summaryrefslogtreecommitdiff
path: root/lib
diff options
context:
space:
mode:
Diffstat (limited to 'lib')
-rw-r--r--lib/util/charset/charcnv.c42
-rw-r--r--lib/util/charset/charset.h125
-rw-r--r--lib/util/charset/iconv.c1
-rw-r--r--lib/util/charset/tests/iconv.c11
-rw-r--r--lib/util/charset/util_unistr.c44
5 files changed, 169 insertions, 54 deletions
diff --git a/lib/util/charset/charcnv.c b/lib/util/charset/charcnv.c
index 18be2bcac2..258730ec82 100644
--- a/lib/util/charset/charcnv.c
+++ b/lib/util/charset/charcnv.c
@@ -186,7 +186,7 @@ convert:
destlen = 2 + (destlen*3);
ob = talloc_realloc(ctx, outbuf, char, destlen);
if (!ob) {
- DEBUG(0, ("convert_string_talloc: realloc failed!\n"));
+ DEBUG(0, ("iconv_talloc: realloc failed!\n"));
talloc_free(outbuf);
return (size_t)-1;
} else {
@@ -237,10 +237,11 @@ convert:
* @param destlen maximal length allowed for string
* @returns the number of bytes occupied in the destination
**/
-_PUBLIC_ ssize_t convert_string_convenience(struct smb_iconv_convenience *ic,
+_PUBLIC_ bool convert_string_convenience(struct smb_iconv_convenience *ic,
charset_t from, charset_t to,
void const *src, size_t srclen,
- void *dest, size_t destlen, bool allow_badcharcnv)
+ void *dest, size_t destlen, size_t *converted_size,
+ bool allow_badcharcnv)
{
size_t i_len, o_len;
size_t retval;
@@ -249,7 +250,8 @@ _PUBLIC_ ssize_t convert_string_convenience(struct smb_iconv_convenience *ic,
smb_iconv_t descriptor;
if (allow_badcharcnv) {
- return -1;
+ /* Not implemented yet */
+ return false;
}
if (srclen == (size_t)-1)
@@ -261,7 +263,8 @@ _PUBLIC_ ssize_t convert_string_convenience(struct smb_iconv_convenience *ic,
/* conversion not supported, use as is */
size_t len = MIN(srclen,destlen);
memcpy(dest,src,len);
- return len;
+ *converted_size = len;
+ return true;
}
i_len=srclen;
@@ -272,7 +275,7 @@ _PUBLIC_ ssize_t convert_string_convenience(struct smb_iconv_convenience *ic,
switch(errno) {
case EINVAL:
reason="Incomplete multibyte sequence";
- return -1;
+ return false;
case E2BIG:
reason="No more room";
if (from == CH_UNIX) {
@@ -285,14 +288,16 @@ _PUBLIC_ ssize_t convert_string_convenience(struct smb_iconv_convenience *ic,
charset_name(ic, from), charset_name(ic, to),
(int)srclen, (int)destlen));
}
- return -1;
+ return false;
case EILSEQ:
reason="Illegal multibyte sequence";
- return -1;
+ return false;
}
/* smb_panic(reason); */
}
- return destlen-o_len;
+ if (converted_size != NULL)
+ *converted_size = destlen-o_len;
+ return true;
}
/**
@@ -305,21 +310,23 @@ _PUBLIC_ ssize_t convert_string_convenience(struct smb_iconv_convenience *ic,
* @returns Size in bytes of the converted string; or -1 in case of error.
