CharsetConverter::findSupportedCharset() - Code Metrics - Inspection of "check if the convert chatset includes UTF-8, take..." - zbateson/mail-mime-parser - Measure and Improve Code Quality continuously with Scrutinizer

Completed

Pull Request — master (#54)

unknown

created 2017-10-31 08:12 UTC

CharsetConverter::findSupportedCharset() A

↳ Parent: CharsetConverter

Complexity

Conditions	4
Paths	6

Size

Total Lines	19
Code Lines	12

Duplication

Lines	0
Ratio	0 %

Code Coverage

Tests	12
CRAP Score	4.0072

Importance

Changes	1
Bugs	0	Features	0

Metric	Value
c	1
b	0
f	0
dl	0
loc	19
ccs	12
cts	13
cp	0.9231
rs	9.2
cc	4
eloc	12
nc	6
nop	2
crap	4.0072

<?php
/**
 * This file is part of the ZBateson\MailMimeParser project.
 *
 * @license http => //opensource.org/licenses/bsd-license.php BSD
 */
namespace ZBateson\MailMimeParser\Stream\Helper;

/**
 * Helper class for converting strings between charsets.
 * 
 * CharasetConverter tries to convert using mb_convert_encoding when possible,
 * defining as many aliases as possible for supported encodings.  If not
 * supported, iconv is attempted.
 *
 * @author Zaahid Bateson
 */
class CharsetConverter
{
    /**
     * @var array aliased charsets supported by mb_convert_encoding.
     *      The alias is stripped of any non-alphanumeric characters (so CP367
     *      is equal to CP-367) when comparing.
     *      Some of these translations are already supported by
     *      mb_convert_encoding on "my" PHP 5.5.9, but may not be supported in
     *      other implementations or versions since they're not part of
     *      documented support.
     */
    public static $mbAliases = [
        // supported but not included in mb_list_encodings for some reason...
        'CP850' => 'CP850',
        'GB2312' => 'GB2312',
        // aliases
        '646' => 'ASCII',
        'ANSIX341968' => 'ASCII',
        'ANSIX341986' => 'ASCII',
        'CP367' => 'ASCII',
        'CSASCII' => 'ASCII',
        'IBM367' => 'ASCII',
        'ISO646US' => 'ASCII',
        'ISO646IRV1991' => 'ASCII',
        'ISOIR6' => 'ASCII',
        'US' => 'ASCII',
        'USASCII' => 'ASCII',
        'BIG5' => 'BIG-5',
        'BIG5TW' => 'BIG-5',
        'CSBIG5' => 'BIG-5',
        '1251' => 'WINDOWS-1251',
        'CP1251' => 'WINDOWS-1251',
        'WINDOWS1251' => 'WINDOWS-1251',
        '1252' => 'WINDOWS-1252',
        'CP1252' => 'WINDOWS-1252',
        'WINDOWS1252' => 'WINDOWS-1252',
        'WE8MSWIN1252' => 'WINDOWS-1252',
        '1254' => 'WINDOWS-1254',
        'CP1254' => 'WINDOWS-1254',
        'WINDOWS1254' => 'WINDOWS-1254',
        '1255' => 'ISO-8859-8',
        'CP1255' => 'ISO-8859-8',
        'ISO88598I' => 'ISO-8859-8',
        'WINDOWS1255' => 'ISO-8859-8',
        '850' => 'CP850',
        'CSPC850MULTILINGUAL' => 'CP850',
        'IBM850' => 'CP850',
        '866' => 'CP866',
        'CSIBM866' => 'CP866',
        'IBM866' => 'CP866',
        '932' => 'CP932',
        'MS932' => 'CP932',
        'MSKANJI' => 'CP932',
        '950' => 'CP950',
        'MS950' => 'CP950',
        'EUCJP' => 'EUC-JP',
        'UJIS' => 'EUC-JP',
        'EUCKR' => 'EUC-KR',
        'KOREAN' => 'EUC-KR',
        'KSC5601' => 'EUC-KR',
        'KSC56011987' => 'EUC-KR',
        'KSX1001' => 'EUC-KR',
        'GB180302000' => 'GB18030',
        // GB2312 not listed but supported
        'CHINESE' => 'GB2312',
        'CSISO58GB231280' => 'GB2312',
        'EUCCN' => 'GB2312',
        'EUCGB2312CN' => 'GB2312',
