charcnv.c

samba-3.0.22.tar.gz 编译smb服务器的源码

/* 
   Unix SMB/CIFS implementation.
   Character set conversion Extensions
   Copyright (C) Igor Vergeichik <[e-mail]> 2001
   Copyright (C) Andrew Tridgell 2001
   Copyright (C) Simo Sorce 2001
   Copyright (C) Martin Pool 2003
   
   This program is free software; you can redistribute it and/or modify
   it under the terms of the GNU General Public License as published by
   the Free Software Foundation; either version 2 of the License, or
   (at your option) any later version.
   
   This program is distributed in the hope that it will be useful,
   but WITHOUT ANY WARRANTY; without even the implied warranty of
   MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
   GNU General Public License for more details.
   
   You should have received a copy of the GNU General Public License
   along with this program; if not, write to the Free Software
   Foundation, Inc., 675 Mass Ave, Cambridge, MA 02139, USA.

*/
#include "includes.h"

/* We can parameterize this if someone complains.... JRA. */

char lp_failed_convert_char(void)
{
	return '_';
}

/**
 * @file
 *
 * @brief Character-set conversion routines built on our iconv.
 * 
 * @note Samba's internal character set (at least in the 3.0 series)
 * is always the same as the one for the Unix filesystem.  It is
 * <b>not</b> necessarily UTF-8 and may be different on machines that
 * need i18n filenames to be compatible with Unix software.  It does
 * have to be a superset of ASCII.  All multibyte sequences must start
 * with a byte with the high bit set.
 *
 * @sa lib/iconv.c
 */


static smb_iconv_t conv_handles[NUM_CHARSETS][NUM_CHARSETS];
static BOOL conv_silent; /* Should we do a debug if the conversion fails ? */

/**
 * Return the name of a charset to give to iconv().
 **/
static const char *charset_name(charset_t ch)
{
	const char *ret = NULL;

	if (ch == CH_UCS2) ret = "UTF-16LE";
	else if (ch == CH_UNIX) ret = lp_unix_charset();
	else if (ch == CH_DOS) ret = lp_dos_charset();
	else if (ch == CH_DISPLAY) ret = lp_display_charset();
	else if (ch == CH_UTF8) ret = "UTF8";

#if defined(HAVE_NL_LANGINFO) && defined(CODESET)
	if (ret && !strcmp(ret, "LOCALE")) {
		const char *ln = NULL;

#ifdef HAVE_SETLOCALE
		setlocale(LC_ALL, "");
#endif
		ln = nl_langinfo(CODESET);
		if (ln) {
			/* Check whether the charset name is supported
			   by iconv */
			smb_iconv_t handle = smb_iconv_open(ln,"UCS-2LE");
			if (handle == (smb_iconv_t) -1) {
				DEBUG(5,("Locale charset '%s' unsupported, using ASCII instead\n", ln));
				ln = NULL;
			} else {
				DEBUG(5,("Substituting charset '%s' for LOCALE\n", ln));
				smb_iconv_close(handle);
			}
		}
		ret = ln;
	}
#ifdef HAVE_SETLOCALE
	/* We set back the locale to C to get ASCII-compatible toupper/lower functions.
	   For now we do not need any other POSIX localisations anyway. When we should
	   really need localized string functions one day we need to write our own
	   ascii_tolower etc.
	*/
	setlocale(LC_ALL, "C");
 #endif

#endif

	if (!ret || !*ret) ret = "ASCII";
	return ret;
}

void lazy_initialize_conv(void)
{
	static int initialized = False;

	if (!initialized) {
		initialized = True;
		load_case_tables();
		init_iconv();
	}
}

/**
 * Initialize iconv conversion descriptors.
 *
 * This is called the first time it is needed, and also called again
 * every time the configuration is reloaded, because the charset or
 * codepage might have changed.
 **/
void init_iconv(void)
{
	int c1, c2;
	BOOL did_reload = False;

	/* so that charset_name() works we need to get the UNIX<->UCS2 going
	   first */
	if (!conv_handles[CH_UNIX][CH_UCS2])
		conv_handles[CH_UNIX][CH_UCS2] = smb_iconv_open(charset_name(CH_UCS2), "ASCII");

	if (!conv_handles[CH_UCS2][CH_UNIX])
		conv_handles[CH_UCS2][CH_UNIX] = smb_iconv_open("ASCII", charset_name(CH_UCS2));

	for (c1=0;c1<NUM_CHARSETS;c1++) {
		for (c2=0;c2<NUM_CHARSETS;c2++) {
			const char *n1 = charset_name((charset_t)c1);
			const char *n2 = charset_name((charset_t)c2);
			if (conv_handles[c1][c2] &&
			    strcmp(n1, conv_handles[c1][c2]->from_name) == 0 &&
			    strcmp(n2, conv_handles[c1][c2]->to_name) == 0)
				continue;

			did_reload = True;

			if (conv_handles[c1][c2])
				smb_iconv_close(conv_handles[c1][c2]);

