Wstęp

While there are many languages in which every necessary character can be represented by a one-to-one mapping to a 8-bit value, there are also several languages which require so many characters for written communication that cannot be contained within the range a mere byte can code. Multibyte character encoding schemes were developed to express that many (more than 256) characters in the regular bytewise coding system.

When you manipulate (trim, split, splice, etc.) strings encoded in a multibyte encoding, you need to use special functions since two or more consecutive bytes may represent a single character in such encoding schemes. Otherwise, if you apply a non-multibyte-aware string function to the string, it probably fails to detect the beginning or ending of the multibyte character and ends up with a corrupted garbage string that most likely loses its original meaning.

mbstring provides these multibyte specific string functions that help you deal with multibyte encodings in PHP, which is basically supposed to be used with single byte encodings. In addition to that, mbstring handles character encoding conversion between the possible encoding pairs.

mbstring is also designed to handle Unicode-based encodings such as UTF-8 and UCS-2 and many single-byte encodings for convenience (listed below), whereas mbstring was originally developed for use in Japanese web pages.

PHP Character Encoding Requirements

Encodings of the following types are safely used with PHP.

A singlebyte encoding,
- which has ASCII-compatible (ISO646 compatible) mappings for the characters in range of 00h to 7fh.
A multibyte encoding,
- which has ASCII-compatible mappings for the characters in range of 00h to 7fh.
- which don't use ISO2022 escape sequences.
- which don't use a value from 00h to 7fh in any of the compounded bytes that represents a single character.

These are examples of character encodings that are unlikely to work with PHP.

JIS, SJIS, ISO-2022-JP, BIG-5

Although PHP scripts written in any of those encodings might not work, especially in the case where encoded strings appear as identifiers or literals in the script, you can almost avoid using these encodings by setting up the mbstring's transparent encoding filter function for incoming HTTP queries.

Notatka: It's highly discouraged to use SJIS, BIG5, CP936, CP949 and GB18030 for the internal encoding unless you are familiar with the parser, the scanner and the character encoding.

Notatka: If you have some database connected with PHP, it is recommended that you use the same character encoding for both database and the internal encoding for ease of use and better performance.
If you are using PostgreSQL, the character encoding used in the database and the one used in the PHP may differ as it supports automatic character set conversion between the backend and the frontend.

Instalacja

mbstring is a non-default extension. This means it is not enabled by default. You must explicitly enable the module with the configure option. See the Install section for details.

The following configure options are related to the mbstring module.

--enable-mbstring=LANG: Enable mbstring functions. This option is required to use mbstring functions.
As of PHP 4.3.0, mbstring extension provides enhanced support for Simplified Chinese, Traditional Chinese, Korean, and Russian in addition to Japanese. To enable that feature, you will have to supply either one of the following options to the LANG parameter; --enable-mbstring=cn for Simplified Chinese support, --enable-mbstring=tw for Traditional Chinese support, --enable-mbstring=kr for Korean support, --enable-mbstring=ru for Russian support, and --enable-mbstring=ja for Japanese support.
Also --enable-mbstring=all is convenient for you to enable all the supported languages listed above.
Notatka: Japanese language support is also enabled by --enable-mbstring without any options for the sake of backwards compatibility.
--enable-mbstr-enc-trans : Enable HTTP input character encoding conversion using mbstring conversion engine. If this feature is enabled, HTTP input character encoding may be converted to mbstring.internal_encoding automatically.
Notatka: As of PHP 4.3.0, the option --enable-mbstr-enc-trans was eliminated and replaced with the runtime setting mbstring.encoding_translation. HTTP input character encoding conversion is enabled when this is set to On (the default is Off).
--enable-mbregex: Enable regular expression functions with multibyte character support.

Konfiguracja czasu wykonywania

Na działanie tych funcji wpływają ustawienia zawarte w pliku php.ini.

