path: root/e-util/e-iconv.c

/* -*- Mode: C; tab-width: 8; indent-tabs-mode: t; c-basic-offset: 8 -*- */
/* 
 * e-iconv.c
 * Copyright 2000, 2001, Ximian, Inc.
 *
 * Authors:
 *   Michael Zucchi <notzed@ximian.com>
 *   Jeffery Stedfast <fejj@ximian.com>
 *
 * This library is free software; you can redistribute it and/or
 * modify it under the terms of the GNU Library General Public
 * License, version 2, as published by the Free Software Foundation.
 *
 * This library is distributed in the hope that it will be useful, but
 * WITHOUT ANY WARRANTY; without even the implied warranty of
 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
 * Library General Public License for more details.
 *
 * You should have received a copy of the GNU Library General Public
 * License along with this library; if not, write to the Free Software
 * Foundation, Inc., 59 Temple Place - Suite 330, Boston, MA
 * 02111-1307, USA.
 */

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

#include <stdio.h>
#include <stdlib.h>
#include <string.h>
#include <errno.h>

#include <glib.h>
#include "e-iconv.h"

#include <locale.h>

#ifdef HAVE_CODESET
#include <langinfo.h>
#endif

#include "iconv-detect.h"

#define cd(x) 

#ifdef G_THREADS_ENABLED
static GStaticMutex lock = G_STATIC_MUTEX_INIT;
#define LOCK() g_static_mutex_lock(&lock)
#define UNLOCK() g_static_mutex_unlock(&lock)
#else
#define LOCK()
#define UNLOCK()
#endif

typedef struct _EDListNode {
    struct _EDListNode *next;
    struct _EDListNode *prev;
} EDListNode;

typedef struct _EDList {
    struct _EDListNode *head;
    struct _EDListNode *tail;
    struct _EDListNode *tailpred;
} EDList;

#define E_DLIST_INITIALISER(l) { (EDListNode *)&l.tail, 0, (EDListNode *)&l.head }

struct _iconv_cache_node {
    struct _iconv_cache_node *next;
    struct _iconv_cache_node *prev;

    struct _iconv_cache *parent;

    int busy;
    iconv_t ip;
};

struct _iconv_cache {
    struct _iconv_cache *next;
    struct _iconv_cache *prev;

    char *conv;

    EDList open;        /* stores iconv_cache_nodes, busy ones up front */
};

#define E_ICONV_CACHE_SIZE (16)

static EDList iconv_cache_list;
static GHashTable *iconv_cache;
static GHashTable *iconv_cache_open;
static unsigned int iconv_cache_size = 0;

static GHashTable *iconv_charsets = NULL;
static char *locale_charset = NULL;
static char *locale_lang = NULL;

struct {
    char *charset;
    char *iconv_name;
} known_iconv_charsets[] = {
#if 0
    /* charset name, iconv-friendly charset name */
    { "iso-8859-1",     "iso-8859-1" },
    { "iso8859-1",      "iso-8859-1" },
    /* the above mostly serves as an example for iso-style charsets,
       but we have code that will populate the iso-*'s if/when they
       show up in e_iconv_charset_name() so I'm
       not going to bother putting them all in here... */
    { "windows-cp1251", "cp1251"     },
    { "windows-1251",   "cp1251"     },
    { "cp1251",         "cp1251"     },
    /* the above mostly serves as an example for windows-style
       charsets, but we have code that will parse and convert them
       to their cp#### equivalents if/when they show up in
       e_iconv_charset_name() so I'm not going to bother
       putting them all in here either... */
#endif
    /* charset name (lowercase!), iconv-friendly name (sometimes case sensitive) */
    { "utf-8",          "UTF-8"      },

    /* 10646 is a special case, its usually UCS-2 big endian */
    /* This might need some checking but should be ok for solaris/linux */
    { "iso-10646-1",    "UCS-2BE"    },
    { "iso_10646-1",    "UCS-2BE"    },
    { "iso10646-1",     "UCS-2BE"    },
    { "iso-10646",      "UCS-2BE"    },
    { "iso_10646",      "UCS-2BE"    },
    { "iso10646",       "UCS-2BE"    },

    { "ks_c_5601-1987", "EUC-KR"     },

    /* FIXME: Japanese/Korean/Chinese stuff needs checking */
    { "euckr-0",        "EUC-KR"     },
    { "5601",           "EUC-KR"     },
    { "zh_TW-euc",      "EUC-TW"     },
    { "zh_CN.euc",      "gb2312"     },
    { "zh_TW-big5",     "BIG5"       },
    { "euc-cn",         "gb2312"     },
    { "big5-0",         "BIG5"       },
    { "big5.eten-0",    "BIG5"       },
    { "big5hkscs-0",    "BIG5HKSCS"  },
    { "gb2312-0",       "gb2312"     },
    { "gb2312.1980-0",  "gb2312"     },
    { "gb-2312",        "gb2312"     },
    { "gb18030-0",      "gb18030"    },
    { "gbk-0",          "GBK"        },

