ar_EG
1 year ago
ar_JO
1 year ago
ar_SA
1 year ago
at_AT
1 year ago
bg_BG
1 year ago
bn_BD
1 year ago
cs_CZ
1 year ago
da_DK
1 year ago
de_AT
1 year ago
de_CH
1 year ago
de_DE
1 year ago
el_CY
1 year ago
el_GR
1 year ago
en_AU
1 year ago
en_CA
1 year ago
en_GB
1 year ago
en_HK
1 year ago
en_IN
1 year ago
en_NG
1 year ago
en_NZ
1 year ago
en_PH
1 year ago
en_SG
1 year ago
en_UG
1 year ago
en_US
1 year ago
en_ZA
1 year ago
es_AR
1 year ago
es_ES
1 year ago
es_PE
1 year ago
es_VE
1 year ago
et_EE
1 year ago
fa_IR
1 year ago
fi_FI
1 year ago
fr_BE
1 year ago
fr_CA
1 year ago
fr_CH
1 year ago
fr_FR
1 year ago
he_IL
1 year ago
hr_HR
1 year ago
hu_HU
1 year ago
hy_AM
1 year ago
id_ID
1 year ago
is_IS
1 year ago
it_CH
1 year ago
it_IT
1 year ago
ja_JP
1 year ago
ka_GE
1 year ago
kk_KZ
1 year ago
ko_KR
1 year ago
lt_LT
1 year ago
lv_LV
1 year ago
me_ME
1 year ago
mn_MN
1 year ago
ms_MY
1 year ago
nb_NO
1 year ago
ne_NP
1 year ago
nl_BE
1 year ago
nl_NL
1 year ago
pl_PL
1 year ago
pt_BR
1 year ago
pt_PT
1 year ago
ro_MD
1 year ago
ro_RO
1 year ago
ru_RU
1 year ago
sk_SK
1 year ago
sl_SI
1 year ago
sr_Cyrl_RS
1 year ago
sr_Latn_RS
1 year ago
sr_RS
1 year ago
sv_SE
1 year ago
th_TH
1 year ago
tr_TR
1 year ago
uk_UA
1 year ago
vi_VN
1 year ago
zh_CN
1 year ago
zh_TW
1 year ago
Address.php
1 year ago
Barcode.php
1 year ago
Base.php
1 year ago
Biased.php
1 year ago
Color.php
1 year ago
Company.php
1 year ago
DateTime.php
1 year ago
File.php
1 year ago
HtmlLorem.php
1 year ago
Image.php
1 year ago
Internet.php
1 year ago
Lorem.php
1 year ago
Medical.php
1 year ago
Miscellaneous.php
1 year ago
Payment.php
1 year ago
Person.php
1 year ago
PhoneNumber.php
1 year ago
Text.php
1 year ago
UserAgent.php
1 year ago
Uuid.php
1 year ago
Text.php
209 lines
| 1 | <?php |
| 2 | /** |
| 3 | * @license MIT |
| 4 | * |
| 5 | * Modified by impress-org on 07-August-2024 using Strauss. |
| 6 | * @see https://github.com/BrianHenryIE/strauss |
| 7 | */ |
| 8 | |
| 9 | namespace Give\Vendors\Faker\Provider; |
| 10 | |
| 11 | abstract class Text extends Base |
| 12 | { |
| 13 | protected static $baseText = ''; |
| 14 | protected static $separator = ' '; |
| 15 | protected static $separatorLen = 1; |
| 16 | protected $explodedText; |
| 17 | protected $consecutiveWords = []; |
| 18 | protected static $textStartsWithUppercase = true; |
| 19 | |
| 20 | /** |
| 21 | * Generate a text string by the Markov chain algorithm. |
| 22 | * |
| 23 | * Depending on the $maxNbChars, returns a random valid looking text. The algorithm |
| 24 | * generates a weighted table with the specified number of words as the index and the |
| 25 | * possible following words as the value. |
| 26 | * |
| 27 | * @example 'Alice, swallowing down her flamingo, and began by taking the little golden key' |
| 28 | * |
| 29 | * @param int $maxNbChars Maximum number of characters the text should contain (minimum: 10) |
| 30 | * @param int $indexSize Determines how many words are considered for the generation of the next word. |
| 31 | * The minimum is 1, and it produces a higher level of randomness, although the |