			conv_handles[c1][c2] = smb_iconv_open(n2,n1);
			if (conv_handles[c1][c2] == (smb_iconv_t)-1) {
				DEBUG(0,("init_iconv: Conversion from %s to %s not supported\n",
					 charset_name((charset_t)c1), charset_name((charset_t)c2)));
				if (c1 != CH_UCS2) {
					n1 = "ASCII";
				}
				if (c2 != CH_UCS2) {
					n2 = "ASCII";
				}
				DEBUG(0,("init_iconv: Attempting to replace with conversion from %s to %s\n",
					n1, n2 ));
				conv_handles[c1][c2] = smb_iconv_open(n2,n1);
				if (!conv_handles[c1][c2]) {
					DEBUG(0,("init_iconv: Conversion from %s to %s failed", n1, n2));
					smb_panic("init_iconv: conv_handle initialization failed.");
				}
			}
		}
	}

	if (did_reload) {
		/* XXX: Does this really get called every time the dos
		 * codepage changes? */
		/* XXX: Is the did_reload test too strict? */
		conv_silent = True;
		init_doschar_table();
		init_valid_table();
		conv_silent = False;
	}
}

/**
 * Convert string from one encoding to another, making error checking etc
 * Slow path version - uses (slow) iconv.
 *
 * @param src pointer to source string (multibyte or singlebyte)
 * @param srclen length of the source string in bytes
 * @param dest pointer to destination string (multibyte or singlebyte)
 * @param destlen maximal length allowed for string
 * @param allow_bad_conv determines if a "best effort" conversion is acceptable (never returns errors)
 * @returns the number of bytes occupied in the destination
 *
 * Ensure the srclen contains the terminating zero.
 *
 **/

static size_t convert_string_internal(charset_t from, charset_t to,
		      void const *src, size_t srclen, 
		      void *dest, size_t destlen, BOOL allow_bad_conv)
{
	size_t i_len, o_len;
	size_t retval;
	const char* inbuf = (const char*)src;
	char* outbuf = (char*)dest;
	smb_iconv_t descriptor;

	lazy_initialize_conv();

	descriptor = conv_handles[from][to];

	if (srclen == (size_t)-1) {
		if (from == CH_UCS2) {
			srclen = (strlen_w((const smb_ucs2_t *)src)+1) * 2;
		} else {
			srclen = strlen((const char *)src)+1;
		}
	}


	if (descriptor == (smb_iconv_t)-1 || descriptor == (smb_iconv_t)0) {
		if (!conv_silent)
			DEBUG(0,("convert_string_internal: Conversion not supported.\n"));
		return (size_t)-1;
	}

	i_len=srclen;
	o_len=destlen;

 again:

	retval = smb_iconv(descriptor, &inbuf, &i_len, &outbuf, &o_len);
	if(retval==(size_t)-1) {
	    	const char *reason="unknown error";
		switch(errno) {
			case EINVAL:
				reason="Incomplete multibyte sequence";
				if (!conv_silent)
					DEBUG(3,("convert_string_internal: Conversion error: %s(%s)\n",reason,inbuf));
				if (allow_bad_conv)
					goto use_as_is;
				break;
			case E2BIG:
				reason="No more room"; 
				if (!conv_silent) {
					if (from == CH_UNIX) {
						DEBUG(3,("E2BIG: convert_string(%s,%s): srclen=%u destlen=%u - '%s'\n",
							charset_name(from), charset_name(to),
							(unsigned int)srclen, (unsigned int)destlen, (const char *)src));
					} else {
						DEBUG(3,("E2BIG: convert_string(%s,%s): srclen=%u destlen=%u\n",
							charset_name(from), charset_name(to),
							(unsigned int)srclen, (unsigned int)destlen));
					}
				}
				break;
			case EILSEQ:
				reason="Illegal multibyte sequence";
				if (!conv_silent)
					DEBUG(3,("convert_string_internal: Conversion error: %s(%s)\n",reason,inbuf));
				if (allow_bad_conv)
					goto use_as_is;
				break;
			default:
				if (!conv_silent)
					DEBUG(0,("convert_string_internal: Conversion error: %s(%s)\n",reason,inbuf));
				break;
		}
		/* smb_panic(reason); */
	}
	return destlen-o_len;

 use_as_is:

	/* 
	 * Conversion not supported. This is actually an error, but there are so
	 * many misconfigured iconv systems and smb.conf's out there we can't just
	 * fail. Do a very bad conversion instead.... JRA.
	 */

	{
		if (o_len == 0 || i_len == 0)
			return destlen - o_len;

		if (from == CH_UCS2 && to != CH_UCS2) {
			/* Can't convert from ucs2 to multibyte. Replace with the default fail char. */
			if (i_len < 2)
				return destlen - o_len;
			if (i_len >= 2) {
				*outbuf = lp_failed_convert_char();

				outbuf++;
				o_len--;

				inbuf += 2;
				i_len -= 2;
			}

			if (o_len == 0 || i_len == 0)
				return destlen - o_len;

			/* Keep trying with the next char... */
			goto again;

