Tsukada
ttsuk****@users*****
2002年 5月 23日 (木) 04:55:52 JST
ttsukada 02/05/23 04:55:52
Modified: ext/mbstring mbfunction.c
Log:
MBFL update
Revision Changes Path
1.3 +290 -16 php4/ext/mbstring/mbfunction.c
Index: mbfunction.c
===================================================================
RCS file: /cvsroot/php-i18n/php4/ext/mbstring/mbfunction.c,v
retrieving revision 1.2
retrieving revision 1.3
diff -u -r1.2 -r1.3
--- mbfunction.c 21 May 2002 20:31:26 -0000 1.2
+++ mbfunction.c 22 May 2002 19:55:52 -0000 1.3
@@ -7,10 +7,6 @@
#include "php.h"
#include "php_mb.h"
-#include "php_mb_str.h"
-#include "php_mb_buf.h"
-#include "php_mb_filter.h"
-#include "php_mb_function.h"
#include "ext/standard/php_string.h"
@@ -59,27 +55,43 @@
}
+/* Return 0 if input contains any illegal encoding, otherwise 1.
+ * Even if any illegal encoding is detected the result may contain a list
+ * of parsed encodings.
+ */
PHPAPI int
-php_mb_parse_encoding_list(const char *string_val, int string_len, php_mb_enc ***return_list, int *return_size, int persistent TSRMLS_DC)
+php_mb_parse_encoding_list(
+ const char *string_val,
+ int string_len,
+ php_mb_enc ***return_list,
+ int *return_size,
+ int persistent TSRMLS_DC)
{
- int n, size;
+ int i, n, size, bauto, result;
char *p, *p1, *p2, *endp, *tmpstr;
php_mb_enc **list, *encoding;
+ result = 0;
+ if (return_list) {
+ *return_list = NULL;
+ }
+ if (return_size) {
+ *return_size = 0;
+ }
if (string_val == NULL) {
MBG(last_errno) = PHP_MB_ERR_NULL_POINTER;
- return FAILURE;
+ return result;
}
if (string_len <= 0) {
MBG(last_errno) = PHP_MB_ERR_ILLEGAL_ARGUMENT;
- return FAILURE;
+ return result;
}
- /* copy the value string for work */
+ /* copy the string for work */
tmpstr = (char *)estrndup(string_val, string_len);
if (tmpstr == NULL) {
MBG(last_errno) = PHP_MB_ERR_NO_MEMORY;
- return FAILURE;
+ return result;
}
/* count the number of listed encoding names */
@@ -90,16 +102,18 @@
p1 = p2 + 1;
n++;
}
- size = n;
+ size = n + MBG(language_r)->auto_detect_count;
/* make list */
list = (php_mb_enc **)pecalloc(size + 1, sizeof(php_mb_enc *), persistent);
if (list == NULL) {
MBG(last_errno) = PHP_MB_ERR_NO_MEMORY;
efree(tmpstr);
- return FAILURE;
+ return result;
}
+ result = 1;
+ bauto = 0;
n = 0;
p1 = tmpstr;
do {
@@ -120,25 +134,52 @@
/* convert to the encoding number and check encoding */
encoding = php_mb_enc_resolve(p1);
if (encoding != NULL) {
- list[n++] = encoding;
+ if (encoding->id == php_mb_encid_auto) {
+ if (!bauto) {
+ bauto = 1;
+ i = 0;
+ while (i < MBG(language_r)->auto_detect_count) {
+ list[n] = php_mb_enc_get(MBG(language_r)->auto_detect_list[i]);
+ i++;
+ n++;
+ }
+ }
+ } else {
+ list[n] = encoding;
+ n++;
+ }
+ } else {
+ result = 0;
}
p1 = p2 + 1;
} while (n < size && p2 != NULL);
list[n] = NULL;
- if (return_list != NULL) {
- *return_list = list;
+ if (n > 0) {
+ if (return_list) {
+ *return_list = list;
+ } else {
+ pefree(list, persistent);
+ }
} else {
pefree(list, persistent);
+ result = 0;
}
if (return_size != NULL) {
*return_size = n;
}
efree(tmpstr);
