class ChatLib {
- public function addMessage($msg) {
- $tokens = $this->tokenize($msg->text_content);
- if (empty($tokens)) return;
- $tokens[] = '';
- foreach ($tokens as $num => $token) {
- if ($num === 0) {
- $this->addTransition([], $token);
- } else {
- $start = max(0, $num - $this->size);
- $end = $num;
- for ($i = $start; $i < $end; ++$i) {
- $this->addTransition(array_slice($tokens, $i, $end - $i), $token);
- if ($end - $i < 3) break;
- }
+ public function __construct($size = 3) {
+ $this->size = $size;
+ $converted = [];
+ foreach ($this->categories as $category => $patterns) {
+ $converted_patterns = [];
+ foreach ($patterns as $pattern) {
+ $converted_patterns[] = '/\b'.$pattern.'\b/u';
}
+ $converted[strtoupper($category)] = $converted_patterns;
}
+ $this->categories = $converted;
}
- public function compile() {
- foreach ($this->transitions as $key => $value) {
- $this->transitions[$key] = $this->index($this->transitions[$key]);
- if (empty($this->transitions[$key])) {
- unset($this->transitions[$key]);
+ public function addMessage(ChatLog $msg, ChatLog $previous = null) {
+ if ($msg->isReply()) {
+ $this->addText($msg->text_content, $msg->getReplyParent());
+ } else if (!is_null($previous)) {
+ $this->addText($msg->text_content, $previous->text_content);
+ } else {
+ $this->addText($msg->text_content);
+ }
+ }
+
+ public function addText($text, $context = '') {
+ $tokens = $this->tokenize($text);
+ for ($i = 0; $i < count($tokens) - $this->size; ++$i) {
+ $this->addTransition(array_slice($tokens, $i, $this->size), $tokens[$i + $this->size]);
+ }
+ if (!empty($context)) {
+ $tokens = $this->tokenizeWithContext($text, $context);
+ $size = min($this->size - 1, count($tokens) - $this->size);
+ for ($i = 0; $i < $size; ++$i) {
+ $this->addTransition(array_slice($tokens, $i, $this->size), $tokens[$i + $this->size]);
}
}
}
- public function generate($limit = 100) {
- $tokens = [];
- $generated = '';
- while (strlen($generated) < $limit) {
- $next = $this->randomNext($tokens);
- if (empty($next)) break;
- $tokens[] = $next;
- $generated .= $next;
+ public function compile() {
+ foreach ($this->transitions as $key => $values) {
+ $this->transitions[$key] = $this->index($values);
}
- return $generated;
+ }
+
+ public function generate($context = null) {
+ if (!is_null($context)) {
+ $tokens = $this->tokenizeWithContext('', $context);
+ $generated = $this->loop($tokens);
+ if (!empty($generated)) {
+ return $generated;
+ }
+ }
+ $tokens = $this->tokenize('');
+ return $this->loop($tokens);
}
public function saveAs($name) {
private function index($arr) {
$result = [];
$sum = 0;
- foreach ($arr as $key => $entry) {
- $weight = $entry['count'];
- if ($weight == 1) continue;
+ foreach ($arr as $key => $weight) {
$lower = $sum;
$sum += $weight;
- $examples = [];
