vendor/symfony/src/Symfony/Component/Yaml/Unescaper.php
author ymh <ymh.work@gmail.com>
Sat, 24 Sep 2011 15:40:41 +0200
changeset 0 7f95f8617b0b
permissions -rwxr-xr-x
first commit
Ignore whitespace changes - Everywhere: Within whitespace: At end of lines:
0
7f95f8617b0b first commit
ymh <ymh.work@gmail.com>
parents:
diff changeset
     1
<?php
7f95f8617b0b first commit
ymh <ymh.work@gmail.com>
parents:
diff changeset
     2
7f95f8617b0b first commit
ymh <ymh.work@gmail.com>
parents:
diff changeset
     3
/*
7f95f8617b0b first commit
ymh <ymh.work@gmail.com>
parents:
diff changeset
     4
 * This file is part of the Symfony package.
7f95f8617b0b first commit
ymh <ymh.work@gmail.com>
parents:
diff changeset
     5
 * (c) Fabien Potencier <fabien@symfony.com>
7f95f8617b0b first commit
ymh <ymh.work@gmail.com>
parents:
diff changeset
     6
 *
7f95f8617b0b first commit
ymh <ymh.work@gmail.com>
parents:
diff changeset
     7
 * For the full copyright and license information, please view the LICENSE
7f95f8617b0b first commit
ymh <ymh.work@gmail.com>
parents:
diff changeset
     8
 * file that was distributed with this source code.
7f95f8617b0b first commit
ymh <ymh.work@gmail.com>
parents:
diff changeset
     9
 */
7f95f8617b0b first commit
ymh <ymh.work@gmail.com>
parents:
diff changeset
    10
7f95f8617b0b first commit
ymh <ymh.work@gmail.com>
parents:
diff changeset
    11
namespace Symfony\Component\Yaml;
7f95f8617b0b first commit
ymh <ymh.work@gmail.com>
parents:
diff changeset
    12
7f95f8617b0b first commit
ymh <ymh.work@gmail.com>
parents:
diff changeset
    13
/**
7f95f8617b0b first commit
ymh <ymh.work@gmail.com>
parents:
diff changeset
    14
 * Unescaper encapsulates unescaping rules for single and double-quoted
7f95f8617b0b first commit
ymh <ymh.work@gmail.com>
parents:
diff changeset
    15
 * YAML strings.
7f95f8617b0b first commit
ymh <ymh.work@gmail.com>
parents:
diff changeset
    16
 *
7f95f8617b0b first commit
ymh <ymh.work@gmail.com>
parents:
diff changeset
    17
 * @author Matthew Lewinski <matthew@lewinski.org>
7f95f8617b0b first commit
ymh <ymh.work@gmail.com>
parents:
diff changeset
    18
 */
7f95f8617b0b first commit
ymh <ymh.work@gmail.com>
parents:
diff changeset
    19
class Unescaper
7f95f8617b0b first commit
ymh <ymh.work@gmail.com>
parents:
diff changeset
    20
{
7f95f8617b0b first commit
ymh <ymh.work@gmail.com>
parents:
diff changeset
    21
    // Parser and Inline assume UTF-8 encoding, so escaped Unicode characters
7f95f8617b0b first commit
ymh <ymh.work@gmail.com>
parents:
diff changeset
    22
    // must be converted to that encoding.
7f95f8617b0b first commit
ymh <ymh.work@gmail.com>
parents:
diff changeset
    23
    const ENCODING = 'UTF-8';
7f95f8617b0b first commit
ymh <ymh.work@gmail.com>
parents:
diff changeset
    24
7f95f8617b0b first commit
ymh <ymh.work@gmail.com>
parents:
diff changeset
    25
    // Regex fragment that matches an escaped character in a double quoted
7f95f8617b0b first commit
ymh <ymh.work@gmail.com>
parents:
diff changeset
    26
    // string.