        'GB23121980' => 'GB2312',
        'GB231280' => 'GB2312',
        'ISOIR58' => 'GB2312',
        'GBK' => 'CP936',
        '936' => 'CP936',
        'ms936' => 'CP936',
        'HZGB' => 'HZ',
        'HZGB2312' => 'HZ',
        'CSISO2022JP' => 'ISO-2022-JP',
        'ISO2022JP' => 'ISO-2022-JP',
        'ISO2022JP2004' => 'ISO-2022-JP-2004',
        'CSISO2022KR' => 'ISO-2022-KR',
        'ISO2022KR' => 'ISO-2022-KR',
        'CSISOLATIN6' => 'ISO-8859-10',
        'ISO885910' => 'ISO-8859-10',
        'ISO8859101992' => 'ISO-8859-10',
        'ISOIR157' => 'ISO-8859-10',
        'L6' => 'ISO-8859-10',
        'LATIN6' => 'ISO-8859-10',
        'ISO885913' => 'ISO-8859-13',
        'ISO885914' => 'ISO-8859-14',
        'ISO8859141998' => 'ISO-8859-14',
        'ISOCELTIC' => 'ISO-8859-14',
        'ISOIR199' => 'ISO-8859-14',
        'L8' => 'ISO-8859-14',
        'LATIN8' => 'ISO-8859-14',
        'ISO885915' => 'ISO-8859-15',
        'ISO885916' => 'ISO-8859-16',
        'ISO8859162001' => 'ISO-8859-16',
        'ISOIR226' => 'ISO-8859-16',
        'L10' => 'ISO-8859-16',
        'LATIN10' => 'ISO-8859-16',
        'CSISOLATIN2' => 'ISO-8859-2',
        'ISO88592' => 'ISO-8859-2',
        'ISO885921987' => 'ISO-8859-2',
        'ISOIR101' => 'ISO-8859-2',
        'L2' => 'ISO-8859-2',
        'LATIN2' => 'ISO-8859-2',
        'CSISOLATIN3' => 'ISO-8859-3',
        'ISO88593' => 'ISO-8859-3',
        'ISO885931988' => 'ISO-8859-3',
        'ISOIR109' => 'ISO-8859-3',
        'L3' => 'ISO-8859-3',
        'LATIN3' => 'ISO-8859-3',
        'CSISOLATIN4' => 'ISO-8859-4',
        'ISO88594' => 'ISO-8859-4',
        'ISO885941988' => 'ISO-8859-4',
        'ISOIR110' => 'ISO-8859-4',
        'L4' => 'ISO-8859-4',
        'LATIN4' => 'ISO-8859-4',
        'CSISOLATINCYRILLIC' => 'ISO-8859-5',
        'CYRILLIC' => 'ISO-8859-5',
        'ISO88595' => 'ISO-8859-5',
        'ISO885951988' => 'ISO-8859-5',
        'ISOIR144' => 'ISO-8859-5',
        'ARABIC' => 'ISO-8859-6',
        'ASMO708' => 'ISO-8859-6',
        'CSISOLATINARABIC' => 'ISO-8859-6',
        'ECMA114' => 'ISO-8859-6',
        'ISO88596' => 'ISO-8859-6',
        'ISO885961987' => 'ISO-8859-6',
        'ISOIR127' => 'ISO-8859-6',
        'CSISOLATINGREEK' => 'ISO-8859-7',
        'ECMA118' => 'ISO-8859-7',
        'ELOT928' => 'ISO-8859-7',
        'GREEK' => 'ISO-8859-7',
        'GREEK8' => 'ISO-8859-7',
        'ISO88597' => 'ISO-8859-7',
        'ISO885971987' => 'ISO-8859-7',
        'ISOIR126' => 'ISO-8859-7',
        'CSISOLATINHEBREW' => 'ISO-8859-8',
        'HEBREW' => 'ISO-8859-8',
        'ISO88598' => 'ISO-8859-8',
        'ISO885981988' => 'ISO-8859-8',
        'ISOIR138' => 'ISO-8859-8',
        'CSISOLATIN5' => 'ISO-8859-9',
        'ISO88599' => 'ISO-8859-9',
        'ISO885991989' => 'ISO-8859-9',
        'ISOIR148' => 'ISO-8859-9',
        'L5' => 'ISO-8859-9',
        'LATIN5' => 'ISO-8859-9',
        'CSKOI8R' => 'KOI8-R',
        'KOI8R' => 'KOI8-R',
        '8859' => 'ISO-8859-1',
        'CP819' => 'ISO-8859-1',
        'CSISOLATIN1' => 'ISO-8859-1',
        'IBM819' => 'ISO-8859-1',
        'ISO8859' => 'ISO-8859-1',
        'ISO88591' => 'ISO-8859-1',
        'ISO885911987' => 'ISO-8859-1',
        'ISOIR100' => 'ISO-8859-1',
        'L1' => 'ISO-8859-1',
        'LATIN' => 'ISO-8859-1',
        'LATIN1' => 'ISO-8859-1',
        'CSSHIFTJIS' => 'SJIS',
        'SHIFTJIS' => 'SJIS',
        'SHIFTJIS2004' => 'SJIS-2004',
        'SJIS2004' => 'SJIS-2004',
    ];
    
