Improve categorization

app/Console/Commands/FixNzbSplitNames.php

@@ -0,0 +1,113 @@
+<?php
+
+declare(strict_types=1);
+
+namespace App\Console\Commands;
+
+use App\Facades\Search;
+use App\Models\Release;
+use App\Services\Categorization\CategorizationService;
+use App\Services\NameFixing\NzbSplitUnwrapper;
+use Illuminate\Console\Command;
+
+class FixNzbSplitNames extends Command
+{
+    protected $signature = 'releases:fix-nzbsplit
+                            {--dry-run : Preview changes without updating the database}';
+
+    protected $description = 'Fix NZBSPLIT-wrapped release names and recategorize the affected releases.';
+
+    public function handle(CategorizationService $categorizationService, NzbSplitUnwrapper $nzbSplitUnwrapper): int
+    {
+        $query = Release::query()
+            ->select(['id', 'name', 'searchname', 'fromname', 'groups_id', 'categories_id'])
+            ->where(function ($builder): void {
+                $builder
+                    ->where('name', 'like', '%__NZBSPLIT__%')
+                    ->orWhere('searchname', 'like', '%__NZBSPLIT__%');
+            })
+            ->orderBy('id');
+
+        $count = (clone $query)->count();
+
+        if ($count === 0) {
+            $this->info('No NZBSPLIT-wrapped releases found.');
+
+            return self::SUCCESS;
+        }
+
+        $dryRun = (bool) $this->option('dry-run');
+        $updated = 0;
+        $bar = $this->output->createProgressBar($count);
+        $bar->start();
+
+        $query->chunkById(250, function ($releases) use ($categorizationService, $nzbSplitUnwrapper, $dryRun, &$updated, $bar): void {
+            foreach ($releases as $release) {
+                $bar->advance();
+
+                $newTitle = $nzbSplitUnwrapper->unwrap((string) $release->searchname)
+                    ?? $nzbSplitUnwrapper->unwrap((string) $release->name);
+
+                if ($newTitle === null) {
+                    continue;
+                }
+
+                $category = $categorizationService->determineCategory(
+                    $release->groups_id,
+                    $newTitle,
+                    (string) ($release->fromname ?? '')
+                );
+
+                $newCategoryId = (int) ($category['categories_id'] ?? $release->categories_id);
+
+                if ($dryRun) {
+                    $this->newLine();
+                    $this->line(sprintf(
+                        '[dry-run] #%d %s -> %s (%d -> %d)',
+                        $release->id,
+                        (string) $release->searchname,
+                        $newTitle,
+                        (int) $release->categories_id,
+                        $newCategoryId
+                    ));
+
+                    continue;
+                }
+
+                Release::query()
+                    ->where('id', $release->id)
+                    ->update([
+                        'name' => $newTitle,
+                        'searchname' => $newTitle,
+                        'categories_id' => $newCategoryId,
+                        'isrenamed' => 1,
+                        'iscategorized' => 1,
+                        'videos_id' => 0,
+                        'tv_episodes_id' => 0,
+                        'imdbid' => null,
+                        'musicinfo_id' => null,
+                        'consoleinfo_id' => null,
+                        'gamesinfo_id' => 0,
+                        'bookinfo_id' => 0,
+                        'anidbid' => null,
+                    ]);
+
+                Search::updateRelease((int) $release->id);
+                $updated++;
+            }
+        });
+
+        $bar->finish();
+        $this->newLine(2);
+
+        if ($dryRun) {
+            $this->info(sprintf('Dry run complete. %d NZBSPLIT-wrapped releases inspected.', $count));
+
+            return self::SUCCESS;
+        }
+
+        $this->info(sprintf('Updated %d NZBSPLIT-wrapped releases.', $updated));
+
+        return self::SUCCESS;
+    }
+}

