Skip to content

Commit

Permalink
Implement RFC https://wiki.php.net/rfc/default_encoding
Browse files Browse the repository at this point in the history
  • Loading branch information
Yasuo Ohgaki committed Feb 13, 2014
1 parent 06bbfe1 commit cbd108a
Show file tree
Hide file tree
Showing 34 changed files with 464 additions and 161 deletions.
44 changes: 39 additions & 5 deletions ext/iconv/iconv.c
View file
Original file line number Diff line number Diff line change
Expand Up @@ -220,21 +220,55 @@ static char _generic_superset_name[] = ICONV_UCS4_ENCODING;
#define GENERIC_SUPERSET_NBYTES 4
/* }}} */

static PHP_INI_MH(OnUpdateStringIconvCharset)

static PHP_INI_MH(OnUpdateInputEncoding)
{
if (new_value_length >= ICONV_CSNMAXLEN) {
return FAILURE;
}
if (new_value_length) {
OnUpdateString(entry, new_value, new_value_length, mh_arg1, mh_arg2, mh_arg3, stage TSRMLS_CC);
} else {
OnUpdateString(entry, PG(input_encoding), strlen(PG(input_encoding))+1, mh_arg1, mh_arg2, mh_arg3, stage TSRMLS_CC);
}
return SUCCESS;
}


static PHP_INI_MH(OnUpdateOutputEncoding)
{
if(new_value_length >= ICONV_CSNMAXLEN) {
return FAILURE;
}
OnUpdateString(entry, new_value, new_value_length, mh_arg1, mh_arg2, mh_arg3, stage TSRMLS_CC);
if (new_value_length) {
OnUpdateString(entry, new_value, new_value_length, mh_arg1, mh_arg2, mh_arg3, stage TSRMLS_CC);
} else {
OnUpdateString(entry, PG(output_encoding), strlen(PG(output_encoding))+1, mh_arg1, mh_arg2, mh_arg3, stage TSRMLS_CC);
}
return SUCCESS;
}


static PHP_INI_MH(OnUpdateInternalEncoding)
{
if(new_value_length >= ICONV_CSNMAXLEN) {
return FAILURE;
}
if (new_value_length) {
OnUpdateString(entry, new_value, new_value_length, mh_arg1, mh_arg2, mh_arg3, stage TSRMLS_CC);
} else {
OnUpdateString(entry, PG(internal_encoding), strlen(PG(internal_encoding))+1, mh_arg1, mh_arg2, mh_arg3, stage TSRMLS_CC);
}
return SUCCESS;
}


/* {{{ PHP_INI
*/
PHP_INI_BEGIN()
STD_PHP_INI_ENTRY("iconv.input_encoding", ICONV_INPUT_ENCODING, PHP_INI_ALL, OnUpdateStringIconvCharset, input_encoding, zend_iconv_globals, iconv_globals)
STD_PHP_INI_ENTRY("iconv.output_encoding", ICONV_OUTPUT_ENCODING, PHP_INI_ALL, OnUpdateStringIconvCharset, output_encoding, zend_iconv_globals, iconv_globals)
STD_PHP_INI_ENTRY("iconv.internal_encoding", ICONV_INTERNAL_ENCODING, PHP_INI_ALL, OnUpdateStringIconvCharset, internal_encoding, zend_iconv_globals, iconv_globals)
STD_PHP_INI_ENTRY("iconv.input_encoding", "", PHP_INI_ALL, OnUpdateInputEncoding, input_encoding, zend_iconv_globals, iconv_globals)
STD_PHP_INI_ENTRY("iconv.output_encoding", "", PHP_INI_ALL, OnUpdateOutputEncoding, output_encoding, zend_iconv_globals, iconv_globals)
STD_PHP_INI_ENTRY("iconv.internal_encoding", "", PHP_INI_ALL, OnUpdateInternalEncoding, internal_encoding, zend_iconv_globals, iconv_globals)
PHP_INI_END()
/* }}} */

Expand Down
6 changes: 0 additions & 6 deletions ext/iconv/php_iconv.h
View file
Original file line number Diff line number Diff line change
Expand Up @@ -79,15 +79,9 @@ ZEND_END_MODULE_GLOBALS(iconv)
#endif

