;
}

/* ok, here's the address stuff, what a mess ... */
struct _header_address *header_address_new(void)
{
    struct _header_address *h;
    h = g_malloc0(sizeof(*h));
    h->type = HEADER_ADDRESS_NONE;
    h->refcount = 1;
    return h;
}

struct _header_address *header_address_new_name(const char *name, const char *addr)
{
    struct _header_address *h;

    h = header_address_new();
    h->type = HEADER_ADDRESS_NAME;
    h->name = g_strdup(name);
    h->v.addr = g_strdup(addr);
    return h;
}

struct _header_address *header_address_new_group(const char *name)
{
    struct _header_address *h;

    h = header_address_new();
    h->type = HEADER_ADDRESS_GROUP;
    h->name = g_strdup(name);
    return h;
}

void header_address_ref(struct _header_address *h)
{
    if (h)
        h->refcount++;
}

void header_address_unref(struct _header_address *h)
{
    if (h) {
        if (h->refcount <= 1) {
            if (h->type == HEADER_ADDRESS_GROUP) {
                header_address_list_clear(&h->v.members);
            } else if (h->type == HEADER_ADDRESS_NAME) {
                g_free(h->v.addr);
            }
            g_free(h->name);
            g_free(h);
        } else {
            h->refcount--;
        }
    }
}

void header_address_set_name(struct _header_address *h, const char *name)
{
    if (h) {
        g_free(h->name);
        h->name = g_strdup(name);
    }
}

void header_address_set_addr(struct _header_address *h, const char *addr)
{
    if (h) {
        if (h->type == HEADER_ADDRESS_NAME
            || h->type == HEADER_ADDRESS_NONE) {
            h->type = HEADER_ADDRESS_NAME;
            g_free(h->v.addr);
            h->v.addr = g_strdup(addr);
        } else {
            g_warning("Trying to set the address on a group");
        }
    }
}

void header_address_set_members(struct _header_address *h, struct _header_address *group)
{
    if (h) {
        if (h->type == HEADER_ADDRESS_GROUP
            || h->type == HEADER_ADDRESS_NONE) {
            h->type = HEADER_ADDRESS_GROUP;
            header_address_list_clear(&h->v.members);
            /* should this ref them? */
            h->v.members = group;
        } else {
            g_warning("Trying to set the members on a name, not group");
        }
    }
}

void header_address_add_member(struct _header_address *h, struct _header_address *member)
{
    if (h) {
        if (h->type == HEADER_ADDRESS_GROUP
            || h->type == HEADER_ADDRESS_NONE) {
            h->type = HEADER_ADDRESS_GROUP;
            header_address_list_append(&h->v.members, member);
        }           
    }
}

void header_address_list_append_list(struct _header_address **l, struct _header_address **h)
{
    if (l) {
        struct _header_address *n = (struct _header_address *)l;

        while (n->next)
            n = n->next;
        n->next = *h;
    }
}


void header_address_list_append(struct _header_address **l, struct _header_address *h)
{
    if (h) {
        header_address_list_append_list(l, &h);
        h->next = NULL;
    }
}

void header_address_list_clear(struct _header_address **l)
{
    struct _header_address *a, *n;
    a = *l;
    while (a) {
        n = a->next;
        header_address_unref(a);
        a = n;
    }
    *l = NULL;
}

/* if encode is true, then the result is suitable for mailing, otherwise
   the result is suitable for display only (and may not even be re-parsable) */
static void
header_address_list_encode_append(GString *out, int encode, struct _header_address *a)
{
    char *text;

    while (a) {
        switch (a->type) {
        case HEADER_ADDRESS_NAME:
            if (encode)
                text = header_encode_phrase (a->name);
            else
                text = a->name;
            if (text && *text)
                g_string_sprintfa(out, "%s <%s>", text, a->v.addr);
            else
                g_string_append(out, a->v.addr);
            if (encode)
                g_free(text);
            break;
        case HEADER_ADDRESS_GROUP:
            if (encode)
                text = header_encode_phrase(a->name);
            else
                text = a->name;
            g_string_sprintfa(out, "%s: ", text);
            header_address_list_encode_append(out, encode, a->v.members);
            g_string_sprintfa(out, ";");
            if (encode)
                g_free(text);
            break;
        default:
            g_warning("Invalid address type");
            break;
        }
        a = a->next;
        if (a)
            g_string_append(out, ", ");
    }
}

char *
header_address_list_encode(struct _header_address *a)
{
    GString *out;
    char *ret;

    if (a == NULL)
        return NULL;

    out = g_string_new("");

    header_address_list_encode_append(out, TRUE, a);
    ret = out->str;
    g_string_free(out, FALSE);
    return ret;
}

char *
header_address_list_format(struct _header_address *a)
{
    GString *out;
    char *ret;

    if (a == NULL)
        return NULL;

    out = g_string_new("");

    header_address_list_encode_append(out, FALSE, a);
    ret = out->str;
    g_string_free(out, FALSE);
    return ret;
}

char *
header_address_fold (const char *in, size_t headerlen)
{
    size_t len, outlen;
    const char *inptr = in, *space, *p, *n;
    GString *out;
    char *ret;
    int i, needunfold = FALSE;
    
    if (in == NULL)
        return NULL;
    