- if (is_array(end($entry['examples']))) {
- // already processed
- $examples = $entry['examples'];
- } else {
- $subsum = 0;
- foreach ($entry['examples'] as $example => $subweight) {
- $sublower = $subsum;
- $subsum += $subweight * $subweight;
- $examples[] = [$example, $sublower, $subsum];
- }
- }
- $result[] = [$key, $lower, $sum, $examples];
+ $result[] = [$key, $lower, $sum];
}
return $result;
}
- private function randomNext($tokens) {
- $cnt = count($tokens);
- for ($size = min($this->size, $cnt); $size >= 0; --$size) {
- $cmb = $this->generalize(array_slice($tokens, $cnt - $size, $size));
- if (isset($this->transitions[$cmb])) {
- $pick = $this->pick($this->transitions[$cmb]);
- if (!is_null($pick)) {
- return $this->exampleOf($pick);
- }
- }
+ private function loop($tokens) {
+ while (count($tokens) < 50) {
+ $next = $this->randomNext($tokens);
+ if ($next === ' ') break;
+ $tokens[] = $next;
}
- return '';
+ return $this->untokenize($tokens);
+ }
+
+ private function randomNext($tokens) {
+ $key = $this->makeKey($tokens);
+ if (!isset($this->transitions[$key])) return ' ';
+ $pick = $this->pick($this->transitions[$key]);
+ return $pick[0];
}
private function pick($options) {
if (empty($options)) return null;
- $max = end($options)[2];
+ $max = end($options)[2] - 1;
$num = random_int(0, $max);
+ return static::search($options, $num);
+ }
+
+ public static function search($options, $num) {
$min_index = 0;
$max_index = count($options) - 1;
while ($min_index < $max_index) {
$cur_index = intval(($min_index + $max_index) / 2);
$cur_low = $options[$cur_index][1];
- $cur_high = $options[$cur_index][2];
+ $cur_high = $options[$cur_index][2] - 1;
if ($cur_low > $num) {
$max_index = $cur_index;
} else if ($cur_high < $num) {
return $options[$min_index];
}
- private function addTransition($state, $next) {
- $cmb = $this->generalize($state);
- if (!isset($this->transitions[$cmb])) {
- $this->transitions[$cmb] = [];
+ private function addTransition($tokens, $next) {
+ $key = $this->makeKey($tokens);
+ if (!isset($this->transitions[$key])) {
+ $this->transitions[$key] = [];
+ }
+ if (!isset($this->transitions[$key][$next])) {
+ $this->transitions[$key][$next] = 1;
+ } else {
+ ++$this->transitions[$key][$next];
}
- $this->increment($this->transitions[$cmb], $next);
}
- private function increment(&$which, $token) {
- $generalized = $this->generalize([$token]);
- if (!isset($which[$generalized])) {
- $which[$generalized] = [
- 'count' => 1,
- 'examples' => [],
- ];
- $which[$generalized]['examples'][$token] = 1;
- } else {
- ++$which[$generalized]['count'];
- if (!isset($which[$generalized]['examples'][$token])) {
- $which[$generalized]['examples'][$token] = 1;
- } else {
- ++$which[$generalized]['examples'][$token];
- }
+ private function splitText($text) {