#ifdef HAVE_IBM_ICONV
# define ICONV_INPUT_ENCODING "ISO8859-1"
# define ICONV_OUTPUT_ENCODING "ISO8859-1"
# define ICONV_INTERNAL_ENCODING "ISO8859-1"
# define ICONV_ASCII_ENCODING "IBM-850"
# define ICONV_UCS4_ENCODING "UCS-4"
#else
# define ICONV_INPUT_ENCODING "ISO-8859-1"
# define ICONV_OUTPUT_ENCODING "ISO-8859-1"
# define ICONV_INTERNAL_ENCODING "ISO-8859-1"
# define ICONV_ASCII_ENCODING "ASCII"
# define ICONV_UCS4_ENCODING "UCS-4LE"
#endif
Expand Down
68 changes: 68 additions & 0 deletions ext/iconv/tests/iconv_ini_encoding.phpt
View file
Original file line number Diff line number Diff line change
@@ -0,0 +1,68 @@
--TEST--
Encoding INI test
--SKIPIF--
<?php extension_loaded('iconv') or die('skip mbstring not available'); ?>
--INI--
default_charset=ISO-8859-1
internal_encoding=
input_encoding=
output_encoding=
iconv.internal_encoding=ISO-8859-1
iconv.http_input=ISO-8859-1
iconv.http_output=ISO-8859-1
--FILE--
<?php
echo "Getting INI\n";
var_dump(ini_get('default_charset'));
var_dump(ini_get('internal_encoding'));
var_dump(ini_get('input_encoding'));
var_dump(ini_get('output_encoding'));

var_dump(ini_get('iconv.internal_encoding'));
var_dump(ini_get('iconv.input_encoding'));
var_dump(ini_get('iconv.output_encoding'));

echo "Setting INI\n";
var_dump(ini_set('default_charset', 'UTF-8'));
var_dump(ini_set('internal_encoding', 'UTF-8'));
var_dump(ini_set('input_encoding', 'UTF-8'));
var_dump(ini_set('output_encoding', 'UTF-8'));
var_dump(ini_set('iconv.internal_encoding', 'UTF-8'));
var_dump(ini_set('iconv.input_encoding', 'UTF-8'));
var_dump(ini_set('iconv.output_encoding', 'UTF-8'));

echo "Getting INI\n";
var_dump(ini_get('default_charset'));
var_dump(ini_get('internal_encoding'));
var_dump(ini_get('input_encoding'));
var_dump(ini_get('output_encoding'));

var_dump(ini_get('iconv.internal_encoding'));
var_dump(ini_get('iconv.input_encoding'));
var_dump(ini_get('iconv.output_encoding'));

--EXPECT--
Getting INI
string(10) "ISO-8859-1"
string(0) ""
string(0) ""
string(0) ""
string(10) "ISO-8859-1"
string(0) ""
string(0) ""
Setting INI
string(10) "ISO-8859-1"
string(0) ""
string(0) ""
string(0) ""
string(10) "ISO-8859-1"
string(0) ""
string(0) ""
Getting INI
string(5) "UTF-8"
string(5) "UTF-8"
string(5) "UTF-8"
string(5) "UTF-8"
string(5) "UTF-8"
string(5) "UTF-8"
string(5) "UTF-8"
54 changes: 27 additions & 27 deletions ext/iconv/tests/iconv_set_encoding_variation.phpt
View file
Original file line number Diff line number Diff line change
Expand Up @@ -183,17 +183,17 @@ string(3) "0.5"
bool(true)
bool(true)
bool(true)
string(0) ""
string(0) ""
string(0) ""
string(5) "UTF-8"
string(5) "UTF-8"
string(5) "UTF-8"

-- Iteration 11 --
bool(true)
bool(true)
bool(true)
string(0) ""
string(0) ""
string(0) ""
string(5) "UTF-8"
string(5) "UTF-8"
string(5) "UTF-8"

-- Iteration 12 --
bool(true)
Expand All @@ -207,9 +207,9 @@ string(1) "1"
bool(true)
bool(true)
bool(true)
string(0) ""
string(0) ""
string(0) ""
string(5) "UTF-8"
string(5) "UTF-8"
string(5) "UTF-8"

-- Iteration 14 --
bool(true)
Expand All @@ -223,25 +223,25 @@ string(1) "1"
bool(true)
bool(true)
bool(true)
string(0) ""
string(0) ""
string(0) ""
string(5) "UTF-8"
string(5) "UTF-8"
string(5) "UTF-8"