Tabela 1. mbstring configuration options

Name	Default	Changeable	Changelog
mbstring.language	"neutral"	PHP_INI_PERDIR	Available since PHP 4.3.0.
mbstring.detect_order	NULL	PHP_INI_ALL	Available since PHP 4.0.6.
mbstring.http_input	"pass"	PHP_INI_ALL	Available since PHP 4.0.6.
mbstring.http_output	"pass"	PHP_INI_ALL	Available since PHP 4.0.6.
mbstring.internal_encoding	NULL	PHP_INI_ALL	Available since PHP 4.0.6.
mbstring.script_encoding	NULL	PHP_INI_ALL	Available since PHP 4.3.0.
mbstring.substitute_character	NULL	PHP_INI_ALL	Available since PHP 4.0.6.
mbstring.func_overload	"0"	PHP_INI_PERDIR	PHP_INI_SYSTEM in PHP <= 4.2.3. Available since PHP 4.2.0.
mbstring.encoding_translation	"0"	PHP_INI_PERDIR	Available since PHP 4.3.0.

For the definition of the PHP_INI_* constants, please refer to ini_set().

Oto krótkie wyjaśnienie dyrektyw konfiguracji.

mbstring.language string: The default national language setting (NLS) used in mbstring. Note that this option automagically defines mbstring.internal_encoding and mbstring.internal_encoding should be placed after mbstring.language in php.ini
mbstring.encoding_translation boolean: Enables the transparent character encoding filter for the incoming HTTP queries, which performs detection and conversion of the input encoding to the internal character encoding.
mbstring.internal_encoding string: Defines the default internal character encoding.
mbstring.http_input string: Defines the default HTTP input character encoding.
mbstring.http_output string: Defines the default HTTP output character encoding.
mbstring.detect_order string: Defines default character code detection order. See also mb_detect_order().
mbstring.substitute_character string: Defines character to substitute for invalid character encoding.
mbstring.func_overload string: Overloads a set of single byte functions by the mbstring counterparts. See Function overloading for more information.

According to the HTML 4.01 specification, Web browsers are allowed to encode a form being submitted with a character encoding different from the one used for the page. See mb_http_input() to detect character encoding used by browsers.

Although popular browsers are capable of giving a reasonably accurate guess to the character encoding of a given HTML document, it would be better to set the charset parameter in the Content-Type HTTP header to the appropriate value by header() or default_charset ini setting.

Przykład 1. php.ini setting examples

; Set default language
mbstring.language        = Neutral; Set default language to Neutral(UTF-8) (default)
mbstring.language        = English; Set default language to English 
mbstring.language        = Japanese; Set default language to Japanese

;; Set default internal encoding
;; Note: Make sure to use character encoding works with PHP
mbstring.internal_encoding    = UTF-8  ; Set internal encoding to UTF-8

;; HTTP input encoding translation is enabled.
mbstring.encoding_translation = On

;; Set default HTTP input character encoding
;; Note: Script cannot change http_input setting.
mbstring.http_input           = pass    ; No conversion. 
mbstring.http_input           = auto    ; Set HTTP input to auto
                                ; "auto" is expanded to "ASCII,JIS,UTF-8,EUC-JP,SJIS"
mbstring.http_input           = SJIS    ; Set HTTP2 input to  SJIS
mbstring.http_input           = UTF-8,SJIS,EUC-JP ; Specify order

;; Set default HTTP output character encoding 
mbstring.http_output          = pass    ; No conversion
mbstring.http_output          = UTF-8   ; Set HTTP output encoding to UTF-8

;; Set default character encoding detection order
mbstring.detect_order         = auto    ; Set detect order to auto
mbstring.detect_order         = ASCII,JIS,UTF-8,SJIS,EUC-JP ; Specify order

;; Set default substitute character
mbstring.substitute_character = 12307   ; Specify Unicode value
mbstring.substitute_character = none    ; Do not print character
mbstring.substitute_character = long    ; Long Example: U+3000,JIS+7E7E

Przykład 2. php.ini setting for EUC-JP users

;; Disable Output Buffering
output_buffering      = Off

;; Set HTTP header charset
default_charset       = EUC-JP    

;; Set default language to Japanese
mbstring.language = Japanese

;; HTTP input encoding translation is enabled.
mbstring.encoding_translation = On

;; Set HTTP input encoding conversion to auto
mbstring.http_input   = auto 

;; Convert HTTP output to EUC-JP
mbstring.http_output  = EUC-JP    

;; Set internal encoding to EUC-JP
mbstring.internal_encoding = EUC-JP    

;; Do not print invalid characters
mbstring.substitute_character = none

Przykład 3. php.ini setting for SJIS users

;; Enable Output Buffering
output_buffering     = On

;; Set mb_output_handler to enable output conversion
output_handler       = mb_output_handler