7f95f8617b0b first commit
ymh <ymh.work@gmail.com>
parents:
diff changeset
    27
    const REGEX_ESCAPED_CHARACTER = "\\\\([0abt\tnvfre \\\"\\/\\\\N_LP]|x[0-9a-fA-F]{2}|u[0-9a-fA-F]{4}|U[0-9a-fA-F]{8})";
7f95f8617b0b first commit
ymh <ymh.work@gmail.com>
parents:
diff changeset
    28
7f95f8617b0b first commit
ymh <ymh.work@gmail.com>
parents:
diff changeset
    29
    /**
7f95f8617b0b first commit
ymh <ymh.work@gmail.com>
parents:
diff changeset
    30
     * Unescapes a single quoted string.
7f95f8617b0b first commit
ymh <ymh.work@gmail.com>
parents:
diff changeset
    31
     *
7f95f8617b0b first commit
ymh <ymh.work@gmail.com>
parents:
diff changeset
    32
     * @param string $value A single quoted string.
7f95f8617b0b first commit
ymh <ymh.work@gmail.com>
parents:
diff changeset
    33
     *
7f95f8617b0b first commit
ymh <ymh.work@gmail.com>
parents:
diff changeset
    34
     * @return string The unescaped string.
7f95f8617b0b first commit
ymh <ymh.work@gmail.com>
parents:
diff changeset
    35
     */
7f95f8617b0b first commit
ymh <ymh.work@gmail.com>
parents:
diff changeset
    36
    public function unescapeSingleQuotedString($value)
7f95f8617b0b first commit
ymh <ymh.work@gmail.com>
parents:
diff changeset
    37
    {
7f95f8617b0b first commit
ymh <ymh.work@gmail.com>
parents:
diff changeset
    38
        return str_replace('\'\'', '\'', $value);
7f95f8617b0b first commit
ymh <ymh.work@gmail.com>
parents:
diff changeset
    39
    }
7f95f8617b0b first commit
ymh <ymh.work@gmail.com>
parents:
diff changeset
    40
7f95f8617b0b first commit
ymh <ymh.work@gmail.com>
parents:
diff changeset
    41
    /**
7f95f8617b0b first commit
ymh <ymh.work@gmail.com>
parents:
diff changeset
    42
     * Unescapes a double quoted string.
7f95f8617b0b first commit
ymh <ymh.work@gmail.com>
parents:
diff changeset
    43
     *
7f95f8617b0b first commit
ymh <ymh.work@gmail.com>
parents:
diff changeset
    44
     * @param string $value A double quoted string.
7f95f8617b0b first commit
ymh <ymh.work@gmail.com>
parents:
diff changeset
    45
     *
7f95f8617b0b first commit
ymh <ymh.work@gmail.com>
parents:
diff changeset
    46
     * @return string The unescaped string.
7f95f8617b0b first commit
ymh <ymh.work@gmail.com>
parents:
diff changeset
    47
     */
7f95f8617b0b first commit
ymh <ymh.work@gmail.com>
parents:
diff changeset
    48
    public function unescapeDoubleQuotedString($value)
7f95f8617b0b first commit
ymh <ymh.work@gmail.com>
parents:
diff changeset
    49
    {
7f95f8617b0b first commit
ymh <ymh.work@gmail.com>
parents:
diff changeset
    50
        $self = $this;
7f95f8617b0b first commit
ymh <ymh.work@gmail.com>
parents:
diff changeset
    51
        $callback = function($match) use($self) {
7f95f8617b0b first commit
ymh <ymh.work@gmail.com>
parents:
diff changeset
    52
            return $self->unescapeCharacter($match[0]);
7f95f8617b0b first commit
ymh <ymh.work@gmail.com>
parents:
diff changeset
    53
        };
7f95f8617b0b first commit
ymh <ymh.work@gmail.com>
parents:
diff changeset
    54
7f95f8617b0b first commit
ymh <ymh.work@gmail.com>
parents:
diff changeset
    55
        // evaluate the string
7f95f8617b0b first commit
ymh <ymh.work@gmail.com>
parents:
diff changeset
    56
        return preg_replace_callback('/'.self::REGEX_ESCAPED_CHARACTER.'/u', $callback, $value);
7f95f8617b0b first commit