| 32 | * generated text usually doesn't make sense. Higher index sizes (up to 5) |
| 33 | * produce more correct text, at the price of less randomness. |
| 34 | * |
| 35 | * @return string |
| 36 | */ |
| 37 | public function realText($maxNbChars = 200, $indexSize = 2) |
| 38 | { |
| 39 | return $this->realTextBetween((int) round($maxNbChars * 0.8), $maxNbChars, $indexSize); |
| 40 | } |
| 41 | |
| 42 | /** |
| 43 | * Generate a text string by the Markov chain algorithm. |
| 44 | * |
| 45 | * Depending on the $maxNbChars, returns a random valid looking text. The algorithm |
| 46 | * generates a weighted table with the specified number of words as the index and the |
| 47 | * possible following words as the value. |
| 48 | * |
| 49 | * @example 'Alice, swallowing down her flamingo, and began by taking the little golden key' |
| 50 | * |
| 51 | * @param int $minNbChars Minimum number of characters the text should contain (maximum: 8) |
| 52 | * @param int $maxNbChars Maximum number of characters the text should contain (minimum: 10) |
| 53 | * @param int $indexSize Determines how many words are considered for the generation of the next word. |
| 54 | * The minimum is 1, and it produces a higher level of randomness, although the |
| 55 | * generated text usually doesn't make sense. Higher index sizes (up to 5) |
| 56 | * produce more correct text, at the price of less randomness. |
| 57 | * |
| 58 | * @return string |
| 59 | */ |
| 60 | public function realTextBetween($minNbChars = 160, $maxNbChars = 200, $indexSize = 2) |
| 61 | { |
| 62 | if ($minNbChars < 1) { |
| 63 | throw new \InvalidArgumentException('minNbChars must be at least 1'); |
| 64 | } |
| 65 | |
| 66 | if ($maxNbChars < 10) { |
| 67 | throw new \InvalidArgumentException('maxNbChars must be at least 10'); |
| 68 | } |
| 69 | |
| 70 | if ($indexSize < 1) { |
| 71 | throw new \InvalidArgumentException('indexSize must be at least 1'); |
| 72 | } |
| 73 | |
| 74 | if ($indexSize > 5) { |
| 75 | throw new \InvalidArgumentException('indexSize must be at most 5'); |
| 76 | } |
| 77 | |
| 78 | if ($minNbChars >= $maxNbChars) { |
| 79 | throw new \InvalidArgumentException('minNbChars must be smaller than maxNbChars'); |
| 80 | } |
| 81 | |
| 82 | $words = $this->getConsecutiveWords($indexSize); |
| 83 | $iterations = 0; |
| 84 | |
| 85 | do { |
| 86 | ++$iterations; |
| 87 | |
| 88 | if ($iterations >= 100) { |
| 89 | throw new \OverflowException(sprintf('Maximum retries of %d reached without finding a valid real text', $iterations)); |
| 90 | } |
| 91 | |
| 92 | $result = $this->generateText($maxNbChars, $words); |
| 93 | } while (static::strlen($result) <= $minNbChars); |
| 94 | |
| 95 | return $result; |
| 96 | } |
| 97 | |
| 98 | /** |
| 99 | * @param int $maxNbChars |
| 100 | * @param array $words |
| 101 | * |
| 102 | * @return string |
| 103 | */ |
| 104 | protected function generateText($maxNbChars, $words) |
| 105 | { |
| 106 | $result = []; |
| 107 | $resultLength = 0; |
| 108 | // take a random starting point |
| 109 | $next = static::randomKey($words); |
| 110 | |