;; Set HTTP header charset
default_charset      = Shift_JIS

;; Set default language to Japanese
mbstring.language = Japanese

;; Set http input encoding conversion to auto
mbstring.http_input  = auto 

;; Convert to SJIS
mbstring.http_output = SJIS    

;; Set internal encoding to EUC-JP
mbstring.internal_encoding = EUC-JP    

;; Do not print invalid characters
mbstring.substitute_character = none

Typy zasobów

To rozszerzenie nie posiada żadnych rodzajów zasobów.

Stałe predefinopwane

Poniższe stałe są zdefiniowane w tym rozszerzeniu i stają się dostępne, gdy rozszerzenie jest dokompilowane do PHP, lub załadowane dynamicznie przy starcie.

MB_OVERLOAD_MAIL (integer)
MB_OVERLOAD_STRING (integer)
MB_OVERLOAD_REGEX (integer)

HTTP Input and Output

HTTP input/output character encoding conversion may convert binary data also. Users are supposed to control character encoding conversion if binary data is used for HTTP input/output.

Notatka: In PHP 4.3.2 or earlier versions, there was a limitation in this functionality that mbstring does not perform character encoding conversion in POST data if the enctype attribute in the form element is set to multipart/form-data. So you have to convert the incoming data by yourself in this case if necessary.
Beginning with PHP 4.3.3, if enctype for HTML form is set to multipart/form-data and mbstring.encoding_translation is set to On in php.ini the POST'ed variables and the names of uploaded files will be converted to the internal character encoding as well. However, the conversion isn't applied to the query keys.

HTTP Input

There is no way to control HTTP input character conversion from PHP script. To disable HTTP input character conversion, it has to be done in php.ini.

Przykład 4. Disable HTTP input conversion in php.ini


;; Disable HTTP Input conversion

mbstring.http_input = pass

;; Disable HTTP Input conversion (PHP 4.3.0 or higher)

mbstring.encoding_translation = Off

When using PHP as an Apache module, it is possible to override those settings in each Virtual Host directive in httpd.conf or per directory with .htaccess. Refer to the Configuration section and Apache Manual for details.

HTTP Output
There are several ways to enable output character encoding conversion. One is using php.ini, another is using ob_start() with mb_output_handler() as ob_start callback function.
Notatka: PHP3-i18n users should note that mbstring's output conversion differs from PHP3-i18n. Character encoding is converted using output buffer.

Przykład 5. php.ini setting example

;; Enable output character encoding conversion for all PHP pages

;; Enable Output Buffering
output_buffering    = On

;; Set mb_output_handler to enable output conversion
output_handler      = mb_output_handler

Przykład 6. Script example


<?php



// Enable output character encoding conversion only for this page



// Set HTTP output character encoding to SJIS

mb_http_output('SJIS');



// Start buffering and specify "mb_output_handler" as

// callback function

ob_start('mb_output_handler');



?>

Supported Character Encodings

Currently the following character encodings are supported by the mbstring module. Any of those Character encodings can be specified in the encoding parameter of mbstring functions.

The following character encoding is supported in this PHP extension:

UCS-4
UCS-4BE
UCS-4LE
UCS-2
UCS-2BE
UCS-2LE
UTF-32
UTF-32BE
UTF-32LE
UTF-16
UTF-16BE
UTF-16LE
UTF-7
UTF7-IMAP
UTF-8
ASCII
EUC-JP
SJIS
eucJP-win
SJIS-win
ISO-2022-JP
JIS
ISO-8859-1
ISO-8859-2
ISO-8859-3
ISO-8859-4
ISO-8859-5
ISO-8859-6
ISO-8859-7
ISO-8859-8
ISO-8859-9
ISO-8859-10
ISO-8859-13
ISO-8859-14
ISO-8859-15
byte2be
byte2le
byte4be
byte4le
BASE64
HTML-ENTITIES
7bit
8bit
EUC-CN
CP936
HZ
EUC-TW
CP950
BIG-5
EUC-KR
UHC (CP949)
ISO-2022-KR
Windows-1251 (CP1251)
Windows-1252 (CP1252)
CP866 (IBM866)
KOI8-R

php.ini entry, which accepts encoding name, accepts "auto" and "pass" also. mbstring functions, which accepts encoding name, and accepts "auto".