    /**
     * @var array aliased charsets supported by iconv.
     */
    public static $iconvAliases = [
        // iconv aliases -- a lot of these may already be supported
        'BIG5HKSCS' => 'BIG5HKSCS',
        'HKSCS' => 'BIG5HKSCS',
        '037' => 'CP037',
        'EBCDICCPCA' => 'CP037',
        'EBCDICCPNL' => 'CP037',
        'EBCDICCPUS' => 'CP037',
        'EBCDICCPWT' => 'CP037',
        'CSIBM037' => 'CP037',
        'IBM037' => 'CP037',
        'IBM039' => 'CP037',
        '1026' => 'CP1026',
        'CSIBM1026' => 'CP1026',
        'IBM1026' => 'CP1026',
        '1140' => 'CP1140',
        'IBM1140' => 'CP1140',
        '1250' => 'CP1250',
        'WINDOWS1250' => 'CP1250',
        '1253' => 'CP1253',
        'WINDOWS1253' => 'CP1253',
        '1256' => 'CP1256',
        'WINDOWS1256' => 'CP1256',
        '1257' => 'CP1257',
        'WINDOWS1257' => 'CP1257',
        '1258' => 'CP1258',
        'WINDOWS1258' => 'CP1258',
        '424' => 'CP424',
        'CSIBM424' => 'CP424',
        'EBCDICCPHE' => 'CP424',
        'IBM424' => 'CP424',
        '437' => 'CP437',
        'CSPC8CODEPAGE437' => 'CP437',
        'IBM437' => 'CP437',
        '500' => 'CP500',
        'CSIBM500' => 'CP500',
        'EBCDICCPBE' => 'CP500',
        'EBCDICCPCH' => 'CP500',
        'IBM500' => 'CP500',
        '775' => 'CP775',
        'CSPC775BALTIC' => 'CP775',
        'IBM775' => 'CP775',
        '860' => 'CP860',
        'CSIBM860' => 'CP860',
        'IBM860' => 'CP860',
        '861' => 'CP861',
        'CPIS' => 'CP861',
        'CSIBM861' => 'CP861',
        'IBM861' => 'CP861',
        '862' => 'CP862',
        'CSPC862LATINHEBREW' => 'CP862',
        'IBM862' => 'CP862',
        '863' => 'CP863',
        'CSIBM863' => 'CP863',
        'IBM863' => 'CP863',
        '864' => 'CP864',
        'CSIBM864' => 'CP864',
        'IBM864' => 'CP864',
        '865' => 'CP865',
        'CSIBM865' => 'CP865',
        'IBM865' => 'CP865',
        '869' => 'CP869',
        'CPGR' => 'CP869',
        'CSIBM869' => 'CP869',
        'IBM869' => 'CP869',
        '949' => 'CP949',
        'MS949' => 'CP949',
        'UHC' => 'CP949',
        'ROMAN8' => 'ROMAN8',
        'HPROMAN8' => 'ROMAN8',
        'R8' => 'ROMAN8',
        'CSHPROMAN8' => 'ROMAN8',
        'ISO2022JP2' => 'ISO2022JP2',
        'THAI' => 'ISO885911',
        'ISO885911' => 'ISO885911',
        'ISO8859112001' => 'ISO885911',
        'JOHAB' => 'CP1361',
        'MS1361' => 'CP1361',
        'MACCYRILLIC' => 'MACCYRILLIC',
        'CSPTCP154' => 'PT154',
        'PTCP154' => 'PT154',
        'CP154' => 'PT154',
        'CYRILLICASIAN' => 'PT154',
        'TIS620' => 'TIS620',
        'TIS6200' => 'TIS620',
        'TIS62025290' => 'TIS620',
        'TIS62025291' => 'TIS620',
        'ISOIR166' => 'TIS620',
    ];
    
    /**
     * @var string charset to convert from
     */
    protected $fromCharset;
    
    /**
     * @var string charset to convert to
     */
    protected $toCharset;
    
    /**
     * @var boolean indicates if $fromCharset is supported by
     * mb_convert_encoding
     */
    protected $fromCharsetMbSupported = true;
    
