class TokenizedMessage {
public function __construct($text, $tags = []) {
- $this->text = $text;
+ $this->text = trim($text);
$this->tags = $tags;
- $this->raw = strtolower(preg_replace('/[^\w]/u', '', $text));
- $this->tokens = preg_split('/\s+/', strtolower(trim($text)));
+ if (isset($tags['reply-parent-display-name'])) {
+ $this->text = mb_substr($text, mb_strlen($tags['reply-parent-display-name']) + 2);
+ }
+ $this->raw = strtolower(preg_replace('/[^\w]/u', '', $this->text));
+ $this->tokens = array_values(array_map('trim', array_filter(preg_split('/\b/u', strtolower($this->text)))));
- $this->emoteless = $this->text;
+ $this->emoteless = $text;
if (isset($this->tags['emotes']) && !empty($this->tags['emotes'])) {
$emotes = explode('/', $this->tags['emotes']);
foreach ($emotes as $emote) {
$positions = explode(',', $set[1]);
foreach ($positions as $position) {
$coords = explode('-', $position);
- $this->emotes[] = preg_replace('/\d+$/', '', strtolower(substr($this->text, $coords[0], $coords[1] - $coords[0] + 1)));
- for ($i = intval($coords[0]); $i <= intval($coords[1]); ++$i) {
- $this->emoteless[$i] = ' ';
- }
+ $emote_text = mb_substr($this->text, $coords[0], $coords[1] - $coords[0] + 1);
+ $this->original_emotes[] = $emote_text;
+ $this->emotes[] = preg_replace('/\d+$/', '', strtolower($emote_text));
+ $this->emoteless = mb_substr($this->emoteless, 0, $coords[0]).str_repeat(' ', $coords[1] - $coords[0] + 1).mb_substr($this->emoteless, $coords[1] + 1);
}
}
- $this->emoteless = trim(preg_replace('/\s+/', ' ', $this->emoteless));
+ $this->emoteless = trim(preg_replace('/\s+/u', ' ', $this->emoteless));
}
- $this->emoteless_raw = strtolower(preg_replace('/[^\w]/', '', $this->emoteless));
- $this->emoteless_tokens = preg_split('/\s+/', strtolower($this->emoteless));
+ $this->emoteless_raw = strtolower(preg_replace('/[^\w]/u', '', $this->emoteless));
+ $this->emoteless_tokens = array_values(array_map('trim', array_filter(preg_split('/\b/u', strtolower($this->emoteless)))));
}
public static function fromIRC(IRCMessage $msg) {
}
+ public function getText() {
+ return $this->text;
+ }
+
+
public function contains($text) {
return Str::contains($this->text, $text);
}
return Str::endsWith($this->emoteless, $text);
}
+ public function endsWithEmotelessToken($text) {
+ return !empty($this->emoteless_tokens) && $this->emoteless_tokens[count($this->emoteless_tokens) - 1] == $text;
+ }
+
public function endsWithRaw($text) {
return Str::endsWith($this->raw, $text);
}
+ public function endsWithToken($text) {
+ return !empty($this->tokens) && $this->tokens[count($this->tokens) - 1] == $text;
+ }
+
+ public function getEmotes() {
+ return $this->emotes;
+ }
+
public function getNumericValue() {
return intval($this->text);
}
+ public function getOriginalEmotes() {
+ return $this->original_emotes;
+ }
+
+ public function hasConsecutiveTokens($tokens) {
+ for ($i = 0; $i < count($this->tokens) - count($tokens) + 1; ++$i) {
+ for ($j = 0; $j < count($tokens); ++$j) {
+ if ($this->tokens[$i + $j] != $tokens[$j]) break;
+ }
+ if ($j == count($tokens)) return true;
+ }
+ return false;
+ }
+
public function hasEmote($text) {
if (is_array($text)) {
foreach ($text as $token) {
return false;
}
+ public function isLong() {
+ return strlen($this->emoteless_raw) > 20;
+ }
+
+ public function isShort() {
+ return strlen($this->emoteless_raw) < 15;
+ }
+
+ public function isVeryLong() {
+ return strlen($this->emoteless_raw) > 40;