If "pass" is set, no character encoding conversion is performed.

If "auto" is set, it is expanded to the list of encodings defined per the NLS. For instance, if the NLS is set to Japanese, the value is assumed to be "ASCII,JIS,UTF-8,EUC-JP,SJIS".

Function Overloading Feature

You might often find it difficult to get an existing PHP application work in a given multibyte environment. That's mostly because lots of PHP applications out there are written with the standard string functions such as substr(), which are known to not properly handle multibyte-encoded strings.

mbstring supports 'function overloading' feature which enables you to add multibyte awareness to such an application without code modification by overloading multibyte counterparts on the standard string functions. For example, mb_substr() is called instead of substr() if function overloading is enabled. This feature makes it easy to port applications that only support single-byte encodings to a multibyte environment in many cases.

To use the function overloading, set mbstring.func_overload in php.ini to a positive value that represents a combination of bitmasks specifying the categories of functions to be overloaded. It should be set to 1 to overload the mail() function. 2 for string functions, 4 for regular expression functions. For example, if is set for 7, mail, strings and regular expression functions should be overloaded. The list of overloaded functions are shown below.

Tabela 2. Functions to be overloaded

value of mbstring.func_overload	original function	overloaded function
1	mail()	mb_send_mail()
2	strlen()	mb_strlen()
2	strpos()	mb_strpos()
2	strrpos()	mb_strrpos()
2	substr()	mb_substr()
2	strtolower()	mb_strtolower()
2	strtoupper()	mb_strtoupper()
2	substr_count()	mb_substr_count()
4	ereg()	mb_ereg()
4	eregi()	mb_eregi()
4	ereg_replace()	mb_ereg_replace()
4	eregi_replace()	mb_eregi_replace()
4	split()	mb_split()

Notatka: It is not recommended to use the function overloading option in the per-directory context, because it's not confirmed yet to be stable enough in a production environment and may lead to undefined behaviour.

Basics of Japanese multi-byte encodings

It is often said quite hard to figure out how Japanese texts are handled in the computer. This is not only because Japanese characters can only be represented by multibyte encodings, but because different encoding standards are adopted for different purposes / platforms. Moreover, not a few character set standards are used there, which are slightly different from one another. Those facts have often led developers to inevitable mess-up.

To create a working web application that would be put in the Japanese environment, it is important to use the proper character encoding and character set for the task in hand.

Storage for a character can be up to six bytes
Most of multibyte characters often appear twice as wide as a single-byte character on display. Those characters are called "zen-kaku" in Japanese which means "full width", and the other (narrower) characters are called "han-kaku" - means half width. However the graphical properties of the characters depend on the glyphs of the type faces used to display them or print them out.
Some character encodings use shift(escape) sequences defined in ISO2022 to switch the code map of the specific code area (00h to 7fh).
ISO-2022-JP should be used in SMTP/NNTP, and headers and entities should be reencoded as per RFC requirements. Although those are not requisites, it's still a good idea because several popular user agents cannot recognize any other encoding methods.
Webpages created for mobile phone services such as i-mode, Vodafone live!, or EZweb are supposed to use Shift_JIS.

References

Multibyte character encoding schemes and the related issues are very complicated. There should be too few space to cover in sufficient details. Please refer to the following URLs and other resources for further readings.

Unicode materials
http://www.unicode.org/
Japanese/Korean/Chinese character information
http://examples.oreilly.com/cjkvinfo/doc/cjk.inf

Summaries of supported encodings

Summaries of supported encodings

Name in the IANA character set registry: ISO-10646-UCS-4

Underlying character set: ISO 10646

Description: The Universal Character Set with 31-bit code space, standardized as UCS-4 by ISO/IEC 10646. It is kept synchronized with the latest version of the Unicode code map.

Additional note: If this name is used in the encoding conversion facility, the converter attempts to identify by the preceding BOM (byte order mark)in which endian the subsequent bytes are represented.

Name in the IANA character set registry: ISO-10646-UCS-4

Underlying character set: UCS-4

Description: See above.

Additional note: In contrast to UCS-4, strings are always assumed to be in big endian form.