    /**
     * @var boolean indicates if $toCharset is supported by mb_convert_encoding
     */
    protected $toCharsetMbSupported = true;
    
    /**
     * Constructs the charset converter with source/destination charsets.
     * 
     * @param string $fromCharset
     * @param string $toCharset
     */
    public function __construct($fromCharset, $toCharset)
    {
        $this->fromCharset = $this->findSupportedCharset($fromCharset, $this->fromCharsetMbSupported);
        $this->toCharset = $this->findSupportedCharset($toCharset, $this->toCharsetMbSupported);
    }
    
    /**
     * Converts the passed string's charset from $this->fromCharset to
     * $this->toCharset.
     * 
     * The function attempts to use mb_convert_encoding if possible, and falls
     * back to iconv if not.  If the source or destination character sets aren't
     * supported, a blank string is returned.
     * 
     * @param string $str
     * @return string
     */
    public function convert($str)
    {
        // there may be some mb-supported encodings not supported by iconv (on my libiconv for instance
        // HZ isn't supported), and so it may happen that failing an mb_convert_encoding, an iconv
        // may also fail even though both support an encoding separately.
        // Unfortunately there's no great way of testing what charsets are available on iconv, and
        // attempting to blindly convert the string may be too costly, as could converting first
        // to an intermediate (ASSUMPTION: may be worth testing converting to an intermediate)
        if ($str !== '') {
            if ($this->fromCharsetMbSupported && $this->toCharsetMbSupported) {
                return mb_convert_encoding($str, $this->toCharset, $this->fromCharset);
            }
            return iconv($this->fromCharset, $this->toCharset . '//TRANSLIT//IGNORE', $str);
        }
        return $str;
    }
    
    /**
     * Looks up the passed $cs in mb_list_encodings, then strips non
     * alpha-numeric characters and tries again, then failing that calls
     * findAliasedCharset.  The method returns the charset name that should be
     * used in calls to mb_convert_encoding or iconv.
     * 
     * If the charset is part of mb_list_encodings, $mbSupported is set to true.
     * 
     * @param string $cs
     * @param boolean $mbSupported
     * @return string the final charset name to use
     */
    private function findSupportedCharset($cs, &$mbSupported)
    {
        /** @see https://github.com/zbateson/MailMimeParser/issues/53 */
        if (preg_match('/UTF\-8/', $cs)) {
            $cs = 'UTF-8';
        }

        $mbSupported = true;
        $comp = strtoupper($cs);
        $available = array_map('strtoupper', mb_list_encodings());
        if (in_array($comp, $available)) {
            return $comp;
        }
        $stripped = preg_replace('/[^A-Z0-9]+/', '', $comp);
        if (in_array($stripped, $available)) {
            return $stripped;
        }
        return $this->findAliasedCharset($comp, $stripped, $mbSupported);
    }
    
    /**
     * Looks up the passed $comp and $stripped strings in self::$mbAliases, and
     * returns the mapped charset if applicable.  Otherwise calls
     * $this->findAliasedIconvCharset.
     * 
     * $mbSupported is set to false if the charset is not located in
     * self::$mbAliases.
     * 
     * @param string $comp
     * @param string $stripped
     * @param boolean $mbSupported
     * @return string the mapped charset
     */
    private function findAliasedCharset($comp, $stripped, &$mbSupported)
    {
        if (array_key_exists($comp, self::$mbAliases)) {
            return self::$mbAliases[$comp];
        } elseif (array_key_exists($stripped, self::$mbAliases)) {
            return self::$mbAliases[$stripped];
        }
        $mbSupported = false;
        return $this->findAliasedIconvCharset($comp, $stripped);
    }
    
    /**
     * Looks up the passed $comp and $stripped strings in self::$iconvAliases,
     * and returns the mapped charset if applicable.  Otherwise returns $comp.
     * 
     * @param string $comp
     * @param string $stripped
     * @return string the mapped charset (if mapped) or $comp otherwise
     */
    private function findAliasedIconvCharset($comp, $stripped)
    {
        if (array_key_exists($comp, self::$iconvAliases)) {
            return static::$iconvAliases[$comp];
        } elseif (array_key_exists($stripped, self::$iconvAliases)) {
            return static::$iconvAliases[$stripped];
        }
        return $comp;
    }
}