app/Services/AdditionalProcessing/ReleaseFileManager.php

@@ -12,6 +12,7 @@ use App\Models\Release;
 use App\Models\ReleaseFile;
 use App\Services\AdditionalProcessing\Config\ProcessingConfiguration;
 use App\Services\AdditionalProcessing\DTO\ReleaseProcessingContext;
+use App\Services\NameFixing\FileNameCleaner;
 use App\Services\NameFixing\NameFixingService;
 use App\Services\NameFixing\ReleaseUpdateService;
 use App\Services\NfoService;
@@ -31,13 +32,22 @@ use Illuminate\Support\Facades\Log;
  */
 class ReleaseFileManager
 {
+    private readonly ReleaseUpdateService $releaseUpdateService;
+
+    private readonly FileNameCleaner $fileNameCleaner;
+
     public function __construct(
         private readonly ProcessingConfiguration $config,
         private readonly ReleaseImageService $releaseImage,
         private readonly NfoService $nfo,
         private readonly NzbService $nzb,
-        private readonly NameFixingService $nameFixingService
-    ) {}
+        private readonly NameFixingService $nameFixingService,
+        ?ReleaseUpdateService $releaseUpdateService = null,
+        ?FileNameCleaner $fileNameCleaner = null
+    ) {
+        $this->releaseUpdateService = $releaseUpdateService ?? new ReleaseUpdateService;
+        $this->fileNameCleaner = $fileNameCleaner ?? new FileNameCleaner;
+    }
 
     /**
      * Add file information to the database.
@@ -152,6 +162,50 @@ class ReleaseFileManager
         Search::updateRelease($releaseId);
     }
 
+    /**
+     * Rename NZBSPLIT-wrapped releases directly from parsed NZB titles.
+     *
+     * @param  list<array<string, mixed>>  $nzbContents
+     */
+    public function processReleaseNameFromNzbContents(array $nzbContents, ReleaseProcessingContext $context): bool
+    {
+        foreach ($nzbContents as $nzbFile) {
+            $title = (string) ($nzbFile['title'] ?? '');
+
+            if ($title === '' || ! str_contains($title, '__NZBSPLIT__')) {
+                continue;
+            }
+
+            $candidate = $this->fileNameCleaner->extractNzbSplitName($title);
+
+            if ($candidate === null) {
+                continue;
+            }
+
+            $candidate = $this->fileNameCleaner->normalizeCandidateTitle($candidate);
+
+            if (! $this->fileNameCleaner->isPlausibleReleaseTitle($candidate)) {
+                continue;
+            }
+
+            $this->releaseUpdateService->updateRelease(
+                $context->release,
+                $candidate,
+                'NZBSPLIT wrapper',
+                true,
+                'Filenames, ',
+                true,
+                true
+            );
+
+            $context->release->searchname = $candidate;
+
+            return true;
+        }
+
+        return false;
+    }
+
     /**
      * Finalize release processing with status updates.
      */
@@ -519,7 +573,7 @@ class ReleaseFileManager
             $candidate = $this->normalizeCandidateTitle($candidate);
 
             if ($this->isPlausibleReleaseTitle($candidate)) {
-                (new ReleaseUpdateService)->updateRelease(
+                $this->releaseUpdateService->updateRelease(
                     $context->release,
                     $candidate,
                     'RarInfo FileName Match',
@@ -545,7 +599,7 @@ class ReleaseFileManager
                 $candidate = $this->normalizeCandidateTitle($candidate);
 
                 if ($this->isPlausibleReleaseTitle($candidate)) {
-                    (new ReleaseUpdateService)->updateRelease(
+                    $this->releaseUpdateService->updateRelease(
                         $context->release,
                         $candidate,
                         'RarInfo FileName Match',
@@ -566,6 +620,12 @@ class ReleaseFileManager
     private function extractReleaseNameFromFile(string $filename): ?string
     {
         $basename = basename($filename);
+
+        $unwrapped = $this->fileNameCleaner->extractNzbSplitName($basename);
+        if ($unwrapped !== null) {
+            return $unwrapped;
+        }
+
         $cleaned = preg_replace(
             '/\.(mkv|avi|mp4|m4v|mpg|mpeg|wmv|flv|mov|ts|vob|iso|divx|par2?|nfo|sfv|nzb|rar|zip|r\d{2,3}|pkg|exe|msi)$/i',
             '',
@@ -588,13 +648,7 @@ class ReleaseFileManager
      */
     private function normalizeCandidateTitle(string $title): string
     {
-        $t = trim($title);
-        $t = preg_replace('/\.(mkv|avi|mp4|m4v|mpg|mpeg|wmv|flv|mov|ts|vob|iso|divx)$/i', '', $t) ?? $t;
-        $t = preg_replace('/\.(par2?|nfo|sfv|nzb|rar|zip|r\d{2,3}|pkg|exe|msi|jpe?g|png|gif|bmp)$/i', '', $t) ?? $t;
-        $t = preg_replace('/[.\-_ ](?:part|vol|r)\d+(?:\+\d+)?$/i', '', $t) ?? $t;
-        $t = preg_replace('/[\s_]+/', ' ', $t) ?? $t;
-
-        return trim($t, " .-_\t\r\n");
+        return $this->fileNameCleaner->normalizeCandidateTitle($title);
     }
 