| 111 | while ($resultLength < $maxNbChars && isset($words[$next])) { |
| 112 | // fetch a random word to append |
| 113 | $word = static::randomElement($words[$next]); |
| 114 | |
| 115 | // calculate next index |
| 116 | $currentWords = static::explode($next); |
| 117 | $currentWords[] = $word; |
| 118 | array_shift($currentWords); |
| 119 | $next = static::implode($currentWords); |
| 120 | |
| 121 | // ensure text starts with an uppercase letter |
| 122 | if ($resultLength == 0 && !static::validStart($word)) { |
| 123 | continue; |
| 124 | } |
| 125 | |
| 126 | // append the element |
| 127 | $result[] = $word; |
| 128 | $resultLength += static::strlen($word) + static::$separatorLen; |
| 129 | } |
| 130 | |
| 131 | // remove the element that caused the text to overflow |
| 132 | array_pop($result); |
| 133 | |
| 134 | // build result |
| 135 | $result = static::implode($result); |
| 136 | |
| 137 | return static::appendEnd($result); |
| 138 | } |
| 139 | |
| 140 | protected function getConsecutiveWords($indexSize) |
| 141 | { |
| 142 | if (!isset($this->consecutiveWords[$indexSize])) { |
| 143 | $parts = $this->getExplodedText(); |
| 144 | $words = []; |
| 145 | $index = []; |
| 146 | |
| 147 | for ($i = 0; $i < $indexSize; ++$i) { |
| 148 | $index[] = array_shift($parts); |
| 149 | } |
| 150 | |
| 151 | for ($i = 0, $count = count($parts); $i < $count; ++$i) { |
| 152 | $stringIndex = static::implode($index); |
| 153 | |
| 154 | if (!isset($words[$stringIndex])) { |
| 155 | $words[$stringIndex] = []; |
| 156 | } |
| 157 | $word = $parts[$i]; |
| 158 | $words[$stringIndex][] = $word; |
| 159 | array_shift($index); |
| 160 | $index[] = $word; |
| 161 | } |
| 162 | // cache look up words for performance |
| 163 | $this->consecutiveWords[$indexSize] = $words; |
| 164 | } |
| 165 | |
| 166 | return $this->consecutiveWords[$indexSize]; |
| 167 | } |
| 168 | |
| 169 | protected function getExplodedText() |
| 170 | { |
| 171 | if ($this->explodedText === null) { |
| 172 | $this->explodedText = static::explode(preg_replace('/\s+/u', ' ', static::$baseText)); |
| 173 | } |
| 174 | |
| 175 | return $this->explodedText; |
| 176 | } |
| 177 | |
| 178 | protected static function explode($text) |
| 179 | { |
| 180 | return explode(static::$separator, $text); |
| 181 | } |
| 182 | |
| 183 | protected static function implode($words) |
| 184 | { |
| 185 | return implode(static::$separator, $words); |
| 186 | } |
| 187 | |
| 188 | protected static function strlen($text) |
| 189 | { |
| 190 | return function_exists('mb_strlen') ? mb_strlen($text, 'UTF-8') : strlen($text); |
| 191 | } |
| 192 | |
| 193 | protected static function validStart($word) |
| 194 | { |
| 195 | $isValid = true; |
| 196 | |
| 197 | if (static::$textStartsWithUppercase) { |
| 198 | $isValid = preg_match('/^\p{Lu}/u', $word); |
| 199 | } |
| 200 | |
| 201 | return $isValid; |
| 202 | } |
| 203 | |
| 204 | protected static function appendEnd($text) |
| 205 | { |
| 206 | return preg_replace("/([ ,-:;\x{2013}\x{2014}]+$)/us", '', $text) . '.'; |
| 207 | } |
| 208 | } |
| 209 |