    { "eucjp-0",        "eucJP"      },
    { "ujis-0",         "ujis"       },
    { "jisx0208.1983-0","SJIS"       },
    { "jisx0212.1990-0","SJIS"       },
    { "pck",        "SJIS"       },
    { NULL,             NULL         }
};



/* Another copy of this trivial list implementation
   Why?  This stuff gets called a lot (potentially), should run fast,
   and g_list's are f@@#$ed up to make this a hassle */
static void e_dlist_init(EDList *v)
{
        v->head = (EDListNode *)&v->tail;
        v->tail = 0;
        v->tailpred = (EDListNode *)&v->head;
}

static EDListNode *e_dlist_addhead(EDList *l, EDListNode *n)
{
        n->next = l->head;
        n->prev = (EDListNode *)&l->head;
        l->head->prev = n;
        l->head = n;
        return n;
}

static EDListNode *e_dlist_addtail(EDList *l, EDListNode *n)
{
        n->next = (EDListNode *)&l->tail;
        n->prev = l->tailpred;
        l->tailpred->next = n;
        l->tailpred = n;
        return n;
}

static EDListNode *e_dlist_remove(EDListNode *n)
{
        n->next->prev = n->prev;
        n->prev->next = n->next;
        return n;
}


/* fucking glib... */
static const char *
e_strdown (char *str)
{
    register char *s = str;
    
    while (*s) {
        if (*s >= 'A' && *s <= 'Z')
            *s += 0x20;
        s++;
    }
    
    return str;
}

static const char *
e_strup (char *str)
{
    register char *s = str;
    
    while (*s) {
        if (*s >= 'a' && *s <= 'z')
            *s -= 0x20;
        s++;
    }
    
    return str;
}


static void
locale_parse_lang (const char *locale)
{
    char *codeset, *lang;
    
    if ((codeset = strchr (locale, '.')))
        lang = g_strndup (locale, codeset - locale);
    else
        lang = g_strdup (locale);
    
    /* validate the language */
    if (strlen (lang) >= 2) {
        if (lang[2] == '-' || lang[2] == '_') {
            /* canonicalise the lang */
            e_strdown (lang);
            
            /* validate the country code */
            if (strlen (lang + 3) > 2) {
                /* invalid country code */
                lang[2] = '\0';
            } else {
                lang[2] = '-';
                e_strup (lang + 3);
            }
        } else if (lang[2] != '\0') {
            /* invalid language */
            g_free (lang);
            lang = NULL;
        }
        
        locale_lang = lang;
    } else {
        /* invalid language */
        locale_lang = NULL;
        g_free (lang);
    }
}

/* NOTE: Owns the lock on return if keep is TRUE ! */
static void
e_iconv_init(int keep)
{
    char *from, *to, *locale;
    int i;

    LOCK();

    if (iconv_charsets != NULL) {
        if (!keep)
            UNLOCK();
        return;
    }

    iconv_charsets = g_hash_table_new(g_str_hash, g_str_equal);
    
    for (i = 0; known_iconv_charsets[i].charset != NULL; i++) {
        from = g_strdup(known_iconv_charsets[i].charset);
        to = g_strdup(known_iconv_charsets[i].iconv_name);
        e_strdown (from);
        g_hash_table_insert(iconv_charsets, from, to);
    }

    e_dlist_init(&iconv_cache_list);
    iconv_cache = g_hash_table_new(g_str_hash, g_str_equal);
    iconv_cache_open = g_hash_table_new(NULL, NULL);
    
    locale = setlocale (LC_ALL, NULL);
    
    if (!locale || !strcmp (locale, "C") || !strcmp (locale, "POSIX")) {
        /* The locale "C"  or  "POSIX"  is  a  portable  locale;  its
         * LC_CTYPE  part  corresponds  to  the 7-bit ASCII character
         * set.
         */
        
        locale_charset = NULL;
        locale_lang = NULL;
    } else {
#ifdef HAVE_CODESET
        locale_charset = g_strdup (nl_langinfo (CODESET));
        e_strdown (locale_charset);
#else
        /* A locale name is typically of  the  form  language[_terri-
         * tory][.codeset][@modifier],  where  language is an ISO 639
         * language code, territory is an ISO 3166 country code,  and
         * codeset  is  a  character  set or encoding identifier like
         * ISO-8859-1 or UTF-8.
         */
        char *codeset, *p;
        
        codeset = strchr (locale, '.');
        if (codeset) {
            codeset++;
            
