diff --git a/.env.dist b/.env.dist index 44c857ea4..983c80104 100644 --- a/.env.dist +++ b/.env.dist @@ -150,6 +150,7 @@ TVDB_APIKEY=${TVDB_APIKEY} TVDB_PIN=${TVDB_PIN} ANIDB_APIKEY=${ANIDB_APIKEY} FANARTTV_APIKEY=${FANARTTV_APIKEY} +GOOGLE_BOOKS_API_KEY=${GOOGLE_BOOKS_API_KEY} OMDB_APIKEY=${OMDB_APIKEY} TRAKTTV_APIKEY=${TRAKTTV_APIKEY} diff --git a/.env.example b/.env.example index d74ce60a1..fbc0331b7 100644 --- a/.env.example +++ b/.env.example @@ -197,6 +197,7 @@ TVDB_APIKEY=ef6fb572-bcee-4d99-9b52-90549ad7553a TVDB_PIN= ANIDB_APIKEY= FANARTTV_APIKEY= +GOOGLE_BOOKS_API_KEY= ISBNDB_API_KEY= OMDB_APIKEY= TRAKTTV_APIKEY= diff --git a/app/Http/Controllers/Admin/AdminSiteController.php b/app/Http/Controllers/Admin/AdminSiteController.php index daff0bf2a..854bfd1f7 100644 --- a/app/Http/Controllers/Admin/AdminSiteController.php +++ b/app/Http/Controllers/Admin/AdminSiteController.php @@ -5,7 +5,6 @@ declare(strict_types=1); namespace App\Http\Controllers\Admin; use App\Http\Controllers\BasePageController; -use App\Models\Category; use App\Models\GrabStat; use App\Models\ReleaseStat; use App\Models\RoleStat; @@ -33,9 +32,6 @@ class AdminSiteController extends BasePageController switch ($action) { case 'submit': - if ($request->missing('book_reqids')) { - $request->merge(['book_reqids' => []]); - } Settings::settingsUpdate($request->all()); return redirect()->to('admin/site-edit')->with('success', 'Settings updated successfully'); @@ -45,27 +41,6 @@ class AdminSiteController extends BasePageController break; } - // return a list of audiobooks, mags, ebooks, technical and foreign books - $result = Category::query()->whereIn('id', [Category::MUSIC_AUDIOBOOK, Category::BOOKS_MAGAZINES, Category::BOOKS_TECHNICAL, Category::BOOKS_FOREIGN])->get(['id', 'title']); - - // setup the display lists for these categories - $book_reqids_ids = []; - $book_reqids_names = []; - foreach ($result as $bookcategory) { - $book_reqids_ids[] = $bookcategory['id']; - $book_reqids_names[] = $bookcategory['title']; - } - - // convert from a string array to an int array as we want to use int - $book_reqids_ids = array_map(fn ($value) => (int) $value, $book_reqids_ids); - - // convert from a list to an array as we need to use an array, but the Settings table only saves strings - $bookReqidsValue = Settings::settingValue('book_reqids') ?? ''; - $books_selected = $bookReqidsValue !== '' ? explode(',', (string) $bookReqidsValue) : []; - - // convert from a string array to an int array, filtering out empty values - $books_selected = array_map(fn ($value) => (int) trim($value), array_filter($books_selected)); - $compress_headers_warning = ! str_contains(config('settings.nntp_server'), 'astra') ? 