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().
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;
+ }
+#endif
+
if (!ret || !*ret) ret = "ASCII";
return ret;
}
/* 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
* @returns the number of bytes occupied in the destination
+ *
+ * Ensure the srclen contains the terminating zero.
+ *
**/
-size_t convert_string(charset_t from, charset_t to,
+
+static size_t convert_string_internal(charset_t from, charset_t to,
void const *src, size_t srclen,
void *dest, size_t destlen)
{
char* outbuf = (char*)dest;
smb_iconv_t descriptor;
- if (srclen == (size_t)-1)
- srclen = strlen(src)+1;
-
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) {
- /* conversion not supported, use as is */
- size_t len = MIN(srclen,destlen);
- memcpy(dest,src,len);
- return len;
+ if (!conv_silent)
+ DEBUG(0,("convert_string_internal: Conversion not supported.\n"));
+ goto use_as_is;
}
i_len=srclen;
o_len=destlen;
- retval = smb_iconv(descriptor, &inbuf, &i_len, &outbuf, &o_len);
+ retval = smb_iconv(descriptor, (char **)&inbuf, &i_len, &outbuf, &o_len);
if(retval==(size_t)-1) {
const char *reason="unknown error";
switch(errno) {
case EINVAL:
reason="Incomplete multibyte sequence";
- break;
+ if (!conv_silent)
+ DEBUG(3,("convert_string_internal: Conversion error: %s(%s)\n",reason,inbuf));
+ goto use_as_is;
case E2BIG:
reason="No more room";
- DEBUG(0, ("convert_string: Required %d, available %d\n",
- srclen, destlen));
+ if (!conv_silent)
+ DEBUG(3, ("convert_string_internal: Required %lu, available %lu\n",
+ (unsigned long)srclen, (unsigned long)destlen));
/* we are not sure we need srclen bytes,
may be more, may be less.
We only know we need more than destlen
bytes ---simo */
break;
case EILSEQ:
- reason="Illegal multibyte sequence";
- break;
+ reason="Illegal multibyte sequence";
+ if (!conv_silent)
+ DEBUG(3,("convert_string_internal: Conversion error: %s(%s)\n",reason,inbuf));
+ goto use_as_is;
+ 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, use as is */
+ {
+ size_t len = MIN(srclen,destlen);
+ if (len)
+ memcpy(dest,src,len);
+ return 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
+ * @param dest pointer to destination string (multibyte or singlebyte)
+ * @param destlen maximal length allowed for string
+ * @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)
+{
+ /*
+ * NB. We deliberately don't do a strlen here is srclen == -1.
+ * This is very expensive over millions of calls and is taken
+ * care of in the slow path in convert_string_internal. JRA.
+ */
+
+ 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;
+ 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);
+#endif
+ }
+ }
+ 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;
+
+ /* If all characters are ascii, fast path here. */
+ while ((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);
+#endif
+ }
+ }
+ 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;
+
+ /* 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;
+ retval += 2;
+ if (!lastp)
+ break;
+ } else {
+#ifdef BROKEN_UNICODE_COMPOSE_CHARACTERS
+ goto general_case;
+#else
+ return retval + convert_string_internal(from, to, p, slen, q, dlen);
+#endif
+ }
+ }
+ return retval;
+ }
+
+#ifdef BROKEN_UNICODE_COMPOSE_CHARACTERS
+ general_case:
+#endif
+ return convert_string_internal(from, to, src, srclen, dest, destlen);
}
/**
* Convert between character sets, allocating a new buffer for the result.
*
+ * @param ctx TALLOC_CTX to use to allocate with. If NULL use malloc.
* @param srclen length of source buffer.
* @param dest always set at least to NULL
* @note -1 is not accepted for srclen.
*
* @returns Size in bytes of the converted string; or -1 in case of error.
+ *
+ * Ensure the srclen contains the terminating zero.