+ }
+
public function startsOrEndsWith($text) {
return $this->startsWith($text) || $this->endsWith($text);
}
+ public function startsOrEndsWithEmotelessToken($text) {
+ return $this->startsWithEmotelessToken($text) || $this->endsWithEmotelessToken($text);
+ }
+
public function startsOrEndsWithRaw($text) {
return $this->startsWithRaw($text) || $this->endsWithRaw($text);
}
+ public function startsOrEndsWithToken($text) {
+ return $this->startsWithToken($text) || $this->endsWithToken($text);
+ }
+
public function startsWith($text) {
return Str::startsWith($this->text, $text);
}
return Str::startsWith($this->emoteless, $text);
}
+ public function startsWithEmotelessToken($text) {
+ return isset($this->emoteless_tokens[0]) && $this->emoteless_tokens[0] == $text;
+ }
+
public function startsWithRaw($text) {
return Str::startsWith($this->raw, $text);
}
+ public function startsWithToken($text) {
+ return isset($this->tokens[0]) && $this->tokens[0] == $text;
+ }
+
public function isSpammy() {
if ($this->startsWith('!')) {
if ($this->contains(['€', '$', '@', '://'])) {
return true;
}
- if ($this->containsRaw(['followers', 'promotion', 'viewers'])) {
+ if ($this->containsRaw(['follow', 'promotion', 'viewer'])) {
return true;
}
- if ($this->containsRaw('horstie')) {
+ if ($this->containsRaw('horsti')) {
return true;
}
- if ($this->containsRaw(['vielendankfürdenraid', 'thanksfortheraid', 'willkommenaufstarbase47'])) {
+ if ($this->containsRaw([
+ 'folgtjetzt',
+ 'hatdeinenkanalgeraided',
+ 'isnowlivestreaming',
+ 'stürmtdenladenmit',
+ 'thanksfortheraid',
+ 'verschwindetfürneweileindenlurk',
+ 'vielendankfürdenraid',
+ 'willkommenaufstarbase47',
+ ])) {
return true;
}
return false;
public function classify() {
if (is_null($this->classification)) {
- if (empty($this->raw)) {
+ if (empty($this->text) || $this->isVeryLong()) {
$this->classification = 'unclassified';
} else if ($this->startsWith('!')) {
$this->classification = 'cmd';
- } else if (is_numeric($this->raw)) {
- $this->classification = 'number';
- } else if ($this->hasTokenThatStartsOrEndsWith(['gg']) || $this->hasEmoteThatEndsWith(['gg'])) {
+ } else if ($this->isShort() && ($this->hasTokenThatStartsOrEndsWith(['gg']) || $this->hasEmoteThatEndsWith(['gg']))) {
$this->classification = 'gg';
- } else if ($this->containsRaw(['glgl', 'glhf', 'goodluck', 'hfgl'])) {
+ } else if ($this->isShort() && ($this->containsRaw(['glgl', 'glhf', 'goodluck', 'hfgl', 'vielglück']) || $this->hasToken('gl'))) {
$this->classification = 'gl';
- } else if ($this->startsWithRaw(['ahoi', 'hallo', 'hello', 'hey', 'huhu', 'moin']) || $this->hasEmoteThatEndsWith(['hello', 'heyguys', 'hi', 'wave']) || $this->hasToken(['hi', 'hey']) || $this->containsRaw(['gutenmorgen', 'gutenabend'])) {
+ } else if ($this->hasToken(['danke', 'thanks', 'thx', 'ty']) && !$this->hasToken(['nah', 'nee', 'nein', 'no'])) {
+ $this->classification = 'thx';
+ } else if (!$this->isLong() && ($this->startsWithRaw(['ahoi', 'hallo', 'hello', 'hey', 'huhu', 'moin']) || $this->hasEmoteThatEndsWith(['hello', 'heyguys', 'hi', 'vohiyo', 'wave']) || $this->hasToken(['hi', 'hey', 'yo']) || $this->containsRaw(['gutenmorgen', 'gutenabend']))) {
$this->classification = 'hi';
- } else if ($this->hasTokenThatStartsOrEndsWith(['pog', 'wow'])) {
+ } else if ($this->isShort() && $this->hasTokenThatStartsOrEndsWith(['pog', 'wow'])) {
$this->classification = 'pog';
- } else if ($this->containsRaw(['hype'])) {