     /**
@@ -602,30 +656,6 @@ class ReleaseFileManager
      */
     private function isPlausibleReleaseTitle(string $title): bool
     {
-        $t = trim($title);
-        if ($t === '' || strlen($t) < 12) {
-            return false;
-        }
-
-        $wordCount = preg_match_all('/[A-Za-z0-9]{3,}/', $t);
-        if ($wordCount < 2) {
-            return false;
-        }
-
-        if (preg_match('/(?:^|[.\-_ ])(?:part|vol|r)\d+(?:\+\d+)?$/i', $t)) {
-            return false;
-        }
-
-        if (preg_match('/^(setup|install|installer|patch|update|crack|keygen)\d*[\s._-]/i', $t)) {
-            return false;
-        }
-
-        $hasGroupSuffix = (bool) preg_match('/[-.][A-Za-z0-9]{2,}$/', $t);
-        $hasYear = (bool) preg_match('/\b(19|20)\d{2}\b/', $t);
-        $hasQuality = (bool) preg_match('/\b(480p|720p|1080p|2160p|4k|webrip|web[ .-]?dl|bluray|bdrip|dvdrip|hdtv|hdrip|xvid|x264|x265|hevc|h\.?264|ts|cam|r5|proper|repack)\b/i', $t);
-        $hasTV = (bool) preg_match('/\bS\d{1,2}[Eex]\d{1,3}\b/i', $t);
-        $hasXXX = (bool) preg_match('/\bXXX\b/i', $t);
-
-        return $hasGroupSuffix || ($hasTV && $hasQuality) || ($hasYear && ($hasQuality || $hasTV)) || $hasXXX;
+        return $this->fileNameCleaner->isPlausibleReleaseTitle($title);
     }
 }

app/Services/AdditionalProcessing/ReleaseProcessor.php

@@ -59,6 +59,7 @@ class ReleaseProcessor
             }
 
             $this->initializeContext($context);
+            $this->releaseManager->processReleaseNameFromNzbContents($context->nzbContents, $context);
             $messageIds = $this->nzbParser->extractMessageIDs($context->nzbContents, $context->releaseGroupName, $this->config);
 
             $context->nzbHasCompressedFile = $messageIds['hasCompressedFile'];

app/Services/Categorization/CategorizationPipeline.php

@@ -8,6 +8,7 @@ use App\Models\Settings;
 use App\Models\UsenetGroup;
 use App\Services\Categorization\Pipes\AbstractCategorizationPipe;
 use App\Services\Categorization\Pipes\CategorizationPassable;
+use App\Services\NameFixing\NzbSplitUnwrapper;
 use Illuminate\Pipeline\Pipeline;
 use Illuminate\Support\Collection;
 use Illuminate\Support\Facades\Log;
@@ -31,10 +32,12 @@ class CategorizationPipeline
 
     protected bool $catWebDL;
 
+    protected NzbSplitUnwrapper $nzbSplitUnwrapper;
+
     /**
      * @param  iterable<AbstractCategorizationPipe>  $pipes
      */
-    public function __construct(iterable $pipes = [])
+    public function __construct(iterable $pipes = [], ?NzbSplitUnwrapper $nzbSplitUnwrapper = null)
     {
         /** @phpstan-ignore argument.templateType */
         $this->pipes = collect($pipes)
@@ -42,6 +45,7 @@ class CategorizationPipeline
 