    /* first, check to see if we even need to fold */
    len = headerlen + 2;
    p = in;
    while (*p) {
        n = strchr (p, '\n');
        if (n == NULL) {
            len += strlen (p);
            break;
        }
        
        needunfold = TRUE;
        len += n-p;
        
        if (len >= CAMEL_FOLD_SIZE)
            break;
        len = 0;
        p = n + 1;
    }
    if (len < CAMEL_FOLD_SIZE)
        return g_strdup (in);
    
    /* we need to fold, so first unfold (if we need to), then process */
    if (needunfold)
        inptr = in = header_unfold (in);
    
    out = g_string_new ("");
    outlen = headerlen + 2;
    while (*inptr) {
        space = strchr (inptr, ' ');
        if (space) {
            len = space - inptr + 1;
        } else {
            len = strlen (inptr);
        }
        
        d(printf("next word '%.*s'\n", len, inptr));
        
        if (outlen + len > CAMEL_FOLD_SIZE) {
            d(printf("outlen = %d wordlen = %d\n", outlen, len));
            /* strip trailing space */
            if (out->len > 0 && out->str[out->len-1] == ' ')
                g_string_truncate (out, out->len-1);
            g_string_append (out, "\n\t");
            outlen = 1;
        }
        
        outlen += len;
        for (i = 0; i < len; i++) {
            g_string_append_c (out, inptr[i]);
        }
        
        inptr += len;
    }
    ret = out->str;
    g_string_free (out, FALSE);
    
    if (needunfold)
        g_free ((char *)in);
    
    return ret; 
}

/* simple header folding */
/* will work even if the header is already folded */
char *
header_fold(const char *in, size_t headerlen)
{
    size_t len, outlen, i;
    const char *inptr = in, *space, *p, *n;
    GString *out;
    char *ret;
    int needunfold = FALSE;

    if (in == NULL)
        return NULL;

    /* first, check to see if we even need to fold */
    len = headerlen + 2;
    p = in;
    while (*p) {
        n = strchr(p, '\n');
        if (n == NULL) {
            len += strlen (p);
            break;
        }

        needunfold = TRUE;
        len += n-p;
        
        if (len >= CAMEL_FOLD_SIZE)
            break;
        len = 0;
        p = n + 1;
    }
    if (len < CAMEL_FOLD_SIZE)
        return g_strdup(in);

    /* we need to fold, so first unfold (if we need to), then process */
    if (needunfold)
        inptr = in = header_unfold(in);

    out = g_string_new("");
    outlen = headerlen+2;
    while (*inptr) {
        space = strchr(inptr, ' ');
        if (space) {
            len = space-inptr+1;
        } else {
            len = strlen(inptr);
        }
        d(printf("next word '%.*s'\n", len, inptr));
        if (outlen + len > CAMEL_FOLD_SIZE) {
            d(printf("outlen = %d wordlen = %d\n", outlen, len));
            /* strip trailing space */
            if (out->len > 0 && out->str[out->len-1] == ' ')
                g_string_truncate(out, out->len-1);
            g_string_append(out, "\n\t");
            outlen = 1;
            /* check for very long words, just cut them up */
            while (outlen+len > CAMEL_FOLD_SIZE) {
                for (i=0;i<CAMEL_FOLD_SIZE-outlen;i++)
                    g_string_append_c(out, inptr[i]);
                inptr += CAMEL_FOLD_SIZE-outlen;
                len -= CAMEL_FOLD_SIZE-outlen;
                g_string_append(out, "\n\t");
                outlen = 1;
            }
        }
        outlen += len;
        for (i=0;i<len;i++) {
            g_string_append_c(out, inptr[i]);
        }
        inptr += len;
    }
    ret = out->str;
    g_string_free(out, FALSE);

    if (needunfold)
        g_free((char *)in);

    return ret; 
}

char *
header_unfold(const char *in)
{
    char *out = g_malloc(strlen(in)+1);
    const char *inptr = in;
    char c, *o = out;

    o = out;
    while ((c = *inptr++)) {
        if (c == '\n') {
            if (is_lwsp(*inptr)) {
                do {
                    inptr++;
                } while (is_lwsp(*inptr));
                *o++ = ' ';
            } else {
                *o++ = c;
            }
        } else {
            *o++ = c;
        }
    }
    *o = 0;

    return out;
}

void
camel_mime_utils_init(void)
{
    int i, errcode, regex_compilation_failed=0;

    /* Init tables */
    header_decode_init();
    base64_init();

    /* precompile regex's for speed at runtime */
    for (i = 0; i < sizeof(mail_list_magic) / sizeof(mail_list_magic[0]); i++) {
        errcode = regcomp(&mail_list_magic[i].regex, mail_list_magic[i].pattern, REG_EXTENDED|REG_ICASE);
        if (errcode != 0) {
            char *errstr;
            size_t len;
        
            len = regerror(errcode, &mail_list_magic[i].regex, NULL, 0);
            errstr = g_malloc0(len + 1);
            regerror(errcode, &mail_list_magic[i].regex, errstr, len);
        
            g_warning("Internal error, compiling regex failed: %s: %s", mail_list_magic[i].pattern, errstr);
            g_free(errstr);
            regex_compilation_failed++;
        }
    }

    g_assert(regex_compilation_failed == 0);
}