Merge pull request #2 from cosmocode/bot/autofix

🤖 Automatic code style fixes
cosmocode · Sep 13, 2023 · d145bc5 · d145bc5
2 parents b377ce3 + 7ebc789
commit d145bc5
Show file tree

Hide file tree

Showing 18 changed files with 81 additions and 74 deletions.
diff --git a/AIChat.php b/AIChat.php
@@ -8,9 +8,9 @@
 class AIChat
 {
     /** @var int preferUIlanguage config: guess language use, all sources */
-    const LANG_AUTO_ALL = 0;
+    public const LANG_AUTO_ALL = 0;
     /** @var int preferUIlanguage config: use UI language, all sources */
-    const LANG_UI_ALL = 1;
+    public const LANG_UI_ALL = 1;
     /** @var int preferUIlanguage config: use UI language, limit sources */
-    const LANG_UI_LIMITED = 2;
+    public const LANG_UI_LIMITED = 2;
 }
diff --git a/Chunk.php b/Chunk.php
@@ -138,9 +138,6 @@ public function setScore($score): void
         $this->score = $score;
     }
 
-    /**
-     * @return string
-     */
     public function getLanguage(): string
     {
         return $this->language;
@@ -182,7 +179,7 @@ protected function determineLanguage()
      * @param string $json
      * @return Chunk
      */
-    static public function fromJSON($json)
+    public static function fromJSON($json)
     {
         $data = json_decode($json, true);
         return new self(

diff --git a/Embeddings.php b/Embeddings.php
@@ -2,6 +2,7 @@
 
 namespace dokuwiki\plugin\aichat;
 
+use dokuwiki\Extension\PluginInterface;
 use dokuwiki\plugin\aichat\Model\AbstractModel;
 use dokuwiki\plugin\aichat\Storage\AbstractStorage;
 use dokuwiki\Search\Indexer;
@@ -18,7 +19,7 @@
 class Embeddings
 {
     /** @var int maximum overlap between chunks in tokens */
-    const MAX_OVERLAP_LEN = 200;
+    public const MAX_OVERLAP_LEN = 200;
 
     /** @var AbstractModel */
     protected $model;
@@ -70,7 +71,7 @@ public function setLogger(CLI $logger)
      */
     public function getTokenEncoder()
     {
-        if ($this->tokenEncoder === null) {
+        if (!$this->tokenEncoder instanceof Encoder) {
             $this->tokenEncoder = new Encoder();
         }
         return $this->tokenEncoder;
@@ -108,7 +109,7 @@ public function createNewIndex($skipRE = '', $clear = false)
             if ($firstChunk && @filemtime(wikiFN($page)) < $firstChunk->getCreated()) {
                 // page is older than the chunks we have, reuse the existing chunks
                 $this->storage->reusePageChunks($page, $chunkID);
-                if ($this->logger) $this->logger->info("Reusing chunks for $page");
+                if ($this->logger instanceof CLI) $this->logger->info("Reusing chunks for $page");
             } else {
                 // page is newer than the chunks we have, create new chunks
                 $this->storage->deletePageChunks($page, $chunkID);
@@ -134,7 +135,7 @@ protected function createPageChunks($page, $firstChunkID)
         $chunkList = [];
 
         $textRenderer = plugin_load('renderer', 'text');
-        if ($textRenderer) {
+        if ($textRenderer instanceof PluginInterface) {
             global $ID;
             $ID = $page;
             $text = p_cached_output(wikiFN($page), 'text', $page);
@@ -149,7 +150,7 @@ protected function createPageChunks($page, $firstChunkID)
             try {
                 $embedding = $this->model->getEmbedding($part);
             } catch (\Exception $e) {
-                if ($this->logger) {
+                if ($this->logger instanceof CLI) {
                     $this->logger->error(
                         'Failed to get embedding for chunk of page {page}: {msg}',
                         ['page' => $page, 'msg' => $e->getMessage()]
@@ -160,8 +161,8 @@ protected function createPageChunks($page, $firstChunkID)
             $chunkList[] = new Chunk($page, $firstChunkID, $part, $embedding);
             $firstChunkID++;
         }
-        if ($this->logger) {
-            if (count($chunkList)) {
+        if ($this->logger instanceof CLI) {
+            if ($chunkList !== []) {
                 $this->logger->success('{id} split into {count} chunks', ['id' => $page, 'count' => count($chunkList)]);
             } else {
                 $this->logger->warning('{id} could not be split into chunks', ['id' => $page]);
@@ -181,7 +182,7 @@ protected function createPageChunks($page, $firstChunkID)
      * @return Chunk[]
      * @throws \Exception
      */
-    public function getSimilarChunks($query, $lang='')
+    public function getSimilarChunks($query, $lang = '')
     {
         global $auth;
         $vector = $this->model->getEmbedding($query);
@@ -193,7 +194,7 @@ public function getSimilarChunks($query, $lang='')
 