         $this->categorizeForeign = (bool) Settings::settingValue('categorizeforeign');
         $this->catWebDL = (bool) Settings::settingValue('catwebdl');
+        $this->nzbSplitUnwrapper = $nzbSplitUnwrapper ?? new NzbSplitUnwrapper;
     }
 
     /**
@@ -70,6 +74,8 @@ class CategorizationPipeline
         ?string $poster = '',
         bool $debug = false
     ): array {
+        $releaseName = $this->nzbSplitUnwrapper->unwrap($releaseName) ?? $releaseName;
+
         $groupName = UsenetGroup::whereId($groupId)->value('name') ?? '';
 
         $context = new ReleaseContext(

app/Services/Categorization/Categorizers/TvCategorizer.php

@@ -13,6 +13,10 @@ use App\Services\Categorization\ReleaseContext;
  */
 class TvCategorizer extends AbstractCategorizer
 {
+    private const string SPORTS_MARKER_REGEX = '/\b(NFL|NBA|NHL|MLB|MLS|EPL|UFC|WWE|Boxing|F1|Formula[._ -]?1|NASCAR|PGA|Tennis|Golf|Soccer|Football|Cricket|Rugby|Olympics?|Olympic[._ -]?Games?|Paralympics?)\b/i';
+
+    private const string SPORTS_EVENT_CONTEXT_REGEX = '/\d{4}|\b(Season|Week|Round|Match|Game|vs|Playoffs?|Finals?|Qualifying|Opening|Closing|Ceremony|Championship)\b/i';
+
     protected int $priority = 20;
 
     public function getName(): string
@@ -97,6 +101,10 @@ class TvCategorizer extends AbstractCategorizer
         if (preg_match('/[._ -](19|20)\d{2}[._ -]\d{2}[._ -]\d{2}[._ -]/i', $name)) {
             return true;
         }
+        // Sports events are TV-like even without SxxExx markers.
+        if ($this->isSportEvent($name)) {
+            return true;
+        }
         // Known anime release groups (should be treated as TV)
         if (preg_match('/(?:^|[.\-_ \[])(URANiME|ANiHLS|HaiKU|ANiURL|SkyAnime|Erai-raws|LostYears|Vodes|SubsPlease|Judas|Ember|EMBER|YuiSubs|ASW|Tsundere-Raws|Anime-Raws|Kekkan)(?:[.\-_ \]]|$)/i', $name)) {
             return true;
@@ -148,8 +156,7 @@ class TvCategorizer extends AbstractCategorizer
 
     protected function checkSport(string $name): ?CategorizationResult
     {
-        if (preg_match('/\b(NFL|NBA|NHL|MLB|MLS|UFC|WWE|Boxing|F1|Formula[._ -]?1|NASCAR|PGA|Tennis|Golf|Soccer|Football|Cricket|Rugby)\b/i', $name) &&
-            preg_match('/\d{4}|\b(Season|Week|Round|Match|Game)\b/i', $name)) {
+        if ($this->isSportEvent($name)) {
             return $this->matched(Category::TV_SPORT, 0.85, 'sport');
         }
 
@@ -250,4 +257,10 @@ class TvCategorizer extends AbstractCategorizer
 
         return null;
     }
+
+    protected function isSportEvent(string $name): bool
+    {
+        return preg_match(self::SPORTS_MARKER_REGEX, $name) === 1
+            && preg_match(self::SPORTS_EVENT_CONTEXT_REGEX, $name) === 1;
+    }
 }

app/Services/NameFixing/Extractors/FileNameExtractor.php

@@ -37,6 +37,16 @@ class FileNameExtractor
 
         // Try each pattern in order of specificity
 
+        if (str_contains($filename, '__NZBSPLIT__')) {
+            $unwrapped = $this->cleaner->extractNzbSplitName($filename);
+
+            if ($unwrapped !== null && $this->cleaner->isPlausibleReleaseTitle($unwrapped)) {
+                return NameFixResult::fromMatch($unwrapped, 'NZBSPLIT wrapper', 'File');
+            }
+
+            return null;
+        }
+
         // Scene TV release with group suffix
         if (preg_match('/^(.+?(x264|x265|HEVC|XviD|H\.?264|H\.?265)\-[A-Za-z0-9]+)\\\\/i', $filename, $result)) {
             return NameFixResult::fromMatch($result[1], 'Scene release with group', 'File');

app/Services/NameFixing/FileNameCleaner.php

@@ -16,6 +16,13 @@ class FileNameCleaner
 {
     use DetectsHashedNames;
 