		} else if (from != CH_UCS2 && to == CH_UCS2) {
			/* Can't convert to ucs2 - just widen by adding the default fail char then zero. */
			if (o_len < 2)
				return destlen - o_len;

			outbuf[0] = lp_failed_convert_char();
			outbuf[1] = '\0';

			inbuf++;
			i_len--;

			outbuf += 2;
			o_len -= 2;

			if (o_len == 0 || i_len == 0)
				return destlen - o_len;

			/* Keep trying with the next char... */
			goto again;

		} else if (from != CH_UCS2 && to != CH_UCS2) {
			/* Failed multibyte to multibyte. Just copy the default fail char and
				try again. */
			outbuf[0] = lp_failed_convert_char();

			inbuf++;
			i_len--;

			outbuf++;
			o_len--;

			if (o_len == 0 || i_len == 0)
				return destlen - o_len;

			/* Keep trying with the next char... */
			goto again;

		} else {
			/* Keep compiler happy.... */
			return destlen - o_len;
		}
	}
}

/**
 * Convert string from one encoding to another, making error checking etc
 * Fast path version - handles ASCII first.
 *
 * @param src pointer to source string (multibyte or singlebyte)
 * @param srclen length of the source string in bytes, or -1 for nul terminated.
 * @param dest pointer to destination string (multibyte or singlebyte)
 * @param destlen maximal length allowed for string - *NEVER* -1.
 * @param allow_bad_conv determines if a "best effort" conversion is acceptable (never returns errors)
 * @returns the number of bytes occupied in the destination
 *
 * Ensure the srclen contains the terminating zero.
 *
 * This function has been hand-tuned to provide a fast path.
 * Don't change unless you really know what you are doing. JRA.
 **/

size_t convert_string(charset_t from, charset_t to,
		      void const *src, size_t srclen, 
		      void *dest, size_t destlen, BOOL allow_bad_conv)
{
	/*
	 * NB. We deliberately don't do a strlen here if srclen == -1.
	 * This is very expensive over millions of calls and is taken
	 * care of in the slow path in convert_string_internal. JRA.
	 */

#ifdef DEVELOPER
	SMB_ASSERT(destlen != (size_t)-1);
#endif

	if (srclen == 0)
		return 0;

	if (from != CH_UCS2 && to != CH_UCS2) {
		const unsigned char *p = (const unsigned char *)src;
		unsigned char *q = (unsigned char *)dest;
		size_t slen = srclen;
		size_t dlen = destlen;
		unsigned char lastp = '\0';
		size_t retval = 0;

		/* If all characters are ascii, fast path here. */
		while (slen && dlen) {
			if ((lastp = *p) <= 0x7f) {
				*q++ = *p++;
				if (slen != (size_t)-1) {
					slen--;
				}
				dlen--;
				retval++;
				if (!lastp)
					break;
			} else {
#ifdef BROKEN_UNICODE_COMPOSE_CHARACTERS
				goto general_case;
#else
				return retval + convert_string_internal(from, to, p, slen, q, dlen, allow_bad_conv);
#endif
			}
		}
		if (!dlen) {
			/* Even if we fast path we should note if we ran out of room. */
			if (((slen != (size_t)-1) && slen) ||
					((slen == (size_t)-1) && lastp)) {
				errno = E2BIG;
			}
		}
		return retval;
	} else if (from == CH_UCS2 && to != CH_UCS2) {
		const unsigned char *p = (const unsigned char *)src;
		unsigned char *q = (unsigned char *)dest;
		size_t retval = 0;
		size_t slen = srclen;
		size_t dlen = destlen;
		unsigned char lastp = '\0';

		/* If all characters are ascii, fast path here. */
		while (((slen == (size_t)-1) || (slen >= 2)) && dlen) {
			if (((lastp = *p) <= 0x7f) && (p[1] == 0)) {
				*q++ = *p;
				if (slen != (size_t)-1) {
					slen -= 2;
				}
				p += 2;
				dlen--;
				retval++;
				if (!lastp)
					break;
			} else {
#ifdef BROKEN_UNICODE_COMPOSE_CHARACTERS
				goto general_case;
#else
				return retval + convert_string_internal(from, to, p, slen, q, dlen, allow_bad_conv);
#endif
			}
		}
		if (!dlen) {
			/* Even if we fast path we should note if we ran out of room. */
			if (((slen != (size_t)-1) && slen) ||
					((slen == (size_t)-1) && lastp)) {
				errno = E2BIG;
			}
		}
		return retval;
	} else if (from != CH_UCS2 && to == CH_UCS2) {
		const unsigned char *p = (const unsigned char *)src;
		unsigned char *q = (unsigned char *)dest;
		size_t retval = 0;
		size_t slen = srclen;
		size_t dlen = destlen;
		unsigned char lastp = '\0';

		/* If all characters are ascii, fast path here. */
		while (slen && (dlen >= 2)) {
			if ((lastp = *p) <= 0x7F) {
				*q++ = *p++;
				*q++ = '\0';
				if (slen != (size_t)-1) {
					slen--;
				}
				dlen -= 2;