+ if (trim($text) === '') return [];
+ return preg_split('/\s+/u', $text);
+ }
+
+ private function makeKey($tokens) {
+ $key = $this->joinText(array_slice($tokens, $this->size * -1));
+ $key = mb_strtolower($key);
+ $key = str_replace(['.', ',', ':', ';', '!', '?', '^', '+', '-', '"', "'", '(', ')', '[', ']'], '', $key);
+ $key = preg_replace('/\d+/u', '0', $key);
+ foreach ($this->categories as $category => $patterns) {
+ $key = preg_replace($patterns, $category, $key);
}
+ return $key;
+ }
+
+ private function joinText($tokens) {
+ return implode(' ', $tokens);
}
- private function tokenize($str) {
- return array_values(array_filter(preg_split('/\b/u', $str), function($token) {
- if (empty($token)) return false;
- if (preg_match('/cheer\d+/u', strtolower($token))) return false;
- return true;
- }));
+ private function untokenize($tokens) {
+ return $this->joinText(array_slice($tokens, $this->size));
}
- private function generalize($tokens) {
- $str = '';
- foreach ($tokens as $token) {
- $replaced = preg_replace('/\d+/', '0', $token);
- $replaced = strtolower($replaced);
- $str .= empty($replaced) ? $token : $replaced;
+ private function tokenize($text) {
+ $tokens = $this->splitText($text);
+ $combined = array_merge(array_fill(0, $this->size, ' '), $tokens);
+ if (!empty($tokens)) {
+ $combined[] = ' ';
}
- return $str;
+ return $combined;
}
- private function exampleOf($pick) {
- $example = $this->pick($pick[3]);
- return $example[0];
+ private function tokenizeWithContext($text, $context) {
+ $combined = $this->tokenize($text);
+ $context_tokens = array_slice($this->splitText($context), $this->size * -1 + 1);
+ for ($i = 0; $i < count($context_tokens); ++$i) {
+ $combined[$this->size - $i - 2] = $context_tokens[count($context_tokens) - $i - 1];
+ }
+ return $combined;
}
- private $size = 5;
+ private $size;
private $transitions = [];
+ private $categories = [
+ 'fail' => [
+ 'failfish',
+ 'holysm0notlikethis',
+ 'notlikethis',
+ 'tetobridge0',
+ 'vinter0clown',
+ ],
+
+ 'hype' => [
+ 'dergoaparty',
+ 'dinodance',
+ 'elemen0party',
+ 'muftaahype',
+ 'luckwuhype',
+ 'olliwahype',
+ 'osora0umbrihype',
+ 'partyhat',
+ 'peepocheer',
+ 'rei0hype',
+ 'sakayahype',
+ 'tetotroete',
+ 'ticknaboargeil0',
+ 'ticknahype0',
+ ],
+
+ 'kappa' => [
+ 'kappa(claus|hd)?',
+ ],
+
+ 'jam' => [
+ '(cat|dog|rat)jam',
+ 'kanash0jam',
+ 'rei0jamers',
+ 'samusdance',
+ ],
+
+ 'lol' => [
+ ':d',
+ 'boothi0lul',
+ 'kekw',
+ 'lol',
+ 'lul',
+ 'rei0lul',
+ 'samusgrin',
+ 'ticknaauslachen',
+ 'xd',
+ ],
+
+ 'love' => [
+ '<3',
+ 'duden0love',
+ 'exec0love',
+ 'krawal0heart',
+ 'lodanzhug',
+ 'luckwulove',
+ 'luvsign',
+ 'muftaal',
+ 'osora0love',
+ 'peepoexcitedhug',
+ 'spirit0love',
+ 'svenkalove',
+ 'ticknaherz',
+ ],
+
+ 'name' => [
+ 'baba',
+ 'baka',
+ 'bobe?r',
