From ac6921da72ff4b0beab9e5f1308788a55aae3ad9 Mon Sep 17 00:00:00 2001
From: Daniel Karbach <daniel.karbach@localhorst.tv>
Date: Thu, 9 May 2024 12:32:01 +0200
Subject: [PATCH] add emote stats command

---
 app/Console/Commands/TwitchTopEmotes.php | 62 ++++++++++++++++++++++++
 app/Models/ChatLog.php                   |  4 +-
 app/TwitchBot/TokenizedMessage.php       | 13 ++++-
 3 files changed, 75 insertions(+), 4 deletions(-)
 create mode 100644 app/Console/Commands/TwitchTopEmotes.php

diff --git a/app/Console/Commands/TwitchTopEmotes.php b/app/Console/Commands/TwitchTopEmotes.php
new file mode 100644
index 0000000..81436d9
--- /dev/null
+++ b/app/Console/Commands/TwitchTopEmotes.php
@@ -0,0 +1,62 @@
+<?php
+
+namespace App\Console\Commands;
+
+use App\Models\ChatLog;
+use Illuminate\Console\Command;
+
+class TwitchTopEmotes extends Command {
+
+	/**
+	 * The name and signature of the console command.
+	 *
+	 * @var string
+	 */
+	protected $signature = 'twitch:top-emotes {amount=10}';
+
+	/**
+	 * The console command description.
+	 *
+	 * @var string
+	 */
+	protected $description = 'Compiles a list of top N emotes in the chatbot catalog';
+
+	/**
+	 * Execute the console command.
+	 *
+	 * @return int
+	 */
+	public function handle() {
+		$counts = [];
+
+		ChatLog::where('type', '=', 'chat')
+			->where('banned', '=', false)
+			->whereNotNull('evaluated_at')
+			->chunk(5000, function ($msgs) use (&$counts) {
+				foreach ($msgs as $msg) {
+					$tokenized = $msg->tokenize();
+					foreach ($tokenized->getOriginalEmotes() as $emote) {
+						if (!isset($counts[$emote])) {
+							$counts[$emote] = 1;
+						} else {
+							++$counts[$emote];
+						}
+					}
+				}
+			});
+
+		arsort($counts);
+
+		$amount = intval($this->argument('amount'));
+		$counts = array_slice($counts, 0, $amount);
+
+		foreach ($counts as $emote => $count) {
+			$this->line(mb_str_pad($emote, 20).$count);
+		}
+
+		return 0;
+	}
+
+}
+
+?>
diff --git a/app/Models/ChatLog.php b/app/Models/ChatLog.php
index 6af1587..2d43b2f 100644
--- a/app/Models/ChatLog.php
+++ b/app/Models/ChatLog.php
@@ -34,9 +34,7 @@ class ChatLog extends Model {
 				$positions = explode(',', $set[1]);
 				foreach ($positions as $position) {
 					$coords = explode('-', $position);
-					for ($i = intval($coords[0]); $i <= intval($coords[1]); ++$i) {
-						$text[$i] = ' ';
-					}
+					$text = mb_substr($text, 0, $coords[0]).str_repeat(' ', $coords[1] - $coords[0] + 1).mb_substr($text, $coords[1] + 1);
 				}
 			}
 		}
diff --git a/app/TwitchBot/TokenizedMessage.php b/app/TwitchBot/TokenizedMessage.php
index 37836a9..4b7d758 100644
--- a/app/TwitchBot/TokenizedMessage.php
+++ b/app/TwitchBot/TokenizedMessage.php
@@ -22,7 +22,9 @@ class TokenizedMessage {
 				$positions = explode(',', $set[1]);
 				foreach ($positions as $position) {
 					$coords = explode('-', $position);
-					$this->emotes[] = preg_replace('/\d+$/', '', strtolower(mb_substr($this->text, $coords[0], $coords[1] - $coords[0] + 1)));
+					$emote_text = mb_substr($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, $coords[1] + 1);
 				}
 			}
@@ -77,10 +79,18 @@ class TokenizedMessage {
 		return !empty($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) {
@@ -344,6 +354,7 @@ class TokenizedMessage {
 	private $tokens;
 
 	private $emotes = [];
+	private $original_emotes = [];
 	private $emoteless = '';
 	private $emoteless_raw = '';
 	private $emoteless_tokens = [];
-- 
2.39.2