**/
-_PUBLIC_ ssize_t convert_string_talloc_convenience(TALLOC_CTX *ctx,
+_PUBLIC_ bool convert_string_talloc_convenience(TALLOC_CTX *ctx,
struct smb_iconv_convenience *ic,
charset_t from, charset_t to,
void const *src, size_t srclen,
- void **dest, bool allow_badcharcnv)
+ void **dest, size_t *converted_size,
+ bool allow_badcharcnv)
{
smb_iconv_t descriptor;
+ ssize_t ret;
if (allow_badcharcnv)
- return (size_t)-1;
+ return false; /* Not implemented yet */
*dest = NULL;
if (src == NULL || srclen == (size_t)-1 || srclen == 0)
- return (size_t)-1;
+ return false;
descriptor = get_conv_handle(ic, from, to);
@@ -328,10 +335,15 @@ _PUBLIC_ ssize_t convert_string_talloc_convenience(TALLOC_CTX *ctx,
DEBUG(3, ("convert_string_talloc: conversion from %s to %s not supported!\n",
charset_name(ic, from),
charset_name(ic, to)));
- return -1;
+ return false;
}
- return iconv_talloc(ctx, descriptor, src, srclen, dest);
+ ret = iconv_talloc(ctx, descriptor, src, srclen, dest);
+ if (ret == -1)
+ return false;
+ if (converted_size != NULL)
+ *converted_size = ret;
+ return true;
}
/*
diff --git a/lib/util/charset/charset.h b/lib/util/charset/charset.h
index 3acdde30ad..655bae7bcd 100644
--- a/lib/util/charset/charset.h
+++ b/lib/util/charset/charset.h
@@ -28,9 +28,16 @@
#include <talloc.h>
/* this defines the charset types used in samba */
-typedef enum {CH_UTF16=0, CH_UNIX, CH_DOS, CH_UTF8, CH_UTF16BE, CH_UTF16MUNGED} charset_t;
+typedef enum {CH_UTF16LE=0, CH_UTF16=0, CH_UNIX, CH_DISPLAY, CH_DOS, CH_UTF8, CH_UTF16BE, CH_UTF16MUNGED} charset_t;
-#define NUM_CHARSETS 6
+#define NUM_CHARSETS 7
+
+/*
+ * SMB UCS2 (16-bit unicode) internal type.
+ * smb_ucs2_t is *always* in little endian format.
+ */
+
+typedef uint16_t smb_ucs2_t;
/*
* for each charset we have a function that pulls from that charset to
@@ -51,6 +58,17 @@ typedef uint32_t codepoint_t;
#define INVALID_CODEPOINT ((codepoint_t)-1)
+/*
+ * This is auxiliary struct used by source/script/gen-8-bit-gap.sh script
+ * during generation of an encoding table for charset module
+ * */
+
+struct charset_gap_table {
+ uint16_t start;
+ uint16_t end;
+ int32_t idx;
+};
+
/* generic iconv conversion structure */
typedef struct smb_iconv_s {
@@ -106,19 +124,20 @@ bool strhaslower(const char *string);
char *strrchr_m(const char *s, char c);
char *strchr_m(const char *s, char c);
-ssize_t push_ascii_talloc(TALLOC_CTX *ctx, char **dest, const char *src);
-ssize_t push_ucs2_talloc(TALLOC_CTX *ctx, void **dest, const char *src);
-ssize_t push_utf8_talloc(TALLOC_CTX *ctx, char **dest, const char *src);
-ssize_t pull_ascii_talloc(TALLOC_CTX *ctx, char **dest, const char *src);
-ssize_t pull_ucs2_talloc(TALLOC_CTX *ctx, char **dest, const void *src);
-ssize_t pull_utf8_talloc(TALLOC_CTX *ctx, char **dest, const char *src);
+bool push_ascii_talloc(TALLOC_CTX *ctx, char **dest, const char *src, size_t *converted_size);
+bool push_ucs2_talloc(TALLOC_CTX *ctx, smb_ucs2_t **dest, const char *src, size_t *converted_size);
+bool push_utf8_talloc(TALLOC_CTX *ctx, char **dest, const char *src, size_t *converted_size);
+bool pull_ascii_talloc(TALLOC_CTX *ctx, char **dest, const char *src, size_t *converted_size);
+bool pull_ucs2_talloc(TALLOC_CTX *ctx, char **dest, const smb_ucs2_t *src, size_t *converted_size);
+bool pull_utf8_talloc(TALLOC_CTX *ctx, char **dest, const char *src, size_t *converted_size);
ssize_t push_string(void *dest, const char *src, size_t dest_len, int flags);
ssize_t pull_string(char *dest, const void *src, size_t dest_len, size_t src_len, int flags);
-ssize_t convert_string_talloc(TALLOC_CTX *ctx,
+bool convert_string_talloc(TALLOC_CTX *ctx,
charset_t from, charset_t to,
void const *src, size_t srclen,
- void **dest, bool allow_badcharcnv);
+ void **dest, size_t *converted_size,
+ bool allow_badcharcnv);
size_t convert_string(charset_t from, charset_t to,
void const *src, size_t srclen,
@@ -148,15 +167,16 @@ struct smb_iconv_convenience *smb_iconv_convenience_init(TALLOC_CTX *mem_ctx,
const char *unix_charset,
bool native_iconv);
-ssize_t convert_string_convenience(struct smb_iconv_convenience *ic,
+bool convert_string_convenience(struct smb_iconv_convenience *ic,
charset_t from, charset_t to,
void const *src, size_t srclen,
- void *dest, size_t destlen, bool allow_badcharcnv);
-ssize_t convert_string_talloc_convenience(TALLOC_CTX *ctx,
+ void *dest, size_t destlen, size_t *converted_size,
+ bool allow_badcharcnv);
+bool convert_string_talloc_convenience(TALLOC_CTX *ctx,
struct smb_iconv_convenience *ic,
charset_t from, charset_t to,
void const *src, size_t srclen,
- void **dest, bool allow_badcharcnv);
+ void **dest, size_t *converted_size, bool allow_badcharcnv);
/* iconv */
smb_iconv_t smb_iconv_open(const char *tocode, const char *fromcode);
int smb_iconv_close(smb_iconv_t cd);
@@ -169,4 +189,81 @@ smb_iconv_t smb_iconv_open_ex(TALLOC_CTX *mem_ctx, const char *tocode,
void load_case_tables(void);
bool charset_register_backend(const void *_funcs);
+/*
+ * Define stub for charset module which implements 8-bit encoding with gaps.