            /* ; is a hack for debian systems and / is a hack for Solaris systems */
            for (p = codeset; *p && !strchr ("@;/", *p); p++);
            locale_charset = g_strndup (codeset, p - codeset);
            e_strdown (locale_charset);
        } else {
            /* charset unknown */
            locale_charset = NULL;
        }
#endif      
        
        /* parse the locale lang */
        locale_parse_lang (locale);

    }

    if (!keep)
        UNLOCK();
}

const char *e_iconv_charset_name(const char *charset)
{
    char *name, *ret, *tmp;

    if (charset == NULL)
        return NULL;

    name = g_alloca (strlen (charset) + 1);
    strcpy (name, charset);
    e_strdown (name);
    
    e_iconv_init(TRUE);
    ret = g_hash_table_lookup(iconv_charsets, name);
    if (ret != NULL) {
        UNLOCK();
        return ret;
    }

    /* Unknown, try canonicalise some basic charset types to something that should work */
    if (strncmp(name, "iso", 3) == 0) {
        /* Convert iso-nnnn-n or isonnnn-n or iso_nnnn-n to iso-nnnn-n or isonnnn-n */
        int iso, codepage;
        char *p;
        
        tmp = name + 3;
        if (*tmp == '-' || *tmp == '_')
            tmp++;
        
        iso = strtoul (tmp, &p, 10);
        
        if (iso == 10646) {
            /* they all become ICONV_10646 */
            ret = g_strdup (ICONV_10646);
        } else {
            tmp = p;
            if (*tmp == '-' || *tmp == '_')
                tmp++;
            
            codepage = strtoul (tmp, &p, 10);
            
            if (p > tmp) {
                /* codepage is numeric */
#ifdef __aix__
                if (codepage == 13)
                    ret = g_strdup ("IBM-921");
                else
#endif /* __aix__ */
                    ret = g_strdup_printf (ICONV_ISO_D_FORMAT, iso, codepage);
            } else {
                /* codepage is a string - probably iso-2022-jp or something */
                ret = g_strdup_printf (ICONV_ISO_S_FORMAT, iso, p);
            }
        }
    } else if (strncmp(name, "windows-", 8) == 0) {
        /* Convert windows-nnnnn or windows-cpnnnnn to cpnnnn */
        tmp = name+8;
        if (!strncmp(tmp, "cp", 2))
            tmp+=2;
        ret = g_strdup_printf("CP%s", tmp);
    } else if (strncmp(name, "microsoft-", 10) == 0) {
        /* Convert microsoft-nnnnn or microsoft-cpnnnnn to cpnnnn */
        tmp = name+10;
        if (!strncmp(tmp, "cp", 2))
            tmp+=2;
        ret = g_strdup_printf("CP%s", tmp); 
    } else {
        /* Just assume its ok enough as is, case and all */
        ret = g_strdup(charset);
    }

    g_hash_table_insert(iconv_charsets, g_strdup(name), ret);
    UNLOCK();

    return ret;
}

static void
flush_entry(struct _iconv_cache *ic)
{
    struct _iconv_cache_node *in, *nn;

    in = (struct _iconv_cache_node *)ic->open.head;
    nn = in->next;
    while (nn) {
        if (in->ip != (iconv_t)-1) {
            g_hash_table_remove(iconv_cache_open, in->ip);
            iconv_close(in->ip);
        }
        g_free(in);
        in = nn;
        nn = in->next;
    }
    g_free(ic->conv);
    g_free(ic);
}

/* This should run pretty quick, its called a lot */
iconv_t e_iconv_open(const char *oto, const char *ofrom)
{
    const char *to, *from;
    char *tofrom;
    struct _iconv_cache *ic;
    struct _iconv_cache_node *in;
    int errnosav;
    iconv_t ip;

    if (oto == NULL || ofrom == NULL) {
        errno = EINVAL;
        return (iconv_t) -1;
    }
    
    to = e_iconv_charset_name (oto);
    from = e_iconv_charset_name (ofrom);
    tofrom = g_alloca (strlen (to) + strlen (from) + 2);
    sprintf(tofrom, "%s%%%s", to, from);

