Text.php
| 1 | <?php |
| 2 | |
| 3 | namespace Faker\Provider; |
| 4 | |
| 5 | abstract class Text extends Base |
| 6 | { |
| 7 | protected static $baseText = ''; |
| 8 | protected static $separator = ' '; |
| 9 | protected static $separatorLen = 1; |
| 10 | protected $explodedText; |
| 11 | protected $consecutiveWords = array(); |
| 12 | protected static $textStartsWithUppercase = true; |
| 13 | |
| 14 | /** |
| 15 | * Generate a text string by the Markov chain algorithm. |
| 16 | * |
| 17 | * Depending on the $maxNbChars, returns a random valid looking text. The algorithm |
| 18 | * generates a weighted table with the specified number of words as the index and the |
| 19 | * possible following words as the value. |
| 20 | * |
| 21 | * @example 'Alice, swallowing down her flamingo, and began by taking the little golden key' |
| 22 | * @param integer $maxNbChars Maximum number of characters the text should contain (minimum: 10) |
| 23 | * @param integer $indexSize Determines how many words are considered for the generation of the next word. |
| 24 | * The minimum is 1, and it produces a higher level of randomness, although the |
| 25 | * generated text usually doesn't make sense. Higher index sizes (up to 5) |
| 26 | * produce more correct text, at the price of less randomness. |
| 27 | * @return string |
| 28 | */ |
| 29 | public function realText($maxNbChars = 200, $indexSize = 2) |
| 30 | { |
| 31 | if ($maxNbChars < 10) { |
| 32 | throw new \InvalidArgumentException('maxNbChars must be at least 10'); |
| 33 | } |
| 34 | |
| 35 | if ($indexSize < 1) { |
| 36 | throw new \InvalidArgumentException('indexSize must be at least 1'); |
| 37 | } |
| 38 | |
| 39 | if ($indexSize > 5) { |
| 40 | throw new \InvalidArgumentException('indexSize must be at most 5'); |
| 41 | } |
| 42 | |
| 43 | $words = $this->getConsecutiveWords($indexSize); |
| 44 | $result = array(); |
| 45 | $resultLength = 0; |
| 46 | // take a random starting point |
| 47 | $next = static::randomKey($words); |
| 48 | while ($resultLength < $maxNbChars && isset($words[$next])) { |
| 49 | // fetch a random word to append |
| 50 | $word = static::randomElement($words[$next]); |
| 51 | |
| 52 | // calculate next index |
| 53 | $currentWords = static::explode($next); |
| 54 | $currentWords[] = $word; |
| 55 | array_shift($currentWords); |
| 56 | $next = static::implode($currentWords); |
| 57 | |
| 58 | // ensure text starts with an uppercase letter |
| 59 | if ($resultLength == 0 && !static::validStart($word)) { |
| 60 | continue; |
| 61 | } |
| 62 | |
| 63 | // append the element |
| 64 | $result[] = $word; |
| 65 | $resultLength += static::strlen($word) + static::$separatorLen; |
| 66 | } |
| 67 | |
| 68 | // remove the element that caused the text to overflow |
| 69 | array_pop($result); |
| 70 | |
| 71 | // build result |
| 72 | $result = static::implode($result); |
| 73 | |
| 74 | return static::appendEnd($result); |
| 75 | } |
| 76 | |
| 77 | protected function getConsecutiveWords($indexSize) |
| 78 | { |
| 79 | if (!isset($this->consecutiveWords[$indexSize])) { |
| 80 | $parts = $this->getExplodedText(); |
| 81 | $words = array(); |
| 82 | $index = array(); |
| 83 | for ($i = 0; $i < $indexSize; $i++) { |
| 84 | $index[] = array_shift($parts); |
| 85 | } |
| 86 | |
| 87 | for ($i = 0, $count = count($parts); $i < $count; $i++) { |
| 88 | $stringIndex = static::implode($index); |
| 89 | if (!isset($words[$stringIndex])) { |
| 90 | $words[$stringIndex] = array(); |
| 91 | } |
| 92 | $word = $parts[$i]; |
| 93 | $words[$stringIndex][] = $word; |
| 94 | array_shift($index); |
| 95 | $index[] = $word; |
| 96 | } |
| 97 | // cache look up words for performance |
| 98 | $this->consecutiveWords[$indexSize] = $words; |
| 99 | } |
| 100 | |
| 101 | return $this->consecutiveWords[$indexSize]; |
| 102 | } |
| 103 | |
| 104 | protected function getExplodedText() |
| 105 | { |
| 106 | if ($this->explodedText === null) { |
| 107 | $this->explodedText = static::explode(preg_replace('/\s+/u', ' ', static::$baseText)); |
| 108 | } |
| 109 | |
| 110 | return $this->explodedText; |
| 111 | } |
| 112 | |
| 113 | protected static function explode($text) |
| 114 | { |
| 115 | return explode(static::$separator, $text); |
| 116 | } |
| 117 | |
| 118 | protected static function implode($words) |
| 119 | { |
| 120 | return implode(static::$separator, $words); |
| 121 | } |
| 122 | |
| 123 | protected static function strlen($text) |
| 124 | { |
| 125 | return function_exists('mb_strlen') ? mb_strlen($text, 'UTF-8') : strlen($text); |
| 126 | } |
| 127 | |
| 128 | protected static function validStart($word) |
| 129 | { |
| 130 | $isValid = true; |
| 131 | if (static::$textStartsWithUppercase) { |
| 132 | $isValid = preg_match('/^\p{Lu}/u', $word); |
| 133 | } |
| 134 | return $isValid; |
| 135 | } |
| 136 | |
| 137 | protected static function appendEnd($text) |
| 138 | { |
| 139 | return preg_replace("/([ ,-:;\x{2013}\x{2014}]+$)/us", '', $text).'.'; |
| 140 | } |
| 141 | } |
| 142 |