+    protected NzbSplitUnwrapper $nzbSplitUnwrapper;
+
+    public function __construct(?NzbSplitUnwrapper $nzbSplitUnwrapper = null)
+    {
+        $this->nzbSplitUnwrapper = $nzbSplitUnwrapper ?? new NzbSplitUnwrapper;
+    }
+
     /**
      * Generic DVD structure filenames and similarly low-information names.
      *
@@ -92,6 +99,8 @@ class FileNameCleaner
         // Extract filename from path
         $fileName = $this->extractFilenameFromPath($fileName);
 
+        $fileName = $this->extractNzbSplitName($fileName) ?? $fileName;
+
         if ($this->isIgnorableForMatching($fileName)) {
             return false;
         }
@@ -153,6 +162,8 @@ class FileNameCleaner
     {
         $t = trim($title);
 
+        $t = $this->extractNzbSplitName($t) ?? $t;
+
         // Remove common video file extensions
         $t = preg_replace('/\.(mkv|avi|mp4|m4v|mpg|mpeg|wmv|flv|mov|ts|vob|iso|divx)$/i', '', $t) ?? $t;
 
@@ -231,6 +242,8 @@ class FileNameCleaner
      */
     public function cleanForTitleMatch(string $filename): string
     {
+        $filename = $this->extractNzbSplitName($filename) ?? $filename;
+
         // Remove file extension
         $clean = preg_replace('/\.(mkv|avi|mp4|m4v|wmv|mpg|mpeg|mov|ts|m2ts|vob|divx|flv|nfo|sfv|nzb|srr|srs|rar|r\d{2,4}|zip|7z|par2?|vol\d+[\+\-]\d+|\d{3})$/i', '', $filename);
 
@@ -271,6 +284,8 @@ class FileNameCleaner
      */
     public function looksLikeSceneRelease(string $filename): bool
     {
+        $filename = $this->extractNzbSplitName($filename) ?? $filename;
+
         $baseName = preg_replace('/\.[a-z0-9]{2,4}$/i', '', $filename);
 
         // Check for group suffix
@@ -313,6 +328,11 @@ class FileNameCleaner
         return false;
     }
 
+    public function extractNzbSplitName(string $value): ?string
+    {
+        return $this->nzbSplitUnwrapper->unwrap($value);
+    }
+
     /**
      * Ignore junk inputs that should never drive PreDB matches.
      */

app/Services/NameFixing/NameFixingService.php

@@ -1200,6 +1200,13 @@ class NameFixingService
                 break;
 
             case 'Filenames, ':
+                // Try direct file name extraction first (handles NZBSPLIT wrappers)
+                if (! $this->updateService->matched) {
+                    $result = $this->fileExtractor->extractFromFile($release->textstring);
+                    if ($result !== null) {
+                        $this->updateService->updateRelease($release, $result->newName, 'fileCheck: '.$result->method, $echo, $type, $nameStatus, $show);
+                    }
+                }
                 // Try PreDB file check
                 if (! $this->updateService->matched) {
                     $this->preDbFileCheck($release, $echo, $type, $nameStatus, $show);