**/
-size_t convert_string_allocate(charset_t from, charset_t to,
+size_t convert_string_allocate(TALLOC_CTX *ctx, charset_t from, charset_t to,
void const *src, size_t srclen, void **dest)
{
- size_t i_len, o_len, destlen;
+ size_t i_len, o_len, destlen = MAX(srclen, 512);
size_t retval;
const char *inbuf = (const char *)src;
- char *outbuf, *ob;
+ char *outbuf = NULL, *ob = NULL;
smb_iconv_t descriptor;
*dest = NULL;
if (src == NULL || srclen == (size_t)-1)
return (size_t)-1;
+ if (srclen == 0)
+ return 0;
lazy_initialize_conv();
descriptor = conv_handles[from][to];
if (descriptor == (smb_iconv_t)-1 || descriptor == (smb_iconv_t)0) {
- /* conversion not supported, return -1*/
- DEBUG(3, ("convert_string_allocate: conversion not supported!\n"));
- return -1;
+ if (!conv_silent)
+ DEBUG(0,("convert_string_allocate: Conversion not supported.\n"));
+ goto use_as_is;
}
- destlen = MAX(srclen, 512);
- outbuf = NULL;
convert:
- destlen = destlen * 2;
- ob = (char *)Realloc(outbuf, destlen);
+ if ((destlen*2) < destlen) {
+ /* wrapped ! abort. */
+ if (!conv_silent)
+ DEBUG(0, ("convert_string_allocate: destlen wrapped !\n"));
+ if (!ctx)
+ SAFE_FREE(outbuf);
+ return (size_t)-1;
+ } else {
+ destlen = destlen * 2;
+ }
+
+ if (ctx)
+ ob = (char *)talloc_realloc(ctx, ob, destlen);
+ else
+ ob = (char *)Realloc(ob, destlen);
+
if (!ob) {
DEBUG(0, ("convert_string_allocate: realloc failed!\n"));
- SAFE_FREE(outbuf);
+ if (!ctx)
+ SAFE_FREE(outbuf);
return (size_t)-1;
} else {
outbuf = ob;
i_len = srclen;
o_len = destlen;
retval = smb_iconv(descriptor,
- &inbuf, &i_len,
+ (char **)&inbuf, &i_len,
&outbuf, &o_len);
if(retval == (size_t)-1) {
const char *reason="unknown error";
switch(errno) {
case EINVAL:
reason="Incomplete multibyte sequence";
- break;
+ if (!conv_silent)
+ DEBUG(3,("convert_string_allocate: Conversion error: %s(%s)\n",reason,inbuf));
+ goto use_as_is;
case E2BIG:
goto convert;
case EILSEQ:
reason="Illegal multibyte sequence";
- break;
+ if (!conv_silent)
+ DEBUG(3,("convert_string_allocate: Conversion error: %s(%s)\n",reason,inbuf));
+ goto use_as_is;
}
- DEBUG(0,("Conversion error: %s(%s)\n",reason,inbuf));
+ if (!conv_silent)
+ DEBUG(0,("Conversion error: %s(%s)\n",reason,inbuf));
/* smb_panic(reason); */
return (size_t)-1;
}
destlen = destlen - o_len;
- *dest = (char *)Realloc(ob,destlen);
+ if (ctx)
+ *dest = (char *)talloc_realloc(ctx,ob,destlen);
+ else
+ *dest = (char *)Realloc(ob,destlen);
if (destlen && !*dest) {
DEBUG(0, ("convert_string_allocate: out of memory!\n"));
- SAFE_FREE(ob);
+ if (!ctx)
+ SAFE_FREE(ob);
return (size_t)-1;
}
return destlen;
-}
+ use_as_is:
+
+ /* conversion not supported, use as is */
+ {
+ if (srclen && (destlen != srclen)) {
+ destlen = srclen;
+ if (ctx)
+ ob = (char *)talloc_realloc(ctx, ob, destlen);
+ else
+ ob = (char *)Realloc(ob, destlen);
+ if (!ob) {
+ DEBUG(0, ("convert_string_allocate: realloc failed!\n"));
+ if (!ctx)
+ SAFE_FREE(outbuf);
+ return (size_t)-1;
+ }
+ }
+ if (srclen && ob)
+ memcpy(ob,(const char *)src,srclen);
+ *dest = (char *)ob;
+ return srclen;
+ }
+}
/**
* Convert between character sets, allocating a new buffer using talloc for the result.