+ * Encoding tables for such module should be produced from glibc's CHARMAPs
+ * using script source/script/gen-8bit-gap.sh
+ * CHARSETNAME is CAPITALIZED charset name
+ *
+ * */
+#define SMB_GENERATE_CHARSET_MODULE_8_BIT_GAP(CHARSETNAME) \
+static size_t CHARSETNAME ## _push(void *cd, const char **inbuf, size_t *inbytesleft, \
+ char **outbuf, size_t *outbytesleft) \
+{ \
+ while (*inbytesleft >= 2 && *outbytesleft >= 1) { \
+ int i; \
+ int done = 0; \
+ \
+ uint16 ch = SVAL(*inbuf,0); \
+ \
+ for (i=0; from_idx[i].start != 0xffff; i++) { \
+ if ((from_idx[i].start <= ch) && (from_idx[i].end >= ch)) { \
+ ((unsigned char*)(*outbuf))[0] = from_ucs2[from_idx[i].idx+ch]; \
+ (*inbytesleft) -= 2; \
+ (*outbytesleft) -= 1; \
+ (*inbuf) += 2; \
+ (*outbuf) += 1; \
+ done = 1; \
+ break; \
+ } \
+ } \
+ if (!done) { \
+ errno = EINVAL; \
+ return -1; \
+ } \
+ \
+ } \
+ \
+ if (*inbytesleft == 1) { \
+ errno = EINVAL; \
+ return -1; \
+ } \
+ \
+ if (*inbytesleft > 1) { \
+ errno = E2BIG; \
+ return -1; \
+ } \
+ \
+ return 0; \
+} \
+ \
+static size_t CHARSETNAME ## _pull(void *cd, const char **inbuf, size_t *inbytesleft, \
+ char **outbuf, size_t *outbytesleft) \
+{ \
+ while (*inbytesleft >= 1 && *outbytesleft >= 2) { \
+ *(uint16*)(*outbuf) = to_ucs2[((unsigned char*)(*inbuf))[0]]; \
+ (*inbytesleft) -= 1; \
+ (*outbytesleft) -= 2; \
+ (*inbuf) += 1; \
+ (*outbuf) += 2; \
+ } \
+ \
+ if (*inbytesleft > 0) { \
+ errno = E2BIG; \
+ return -1; \
+ } \
+ \
+ return 0; \
+} \
+ \
+struct charset_functions CHARSETNAME ## _functions = \
+ {#CHARSETNAME, CHARSETNAME ## _pull, CHARSETNAME ## _push}; \
+ \
+NTSTATUS charset_ ## CHARSETNAME ## _init(void); \
+NTSTATUS charset_ ## CHARSETNAME ## _init(void) \
+{ \
+ return smb_register_charset(& CHARSETNAME ## _functions); \
+} \
+
+
#endif /* __CHARSET_H__ */
diff --git a/lib/util/charset/iconv.c b/lib/util/charset/iconv.c
index b6842a49aa..98284ce9bd 100644
--- a/lib/util/charset/iconv.c
+++ b/lib/util/charset/iconv.c
@@ -22,6 +22,7 @@
#include "../lib/util/dlinklist.h"
#include "system/iconv.h"
#include "system/filesys.h"
+#undef strcasecmp
/**
diff --git a/lib/util/charset/tests/iconv.c b/lib/util/charset/tests/iconv.c
index fbe7b103ab..091876f63b 100644
--- a/lib/util/charset/tests/iconv.c
+++ b/lib/util/charset/tests/iconv.c
@@ -404,12 +404,12 @@ static bool test_string2key(struct torture_context *tctx)
uint16_t *buf;
char *dest = NULL;
TALLOC_CTX *mem_ctx = talloc_new(tctx);
- ssize_t ret;
size_t len = (random()%1000)+1;