ymh <ymh.work@gmail.com>
parents:
diff changeset
    57
    }
7f95f8617b0b first commit
ymh <ymh.work@gmail.com>
parents:
diff changeset
    58
7f95f8617b0b first commit
ymh <ymh.work@gmail.com>
parents:
diff changeset
    59
    /**
7f95f8617b0b first commit
ymh <ymh.work@gmail.com>
parents:
diff changeset
    60
     * Unescapes a character that was found in a double-quoted string
7f95f8617b0b first commit
ymh <ymh.work@gmail.com>
parents:
diff changeset
    61
     *
7f95f8617b0b first commit
ymh <ymh.work@gmail.com>
parents:
diff changeset
    62
     * @param string $value An escaped character
7f95f8617b0b first commit
ymh <ymh.work@gmail.com>
parents:
diff changeset
    63
     *
7f95f8617b0b first commit
ymh <ymh.work@gmail.com>
parents:
diff changeset
    64
     * @return string The unescaped character
7f95f8617b0b first commit
ymh <ymh.work@gmail.com>
parents:
diff changeset
    65
     */
7f95f8617b0b first commit
ymh <ymh.work@gmail.com>
parents:
diff changeset
    66
    public function unescapeCharacter($value)
7f95f8617b0b first commit
ymh <ymh.work@gmail.com>
parents:
diff changeset
    67
    {
7f95f8617b0b first commit
ymh <ymh.work@gmail.com>
parents:
diff changeset
    68
        switch ($value{1}) {
7f95f8617b0b first commit
ymh <ymh.work@gmail.com>
parents:
diff changeset
    69
            case '0':
7f95f8617b0b first commit
ymh <ymh.work@gmail.com>
parents:
diff changeset
    70
                return "\x0";
7f95f8617b0b first commit
ymh <ymh.work@gmail.com>
parents:
diff changeset
    71
            case 'a':
7f95f8617b0b first commit
ymh <ymh.work@gmail.com>
parents:
diff changeset
    72
                return "\x7";
7f95f8617b0b first commit
ymh <ymh.work@gmail.com>
parents:
diff changeset
    73
            case 'b':
7f95f8617b0b first commit
ymh <ymh.work@gmail.com>
parents:
diff changeset
    74
                return "\x8";
7f95f8617b0b first commit
ymh <ymh.work@gmail.com>
parents:
diff changeset
    75
            case 't':
7f95f8617b0b first commit
ymh <ymh.work@gmail.com>
parents:
diff changeset
    76
                return "\t";
7f95f8617b0b first commit
ymh <ymh.work@gmail.com>
parents:
diff changeset
    77
            case "\t":
7f95f8617b0b first commit
ymh <ymh.work@gmail.com>
parents:
diff changeset
    78
                return "\t";
7f95f8617b0b first commit
ymh <ymh.work@gmail.com>
parents:
diff changeset
    79
            case 'n':
7f95f8617b0b first commit
ymh <ymh.work@gmail.com>
parents:
diff changeset
    80
                return "\n";
7f95f8617b0b first commit
ymh <ymh.work@gmail.com>
parents:
diff changeset
    81
            case 'v':
7f95f8617b0b first commit
ymh <ymh.work@gmail.com>
parents:
diff changeset
    82
                return "\xb";
7f95f8617b0b first commit
ymh <ymh.work@gmail.com>
parents:
diff changeset
    83
            case 'f':
7f95f8617b0b first commit
ymh <ymh.work@gmail.com>
parents:
diff changeset
    84
                return "\xc";
7f95f8617b0b first commit
ymh <ymh.work@gmail.com>
parents:
diff changeset
    85
            case 'r':
7f95f8617b0b first commit
ymh <ymh.work@gmail.com>
parents:
diff changeset
    86
                return "\xd";
7f95f8617b0b first commit
ymh <ymh.work@gmail.com>
parents:
diff changeset
    87
            case 'e':
7f95f8617b0b first commit
ymh <ymh.work@gmail.com>
parents:
diff changeset
    88
                return "\x1b";