-- Iteration 16 --
bool(true)
bool(true)
bool(true)
string(0) ""
string(0) ""
string(0) ""
string(5) "UTF-8"
string(5) "UTF-8"
string(5) "UTF-8"

-- Iteration 17 --
bool(true)
bool(true)
bool(true)
string(0) ""
string(0) ""
string(0) ""
string(5) "UTF-8"
string(5) "UTF-8"
string(5) "UTF-8"

-- Iteration 18 --
bool(true)
Expand Down Expand Up @@ -279,17 +279,17 @@ string(5) "UTF-8"
bool(true)
bool(true)
bool(true)
string(0) ""
string(0) ""
string(0) ""
string(5) "UTF-8"
string(5) "UTF-8"
string(5) "UTF-8"

-- Iteration 23 --
bool(true)
bool(true)
bool(true)
string(0) ""
string(0) ""
string(0) ""
string(5) "UTF-8"
string(5) "UTF-8"
string(5) "UTF-8"

-- Iteration 24 --

Expand All @@ -301,7 +301,7 @@ NULL

Warning: iconv_set_encoding() expects parameter 2 to be string, resource given in %s on line %d
NULL
string(0) ""
string(0) ""
string(0) ""
string(5) "UTF-8"
string(5) "UTF-8"
string(5) "UTF-8"
Done
85 changes: 34 additions & 51 deletions ext/mbstring/mbstring.c
View file
Original file line number Diff line number Diff line change
Expand Up @@ -1236,6 +1236,11 @@ static PHP_INI_MH(OnUpdate_mbstring_http_input)
if (MBSTRG(http_input_list)) {
pefree(MBSTRG(http_input_list), 1);
}
if (SUCCESS == php_mb_parse_encoding_list(PG(input_encoding), strlen(PG(input_encoding))+1, &list, &size, 1 TSRMLS_CC)) {
MBSTRG(http_input_list) = list;
MBSTRG(http_input_list_size) = 0;
return SUCCESS;
}
MBSTRG(http_input_list) = NULL;
MBSTRG(http_input_list_size) = 0;
return SUCCESS;
Expand All @@ -1261,18 +1266,20 @@ static PHP_INI_MH(OnUpdate_mbstring_http_output)
const mbfl_encoding *encoding;

if (new_value == NULL || new_value_length == 0) {
MBSTRG(http_output_encoding) = &mbfl_encoding_pass;
MBSTRG(current_http_output_encoding) = &mbfl_encoding_pass;
return SUCCESS;
}

encoding = mbfl_name2encoding(new_value);
if (!encoding) {
MBSTRG(http_output_encoding) = &mbfl_encoding_pass;
MBSTRG(current_http_output_encoding) = &mbfl_encoding_pass;
return FAILURE;
encoding = mbfl_name2encoding(PG(output_encoding));
if (!encoding) {
MBSTRG(http_output_encoding) = &mbfl_encoding_pass;
MBSTRG(current_http_output_encoding) = &mbfl_encoding_pass;
return SUCCESS;
}
} else {
encoding = mbfl_name2encoding(new_value);
if (!encoding) {
MBSTRG(http_output_encoding) = &mbfl_encoding_pass;
MBSTRG(current_http_output_encoding) = &mbfl_encoding_pass;
return FAILURE;
}
}

MBSTRG(http_output_encoding) = encoding;
MBSTRG(current_http_output_encoding) = encoding;
return SUCCESS;
Expand All @@ -1285,47 +1292,17 @@ int _php_mb_ini_mbstring_internal_encoding_set(const char *new_value, uint new_v
const mbfl_encoding *encoding;