    LOCK();

    ic = g_hash_table_lookup(iconv_cache, tofrom);
    if (ic) {
        e_dlist_remove((EDListNode *)ic);
    } else {
        struct _iconv_cache *last = (struct _iconv_cache *)iconv_cache_list.tailpred;
        struct _iconv_cache *prev;

        prev = last->prev;
        while (prev && iconv_cache_size > E_ICONV_CACHE_SIZE) {
            in = (struct _iconv_cache_node *)last->open.head;
            if (in->next && !in->busy) {
                cd(printf("Flushing iconv converter '%s'\n", last->conv));
                e_dlist_remove((EDListNode *)last);
                g_hash_table_remove(iconv_cache, last->conv);
                flush_entry(last);
                iconv_cache_size--;
            }
            last = prev;
            prev = last->prev;
        }

        iconv_cache_size++;
        
        ic = g_malloc(sizeof(*ic));
        e_dlist_init(&ic->open);
        ic->conv = g_strdup(tofrom);
        g_hash_table_insert(iconv_cache, ic->conv, ic);

        cd(printf("Creating iconv converter '%s'\n", ic->conv));
    }
    e_dlist_addhead(&iconv_cache_list, (EDListNode *)ic);

    /* If we have a free iconv, use it */
    in = (struct _iconv_cache_node *)ic->open.tailpred;
    if (in->prev && !in->busy) {
        cd(printf("using existing iconv converter '%s'\n", ic->conv));
        ip = in->ip;
        if (ip != (iconv_t)-1) {
            /* work around some broken iconv implementations 
             * that die if the length arguments are NULL 
             */
            size_t buggy_iconv_len = 0;
            char *buggy_iconv_buf = NULL;

            /* resets the converter */
            iconv(ip, &buggy_iconv_buf, &buggy_iconv_len, &buggy_iconv_buf, &buggy_iconv_len);
            in->busy = TRUE;
            e_dlist_remove((EDListNode *)in);
            e_dlist_addhead(&ic->open, (EDListNode *)in);
        }
    } else {
        cd(printf("creating new iconv converter '%s'\n", ic->conv));
        ip = iconv_open(to, from);
        in = g_malloc(sizeof(*in));
        in->ip = ip;
        in->parent = ic;
        e_dlist_addhead(&ic->open, (EDListNode *)in);
        if (ip != (iconv_t)-1) {
            g_hash_table_insert(iconv_cache_open, ip, in);
            in->busy = TRUE;
        } else {
            errnosav = errno;
            g_warning("Could not open converter for '%s' to '%s' charset", from, to);
            in->busy = FALSE;
            errno = errnosav;
        }
    }

    UNLOCK();

    return ip;
}

size_t e_iconv(iconv_t cd, const char **inbuf, size_t *inbytesleft, char ** outbuf, size_t *outbytesleft)
{
    return iconv(cd, (char **) inbuf, inbytesleft, outbuf, outbytesleft);
}

void
e_iconv_close(iconv_t ip)
{
    struct _iconv_cache_node *in;

    if (ip == (iconv_t)-1)
        return;

    LOCK();
    in = g_hash_table_lookup(iconv_cache_open, ip);
    if (in) {
        cd(printf("closing iconv converter '%s'\n", in->parent->conv));
        e_dlist_remove((EDListNode *)in);
        in->busy = FALSE;
        e_dlist_addtail(&in->parent->open, (EDListNode *)in);
    } else {
        g_warning("trying to close iconv i dont know about: %p", ip);
        iconv_close(ip);
    }
    UNLOCK();

}

const char *e_iconv_locale_charset(void)
{
    e_iconv_init(FALSE);

    return locale_charset;
}


const char *
e_iconv_locale_language (void)
{
    e_iconv_init (FALSE);
    
    return locale_lang;
}

/* map CJKR charsets to their language code */
/* NOTE: only support charset names that will be returned by
 * e_iconv_charset_name() so that we don't have to keep track of all
 * the aliases too. */
static struct {
    char *charset;
    char *lang;
} cjkr_lang_map[] = {
    { "Big5",        "zh" },
    { "BIG5HKSCS",   "zh" },
    { "gb2312",      "zh" },
    { "gb18030",     "zh" },
    { "gbk",         "zh" },
    { "euc-tw",      "zh" },
    { "iso-2022-jp", "ja" },
    { "sjis",        "ja" },
    { "ujis",        "ja" },
    { "eucJP",       "ja" },
    { "euc-jp",      "ja" },
    { "euc-kr",      "ko" },
    { "koi8-r",      "ru" },
    { "koi8-u",      "uk" }
};

#define NUM_CJKR_LANGS (sizeof (cjkr_lang_map) / sizeof (cjkr_lang_map[0]))

const char *
e_iconv_charset_language (const char *charset)
{
    int i;
    
    if (!charset)
        return NULL;
    
    charset = e_iconv_charset_name (charset);
    for (i = 0; i < NUM_CJKR_LANGS; i++) {
        if (!strcasecmp (cjkr_lang_map[i].charset, charset))
            return cjkr_lang_map[i].lang;
    }
    
    return NULL;
}