1		<?php
2		/**
3		* This file is part of the ZBateson\MailMimeParser project.
4		*
5		* @license http => //opensource.org/licenses/bsd-license.php BSD
6		*/
7		namespace ZBateson\MailMimeParser\Stream\Helper;
8
9		/**
10		* Helper class for converting strings between charsets.
11		*
12		* CharasetConverter tries to convert using mb_convert_encoding when possible,
13		* defining as many aliases as possible for supported encodings. If not
14		* supported, iconv is attempted.
15		*
16		* @author Zaahid Bateson
17		*/
18		class CharsetConverter
19		{
20		/**
21		* @var array aliased charsets supported by mb_convert_encoding.
22		* The alias is stripped of any non-alphanumeric characters (so CP367
23		* is equal to CP-367) when comparing.
24		* Some of these translations are already supported by
25		* mb_convert_encoding on "my" PHP 5.5.9, but may not be supported in
26		* other implementations or versions since they're not part of
27		* documented support.
28		*/
29		public static $mbAliases = [
30		// supported but not included in mb_list_encodings for some reason...
31		'CP850' => 'CP850',
32		'GB2312' => 'GB2312',
33		// aliases
34		'646' => 'ASCII',
35		'ANSIX341968' => 'ASCII',
36		'ANSIX341986' => 'ASCII',
37		'CP367' => 'ASCII',
38		'CSASCII' => 'ASCII',
39		'IBM367' => 'ASCII',
40		'ISO646US' => 'ASCII',
41		'ISO646IRV1991' => 'ASCII',
42		'ISOIR6' => 'ASCII',
43		'US' => 'ASCII',
44		'USASCII' => 'ASCII',
45		'BIG5' => 'BIG-5',
46		'BIG5TW' => 'BIG-5',
47		'CSBIG5' => 'BIG-5',
48		'1251' => 'WINDOWS-1251',
49		'CP1251' => 'WINDOWS-1251',
50		'WINDOWS1251' => 'WINDOWS-1251',
51		'1252' => 'WINDOWS-1252',
52		'CP1252' => 'WINDOWS-1252',
53		'WINDOWS1252' => 'WINDOWS-1252',
54		'WE8MSWIN1252' => 'WINDOWS-1252',
55		'1254' => 'WINDOWS-1254',
56		'CP1254' => 'WINDOWS-1254',
57		'WINDOWS1254' => 'WINDOWS-1254',
58		'1255' => 'ISO-8859-8',
59		'CP1255' => 'ISO-8859-8',
60		'ISO88598I' => 'ISO-8859-8',
61		'WINDOWS1255' => 'ISO-8859-8',
62		'850' => 'CP850',
63		'CSPC850MULTILINGUAL' => 'CP850',
64		'IBM850' => 'CP850',
65		'866' => 'CP866',
66		'CSIBM866' => 'CP866',
67		'IBM866' => 'CP866',
68		'932' => 'CP932',
69		'MS932' => 'CP932',
70		'MSKANJI' => 'CP932',
71		'950' => 'CP950',
72		'MS950' => 'CP950',
73		'EUCJP' => 'EUC-JP',
74		'UJIS' => 'EUC-JP',
75		'EUCKR' => 'EUC-KR',
76		'KOREAN' => 'EUC-KR',
77		'KSC5601' => 'EUC-KR',
78		'KSC56011987' => 'EUC-KR',
79		'KSX1001' => 'EUC-KR',
80		'GB180302000' => 'GB18030',
81		// GB2312 not listed but supported
82		'CHINESE' => 'GB2312',
83		'CSISO58GB231280' => 'GB2312',
84		'EUCCN' => 'GB2312',
85		'EUCGB2312CN' => 'GB2312',
86		'GB23121980' => 'GB2312',
87		'GB231280' => 'GB2312',
88		'ISOIR58' => 'GB2312',
89		'GBK' => 'CP936',
90		'936' => 'CP936',
91		'ms936' => 'CP936',
92		'HZGB' => 'HZ',
93		'HZGB2312' => 'HZ',
94		'CSISO2022JP' => 'ISO-2022-JP',
95		'ISO2022JP' => 'ISO-2022-JP',
96		'ISO2022JP2004' => 'ISO-2022-JP-2004',
97		'CSISO2022KR' => 'ISO-2022-KR',
98		'ISO2022KR' => 'ISO-2022-KR',
99		'CSISOLATIN6' => 'ISO-8859-10',
100		'ISO885910' => 'ISO-8859-10',
101		'ISO8859101992' => 'ISO-8859-10',
102		'ISOIR157' => 'ISO-8859-10',
103		'L6' => 'ISO-8859-10',
104		'LATIN6' => 'ISO-8859-10',
105		'ISO885913' => 'ISO-8859-13',
106		'ISO885914' => 'ISO-8859-14',
107		'ISO8859141998' => 'ISO-8859-14',
108		'ISOCELTIC' => 'ISO-8859-14',