if (!new_value || new_value_length == 0 || !(encoding = mbfl_name2encoding(new_value))) {
switch (MBSTRG(language)) {
case mbfl_no_language_uni:
encoding = mbfl_no2encoding(mbfl_no_encoding_utf8);
break;
case mbfl_no_language_japanese:
encoding = mbfl_no2encoding(mbfl_no_encoding_euc_jp);
break;
case mbfl_no_language_korean:
encoding = mbfl_no2encoding(mbfl_no_encoding_euc_kr);
break;
case mbfl_no_language_simplified_chinese:
encoding = mbfl_no2encoding(mbfl_no_encoding_euc_cn);
break;
case mbfl_no_language_traditional_chinese:
encoding = mbfl_no2encoding(mbfl_no_encoding_euc_tw);
break;
case mbfl_no_language_russian:
encoding = mbfl_no2encoding(mbfl_no_encoding_koi8r);
break;
case mbfl_no_language_german:
encoding = mbfl_no2encoding(mbfl_no_encoding_8859_15);
break;
case mbfl_no_language_armenian:
encoding = mbfl_no2encoding(mbfl_no_encoding_armscii8);
break;
case mbfl_no_language_turkish:
encoding = mbfl_no2encoding(mbfl_no_encoding_8859_9);
break;
default:
encoding = mbfl_no2encoding(mbfl_no_encoding_8859_1);
break;
}
}
/* falls back to UTF-8 if an unkown encoding name is given */
encoding = mbfl_no2encoding(mbfl_no_encoding_utf8);
}
MBSTRG(internal_encoding) = encoding;
MBSTRG(current_internal_encoding) = encoding;
#if HAVE_MBREGEX
{
const char *enc_name = new_value;
if (FAILURE == php_mb_regex_set_default_mbctype(enc_name TSRMLS_CC)) {
/* falls back to EUC-JP if an unknown encoding name is given */
enc_name = "EUC-JP";
/* falls back to UTF-8 if an unknown encoding name is given */
enc_name = "UTF-8";
php_mb_regex_set_default_mbctype(enc_name TSRMLS_CC);
}
php_mb_regex_set_mbctype(new_value TSRMLS_CC);
Expand All @@ -1343,7 +1320,11 @@ static PHP_INI_MH(OnUpdate_mbstring_internal_encoding)
}
if (stage == PHP_INI_STAGE_STARTUP || stage == PHP_INI_STAGE_SHUTDOWN
|| stage == PHP_INI_STAGE_RUNTIME) {
return _php_mb_ini_mbstring_internal_encoding_set(new_value, new_value_length TSRMLS_CC);
if (new_value_length) {
return _php_mb_ini_mbstring_internal_encoding_set(new_value, new_value_length TSRMLS_CC);
} else {
return _php_mb_ini_mbstring_internal_encoding_set(PG(internal_encoding), strlen(PG(internal_encoding))+1 TSRMLS_CC);
}
} else {
/* the corresponding mbstring globals needs to be set according to the
* ini value in the later stage because it never falls back to the
Expand Down Expand Up @@ -1450,8 +1431,8 @@ static PHP_INI_MH(OnUpdate_mbstring_http_output_conv_mimetypes)
PHP_INI_BEGIN()
PHP_INI_ENTRY("mbstring.language", "neutral", PHP_INI_ALL, OnUpdate_mbstring_language)
PHP_INI_ENTRY("mbstring.detect_order", NULL, PHP_INI_ALL, OnUpdate_mbstring_detect_order)
PHP_INI_ENTRY("mbstring.http_input", "pass", PHP_INI_ALL, OnUpdate_mbstring_http_input)
PHP_INI_ENTRY("mbstring.http_output", "pass", PHP_INI_ALL, OnUpdate_mbstring_http_output)
PHP_INI_ENTRY("mbstring.http_input", NULL, PHP_INI_ALL, OnUpdate_mbstring_http_input)
PHP_INI_ENTRY("mbstring.http_output", NULL, PHP_INI_ALL, OnUpdate_mbstring_http_output)
STD_PHP_INI_ENTRY("mbstring.internal_encoding", NULL, PHP_INI_ALL, OnUpdate_mbstring_internal_encoding, internal_encoding_name, zend_mbstring_globals, mbstring_globals)
PHP_INI_ENTRY("mbstring.substitute_character", NULL, PHP_INI_ALL, OnUpdate_mbstring_substitute_character)
STD_PHP_INI_ENTRY("mbstring.func_overload", "0",
Expand Down Expand Up @@ -2162,8 +2143,10 @@ PHP_FUNCTION(mb_output_handler)

/* feed the string */
mbfl_string_init(&string);
string.no_language = MBSTRG(language);
string.no_encoding = MBSTRG(current_internal_encoding)->no_encoding;
/* these are not needed. convd has encoding info.
string.no_language = MBSTRG(language);
string.no_encoding = MBSTRG(current_internal_encoding)->no_encoding;
*/
string.val = (unsigned char *)arg_string;
string.len = arg_string_len;
mbfl_buffer_converter_feed(MBSTRG(outconv), &string);
Expand Down
Loading

0 comments on commit cbd108a

Please sign in to comment.