refined classification

diff --git a/app/TwitchBot/TokenizedMessage.php b/app/TwitchBot/TokenizedMessage.php
index 82b4d473f1c0761c8b14ac36c2861681561fdabe..42d6ca8ca8f08c05cc625e3b32df236ef4683932 100644 (file)
--- a/app/TwitchBot/TokenizedMessage.php
+++ b/app/TwitchBot/TokenizedMessage.php
@@ -12,7 +12,7 @@ class TokenizedMessage {
                $this->text = $text;
                $this->tags = $tags;
                $this->raw = strtolower(preg_replace('/[^\w]/u', '', $text));
-               $this->tokens = preg_split('/\s+/', strtolower(trim($text)));
+               $this->tokens = array_values(array_map('trim', array_filter(preg_split('/\b/', strtolower($text)))));
 
                $this->emoteless = $this->text;
                if (isset($this->tags['emotes']) && !empty($this->tags['emotes'])) {
@@ -31,7 +31,7 @@ class TokenizedMessage {
                        $this->emoteless = trim(preg_replace('/\s+/', ' ', $this->emoteless));
                }
                $this->emoteless_raw = strtolower(preg_replace('/[^\w]/', '', $this->emoteless));
-               $this->emoteless_tokens = preg_split('/\s+/', strtolower($this->emoteless));
+               $this->emoteless_tokens = array_values(array_map('trim', array_filter(preg_split('/\b/', strtolower($this->emoteless)))));
        }
 
        public static function fromIRC(IRCMessage $msg) {
@@ -75,6 +75,16 @@ class TokenizedMessage {
                return intval($this->text);
        }
 
+       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) {
@@ -224,23 +234,23 @@ class TokenizedMessage {
                                $this->classification = 'gg';
                        } else if ($this->containsRaw(['glgl', 'glhf', 'goodluck', 'hfgl'])) {
                                $this->classification = 'gl';
-                       } else if ($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->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'])) {
                                $this->classification = 'pog';
-                       } else if ($this->containsRaw(['hype'])) {
+                       } else if ($this->containsRaw(['hype']) || $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->hasToken(['wat', 'wat?']) || $this->hasTokenThatStartsWith(['wtf']) || $this->hasEmoteThatEndsWith(['wat', 'wtf'])) {
+                       } else if ($this->hasToken(['nani', 'wat', 'wtf']) || $this->hasEmoteThatEndsWith(['wat', 'wtf'])) {
                                $this->classification = 'wtf';
                        } else if ($this->endsWithEmoteless('?')) {
                                $this->classification = 'question';
                        } else if ($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->containsRaw(['haha', 'hehe', 'hihi', 'kekw', 'lol', 'lul']) || $this->hasTokenThatStartsWith(['xd']) || $this->hasConsecutiveTokens([':', 'd'])) {
                                $this->classification = 'lol';
                        } else {
                                $this->classification = 'unclassified';