'compress_headers_warning' : ''; $this->viewData = array_merge($this->viewData, [ @@ -127,11 +102,6 @@ class AdminSiteController extends BasePageController 'ids' => [0, 1, 2], 'names' => ['Disabled', 'Lookup Request IDs', 'Lookup Request IDs Threaded'], ], - 'book_reqids' => [ - 'ids' => $book_reqids_ids, - 'names' => $book_reqids_names, - 'selected' => $books_selected, - ], 'compress_headers_warning' => $compress_headers_warning, 'title' => $title, 'meta_title' => $meta_title, diff --git a/app/Services/BookService.php b/app/Services/BookService.php index 83a9d81cb..59a6484c5 100644 --- a/app/Services/BookService.php +++ b/app/Services/BookService.php @@ -11,6 +11,8 @@ use App\Models\Category; use App\Models\Release; use App\Models\Settings; use App\Services\Releases\ReleaseBrowseService; +use App\Support\BookMatchScorer; +use App\Support\DTOs\BookParseResult; use App\Support\MetadataSearchLookup; use Illuminate\Database\Eloquent\Model; use Illuminate\Support\Facades\Cache; @@ -29,17 +31,12 @@ class BookService public string $imgSavePath; - public ?string $bookreqids; - public string $renamed; - /** - * @var array - */ - public array $failCache; - public ?string $parsedIsbn; + public ?BookParseResult $parsedBookResult; + /** * @throws \Exception */ @@ -51,11 +48,10 @@ class BookService $this->sleeptime = Settings::settingValue('amazonsleep') !== '' ? (int) Settings::settingValue('amazonsleep') : 1000; $this->imgSavePath = storage_path('covers/book/'); - $this->bookreqids = (string) Category::BOOKS_EBOOK; $this->renamed = (int) Settings::settingValue('lookupbooks') === 2 ? 'AND isrenamed = 1' : ''; - $this->failCache = []; $this->parsedIsbn = null; + $this->parsedBookResult = null; } /** @@ -73,8 +69,9 @@ class BookService /** * Get book info by name using full-text search. */ - public function getBookInfoByName(string $title): ?Model + public function getBookInfoByName(string $title, ?BookParseResult $parsed = null): ?Model { + $parsed ??= $this->parsedBookResult; $searchWords = ''; $title = preg_replace(['/( - | -|\(.+\)|\(|\))/', '/[^\w ]+/'], [' ', ''], $title); $title = trim(trim(preg_replace('/\s\s+/i', ' ', $title))); @@ -97,14 +94,9 @@ class BookService ->whereIn('id', $bookIds) ->get() ->keyBy('id'); - - foreach ($bookIds as $bookId) { - if ($rowsById->has($bookId)) { - /** @var BookInfo $book */ - $book = $rowsById->get($bookId); - - return $book; - } + $best = $this->pickBestExistingBook($rowsById->values()->all(), $parsed); + if ($best !== null) { + return $best; } } } @@ -112,9 +104,12 @@ class BookService return null; } - return BookInfo::query() + $results = BookInfo::query() ->whereRaw('MATCH (author, title) AGAINST (? IN BOOLEAN MODE)', [$searchWords]) - ->first(); + ->limit(25) + ->get(); + + return $this->pickBestExistingBook($results->all(), $parsed); } /** @@ -366,35 +361,24 @@ class BookService */ public function processBookReleases(string $groupID = '', string $guidChar = ''): void { - $bookids = []; - if (ctype_digit((string) $this->bookreqids)) { - $bookids[] = $this->bookreqids; - } else { - $bookids = explode(', ', $this->bookreqids); + $query = Release::query() + ->whereNull('bookinfo_id') + ->whereBetween('categories_id', [Category::BOOKS_ROOT, Category::BOOKS_UNKNOWN]) + ->orderByDesc('postdate') + ->limit($this->bookqty); + + if ($guidChar !== '') { + $query->where('leftguid', 'like', $guidChar.'%'); } - $total = \count($bookids); - if ($total > 0) { - foreach ($bookids as $i => $iValue) { - $query = Release::query() - ->whereNull('bookinfo_id') - ->whereIn('categories_id', [$iValue]) - ->orderByDesc('postdate') - ->limit($this->bookqty); - - if ($guidChar !