         $time = microtime(true);
         $chunks = $this->storage->getSimilarChunks($vector, $lang, $fetch);
-        if ($this->logger) {
+        if ($this->logger instanceof CLI) {
             $this->logger->info(
                 'Fetched {count} similar chunks from store in {time} seconds',
                 ['count' => count($chunks), 'time' => round(microtime(true) - $time, 2)]
@@ -236,7 +237,7 @@ public function splitIntoChunks($text)
             $slen = count($tiktok->encode($sentence));
             if ($slen > $this->model->getMaxEmbeddingTokenLength()) {
                 // sentence is too long, we need to split it further
-                if ($this->logger) $this->logger->warning('Sentence too long, splitting not implemented yet');
+                if ($this->logger instanceof CLI) $this->logger->warning('Sentence too long, splitting not implemented yet');
                 continue;
             }
 
@@ -251,7 +252,7 @@ public function splitIntoChunks($text)
                 $chunks[] = $chunk;
 
                 // start new chunk with remembered sentences
-                $chunk = join(' ', $this->sentenceQueue);
+                $chunk = implode(' ', $this->sentenceQueue);
                 $chunk .= $sentence;
                 $chunklen = count($tiktok->encode($chunk));
             }
@@ -274,7 +275,7 @@ protected function rememberSentence($sentence)
 
         // remove oldest sentences from queue until we are below the max overlap
         $encoder = $this->getTokenEncoder();
-        while (count($encoder->encode(join(' ', $this->sentenceQueue))) > self::MAX_OVERLAP_LEN) {
+        while (count($encoder->encode(implode(' ', $this->sentenceQueue))) > self::MAX_OVERLAP_LEN) {
             array_shift($this->sentenceQueue);
         }
     }

diff --git a/Model/AbstractModel.php b/Model/AbstractModel.php
@@ -2,10 +2,8 @@
 
 namespace dokuwiki\plugin\aichat\Model;
 
-
-abstract class AbstractModel {
-
-
+abstract class AbstractModel
+{
     /** @var int total tokens used by this instance */
     protected $tokensUsed = 0;
     /** @var int total cost used by this instance (multiplied by 1000*10000) */
@@ -43,7 +41,8 @@ abstract public function getMaxContextTokenLength();
      *
      * @return int
      */
-    public function getMaxRephrasingTokenLength() {
+    public function getMaxRephrasingTokenLength()
+    {
         return $this->getMaxContextTokenLength();
     }
 
@@ -77,7 +76,8 @@ abstract public function getAnswer($messages);
      * @return string The new question
      * @throws \Exception
      */
-    public function getRephrasedQuestion($messages) {
+    public function getRephrasedQuestion($messages)
+    {
         return $this->getAnswer($messages);
     }
 

diff --git a/Model/OpenAI/GPT35Turbo.php b/Model/OpenAI/GPT35Turbo.php
@@ -13,7 +13,7 @@
 class GPT35Turbo extends AbstractModel
 {
     /** @var int[] real 1K cost multiplied by 10000 to avoid floating point issues, as of 2023-06-14 */
-    static protected $prices = [
+    protected static $prices = [
         'text-embedding-ada-002' => 1, // $0.0001 per 1k token
         'gpt-3.5-turbo' => 15, // $0.0015 per 1k token
         'gpt-3.5-turbo-16k' => 30, // $0.003 per 1k token
@@ -22,14 +22,14 @@ class GPT35Turbo extends AbstractModel
     ];
 