static size_t convert_string_talloc(TALLOC_CTX *ctx, charset_t from, charset_t to,
void const *src, size_t srclen, void **dest)
{
- void *alloced_string;
size_t dest_len;
- /* FIXME: Ridiculous to allocate two buffers and then copy the string! */
-
*dest = NULL;
- dest_len=convert_string_allocate(from, to, src, srclen, &alloced_string);
+ dest_len=convert_string_allocate(ctx, from, to, src, srclen, dest);
if (dest_len == (size_t)-1)
return (size_t)-1;
- *dest = talloc_memdup(ctx, alloced_string, dest_len);
- SAFE_FREE(alloced_string);
if (*dest == NULL)
return (size_t)-1;
return dest_len;
size_t size;
smb_ucs2_t *buffer;
- size = convert_string_allocate(CH_UNIX, CH_UCS2, src, srclen,
- (void **) &buffer);
+ size = push_ucs2_allocate(&buffer, src);
if (size == -1) {
smb_panic("failed to create UCS2 buffer");
}
return size;
}
+/**
+ strdup() a unix string to upper case.
+ Max size is pstring.
+**/
+
+char *strdup_upper(const char *s)
+{
+ pstring out_buffer;
+ const unsigned char *p = (const unsigned char *)s;
+ unsigned char *q = (unsigned char *)out_buffer;
+
+ /* this is quite a common operation, so we want it to be
+ fast. We optimise for the ascii case, knowing that all our
+ supported multi-byte character sets are ascii-compatible
+ (ie. they match for the first 128 chars) */
+
+ while (1) {
+ if (*p & 0x80)
+ break;
+ *q++ = toupper(*p);
+ if (!*p)
+ break;
+ p++;
+ if (p - ( const unsigned char *)s >= sizeof(pstring))
+ break;
+ }
+
+ if (*p) {
+ /* MB case. */
+ size_t size;
+ wpstring buffer;
+ size = convert_string(CH_UNIX, CH_UCS2, s, -1, buffer, sizeof(buffer));
+ if (size == -1) {
+ return NULL;
+ }
+
+ strupper_w(buffer);
+
+ size = convert_string(CH_UCS2, CH_UNIX, buffer, -1, out_buffer, sizeof(out_buffer));
+ if (size == -1) {
+ return NULL;
+ }
+ }
+
+ return strdup(out_buffer);
+}
+
size_t unix_strlower(const char *src, size_t srclen, char *dest, size_t destlen)
{
size_t size;
- smb_ucs2_t *buffer;
+ smb_ucs2_t *buffer = NULL;
- size = convert_string_allocate(CH_UNIX, CH_UCS2, src, srclen,
+ size = convert_string_allocate(NULL, CH_UNIX, CH_UCS2, src, srclen,
(void **) &buffer);
- if (size == -1) {
+ if (size == -1 || !buffer) {
smb_panic("failed to create UCS2 buffer");
}
if (!strlower_w(buffer) && (dest == src)) {
- free(buffer);
+ SAFE_FREE(buffer);
return srclen;
}
size = convert_string(CH_UCS2, CH_UNIX, buffer, size, dest, destlen);
- free(buffer);
+ SAFE_FREE(buffer);
return size;
}
+/**
+ strdup() a unix string to lower case.