const uint16_t in1[10] = { 'a', 0xd805, 'b', 0xdcf0, 'c', 0, 'd', 'e', 'f', 'g' };
uint8_t le1[20];
uint8_t *munged1;
uint8_t *out1;
+ size_t ret;
int i;
const char *correct = "a\357\277\275b\357\277\275c\001defg";
@@ -418,8 +418,7 @@ static bool test_string2key(struct torture_context *tctx)
torture_comment(tctx, "converting random buffer\n");
- ret = convert_string_talloc(mem_ctx, CH_UTF16MUNGED, CH_UTF8, (void *)buf, len*2, (void**)&dest, false);
- if (ret == -1) {
+ if (!convert_string_talloc(mem_ctx, CH_UTF16MUNGED, CH_UTF8, (void *)buf, len*2, (void**)&dest, &ret, false)) {
torture_fail(tctx, "Failed to convert random buffer\n");
}
@@ -429,8 +428,7 @@ static bool test_string2key(struct torture_context *tctx)
torture_comment(tctx, "converting fixed buffer to UTF16\n");
- ret = convert_string_talloc(mem_ctx, CH_UTF16MUNGED, CH_UTF16, (void *)le1, 20, (void**)&munged1, false);
- if (ret == -1) {
+ if (!convert_string_talloc(mem_ctx, CH_UTF16MUNGED, CH_UTF16, (void *)le1, 20, (void**)&munged1, &ret, false)) {
torture_fail(tctx, "Failed to convert fixed buffer to UTF16_MUNGED\n");
}
@@ -438,8 +436,7 @@ static bool test_string2key(struct torture_context *tctx)
torture_comment(tctx, "converting fixed buffer to UTF8\n");
- ret = convert_string_talloc(mem_ctx, CH_UTF16MUNGED, CH_UTF8, (void *)le1, 20, (void**)&out1, false);
- if (ret == -1) {
+ if (!convert_string_talloc(mem_ctx, CH_UTF16MUNGED, CH_UTF8, (void *)le1, 20, (void**)&out1, &ret, false)) {
torture_fail(tctx, "Failed to convert fixed buffer to UTF8\n");
}
diff --git a/lib/util/charset/util_unistr.c b/lib/util/charset/util_unistr.c
index 96c987393f..ec88e784d0 100644
--- a/lib/util/charset/util_unistr.c
+++ b/lib/util/charset/util_unistr.c
@@ -668,11 +668,11 @@ static ssize_t push_ascii(void *dest, const char *src, size_t dest_len, int flag
* @returns The number of bytes occupied by the string in the destination
* or -1 in case of error.
**/
-_PUBLIC_ ssize_t push_ascii_talloc(TALLOC_CTX *ctx, char **dest, const char *src)
+_PUBLIC_ bool push_ascii_talloc(TALLOC_CTX *ctx, char **dest, const char *src, size_t *converted_size)
{
size_t src_len = strlen(src)+1;
*dest = NULL;
- return convert_string_talloc(ctx, CH_UNIX, CH_DOS, src, src_len, (void **)dest, false);
+ return convert_string_talloc(ctx, CH_UNIX, CH_DOS, src, src_len, (void **)dest, converted_size, false);
}
@@ -779,11 +779,11 @@ static ssize_t push_ucs2(void *dest, const char *src, size_t dest_len, int flags
* @returns The number of bytes occupied by the string in the destination
* or -1 in case of error.