109		'ISOIR199' => 'ISO-8859-14',
110		'L8' => 'ISO-8859-14',
111		'LATIN8' => 'ISO-8859-14',
112		'ISO885915' => 'ISO-8859-15',
113		'ISO885916' => 'ISO-8859-16',
114		'ISO8859162001' => 'ISO-8859-16',
115		'ISOIR226' => 'ISO-8859-16',
116		'L10' => 'ISO-8859-16',
117		'LATIN10' => 'ISO-8859-16',
118		'CSISOLATIN2' => 'ISO-8859-2',
119		'ISO88592' => 'ISO-8859-2',
120		'ISO885921987' => 'ISO-8859-2',
121		'ISOIR101' => 'ISO-8859-2',
122		'L2' => 'ISO-8859-2',
123		'LATIN2' => 'ISO-8859-2',
124		'CSISOLATIN3' => 'ISO-8859-3',
125		'ISO88593' => 'ISO-8859-3',
126		'ISO885931988' => 'ISO-8859-3',
127		'ISOIR109' => 'ISO-8859-3',
128		'L3' => 'ISO-8859-3',
129		'LATIN3' => 'ISO-8859-3',
130		'CSISOLATIN4' => 'ISO-8859-4',
131		'ISO88594' => 'ISO-8859-4',
132		'ISO885941988' => 'ISO-8859-4',
133		'ISOIR110' => 'ISO-8859-4',
134		'L4' => 'ISO-8859-4',
135		'LATIN4' => 'ISO-8859-4',
136		'CSISOLATINCYRILLIC' => 'ISO-8859-5',
137		'CYRILLIC' => 'ISO-8859-5',
138		'ISO88595' => 'ISO-8859-5',
139		'ISO885951988' => 'ISO-8859-5',
140		'ISOIR144' => 'ISO-8859-5',
141		'ARABIC' => 'ISO-8859-6',
142		'ASMO708' => 'ISO-8859-6',
143		'CSISOLATINARABIC' => 'ISO-8859-6',
144		'ECMA114' => 'ISO-8859-6',
145		'ISO88596' => 'ISO-8859-6',
146		'ISO885961987' => 'ISO-8859-6',
147		'ISOIR127' => 'ISO-8859-6',
148		'CSISOLATINGREEK' => 'ISO-8859-7',
149		'ECMA118' => 'ISO-8859-7',
150		'ELOT928' => 'ISO-8859-7',
151		'GREEK' => 'ISO-8859-7',
152		'GREEK8' => 'ISO-8859-7',
153		'ISO88597' => 'ISO-8859-7',
154		'ISO885971987' => 'ISO-8859-7',
155		'ISOIR126' => 'ISO-8859-7',
156		'CSISOLATINHEBREW' => 'ISO-8859-8',
157		'HEBREW' => 'ISO-8859-8',
158		'ISO88598' => 'ISO-8859-8',
159		'ISO885981988' => 'ISO-8859-8',
160		'ISOIR138' => 'ISO-8859-8',
161		'CSISOLATIN5' => 'ISO-8859-9',
162		'ISO88599' => 'ISO-8859-9',
163		'ISO885991989' => 'ISO-8859-9',
164		'ISOIR148' => 'ISO-8859-9',
165		'L5' => 'ISO-8859-9',
166		'LATIN5' => 'ISO-8859-9',
167		'CSKOI8R' => 'KOI8-R',
168		'KOI8R' => 'KOI8-R',
169		'8859' => 'ISO-8859-1',
170		'CP819' => 'ISO-8859-1',
171		'CSISOLATIN1' => 'ISO-8859-1',
172		'IBM819' => 'ISO-8859-1',
173		'ISO8859' => 'ISO-8859-1',
174		'ISO88591' => 'ISO-8859-1',
175		'ISO885911987' => 'ISO-8859-1',
176		'ISOIR100' => 'ISO-8859-1',
177		'L1' => 'ISO-8859-1',
178		'LATIN' => 'ISO-8859-1',
179		'LATIN1' => 'ISO-8859-1',
180		'CSSHIFTJIS' => 'SJIS',
181		'SHIFTJIS' => 'SJIS',
182		'SHIFTJIS2004' => 'SJIS-2004',
183		'SJIS2004' => 'SJIS-2004',
184		];
185
186		/**
187		* @var array aliased charsets supported by iconv.
188		*/
189		public static $iconvAliases = [
190		// iconv aliases -- a lot of these may already be supported
191		'BIG5HKSCS' => 'BIG5HKSCS',
192		'HKSCS' => 'BIG5HKSCS',
193		'037' => 'CP037',
194		'EBCDICCPCA' => 'CP037',
195		'EBCDICCPNL' => 'CP037',
196		'EBCDICCPUS' => 'CP037',
197		'EBCDICCPWT' => 'CP037',
198		'CSIBM037' => 'CP037',
199		'IBM037' => 'CP037',
200		'IBM039' => 'CP037',
201		'1026' => 'CP1026',
202		'CSIBM1026' => 'CP1026',
203		'IBM1026' => 'CP1026',
204		'1140' => 'CP1140',
205		'IBM1140' => 'CP1140',
206		'1250' => 'CP1250',
207		'WINDOWS1250' => 'CP1250',
208		'1253' => 'CP1253',
209		'WINDOWS1253' => 'CP1253',
210		'1256' => 'CP1256',
211		'WINDOWS1256' => 'CP1256',
212		'1257' => 'CP1257',
213		'WINDOWS1257' => 'CP1257',
214		'1258' => 'CP1258',
215		'WINDOWS1258' => 'CP1258',