+**/
+
+char *strdup_lower(const char *s)
+{
+ size_t size;
+ smb_ucs2_t *buffer = NULL;
+ char *out_buffer;
+
+ size = push_ucs2_allocate(&buffer, s);
+ if (size == -1 || !buffer) {
+ return NULL;
+ }
+
+ strlower_w(buffer);
+
+ size = pull_ucs2_allocate(&out_buffer, buffer);
+ SAFE_FREE(buffer);
+
+ if (size == -1) {
+ return NULL;
+ }
+
+ return out_buffer;
+}
static size_t ucs2_align(const void *base_ptr, const void *p, int flags)
{
if (flags & STR_UPPER) {
pstrcpy(tmpbuf, src);
- strupper(tmpbuf);
+ strupper_m(tmpbuf);
src = tmpbuf;
}
- if (flags & STR_TERMINATE)
+ if (flags & (STR_TERMINATE | STR_TERMINATE_ASCII))
src_len++;
return convert_string(CH_UNIX, CH_DOS, src, src_len, dest, dest_len);
return push_ascii(dest, src, sizeof(pstring), STR_TERMINATE);
}
+size_t push_ascii_nstring(void *dest, const char *src)
+{
+ return push_ascii(dest, src, sizeof(nstring), STR_TERMINATE);
+}
+
/**
* Copy a string from a dos codepage source to a unix char* destination.
*
return pull_ascii(dest, src, sizeof(fstring), -1, STR_TERMINATE);
}
+size_t pull_ascii_nstring(char *dest, const void *src)
+{
+ return pull_ascii(dest, src, sizeof(nstring), sizeof(nstring), STR_TERMINATE);
+}
+
/**
* Copy a string from a char* src to a unicode destination.
*
* @param dest_len is the maximum length allowed in the
* destination. If dest_len is -1 then no maxiumum is used.
**/
+
size_t push_ucs2(const void *base_ptr, void *dest, const char *src, size_t dest_len, int flags)
{
size_t len=0;
- size_t src_len = strlen(src);
- pstring tmpbuf;
+ size_t src_len;
/* treat a pstring as "unlimited" length */
if (dest_len == (size_t)-1)
dest_len = sizeof(pstring);
- if (flags & STR_UPPER) {
- pstrcpy(tmpbuf, src);
- strupper(tmpbuf);
- src = tmpbuf;
- }
-
if (flags & STR_TERMINATE)
- src_len++;
+ src_len = (size_t)-1;
+ else
+ src_len = strlen(src);
if (ucs2_align(base_ptr, dest, flags)) {
*(char *)dest = 0;
dest = (void *)((char *)dest + 1);
- if (dest_len) dest_len--;
+ if (dest_len)
+ dest_len--;
len++;
}
dest_len &= ~1;
len += convert_string(CH_UNIX, CH_UCS2, src, src_len, dest, dest_len);
+
+ if (flags & STR_UPPER) {
+ smb_ucs2_t *dest_ucs2 = dest;
+ size_t i;
+ for (i = 0; i < (dest_len / 2) && dest_ucs2[i]; i++) {
+ smb_ucs2_t v = toupper_w(dest_ucs2[i]);
+ if (v != dest_ucs2[i]) {
+ dest_ucs2[i] = v;
+ }
+ }
+ }
+
return len;
}
size_t src_len = strlen(src)+1;
*dest = NULL;
- return convert_string_allocate(CH_UNIX, CH_UCS2, src, src_len, (void **)dest);
+ return convert_string_allocate(NULL, CH_UNIX, CH_UCS2, src, src_len, (void **)dest);
}
/**
if (flags & STR_UPPER) {
pstrcpy(tmpbuf, src);
- strupper(tmpbuf);
+ strupper_m(tmpbuf);
src = tmpbuf;
}
size_t src_len = strlen(src)+1;
*dest = NULL;
- return convert_string_allocate(CH_UNIX, CH_UTF8, src, src_len, (void **)dest);
+ return convert_string_allocate(NULL, CH_UNIX, CH_UTF8, src, src_len, (void **)dest);
}
/**
}
if (flags & STR_TERMINATE) {
- if (src_len == (size_t)-1) {