     /** @var array[] The models and limits for the different use cases */
-    static protected $setup = [
+    protected static $setup = [
         'embedding' => ['text-embedding-ada-002', 1000], // chunk size
         'rephrase' => ['gpt-3.5-turbo', 3500], // rephrasing context size
         'chat' => ['gpt-3.5-turbo', 3500], // question context size
     ];
 
     /** @var int How often to retry a request if it fails */
-    const MAX_RETRIES = 3;
+    public const MAX_RETRIES = 3;
 
     /** @var DokuHTTPClient */
     protected $http;
@@ -97,7 +97,8 @@ public function getRephrasedQuestion($messages)
     /**
      * @internal for checking available models
      */
-    public function listUpstreamModels() {
+    public function listUpstreamModels()
+    {
         $url = 'https://api.openai.com/v1/models';
         $result = $this->http->get($url);
         return $result;
@@ -167,7 +168,7 @@ protected function request($endpoint, $data, $retry = 0)
         }
 
         // update usage statistics
-        if(isset($result['usage'])) {
+        if (isset($result['usage'])) {
             $price = self::$prices[$data['model']] ?? 0;
             $this->tokensUsed += $result['usage']['total_tokens'];
             $this->costEstimate += $result['usage']['total_tokens'] * $price;

diff --git a/Model/OpenAI/GPT35Turbo16k.php b/Model/OpenAI/GPT35Turbo16k.php
@@ -9,7 +9,7 @@
  */
 class GPT35Turbo16K extends GPT35Turbo
 {
-    static protected $setup = [
+    protected static $setup = [
         'embedding' => ['text-embedding-ada-002', 3000],
         'rephrase' => ['gpt-3.5-turbo', 3500],
         'chat' => ['gpt-3.5-turbo-16k', 6000],

diff --git a/Model/OpenAI/GPT4.php b/Model/OpenAI/GPT4.php
@@ -9,7 +9,7 @@
  */
 class GPT4 extends GPT35Turbo
 {
-    static protected $setup = [
+    protected static $setup = [
         'embedding' => ['text-embedding-ada-002', 2000],
         'rephrase' => ['gpt-4', 3500],
         'chat' => ['gpt-4', 3000],

diff --git a/Storage/AbstractStorage.php b/Storage/AbstractStorage.php
@@ -21,7 +21,8 @@ abstract class AbstractStorage
      * @param CLIPlugin $logger
      * @return void
      */
-    public function setLogger($logger) {
+    public function setLogger($logger)
+    {
         $this->logger = $logger;
     }
 
@@ -112,7 +113,7 @@ abstract public function getPageChunks($page, $firstChunkID);
      * @param int $limit The number of results to return, see note above
      * @return Chunk[]
      */
-    abstract public function getSimilarChunks($vector, $lang='', $limit = 4);
+    abstract public function getSimilarChunks($vector, $lang = '', $limit = 4);
 
     /**
      * Get information about the storage
@@ -130,7 +131,8 @@ abstract public function statistics();
      * @param string $metafile path to the file with the metadata
      * @return void
      */
-    public function dumpTSV($vectorfile, $metafile) {
+    public function dumpTSV($vectorfile, $metafile)
+    {
         throw new \RuntimeException('Not implemented for current storage');
     }
 }
diff --git a/Storage/PineconeStorage.php b/Storage/PineconeStorage.php
@@ -45,7 +45,7 @@ protected function runQuery($endpoint, $data, $method = 'POST')
     {
         $url = $this->baseurl . $endpoint;
 
-        if (is_array($data) && !count($data)) {
+        if (is_array($data) && $data === []) {
             $json = '{}';
         } else {
             $json = json_encode($data);
@@ -197,7 +197,7 @@ public function getPageChunks($page, $firstChunkID)
     /** @inheritdoc */
     public function getSimilarChunks($vector, $lang = '', $limit = 4)
     {
-        $limit = $limit * 2; // we can't check ACLs, so we return more than requested
+        $limit *= 2; // we can't check ACLs, so we return more than requested
 
         if ($lang) {
             $filter = ['language' => ['$eq', $lang]];

diff --git a/Storage/SQLiteStorage.php b/Storage/SQLiteStorage.php
@@ -1,4 +1,6 @@
-<?php /** @noinspection SqlResolve */
+<?php
+
+/** @noinspection SqlResolve */
 
 namespace dokuwiki\plugin\aichat\Storage;
 