**/
-_PUBLIC_ ssize_t push_ucs2_talloc(TALLOC_CTX *ctx, void **dest, const char *src)
+_PUBLIC_ bool push_ucs2_talloc(TALLOC_CTX *ctx, smb_ucs2_t **dest, const char *src, size_t *converted_size)
{
size_t src_len = strlen(src)+1;
*dest = NULL;
- return convert_string_talloc(ctx, CH_UNIX, CH_UTF16, src, src_len, dest, false);
+ return convert_string_talloc(ctx, CH_UNIX, CH_UTF16, src, src_len, (void **)dest, converted_size, false);
}
@@ -795,11 +795,11 @@ _PUBLIC_ ssize_t push_ucs2_talloc(TALLOC_CTX *ctx, void **dest, const char *src)
* @returns The number of bytes occupied by the string in the destination
**/
-_PUBLIC_ ssize_t push_utf8_talloc(TALLOC_CTX *ctx, char **dest, const char *src)
+_PUBLIC_ bool push_utf8_talloc(TALLOC_CTX *ctx, char **dest, const char *src, size_t *converted_size)
{
size_t src_len = strlen(src)+1;
*dest = NULL;
- return convert_string_talloc(ctx, CH_UNIX, CH_UTF8, src, src_len, (void **)dest, false);
+ return convert_string_talloc(ctx, CH_UNIX, CH_UTF8, src, src_len, (void **)dest, converted_size, false);
}
/**
@@ -850,11 +850,11 @@ static size_t pull_ucs2(char *dest, const void *src, size_t dest_len, size_t src
* @returns The number of bytes occupied by the string in the destination
**/
-_PUBLIC_ ssize_t pull_ascii_talloc(TALLOC_CTX *ctx, char **dest, const char *src)
+_PUBLIC_ bool pull_ascii_talloc(TALLOC_CTX *ctx, char **dest, const char *src, size_t *converted_size)
{
size_t src_len = strlen(src)+1;
*dest = NULL;
- return convert_string_talloc(ctx, CH_DOS, CH_UNIX, src, src_len, (void **)dest, false);
+ return convert_string_talloc(ctx, CH_DOS, CH_UNIX, src, src_len, (void **)dest, converted_size, false);
}
/**
@@ -865,11 +865,11 @@ _PUBLIC_ ssize_t pull_ascii_talloc(TALLOC_CTX *ctx, char **dest, const char *src
* @returns The number of bytes occupied by the string in the destination
**/
-_PUBLIC_ ssize_t pull_ucs2_talloc(TALLOC_CTX *ctx, char **dest, const void *src)
+_PUBLIC_ bool pull_ucs2_talloc(TALLOC_CTX *ctx, char **dest, const smb_ucs2_t *src, size_t *converted_size)
{
size_t src_len = utf16_len(src);
*dest = NULL;
- return convert_string_talloc(ctx, CH_UTF16, CH_UNIX, src, src_len, (void **)dest, false);
+ return convert_string_talloc(ctx, CH_UTF16, CH_UNIX, src, src_len, (void **)dest, converted_size, false);
}
/**
@@ -880,11 +880,11 @@ _PUBLIC_ ssize_t pull_ucs2_talloc(TALLOC_CTX *ctx, char **dest, const void *src)
* @returns The number of bytes occupied by the string in the destination
**/
-_PUBLIC_ ssize_t pull_utf8_talloc(TALLOC_CTX *ctx, char **dest, const char *src)
+_PUBLIC_ bool pull_utf8_talloc(TALLOC_CTX *ctx, char **dest, const char *src, size_t *converted_size)
{
size_t src_len = strlen(src)+1;
*dest = NULL;
- return convert_string_talloc(ctx, CH_UTF8, CH_UNIX, src, src_len, (void **)dest, false);
+ return convert_string_talloc(ctx, CH_UTF8, CH_UNIX, src, src_len, (void **)dest, converted_size, false);
}
/**
@@ -952,11 +952,16 @@ _PUBLIC_ ssize_t pull_string(char *dest, const void *src, size_t dest_len, size_
**/
_PUBLIC_ size_t convert_string(charset_t from, charset_t to,
void const *src, size_t srclen,
- void *dest, size_t destlen, bool allow_badcharcnv)
+ void *dest, size_t destlen,
+ bool allow_badcharcnv)
{
- return convert_string_convenience(get_iconv_convenience(), from, to,
+ size_t ret;
+ if (!convert_string_convenience(get_iconv_convenience(), from, to,
src, srclen,
- dest, destlen, allow_badcharcnv);
+ dest, destlen, &ret,
+ allow_badcharcnv))
+ return -1;
+ return ret;
}
/**
@@ -964,18 +969,21 @@ _PUBLIC_ size_t convert_string(charset_t from, charset_t to,
*
* @param srclen length of source buffer.
* @param dest always set at least to NULL
+ * @param converted_size Size in bytes of the converted string
* @note -1 is not accepted for srclen.
*
- * @returns Size in bytes of the converted string; or -1 in case of error.
+ * @returns boolean indication whether the conversion succeeded
**/
-_PUBLIC_ ssize_t convert_string_talloc(TALLOC_CTX *ctx,
+_PUBLIC_ bool convert_string_talloc(TALLOC_CTX *ctx,
charset_t from, charset_t to,
void const *src, size_t srclen,
- void **dest, bool allow_badcharcnv)
+ void **dest, size_t *converted_size,
+ bool allow_badcharcnv)
{
return convert_string_talloc_convenience(ctx, get_iconv_convenience(),
from, to, src, srclen, dest,
+ converted_size,
allow_badcharcnv);
}
generated by cgit (git 2.34.1) at 2025-04-07 16:11:04 +0000