1<?php 2 3namespace Vanderlee\Sentence; 4 5/** 6 * Multibyte-safe utility functions 7 */ 8class Multibyte 9{ 10 //https://stackoverflow.com/questions/20025030/convert-all-types-of-smart-quotes-with-php 11 private static $unicodeCharacterMap = [ 12 // Windows codepage 1252 13 "\xC2\x82" => "'", // U+0082⇒U+201A single low-9 quotation mark 14 "\xC2\x84" => '"', // U+0084⇒U+201E double low-9 quotation mark 15 "\xC2\x8B" => "'", // U+008B⇒U+2039 single left-pointing angle quotation mark 16 "\xC2\x91" => "'", // U+0091⇒U+2018 left single quotation mark 17 "\xC2\x92" => "'", // U+0092⇒U+2019 right single quotation mark 18 "\xC2\x93" => '"', // U+0093⇒U+201C left double quotation mark 19 "\xC2\x94" => '"', // U+0094⇒U+201D right double quotation mark 20 "\xC2\x9B" => "'", // U+009B⇒U+203A single right-pointing angle quotation mark 21 // Regular Unicode // U+0022 quotation mark (") 22 // U+0027 apostrophe (') 23 "\xC2\xAB" => '"', // U+00AB left-pointing double angle quotation mark 24 "\xC2\xBB" => '"', // U+00BB right-pointing double angle quotation mark 25 "\xE2\x80\x98" => "'", // U+2018 left single quotation mark 26 "\xE2\x80\x99" => "'", // U+2019 right single quotation mark 27 "\xE2\x80\x9A" => "'", // U+201A single low-9 quotation mark 28 "\xE2\x80\x9B" => "'", // U+201B single high-reversed-9 quotation mark 29 "\xE2\x80\x9C" => '"', // U+201C left double quotation mark 30 "\xE2\x80\x9D" => '"', // U+201D right double quotation mark 31 "\xE2\x80\x9E" => '"', // U+201E double low-9 quotation mark 32 "\xE2\x80\x9F" => '"', // U+201F double high-reversed-9 quotation mark 33 "\xE2\x80\xB9" => "'", // U+2039 single left-pointing angle quotation mark 34 "\xE2\x80\xBA" => "'", // U+203A single right-pointing angle quotation mark 35 ]; 36 37 /** 38 * Replace 39 * 40 * @staticvar array $chr_map 41 * @param string $string 42 * @return string 43 */ 44 public static function cleanUnicode($string) 45 { 46 $character = array_keys(self::$unicodeCharacterMap); // but: for efficiency you should 47 $replace = array_values(self::$unicodeCharacterMap); // pre-calculate these two arrays 48 return str_replace($character, $replace, html_entity_decode($string, ENT_QUOTES, "UTF-8")); 49 } 50 51 /** 52 * Multibyte.php safe version of standard trim() function. 53 * 54 * @param string $string 55 * @return string 56 */ 57 public static function trim($string) 58 { 59 return mb_ereg_replace('^\s*([\s\S]*?)\s*$', '\1', $string); 60 } 61 62 /** 63 * A cross between mb_split and preg_split, adding the preg_split flags 64 * to mb_split. 65 * 66 * @param string $pattern 67 * @param string $string 68 * @param int $limit 69 * @param int $flags 70 * @return array 71 */ 72 public static function split($pattern, $string, $limit = -1, $flags = 0) 73 { 74 $offset_capture = (bool)($flags & PREG_SPLIT_OFFSET_CAPTURE); 75 76 $lengths = self::getSplitLengths($pattern, $string); 77 78 // Substrings 79 $parts = []; 80 $position = 0; 81 $count = 1; 82 foreach ($lengths as $length) { 83 if (self::isLastPart($length, $flags, $limit, $count)) { 84 $parts[] = self::makePart($string, $position, null, $offset_capture); 85 return $parts; 86 } 87 88 if (self::isPart($length, $flags)) { 89 $parts[] = self::makePart($string, $position, $length[0], $offset_capture); 90 } 91 92 $position += $length[0]; 93 } 94 95 return $parts; 96 } 97 98 /** 99 * @param $length 100 * @param $flags 101 * @param $limit 102 * @param $count 103 * @return bool 104 */ 105 private static function isLastPart($length, $flags, $limit, &$count) 106 { 107 $split_empty = !($flags & PREG_SPLIT_NO_EMPTY) || $length[0]; 108 $is_delimiter = $length[1]; 109 110 return $limit > 0 111 && !$is_delimiter 112 && $split_empty 113 && ++$count > $limit; 114 } 115 116 /** 117 * @param $length 118 * @param $flags 119 * @return bool 120 */ 121 private static function isPart($length, $flags) 122 { 123 $split_empty = !($flags & PREG_SPLIT_NO_EMPTY) || $length[0]; 124 $is_delimiter = $length[1]; 125 $is_captured = ($flags & PREG_SPLIT_DELIM_CAPTURE) && $length[2]; 126 127 return (!$is_delimiter 128 || $is_captured) 129 && $split_empty; 130 } 131 132 /** 133 * Make part 134 * @param string $string 135 * @param integer $position 136 * @param integer|null $length 137 * @param bool $offset_capture 138 * @return array|string 139 */ 140 private static function makePart($string, $position, $length = null, $offset_capture = false) 141 { 142 $cut = mb_strcut($string, $position, $length); 143 144 return $offset_capture 145 ? [$cut, $position] 146 : $cut; 147 } 148 149 /** 150 * Splits the string by pattern and for each element (part or split) returns: 151 * [ 0 => length, 1 => is_delimiter?, 2 => 152 * 153 * @param $pattern 154 * @param $string 155 * @return array 156 */ 157 private static function getSplitLengths($pattern, $string) 158 { 159 $strlen = strlen($string); // bytes! 160 $lengths = []; 161 162 mb_ereg_search_init($string); 163 164 $position = 0; 165 while ($position < $strlen 166 && ($array = mb_ereg_search_pos($pattern, '')) !== false) { 167 // capture split 168 $lengths[] = [$array[0] - $position, false, null]; 169 170 // move position 171 $position = $array[0] + $array[1]; 172 173 // capture delimiter 174 $regs = mb_ereg_search_getregs(); 175 $lengths[] = [$array[1], true, isset($regs[1]) && $regs[1]]; 176 } 177 178 // Add last bit, if not ending with split 179 $lengths[] = [$strlen - $position, false, null]; 180 181 return $lengths; 182 } 183} 184