- return SUCCESS;
+ return result;
}
+/* {{{ php_mb_check_encoding_list */
+PHPAPI int
+php_mb_check_encoding_list(const char *encoding_list TSRMLS_DC) {
+ return php_mb_parse_encoding_list(encoding_list, strlen(encoding_list), NULL, NULL, 0 TSRMLS_CC);
+}
+/* }}} */
+
/*
* encoding converter
@@ -228,6 +269,82 @@
}
+/* {{{ php_mb_convert_encoding */
+PHPAPI char *
+_php_mb_det_conv_encoding(
+ const char *input,
+ size_t length,
+ const char *_to_encoding,
+ const char *_from_encodings,
+ size_t *output_len TSRMLS_DC)
+{
+ int size;
+ char *output=NULL;
+ php_mb_enc *from_encoding, *to_encoding, **list;
+ php_mb_bufconv *convd;
+
+ if (output_len != NULL) {
+ *output_len = 0;
+ }
+ if ( !input || !length) {
+ return NULL;
+ }
+ /* new encoding */
+ if (_to_encoding && strlen(_to_encoding)) {
+ to_encoding = php_mb_enc_resolve(_to_encoding);
+ if (to_encoding == NULL) {
+ php_error(E_WARNING, "%s() unknown encoding \"%s\"",
+ get_active_function_name(TSRMLS_C), _to_encoding);
+ return NULL;
+ }
+ } else {
+ to_encoding = MBG(internal_encoding_r);
+ }
+
+ /* pre-conversion encoding */
+ if (_from_encodings != NULL) {
+ list = NULL;
+ size = 0;
+ php_mb_parse_encoding_list(_from_encodings, strlen(_from_encodings), &list, &size, 0 TSRMLS_CC);
+ if (size == 1) {
+ from_encoding = list[0];
+ } else if (size > 1) {
+ /* auto detect */
+ from_encoding = php_mb_detect_encoding(input, length, list, size TSRMLS_CC);
+ if (from_encoding == NULL) {
+ php_error(E_WARNING, "%s() unable to detect character encoding",
+ get_active_function_name(TSRMLS_C));
+ from_encoding = php_mb_enc_get(php_mb_encid_pass);
+ to_encoding = from_encoding;
+ }
+ } else {
+ php_error(E_WARNING, "$s() illegal character encoding specified",
+ get_active_function_name(TSRMLS_C));
+ }
+ if (list != NULL) {
+ efree(list);
+ }
+ }
+
+ /* initialize converter */
+ convd = php_mb_bufconv_create(to_encoding, from_encoding, length TSRMLS_CC);
+ if (convd == NULL) {
+ php_error(E_WARNING, "%s() unable to create character encoding converter",
+ get_active_function_name(TSRMLS_C));
+ return NULL;
+ }
+
+ /* do it */
+ php_mb_bufconv_feed(convd, input, length TSRMLS_CC);
+ php_mb_bufconv_flush(convd TSRMLS_CC);
+ php_mb_bufconv_extract(convd, &output, output_len TSRMLS_CC);
+ php_mb_bufconv_free(convd TSRMLS_CC);
+
+ return output;
+}
+/* }}} */
+
+
PHPAPI php_mb_buf *
_php_mb_convert_to_wchar(const char *string_val, int string_len, php_mb_enc *from TSRMLS_DC)
{
@@ -2983,4 +3100,161 @@
return result;
}
+
+
+#ifdef ZEND_MULTIBYTE
+
+PHPAPI int
+php_mbstring_set_zend_encoding(TSRMLS_D)
+{
+ zend_encoding_detector encoding_detector;
+ zend_encoding_converter encoding_converter;
+ zend_multibyte_oddlen multibyte_oddlen;
+
+ /* notify script encoding to Zend Engine */
+ zend_multibyte_set_script_encoding(MBG(script_encoding), strlen(MBG(script_encoding)) TSRMLS_CC);
+
+ encoding_detector = php_mbstring_encoding_detector;
+ encoding_converter = NULL;
+ multibyte_oddlen = php_mbstring_oddlen;
+
+#if defined(MBSTR_ENC_TRANS)
+ /* notify internal encoding to Zend Engine */
