tag analyse v1

composer.development.json

@@ -21,7 +21,8 @@
         "teewurst\/pipeline": "^3.0",
         "guzzlehttp\/guzzle": "^7.8",
         "micilini\/video-stream": "^1.0",
-        "nesbot\/carbon": "^3.0"
+        "nesbot\/carbon": "^3.0",
+        "ext-iconv": "*"
     },
     "autoload": {
         "psr-4": {

composer.json

@@ -21,7 +21,8 @@
         "teewurst/pipeline": "^3.0",
         "guzzlehttp/guzzle": "^7.8",
         "micilini/video-stream": "^1.0",
-        "nesbot/carbon": "^3.0"
+        "nesbot/carbon": "^3.0",
+      "ext-iconv": "*"
     },
     "autoload": {
         "psr-4": {

src/ApiDomain/Console/config/console.php

@@ -1,5 +1,6 @@
 <?php
 
+use MyTube\API\Console\Command\AnalyzeTagsCommand;
 use MyTube\API\Console\Command\AnalyzeVideoTitlesCommand;
 use MyTube\API\Console\Command\InitializeDataCommand;
 use MyTube\API\Console\Command\RbacUpdateCommand;
@@ -11,5 +12,6 @@ return [
         RbacUpdateCommand::class,
         AnalyzeVideoTitlesCommand::class,
         ReadUntaggedVideosCommand::class,
+        AnalyzeTagsCommand::class,
     ]
 ];

src/ApiDomain/Console/config/service_manager.php

@@ -1,5 +1,6 @@
 <?php
 
+use MyTube\API\Console\Command\AnalyzeTagsCommand;
 use MyTube\API\Console\Command\AnalyzeVideoTitlesCommand;
 use MyTube\API\Console\Command\InitializeDataCommand;
 use MyTube\API\Console\Command\RbacUpdateCommand;
@@ -11,6 +12,7 @@ return [
         InitializeDataCommand::class => AutoWiringFactory::class,
         RbacUpdateCommand::class => AutoWiringFactory::class,
         AnalyzeVideoTitlesCommand::class => AutoWiringFactory::class,
+        AnalyzeTagsCommand::class => AutoWiringFactory::class,
         ReadUntaggedVideosCommand::class => AutoWiringFactory::class,
     ],
 ];

src/ApiDomain/Console/src/Command/AnalyzeTagsCommand.php

@@ -0,0 +1,133 @@
+<?php
+
+namespace MyTube\API\Console\Command;
+
+use MyTube\Data\Business\Entity\Tag;
+use MyTube\Data\Business\Entity\Video;
+use MyTube\Data\Business\Manager\MyTubeEntityManager;
+use MyTube\Data\Business\Repository\VideoRepository;
+use MyTube\Handling\Tag\Rule\IsTagSubstringRule;
+use MyTube\Handling\Video\Analyzer\VideoDurationAnalyzer;
+use MyTube\Handling\Video\Analyzer\VideoTitleAnalyzer;
+use MyTube\Infrastructure\Logging\Logger\Logger;
+use Symfony\Component\Console\Attribute\AsCommand;
+use Symfony\Component\Console\Command\Command;
+use Symfony\Component\Console\Input\InputInterface;
+use Symfony\Component\Console\Output\OutputInterface;
+use Symfony\Component\Console\Style\SymfonyStyle;
+use function Webmozart\Assert\Tests\StaticAnalysis\length;
+
+#[AsCommand(name: 'analyze:tags', description: 'Analyzes video titles and add tags')]
+class AnalyzeTagsCommand extends Command
+{
+    private readonly VideoRepository $videoRepository;
+
+    public function __construct(
+        private readonly MyTubeEntityManager $entityManager,
+        private readonly IsTagSubstringRule $isTagSubstringRule,
+        private readonly Logger $logger,
+    ) {
+        parent::__construct($this->getName());
+
+        $this->videoRepository = $this->entityManager->getRepository(Video::class);
+    }
+
+    protected function execute(
+        InputInterface $input,
+        OutputInterface $output
+    ): int {
+        $io = new SymfonyStyle($input, $output);
+
+        try {
+            $videos = $this->videoRepository->findAll();
+
+            /** @var Video $video */
+            foreach ($videos as $video) {
+                $comments[] = $video->getTitle();
+            }
+
+
+            // Normalisierte Kommentare
+            $normalized_comments = array_map([$this, 'normalize'], $comments);
+
+            // Tokenisierung und Wortzählung
+            $word_counts = [];
+            foreach ($normalized_comments as $comment) {
+                $words = explode(' ', $comment);
+                foreach ($words as $word) {
+                    if ($word) {
+                        if (!isset($word_counts[$word])) {
+                            $word_counts[$word] = 0;
+                        }
+                        $word_counts[$word]++;
+                    }
+                }
+            }
+
+            // Konsolidierung der Wörter unter Berücksichtigung von Tippfehlern
+            $corrected_word_counts = [];
+            $dictionary = array_keys($word_counts);
+            foreach ($word_counts as $word => $count) {
+                $correct_word = $this->correct_typo($word, $dictionary);
+                if (!isset($corrected_word_counts[$correct_word])) {
+                    $corrected_word_counts[$correct_word] = 0;
+                }
+                $corrected_word_counts[$correct_word] += $count;
+            }
+
+            // Sortieren nach Häufigkeit
+            arsort($corrected_word_counts);
+            $corrected_word_counts = array_reverse($corrected_word_counts);
+
+            // Ausgabe der häufigsten Wörter
+            foreach ($corrected_word_counts as $word => $count) {
+                if ($count > 3 && !$this->isTagSubstringRule->appliesTo($word)) {
+                    echo $word . ": " . $count . "\n";
+                }
+            }
+
+
+            $io->success('OK!');
+        } catch (\Throwable $e) {
+            $io->error($e->getMessage());
+            $io->error($e->getTraceAsString());
+            $this->logger->error($e->getMessage(), ['exception' => $e]);
+            return Command::FAILURE;
+        }
+
+        return Command::SUCCESS;
+    }
+
+
+    function normalize($text) {
+        // Kleinbuchstaben
+        $text = mb_strtolower($text);
+        // Akzente entfernen
+        $text = iconv('UTF-8', 'ASCII//TRANSLIT', $text);
+        // Interpunktion entfernen
+        $text = preg_replace("/[^a-z\s]/", "", $text);
+        // Trimmen
+        $text = trim($text);
+        return $text;
+    }
+
+
+// Tippfehlerkorrektur mit Levenshtein-Distanz
+    function correct_typo($word, $dictionary) {
+        $closest_word = $word;
+        $shortest_distance = -1;
+        foreach ($dictionary as $dict_word) {
+            $lev = levenshtein($word, $dict_word);
+            if ($lev == 0) {
+                $closest_word = $word;
+                $shortest_distance = 0;
+                break;
+            }
+            if ($lev <= 2 && ($lev < $shortest_distance || $shortest_distance < 0)) {
+                $closest_word = $dict_word;
+                $shortest_distance = $lev;
+            }
+        }
+        return $closest_word;
+    }
+}