216		'424' => 'CP424',
217		'CSIBM424' => 'CP424',
218		'EBCDICCPHE' => 'CP424',
219		'IBM424' => 'CP424',
220		'437' => 'CP437',
221		'CSPC8CODEPAGE437' => 'CP437',
222		'IBM437' => 'CP437',
223		'500' => 'CP500',
224		'CSIBM500' => 'CP500',
225		'EBCDICCPBE' => 'CP500',
226		'EBCDICCPCH' => 'CP500',
227		'IBM500' => 'CP500',
228		'775' => 'CP775',
229		'CSPC775BALTIC' => 'CP775',
230		'IBM775' => 'CP775',
231		'860' => 'CP860',
232		'CSIBM860' => 'CP860',
233		'IBM860' => 'CP860',
234		'861' => 'CP861',
235		'CPIS' => 'CP861',
236		'CSIBM861' => 'CP861',
237		'IBM861' => 'CP861',
238		'862' => 'CP862',
239		'CSPC862LATINHEBREW' => 'CP862',
240		'IBM862' => 'CP862',
241		'863' => 'CP863',
242		'CSIBM863' => 'CP863',
243		'IBM863' => 'CP863',
244		'864' => 'CP864',
245		'CSIBM864' => 'CP864',
246		'IBM864' => 'CP864',
247		'865' => 'CP865',
248		'CSIBM865' => 'CP865',
249		'IBM865' => 'CP865',
250		'869' => 'CP869',
251		'CPGR' => 'CP869',
252		'CSIBM869' => 'CP869',
253		'IBM869' => 'CP869',
254		'949' => 'CP949',
255		'MS949' => 'CP949',
256		'UHC' => 'CP949',
257		'ROMAN8' => 'ROMAN8',
258		'HPROMAN8' => 'ROMAN8',
259		'R8' => 'ROMAN8',
260		'CSHPROMAN8' => 'ROMAN8',
261		'ISO2022JP2' => 'ISO2022JP2',
262		'THAI' => 'ISO885911',
263		'ISO885911' => 'ISO885911',
264		'ISO8859112001' => 'ISO885911',
265		'JOHAB' => 'CP1361',
266		'MS1361' => 'CP1361',
267		'MACCYRILLIC' => 'MACCYRILLIC',
268		'CSPTCP154' => 'PT154',
269		'PTCP154' => 'PT154',
270		'CP154' => 'PT154',
271		'CYRILLICASIAN' => 'PT154',
272		'TIS620' => 'TIS620',
273		'TIS6200' => 'TIS620',
274		'TIS62025290' => 'TIS620',
275		'TIS62025291' => 'TIS620',
276		'ISOIR166' => 'TIS620',
277		];
278
279		/**
280		* @var string charset to convert from
281		*/
282		protected $fromCharset;
283
284		/**
285		* @var string charset to convert to
286		*/
287		protected $toCharset;
288
289		/**
290		* @var boolean indicates if $fromCharset is supported by
291		* mb_convert_encoding
292		*/
293		protected $fromCharsetMbSupported = true;
294
295		/**
296		* @var boolean indicates if $toCharset is supported by mb_convert_encoding
297		*/
298		protected $toCharsetMbSupported = true;
299
300		/**
301		* Constructs the charset converter with source/destination charsets.
302		*
303		* @param string $fromCharset
304		* @param string $toCharset
305		*/
306	3	public function __construct($fromCharset, $toCharset)
307		{
308	3	$this->fromCharset = $this->findSupportedCharset($fromCharset, $this->fromCharsetMbSupported);
309	3	$this->toCharset = $this->findSupportedCharset($toCharset, $this->toCharsetMbSupported);
310	3	}
311
312		/**
313		* Converts the passed string's charset from $this->fromCharset to
314		* $this->toCharset.
315		*
316		* The function attempts to use mb_convert_encoding if possible, and falls
317		* back to iconv if not. If the source or destination character sets aren't
318		* supported, a blank string is returned.
319		*
320		* @param string $str
321		* @return string
322		*/
323	3	public function convert($str)
324		{
325		// there may be some mb-supported encodings not supported by iconv (on my libiconv for instance
326		// HZ isn't supported), and so it may happen that failing an mb_convert_encoding, an iconv
327		// may also fail even though both support an encoding separately.