7f95f8617b0b first commit
ymh <ymh.work@gmail.com>
parents:
diff changeset
    89
            case ' ':
7f95f8617b0b first commit
ymh <ymh.work@gmail.com>
parents:
diff changeset
    90
                return ' ';
7f95f8617b0b first commit
ymh <ymh.work@gmail.com>
parents:
diff changeset
    91
            case '"':
7f95f8617b0b first commit
ymh <ymh.work@gmail.com>
parents:
diff changeset
    92
                return '"';
7f95f8617b0b first commit
ymh <ymh.work@gmail.com>
parents:
diff changeset
    93
            case '/':
7f95f8617b0b first commit
ymh <ymh.work@gmail.com>
parents:
diff changeset
    94
                return '/';
7f95f8617b0b first commit
ymh <ymh.work@gmail.com>
parents:
diff changeset
    95
            case '\\':
7f95f8617b0b first commit
ymh <ymh.work@gmail.com>
parents:
diff changeset
    96
                return '\\';
7f95f8617b0b first commit
ymh <ymh.work@gmail.com>
parents:
diff changeset
    97
            case 'N':
7f95f8617b0b first commit
ymh <ymh.work@gmail.com>
parents:
diff changeset
    98
                // U+0085 NEXT LINE
7f95f8617b0b first commit
ymh <ymh.work@gmail.com>
parents:
diff changeset
    99
                return $this->convertEncoding("\x00\x85", self::ENCODING, 'UCS-2BE');
7f95f8617b0b first commit
ymh <ymh.work@gmail.com>
parents:
diff changeset
   100
            case '_':
7f95f8617b0b first commit
ymh <ymh.work@gmail.com>
parents:
diff changeset
   101
                // U+00A0 NO-BREAK SPACE
7f95f8617b0b first commit
ymh <ymh.work@gmail.com>
parents:
diff changeset
   102
                return $this->convertEncoding("\x00\xA0", self::ENCODING, 'UCS-2BE');
7f95f8617b0b first commit
ymh <ymh.work@gmail.com>
parents:
diff changeset
   103
            case 'L':
7f95f8617b0b first commit
ymh <ymh.work@gmail.com>
parents:
diff changeset
   104
                // U+2028 LINE SEPARATOR
7f95f8617b0b first commit
ymh <ymh.work@gmail.com>
parents:
diff changeset
   105
                return $this->convertEncoding("\x20\x28", self::ENCODING, 'UCS-2BE');
7f95f8617b0b first commit
ymh <ymh.work@gmail.com>
parents:
diff changeset
   106
            case 'P':
7f95f8617b0b first commit
ymh <ymh.work@gmail.com>
parents:
diff changeset
   107
                // U+2029 PARAGRAPH SEPARATOR
7f95f8617b0b first commit
ymh <ymh.work@gmail.com>
parents:
diff changeset
   108
                return $this->convertEncoding("\x20\x29", self::ENCODING, 'UCS-2BE');
7f95f8617b0b first commit
ymh <ymh.work@gmail.com>
parents:
diff changeset
   109
            case 'x':
7f95f8617b0b first commit
ymh <ymh.work@gmail.com>
parents:
diff changeset
   110
                $char = pack('n', hexdec(substr($value, 2, 2)));
7f95f8617b0b first commit
ymh <ymh.work@gmail.com>
parents:
diff changeset
   111
7f95f8617b0b first commit
ymh <ymh.work@gmail.com>
parents:
diff changeset
   112
                return $this->convertEncoding($char, self::ENCODING, 'UCS-2BE');
7f95f8617b0b first commit
ymh <ymh.work@gmail.com>
parents:
diff changeset
   113
            case 'u':
7f95f8617b0b first commit
ymh <ymh.work@gmail.com>
parents:
diff changeset
   114
                $char = pack('n', hexdec(substr($value, 2, 4)));
7f95f8617b0b first commit
ymh <ymh.work@gmail.com>
parents:
diff changeset
   115
7f95f8617b0b first commit
ymh <ymh.work@gmail.com>
parents:
diff changeset
   116
                return $this->convertEncoding($char, self::ENCODING, 'UCS-2BE');