app/Services/NameFixing/NzbSplitUnwrapper.php

@@ -0,0 +1,75 @@
+<?php
+
+declare(strict_types=1);
+
+namespace App\Services\NameFixing;
+
+final class NzbSplitUnwrapper
+{
+    private const string WRAPPER_PATTERN = '/__NZBSPLIT__([a-f0-9]{8,})__NZBSPLIT__(.+)$/i';
+
+    /**
+     * @var list<string>
+     */
+    private const array TRAILING_ARCHIVE_PATTERNS = [
+        '/\.part\d+\.rar$/i',
+        '/\.r\d{2,4}$/i',
+        '/\.7z\.\d{2,4}$/i',
+        '/\.7z$/i',
+        '/\.rar$/i',
+        '/\.par2?$/i',
+        '/\.vol\d+[+\-]\d+\.par2?$/i',
+        '/\.\d{3}$/',
+    ];
+
+    /**
+     * @var list<string>
+     */
+    private const array TRAILING_SUBJECT_PATTERNS = [
+        '/"\s*yEnc(?:\s*\([^)]*\))?$/i',
+        '/\s+yEnc(?:\s*\([^)]*\))?$/i',
+        '/"\s*yEnc.*$/i',
+        '/\s+yEnc.*$/i',
+    ];
+
+    public function unwrap(string $value): ?string
+    {
+        if (! preg_match(self::WRAPPER_PATTERN, $value, $matches)) {
+            return null;
+        }
+
+        $title = $matches[2];
+        $didStrip = true;
+
+        while ($didStrip) {
+            $didStrip = false;
+
+            foreach (self::TRAILING_SUBJECT_PATTERNS as $pattern) {
+                $updated = preg_replace($pattern, '', $title);
+
+                if ($updated !== null && $updated !== $title) {
+                    $title = rtrim($updated, " \t\n\r\0\x0B\"'");
+                    $didStrip = true;
+                }
+            }
+
+            foreach (self::TRAILING_ARCHIVE_PATTERNS as $pattern) {
+                $updated = preg_replace($pattern, '', $title);
+
+                if ($updated !== null && $updated !== $title) {
+                    $title = $updated;
+                    $didStrip = true;
+                }
+            }
+        }
+
+        $title = str_replace(['_', '+'], ['.', ' '], $title);
+        $title = preg_replace('/\.{2,}/', '.', $title) ?? $title;
+        $title = preg_replace('/\s{2,}/', ' ', $title) ?? $title;
+        $title = preg_replace('/\.\s+/', '.', $title) ?? $title;
+        $title = preg_replace('/\s+\./', '.', $title) ?? $title;
+        $title = trim($title, " \t\n\r\0\x0B.-_");
+
+        return $title !== '' ? $title : null;
+    }
+}

app/Services/ReleaseCleaningService.php

@@ -5,6 +5,7 @@ declare(strict_types=1);
 namespace App\Services;
 
 use App\Models\Predb;
+use App\Services\NameFixing\NzbSplitUnwrapper;
 
 /**
  * Cleans names for releases/imports/namefixer.
@@ -48,6 +49,8 @@ class ReleaseCleaningService
 
     protected RegexService $_regexes;
 
+    protected NzbSplitUnwrapper $nzbSplitUnwrapper;
+
     /**
      * ReleaseCleaningService constructor.
      *
@@ -60,6 +63,7 @@ class ReleaseCleaningService
         $this->e1 = CollectionsCleaningService::REGEX_FILE_EXTENSIONS.CollectionsCleaningService::REGEX_END;
         $this->e2 = CollectionsCleaningService::REGEX_FILE_EXTENSIONS.CollectionsCleaningService::REGEX_SUBJECT_SIZE.CollectionsCleaningService::REGEX_END;
         $this->_regexes = new RegexService('release_naming_regexes');
+        $this->nzbSplitUnwrapper = new NzbSplitUnwrapper;
     }
 
     /**
@@ -73,6 +77,19 @@ class ReleaseCleaningService
         $this->fromName = $fromName;
         $this->groupName = $groupName;
 
+        $unwrappedSubject = $this->nzbSplitUnwrapper->unwrap($subject);
+        if ($unwrappedSubject !== null) {
+            $this->subject = $unwrappedSubject;
+
+            return [
+                'cleansubject' => $unwrappedSubject,
+                'properlynamed' => true,
+                'increment' => false,
+                'predb' => 0,
+                'requestid' => false,
+            ];
+        }
+
         $hit = $hits = [];
         // Get pre style name from releases.name
         if (preg_match_all(
@@ -429,7 +446,7 @@ class ReleaseCleaningService
      */
     public function fixerCleaner(string $name)
     {
-        $cleanerName = $name;
+        $cleanerName = $this->nzbSplitUnwrapper->unwrap($name) ?? $name;
 
         // Remove sample/proof/thumbs markers from the end
         $cleanerName = preg_replace('/[.\-_](sample|proof|thumbs?)$/i', '', $cleanerName);