+ 'brog(i|or)',
+ 'cfate',
+ 'danny',
+ 'danzi+',
+ 'daruck',
+ 'dennsen',
+ 'dimez',
+ 'divi',
+ 'dud(en|i+)',
+ 'ele',
+ 'eri(ror)?',
+ '(name)?faker',
+ 'fetti+',
+ 'gamma(chuu)?',
+ 'goat(buster|ie?|y)?',
+ 'hitsu(yan)?',
+ 'holy',
+ 'jem',
+ 'kala(marino)?',
+ 'kromb',
+ 'koval',
+ 'kum(i|o|p)',
+ 'lanux',
+ 'len(esha|chen)',
+ 'leya+',
+ 'magno',
+ 'malmo',
+ 'markam',
+ 'micha',
+ 'mimsy',
+ 'muf(fy|taay)',
+ 'murd(elizer|i+)',
+ 'nami',
+ 'nula',
+ 'onio',
+ 'paulinche',
+ 'phaaze',
+ 'ralen',
+ 'ramond',
+ 'ray(vis)?',
+ 'schulzer',
+ 'skunk(ner)?',
+ 'skipsy',
+ 'soli+',
+ 'sven(ka+)?',
+ 'tantalus',
+ 'teto',
+ 'thalanee?',
+ 'tick(i+|naldo|y+)',
+ 'tofu',
+ 'tr[i0]x+',
+ 'vin(nie?|ny|ter)',
+ 'xall',
+ 'yasi',
+ ],
+
+ 'pog' => [
+ 'bumble0Pog',
+ 'komodohype',
+ 'pog',
+ 'pogchamp',
+ 'poggers',
+ 'satono0pog',
+ ],
+
+ 'run' => [
+ 'dennsenboots',
+ 'lodanzrun',
+ 'ticknaldosprint',
+ 'vinter0run',
+ ],
+
+ 'wave' => [
+ 'dennsenhi',
+ 'dergoawave',
+ 'falcnwavehi',
+ 'heyguys',
+ 'holysm0heyguys',
+ 'muftaahey',
+ 'rei0wave',
+ 'sayuri0wave',
+ 'shindi0wave',
+ 'svenkawave',
+ 'wuschlwave',
+ ],
+
+ 'zb' => [
+ 'aga(hnim)?',
+ 'armos( knights)?',
+ 'arrghus',
+ 'blind',
+ 'ganon(dorf)?',
+ 'helma',
+ 'kholdstare',
+ 'lanmo(las)?',
+ 'moldorm',
+ 'mothula',
+ 'mott[ei]',
+ 'trinexx',
+ 'vit(reous|ty)',
+ ],
+
+ 'zd' => [
+ 'eastern',
+ 'desert( palace)?',
+ 'gt',
+ 'hera',
+ 'ice ?(palace)?',
+ '(misery )?mire',
+ 'pod',
+ 'skull ?woods',
+ 'swamp',
+ 'thieve\'?s\'? ?town',
+ 'tr',
+ 'tt',
+ ],
+
+ 'zi' => [
+ '(big|small|retro|generic) ?keys?',
+ 'b[oö]gen',
+ 'bombos',
+ 'boots',
+ 'bottle',
+ 'bows?',
+ 'bugnet',
+ 'byrna',
+ 'cape',
+ 'ether',
+ 'flasche',
+ 'flippers',
+ 'fl[uö]te',
+ 'frod',
+ '(gloves?|mitts|handschuhe?)',
+ '(half|quarter) ?magic',
+ 'hammer',
+ 'hookshot',
+ '(ice|fire) ?rod',
+ 'lampe?',
+ 'laser ?bridge',
+ 'mearl',
+ 'mirror',
+ 'moon ?pearl',
+ 'mushroom',
+ 'ocarina',
+ 'pilz',
+ 'powder',
+ 'puder',
+ 'quake',
+ '(red|blue) ?cane',
+ '(red|green|blue) ?(goo|potion)',
+ '(red|green|blue|baby) ?mail',
+ '(red|blue|bu|boo|good|bad|both)merang',
+ 'schaufel',
+ '(gro(ss|ß)er? |kleiner? )?schlüssel',
+ 'schwert',
+ 'shovel',
+ 'silvers',
+ 'somaria',
+ 'spiegel',
+ 'sword',
+ ],
+
+ 'zl' => [
+ 'big chest',
+ 'bumper( cave)?( ledge)?',
+ '(hyrule)? ?castle ?(tower)?',
+ 'catfish',
+ 'cave 0?',
+ 'chest ?game',
+ 'cutscene ?chest',
+ 'damm',
+ 'desert( ledge)?',
+ 'dig(ging)? ?game',
+ '((back|front) of )?escape',
+ 'gyl',
+ 'hobo',
+ 'hook ?(shot) cave',
+ 'lava ?chest',
+ '(light|dark) ?world',
+ 'lss',
+ 'magic bat',
+ '(dark )?(death )?mountain',
+ 'ped(estal)?',
+ 'pyramid( fairy)?( ledge)?',
+ 'red bomb',
+ 'sahasrahla',
+ 'sasha',
+ 'sick kid',
+ 'stumpy',
+ 'tile ?room',
+ 'torch',
+ 'zora( ledge)?',
+ ],
+ ];
+
}