- src_len = strlen_w(src)*2 + 2;
- } else {
+ /* src_len -1 is the default for null terminated strings. */
+ if (src_len != (size_t)-1) {
size_t len = strnlen_w(src, src_len/2);
if (len < src_len/2)
len++;
src_len &= ~1;
ret = convert_string(CH_UCS2, CH_UNIX, src, src_len, dest, dest_len);
+
+ if (src_len == (size_t)-1)
+ src_len = ret*2;
+
if (dest_len)
dest[MIN(ret, dest_len-1)] = 0;
else
{
size_t src_len = (strlen_w(src)+1) * sizeof(smb_ucs2_t);
*dest = NULL;
- return convert_string_allocate(CH_UCS2, CH_UNIX, src, src_len, (void **)dest);
+ return convert_string_allocate(NULL, CH_UCS2, CH_UNIX, src, src_len, (void **)dest);
}
/**
* @returns The number of bytes occupied by the string in the destination
**/
-size_t pull_utf8_allocate(void **dest, const char *src)
+size_t pull_utf8_allocate(char **dest, const char *src)
{
size_t src_len = strlen(src)+1;
*dest = NULL;
- return convert_string_allocate(CH_UTF8, CH_UNIX, src, src_len, dest);
+ return convert_string_allocate(NULL, CH_UTF8, CH_UNIX, src, src_len, (void **)dest);
}
/**
size_t push_string_fn(const char *function, unsigned int line, const void *base_ptr, void *dest, const char *src, size_t dest_len, int flags)
{
+#ifdef DEVELOPER
+ /* We really need to zero fill here, not clobber
+ * region, as we want to ensure that valgrind thinks
+ * all of the outgoing buffer has been written to
+ * so a send() or write() won't trap an error.
+ * JRA.
+ */
+#if 0
if (dest_len != (size_t)-1)
clobber_region(function, line, dest, dest_len);
+#else
+ if (dest_len != (size_t)-1)
+ memset(dest, '\0', dest_len);
+#endif
+#endif
if (!(flags & STR_ASCII) && \
((flags & STR_UNICODE || \
size_t pull_string_fn(const char *function, unsigned int line, const void *base_ptr, char *dest, const void *src, size_t dest_len, size_t src_len, int flags)
{
+#ifdef DEVELOPER
if (dest_len != (size_t)-1)
clobber_region(function, line, dest, dest_len);
+#endif
if (!(flags & STR_ASCII) && \
((flags & STR_UNICODE || \
}
return 0;
}
-
-/**
- Convert from unix to ucs2 charset and return the
- allocated and converted string or NULL if an error occurred.
- You must provide a zero terminated string.
- The returning string will be zero terminated.
-**/
-
-smb_ucs2_t *acnv_uxu2(const char *src)
-{
- size_t slen;
- size_t dlen;
- void *dest;
-
- slen = strlen(src) + 1;
- dlen = convert_string_allocate(CH_UNIX, CH_UCS2, src, slen, &dest);
- if (dlen == (size_t)-1)
- return NULL;
- else
- return dest;
-}
-
-/**
- Convert from dos to ucs2 charset and return the
- allocated and converted string or NULL if an error occurred.
- You must provide a zero terminated string.
- The returning string will be zero terminated.
-**/
-
-smb_ucs2_t *acnv_dosu2(const char *src)
-{
- size_t slen;
- size_t dlen;
- void *dest;
-
- slen = strlen(src) + 1;
- dlen = convert_string_allocate(CH_DOS, CH_UCS2, src, slen, &dest);
- if (dlen == (size_t)-1)
- return NULL;
- else
- return dest;
-}
git.samba.org / samba.git / blobdiff