+ } else if ($this->containsRaw(['hype', 'letsgo']) || $this->hasEmoteThatEndsWith(['dance', 'jam', 'party', 'rave', 'troete'])) {
$this->classification = 'hype';
- } else if ($this->hasToken(['danke', 'thanks', 'thx', 'ty'])) {
- $this->classification = 'thx';
- } else if ($this->hasToken(['<3']) || $this->hasEmoteThatEndsWith(['herz', 'hug', 'love'])) {
+ } else if ($this->hasToken(['<3']) || $this->hasEmoteThatEndsWith(['heart', 'herz', 'hug', 'love'])) {
$this->classification = 'love';
- } else if ($this->hasTokenThatStartsWith(['wat', 'wtf']) || $this->hasEmoteThatStartsWith(['wat', 'wtf'])) {
+ } else if ($this->hasToken(['nani', 'wat', 'wtf']) || $this->hasEmoteThatEndsWith(['wat', 'wtf'])) {
$this->classification = 'wtf';
+ } else if ($this->hasConsecutiveTokens([':', 'eyes', ':']) || $this->hasEmoteThatEndsWith(['eyes'])) {
+ $this->classification = 'eyes';
+ } else if ($this->hasEmoteThatEndsWith(['angry', 'rage', 'ree'])) {
+ $this->classification = 'rage';
+ } else if ($this->hasToken([':(']) || $this->hasEmoteThatEndsWith(['cry', 'sad'])) {
+ $this->classification = 'sad';
+ } else if ($this->hasToken(['monkas', 'sweat_smile']) || $this->hasEmoteThatEndsWith(['sweat'])) {
+ $this->classification = 'sweat';
} else if ($this->endsWithEmoteless('?')) {
$this->classification = 'question';
+ } else if ($this->hasToken(['jo', 'yep', 'yes']) || $this->startsOrEndsWithEmotelessToken('ja') || $this->containsRaw('nodders') || $this->hasEmoteThatEndsWith(['nod', 'nodders', 'yea'])) {
+ $this->classification = 'yes';
+ } else if ($this->hasToken(['nah', 'nee', 'nein', 'no']) || $this->containsRaw('nopers') || $this->hasEmoteThatEndsWith(['nay', 'nope', 'nopers'])) {
+ $this->classification = 'no';
+ } else if ($this->hasEmoteThatContains(['kappa', 'keepo'])) {
+ $this->classification = 'kappa';
} else if ($this->startsOrEndsWithRaw(['o7']) || $this->hasEmoteThatContains('salut')) {
$this->classification = 'o7';
- } else if ($this->containsRaw(['haha', 'hehe', 'hihi', 'kekw', 'lol', 'lul']) || $this->hasTokenThatStartsWith([':d', 'xd'])) {
+ } else if (!$this->isLong() && ($this->containsRaw(['haha', 'hehe', 'hihi', 'kekw', 'lol', 'lul']) || $this->hasTokenThatStartsWith(['xd']) || $this->hasConsecutiveTokens([':', 'd']))) {
$this->classification = 'lol';
+ } else if (is_numeric($this->raw)) {
+ $this->classification = 'number';
} else {
$this->classification = 'unclassified';
}
return $this->classification;
}
+ public function getResponseCategory() {
+ switch ($this->classify()) {
+ case 'gg':
+ return ['love', 'eyes', 'thx', 'pog', 'kappa'];
+ case 'gl':
+ return ['love', 'eyes', 'thx'];
+ case 'hi':
+ return ['hi', 'love', 'eyes', 'hype', 'pog'];
+ case 'kappa':
+ return ['kappa', 'lol', 'eyes'];
+ case 'love':
+ return ['hi', 'love', 'eyes', 'thx'];
+ case 'question':
+ if (
+ $this->hasToken(['number', 'nummer', 'wieviel', 'zahl']) ||
+ $this->hasConsecutiveTokens(['how', 'many']) ||
+ $this->hasConsecutiveTokens(['how', 'much']) ||
+ $this->hasConsecutiveTokens(['wie', 'viele'])
+ ) {
+ return ['yes', 'no', 'kappa', 'wtf', 'number'];
+ }
+ return ['yes', 'no', 'kappa', 'wtf'];
+ case 'rage':
+ return ['kappa', 'lol', 'rage'];
+ case 'wtf':
+ return ['kappa', 'lol', 'rage'];
+ }
+ return false;
+ }
+
private $text;
private $tags;
private $tokens;
private $emotes = [];
+ private $original_emotes = [];
private $emoteless = '';
private $emoteless_raw = '';
private $emoteless_tokens = [];
projects / alttp.git / blobdiff