src/HandlingDomain/Analyze/src/Repository/AnalyzeVideoRepository.php

@@ -10,17 +10,17 @@ use MyTube\Data\Business\Manager\MyTubeEntityManager;
 
 class AnalyzeVideoRepository
 {
-    public function __construct(
-        private readonly MyTubeEntityManager $entityManager
-    ) {
-    }
-
     private const FIELD_MAP = [
         'duration' => 'v.duration',
         'title' => 'v.title',
         'createdAt' => 'v.createdAt'
     ];
 
+    public function __construct(
+        private readonly MyTubeEntityManager $entityManager
+    ) {
+    }
+
     public function findByFilter(
         ?string $query,
         int $page,

src/HandlingDomain/Tag/config/service_manager.php

@@ -12,6 +12,7 @@ use MyTube\Handling\Tag\Handler\Query\ReadThumbnail\ReadThumbnailQueryBuilder;
 use MyTube\Handling\Tag\Handler\Query\ReadThumbnail\ReadThumbnailQueryHandler;
 use MyTube\Handling\Tag\Handler\Query\ReadVideoList\ReadVideoListQueryBuilder;
 use MyTube\Handling\Tag\Handler\Query\ReadVideoList\ReadVideoListQueryHandler;
+use MyTube\Handling\Tag\Rule\IsTagSubstringRule;
 use MyTube\Handling\Tag\Rule\TagAliasExistsRule;
 use MyTube\Handling\Tag\Rule\TagExistsRule;
 use Reinfi\DependencyInjection\Factory\AutoWiringFactory;
@@ -24,6 +25,7 @@ return [
         /// Rule
         TagExistsRule::class => InjectionFactory::class,
         TagAliasExistsRule::class => InjectionFactory::class,
+        IsTagSubstringRule::class => InjectionFactory::class,
 
         /// Builder
         TagBuilder::class => AutoWiringFactory::class,

src/HandlingDomain/Tag/src/Rule/IsTagSubstringRule.php

@@ -0,0 +1,34 @@
+<?php
+
+namespace MyTube\Handling\Tag\Rule;
+
+use MyTube\Data\Business\Repository\TagRepository;
+use Reinfi\DependencyInjection\Annotation\InjectDoctrineRepository;
+
+class IsTagSubstringRule
+{
+    /**
+     * @InjectDoctrineRepository(
+     *      entityManager="MyTube\Data\Business\Manager\MyTubeEntityManager",
+     *      entity="MyTube\Data\Business\Entity\Tag"
+     * )
+     */
+    public function __construct(
+        private readonly TagRepository $tagRepository,
+    ) {
+    }
+
+    public function appliesTo(
+        string $substring,
+    ): bool
+    {
+        $substring = "%" . $substring . "%";
+
+        $qb = $this->tagRepository->createQueryBuilder('t')
+            ->where('t.description like :substring')
+            ->setParameter('substring', $substring);
+
+
+        return count($qb->getQuery()->getResult()) !== 0;
+    }
+}