328		// Unfortunately there's no great way of testing what charsets are available on iconv, and
329		// attempting to blindly convert the string may be too costly, as could converting first
330		// to an intermediate (ASSUMPTION: may be worth testing converting to an intermediate)
331	3	if ($str !== '') {
332	3	if ($this->fromCharsetMbSupported && $this->toCharsetMbSupported) {
333	2	return mb_convert_encoding($str, $this->toCharset, $this->fromCharset);
334		}
335	1	return iconv($this->fromCharset, $this->toCharset . '//TRANSLIT//IGNORE', $str);
336		}
337		return $str;
338		}
339
340		/**
341		* Looks up the passed $cs in mb_list_encodings, then strips non
342		* alpha-numeric characters and tries again, then failing that calls
343		* findAliasedCharset. The method returns the charset name that should be
344		* used in calls to mb_convert_encoding or iconv.
345		*
346		* If the charset is part of mb_list_encodings, $mbSupported is set to true.
347		*
348		* @param string $cs
349		* @param boolean $mbSupported
350		* @return string the final charset name to use
351		*/
352	3	private function findSupportedCharset($cs, &$mbSupported)
353		{
354		/** @see https://github.com/zbateson/MailMimeParser/issues/53 */
355	3	if (preg_match('/UTF\-8/', $cs)) {
356	3	$cs = 'UTF-8';
357	3	}
358
359	3	$mbSupported = true;
360	3	$comp = strtoupper($cs);
361	3	$available = array_map('strtoupper', mb_list_encodings());
362	3	if (in_array($comp, $available)) {
363	3	return $comp;
364		}
365	3	$stripped = preg_replace('/[^A-Z0-9]+/', '', $comp);
366	3	if (in_array($stripped, $available)) {
367		return $stripped;
368		}
369	3	return $this->findAliasedCharset($comp, $stripped, $mbSupported);
370		}
371
372		/**
373		* Looks up the passed $comp and $stripped strings in self::$mbAliases, and
374		* returns the mapped charset if applicable. Otherwise calls
375		* $this->findAliasedIconvCharset.
376		*
377		* $mbSupported is set to false if the charset is not located in
378		* self::$mbAliases.
379		*
380		* @param string $comp
381		* @param string $stripped
382		* @param boolean $mbSupported
383		* @return string the mapped charset
384		*/
385	3	private function findAliasedCharset($comp, $stripped, &$mbSupported)
386		{
387	3	if (array_key_exists($comp, self::$mbAliases)) {
388	1	return self::$mbAliases[$comp];
389	2	} elseif (array_key_exists($stripped, self::$mbAliases)) {
390	1	return self::$mbAliases[$stripped];
391		}
392	1	$mbSupported = false;
393	1	return $this->findAliasedIconvCharset($comp, $stripped);
394		}
395
396		/**
397		* Looks up the passed $comp and $stripped strings in self::$iconvAliases,
398		* and returns the mapped charset if applicable. Otherwise returns $comp.
399		*
400		* @param string $comp
401		* @param string $stripped
402		* @return string the mapped charset (if mapped) or $comp otherwise
403		*/
404	1	private function findAliasedIconvCharset($comp, $stripped)
405		{
406	1	if (array_key_exists($comp, self::$iconvAliases)) {
407	1	return static::$iconvAliases[$comp];
408	1	} elseif (array_key_exists($stripped, self::$iconvAliases)) {
409		return static::$iconvAliases[$stripped];
410		}
411	1	return $comp;
412		}
413		}
414

zbateson / mail-mime-parser

Pull Request — master (#54)

CharsetConverter::findSupportedCharset() A

Complexity

Size

Duplication

Code Coverage

Importance

Duplication Side-by-Side

Filter issues like