+ zend_multibyte_set_internal_encoding((char*)MBG(internal_encoding_r)->name, strlen(MBG(internal_encoding_r)->name) TSRMLS_CC);
+
+ encoding_converter = php_mbstring_encoding_converter;
+#endif /* defined(MBSTR_ENC_TRANS) */
+
+ zend_multibyte_set_functions(encoding_detector, encoding_converter,
+ multibyte_oddlen TSRMLS_CC);
+
+ return 0;
+}
+
+/*
+ * mb_detect_encoding (interface for Zend Engine)
+ */
+char *
+php_mbstring_encoding_detector(char *arg_string, int arg_length, char *arg_list TSRMLS_DC)
+{
+ php_mb_enc *encoding, **elist;
+ int size;
+
+ /* make encoding list */
+ list = NULL;
+ size = 0;
+ php_mb_parse_encoding_list(arg_list, strlen(arg_list), &elist, &size, 0 TSRMLS_CC);
+ if (size <= 0) {
+ return NULL;
+ }
+
+ if (elist == NULL) {
+ php_mb_parse_encoding_list(MBG(detect_order), strlen(MBG(detect_order)), &elist, &size, 0 TSRMLS_CC);
+ }
+
+ encoding = php_mb_detect_encoding(arg_string, arg_length, elist, size TSRMLS_CC);
+ if (elist != NULL) {
+ efree(elist);
+ }
+ if (encoding != NULL) {
+ return estrdup(encoding->name);
+ } else {
+ return NULL;
+ }
+}
+
+
+/*
+ * mb_convert_encoding (interface for Zend Engine)
+ */
+int
+php_mbstring_encoding_converter(
+ char **to,
+ int *to_length,
+ char *from,
+ int from_length,
+ const char *encoding_to,
+ const char *encoding_from
+ TSRMLS_DC)
+{
+ int result;
+ php_mb_enc *from_encoding, *to_encoding;
+ php_mb_bufconv *convd;
+
+ /* new encoding */
+ to_encoding = php_mb_enc_resolve(encoding_to);
+ if (to_encoding == NULL)
+ return -1;
+
+ /* old encoding */
+ from_encoding = php_mb_enc_resolve(encoding_from);
+ if (from_encoding == NULL)
+ return -1;
+
+ /* initialize converter */
+ convd = php_mb_bufconv_create(to_encoding, from_encoding, from_length TSRMLS_CC);
+ if (convd == NULL)
+ return -1;
+
+ /* do it */
+ php_mb_bufconv_feed(convd, from, from_length TSRMLS_CC);
+ php_mb_bufconv_flush(convd TSRMLS_CC);
+ php_mb_bufconv_extruct(convd, to, to_length TSRMLS_CC);
+ php_mb_bufconv_free(convd);
+
+ return 0;
+}
+
+
+/*
+ * returns number of odd (e.g. appears only first byte of multibyte
+ * character) chars
+ */
+int php_mbstring_oddlen(char *string, int length, const char *enc_name TSRMLS_DC)
+{
+ int n, m, k;
+ unsigned char *p;
+ const unsigned char *mbtab;
+ php_mb_enc *encoding;
+
+ encoding = php_mb_enc_resolve(enc_name);
+ if (encoding == NULL) {
+ return 0;
+ }
+
+ if (encoding->flag & MBFL_ENCTYPE_SBCS) {
+ return 0;
+ } else if (encoding->flag & (MBFL_ENCTYPE_WCS2BE | MBFL_ENCTYPE_WCS2LE)) {
+ return length % 2;
+ } else if (encoding->flag & (MBFL_ENCTYPE_WCS4BE | MBFL_ENCTYPE_WCS4LE)) {
+ return length % 4;
+ } else if (encoding->mblen_table != NULL) {
+ mbtab = encoding->mblen_table;
+ n = 0;
+ p = string;
+ k = length;
+ /* count */
+ if (p != NULL) {
+ while (n < k) {
+ m = mbtab[*p];
+ n += m;
+ p += m;
+ }
+ }
+ return n-k;
+ } else {
+ /* how can i do ? */
+ return 0;
+ }
+ /* NOT REACHED */
+}
+
+#else
+
+PHPAPI int
+php_mbstring_set_zend_encoding(TSRMLS_D){
+ return 0;
+}
+
+#endif /* ZEND_MULTIBYTE */