7f95f8617b0b first commit
ymh <ymh.work@gmail.com>
parents:
diff changeset
   117
            case 'U':
7f95f8617b0b first commit
ymh <ymh.work@gmail.com>
parents:
diff changeset
   118
                $char = pack('N', hexdec(substr($value, 2, 8)));
7f95f8617b0b first commit
ymh <ymh.work@gmail.com>
parents:
diff changeset
   119
7f95f8617b0b first commit
ymh <ymh.work@gmail.com>
parents:
diff changeset
   120
                return $this->convertEncoding($char, self::ENCODING, 'UCS-4BE');
7f95f8617b0b first commit
ymh <ymh.work@gmail.com>
parents:
diff changeset
   121
        }
7f95f8617b0b first commit
ymh <ymh.work@gmail.com>
parents:
diff changeset
   122
    }
7f95f8617b0b first commit
ymh <ymh.work@gmail.com>
parents:
diff changeset
   123
7f95f8617b0b first commit
ymh <ymh.work@gmail.com>
parents:
diff changeset
   124
    /**
7f95f8617b0b first commit
ymh <ymh.work@gmail.com>
parents:
diff changeset
   125
     * Convert a string from one encoding to another.
7f95f8617b0b first commit
ymh <ymh.work@gmail.com>
parents:
diff changeset
   126
     *
7f95f8617b0b first commit
ymh <ymh.work@gmail.com>
parents:
diff changeset
   127
     * @param string $value The string to convert
7f95f8617b0b first commit
ymh <ymh.work@gmail.com>
parents:
diff changeset
   128
     * @param string $to    The input encoding
7f95f8617b0b first commit
ymh <ymh.work@gmail.com>
parents:
diff changeset
   129
     * @param string $from  The output encoding
7f95f8617b0b first commit
ymh <ymh.work@gmail.com>
parents:
diff changeset
   130
     *
7f95f8617b0b first commit
ymh <ymh.work@gmail.com>
parents:
diff changeset
   131
     * @return string The string with the new encoding
7f95f8617b0b first commit
ymh <ymh.work@gmail.com>
parents:
diff changeset
   132
     *
7f95f8617b0b first commit
ymh <ymh.work@gmail.com>
parents:
diff changeset
   133
     * @throws \RuntimeException if no suitable encoding function is found (iconv or mbstring)
7f95f8617b0b first commit
ymh <ymh.work@gmail.com>
parents:
diff changeset
   134
     */
7f95f8617b0b first commit
ymh <ymh.work@gmail.com>
parents:
diff changeset
   135
    private function convertEncoding($value, $to, $from)
7f95f8617b0b first commit
ymh <ymh.work@gmail.com>
parents:
diff changeset
   136
    {
7f95f8617b0b first commit
ymh <ymh.work@gmail.com>
parents:
diff changeset
   137
        if (function_exists('iconv')) {
7f95f8617b0b first commit
ymh <ymh.work@gmail.com>
parents:
diff changeset
   138
            return iconv($from, $to, $value);
7f95f8617b0b first commit
ymh <ymh.work@gmail.com>
parents:
diff changeset
   139
        } elseif (function_exists('mb_convert_encoding')) {
7f95f8617b0b first commit
ymh <ymh.work@gmail.com>
parents:
diff changeset
   140
            return mb_convert_encoding($value, $to, $from);
7f95f8617b0b first commit
ymh <ymh.work@gmail.com>
parents:
diff changeset
   141
        }
7f95f8617b0b first commit
ymh <ymh.work@gmail.com>
parents:
diff changeset
   142
7f95f8617b0b first commit
ymh <ymh.work@gmail.com>
parents:
diff changeset
   143
        throw new \RuntimeException('No suitable convert encoding function (install the iconv or mbstring extension).');
7f95f8617b0b first commit
ymh <ymh.work@gmail.com>
parents:
diff changeset
   144
    }
7f95f8617b0b first commit
ymh <ymh.work@gmail.com>
parents:
diff changeset
   145
}