@@ -16,12 +18,12 @@
 class SQLiteStorage extends AbstractStorage
 {
     /** @var float minimum similarity to consider a chunk a match */
-    const SIMILARITY_THRESHOLD = 0.75;
+    public const SIMILARITY_THRESHOLD = 0.75;
 
     /** @var int Number of documents to randomly sample to create the clusters */
-    const SAMPLE_SIZE = 2000;
+    public const SAMPLE_SIZE = 2000;
     /** @var int The average size of each cluster */
-    const CLUSTER_SIZE = 400;
+    public const CLUSTER_SIZE = 400;
 
     /** @var SQLiteDB */
     protected $db;
@@ -138,7 +140,8 @@ public function getSimilarChunks($vector, $lang = '', $limit = 4)
     {
         $cluster = $this->getCluster($vector, $lang);
         if ($this->logger) $this->logger->info(
-            'Using cluster {cluster} for similarity search', ['cluster' => $cluster]
+            'Using cluster {cluster} for similarity search',
+            ['cluster' => $cluster]
         );
 
         $result = $this->db->queryAll(
@@ -282,15 +285,15 @@ protected function createLanguageClusters($lang)
                 static $iterations = 0;
                 ++$iterations;
                 if ($this->logger) {
-                    $clustercounts = join(',', array_map('count', $clusters));
+                    $clustercounts = implode(',', array_map('count', $clusters));
                     $this->logger->info('Iteration {iteration}: [{clusters}]', [
                         'iteration' => $iterations, 'clusters' => $clustercounts
                     ]);
                 }
             }, Cluster::INIT_KMEANS_PLUS_PLUS);
 
             // store the clusters
-            foreach ($clusters as $clusterID => $cluster) {
+            foreach ($clusters as $cluster) {
                 /** @var Cluster $cluster */
                 $centroid = $cluster->getCoordinates();
                 $query = 'INSERT INTO clusters (lang, centroid) VALUES (?, ?)';
@@ -322,7 +325,8 @@ protected function setChunkClusters()
             $query = 'UPDATE embeddings SET cluster = ? WHERE id = ?';
             $this->db->exec($query, [$cluster, $record['id']]);
             if ($this->logger) $this->logger->success(
-                'Chunk {id} assigned to cluster {cluster}', ['id' => $record['id'], 'cluster' => $cluster]
+                'Chunk {id} assigned to cluster {cluster}',
+                ['id' => $record['id'], 'cluster' => $cluster]
             );
         }
         $handle->closeCursor();

diff --git a/action.php b/action.php
@@ -1,5 +1,9 @@
 <?php
 
+use dokuwiki\Extension\ActionPlugin;
+use dokuwiki\Extension\EventHandler;
+use dokuwiki\Extension\Event;
+use dokuwiki\Logger;
 use dokuwiki\ErrorHandler;
 use dokuwiki\plugin\aichat\Chunk;
 
@@ -9,11 +13,10 @@
  * @license GPL 2 http://www.gnu.org/licenses/gpl-2.0.html
  * @author  Andreas Gohr <[email protected]>
  */
-class action_plugin_aichat extends \dokuwiki\Extension\ActionPlugin
+class action_plugin_aichat extends ActionPlugin
 {
-
     /** @inheritDoc */
-    public function register(Doku_Event_Handler $controller)
+    public function register(EventHandler $controller)
     {
         $controller->register_hook('AJAX_CALL_UNKNOWN', 'BEFORE', $this, 'handleQuestion');
     }
@@ -27,7 +30,7 @@ public function register(Doku_Event_Handler $controller)
      * @param mixed $param optional parameter passed when event was registered
      * @return void
      */
-    public function handleQuestion(Doku_Event $event, $param)
+    public function handleQuestion(Event $event, $param)
     {
         if ($event->data !== 'aichat') return;
         $event->preventDefault();
@@ -67,7 +70,7 @@ public function handleQuestion(Doku_Event $event, $param)
             ]);
 
             if ($this->getConf('logging')) {
-                \dokuwiki\Logger::getInstance('aichat')->log(
+                Logger::getInstance('aichat')->log(
                     $question,
                     [
                         'interpretation' => $result['question'],
@@ -88,6 +91,4 @@ public function handleQuestion(Doku_Event $event, $param)
             ]);
         }
     }
-
 }
-