Name in the IANA character set registry: ISO-10646-UCS-4

Underlying character set: UCS-4

Description: See above.

Additional note: In contrast to UCS-4, strings are always assumed to be in little endian form.

Name in the IANA character set registry: ISO-10646-UCS-2

Underlying character set: UCS-2

Description: The Universal Character Set with 16-bit code space, standardized as UCS-2 by ISO/IEC 10646. It is kept synchronized with the latest version of the unicode code map.

Additional note: If this name is used in the encoding conversion facility, the converter attempts to identify by the preceding BOM (byte order mark)in which endian the subsequent bytes are represented.

Name in the IANA character set registry: ISO-10646-UCS-2

Underlying character set: UCS-2

Description: See above.

Additional note: In contrast to UCS-2, strings are always assumed to be in big endian form.

Name in the IANA character set registry: ISO-10646-UCS-2

Underlying character set: UCS-2

Description: See above.

Additional note: In contrast to UCS-2, strings are always assumed to be in little endian form.

Name in the IANA character set registry: UTF-32

Underlying character set: Unicode

Description: Unicode Transformation Format of 32-bit unit width, whose encoding space refers to the Unicode's codeset standard. This encoding scheme wasn't identical to UCS-4 because the code space of Unicode were limited to a 21-bit value.

Additional note: If this name is used in the encoding conversion facility, the converter attempts to identify by the preceding BOM (byte order mark)in which endian the subsequent bytes are represented.

Name in the IANA character set registry: UTF-32BE

Underlying character set: Unicode

Description: See above

Additional note: In contrast to UTF-32, strings are always assumed to be in big endian form.

Name in the IANA character set registry: UTF-32LE

Underlying character set: Unicode

Description: See above

Additional note: In contrast to UTF-32, strings are always assumed to be in little endian form.

Name in the IANA character set registry: UTF-16

Underlying character set: Unicode

Description: Unicode Transformation Format of 16-bit unit width. It's worth a note that UTF-16 is no longer the same specification as UCS-2 because the surrogate mechanism has been introduced since Unicode 2.0 and UTF-16 now refers to a 21-bit code space.

Additional note: If this name is used in the encoding conversion facility, the converter attempts to identify by the preceding BOM (byte order mark)in which endian the subsequent bytes are represented.

Name in the IANA character set registry: UTF-16BE

Underlying character set: Unicode

Description: See above.

Additional note: In contrast to UTF-16, strings are always assumed to be in big endian form.

Name in the IANA character set registry: UTF-16BE

Underlying character set: Unicode

Description: See above.

Additional note: In contrast to UTF-16, strings are always assumed to be in big endian form.

Name in the IANA character set registry: UTF-8

Underlying character set: Unicode / UCS

Description: Unicode Transformation Format of 8-bit unit width.

Additional note: none

Name in the IANA character set registry: UTF-7

Underlying character set: Unicode

Description: A mail-safe transformation format of Unicode, specified in RFC2152.

Additional note: none

Name in the IANA character set registry: (none)

Underlying character set: Unicode

Description: A variant of UTF-7 which is specialized for use in the IMAP protocol.

Additional note: none

Name in the IANA character set registry: US-ASCII (preferred MIME name) / iso-ir-6 / ANSI_X3.4-1986 / ISO_646.irv:1991 / ASCII / ISO646-US / us / IBM367 / CP367 / csASCII

Underlying character set: ASCII / ISO 646

Description: American Standard Code for Information Interchange is a commonly-used 7-bit encoding. Also standardized as an international standard, ISO 646.

Additional note: (none)

Name in the IANA character set registry: EUC-JP (preferred MIME name) / Extended_UNIX_Code_Packed_Format_for_Japanese / csEUCPkdFmtJapanese

Underlying character set: Compound of US-ASCII / JIS X0201:1997 (hankaku kana part) / JIS X0208:1990 / JIS X0212:1990

Description: As you see the name is derived from an abbreviation of Extended UNIX Code Packed Format for Japanese, this encoding is mostly used on UNIX or alike platforms. The original encoding scheme, Extended UNIX Code, is designed on the basis of ISO 2022.

Additional note: The character set referred to by EUC-JP is different to IBM932 / CP932, which are used by OS/2

LXV. Multibyte String Functions