== '') { - $query->where('leftguid', 'like', $guidChar.'%'); - } - - if ($groupID !== '') { - $query->where('groups_id', $groupID); - } - - $this->processBookReleasesHelper( - $query->get(['searchname', 'id', 'categories_id']), $iValue - ); - } + if ($groupID !== '') { + $query->where('groups_id', $groupID); } + + $this->processBookReleasesHelper( + $query->get(['searchname', 'id', 'categories_id']), + sprintf('%d-%d', Category::BOOKS_ROOT, Category::BOOKS_UNKNOWN) + ); } /** @@ -428,9 +412,10 @@ class BookService } // Do a local lookup first - $bookCheck = $this->getBookInfoByName($bookInfo); + $bookCheck = $this->getBookInfoByName($bookInfo, $this->parsedBookResult); + $failCacheKey = 'book_lookup_fail_'.md5(strtolower($bookInfo)); - if ($bookCheck === null && \in_array($bookInfo, $this->failCache, false)) { + if ($bookCheck === null && Cache::has($failCacheKey)) { // Lookup recently failed, no point trying again if ($this->echooutput) { cli()->info('Cached previous failure. Skipping.'); @@ -440,7 +425,9 @@ class BookService $bookId = $this->updateBookInfo($bookInfo, $this->parsedIsbn); $usedExternalApi = true; if ($bookId === -2) { - $this->failCache[] = $bookInfo; + Cache::put($failCacheKey, true, now()->addDays(7)); + } else { + Cache::forget($failCacheKey); } } else { $bookId = $bookCheck['id']; @@ -472,21 +459,14 @@ class BookService */ public function parseTitle(mixed $release_name, mixed $releaseID, mixed $releasetype) { - $this->parsedIsbn = $this->extractIsbn((string) $release_name); - - $a = preg_replace('/\d{1,2} \d{1,2} \d{2,4}|(19|20)\d\d|anybody got .+?[a-z]\? |[ ._-](Novel|TIA)([ ._-]|$)|([ \.])HQ([-\. ])|[\(\)\.\-_ ](AVI|AZW3?|DOC|EPUB|LIT|MOBI|NFO|RETAIL|(si)?PDF|RTF|TXT)[\)\]\.\-_ ](?![a-z0-9])|compleet|DAGSTiDNiNGEN|DiRFiX|\+ extra|r?e ?Books?([\.\-_ ]English|ers)?|azw3?|ePu([bp])s?|html|mobi|^NEW[\.\-_ ]|PDF([\.\-_ ]English)?|Please post more|Post description|Proper|Repack(fix)?|[\.\-_ ](Chinese|English|French|German|Italian|Retail|Scan|Swedish)|^R4 |Repost|Skytwohigh|TIA!+|TruePDF|V413HAV|(would someone )?please (re)?post.+? "|with the authors name right/i', '', $release_name); - $b = preg_replace('/^(As Req |conversion |eq |Das neue Abenteuer \d+|Fixed version( ignore previous post)?|Full |Per Req As Found|(\s+)?R4 |REQ |revised |version |\d+(\s+)?$)|(COMPLETE|INTERNAL|RELOADED| (AZW3|eB|docx|ENG?|exe|FR|Fix|gnv64|MU|NIV|R\d\s+\d{1,2} \d{1,2}|R\d|Req|TTL|UC|v(\s+)?\d))(\s+)?$/i', '', $a); - - // Remove book series from title to improve metadata matching. - $c = preg_replace('/ - \[.+\]|\[.+\]/', '', $b); - - // remove any brackets left behind - $d = preg_replace('/(\(\)|\[\])/', '', $c); - $releasename = trim(preg_replace('/\s\s+/i', ' ', $d)); + $parsed = $this->parseReleaseName((string) $release_name, (string) $releasetype); + $this->parsedBookResult = $parsed; + $this->parsedIsbn = $parsed->isbn; + $releasename = $parsed->title; // the default existing type was ebook, this handles that in the same manor as before if ($releasetype === 'ebook') { - if (preg_match('/^([a-z0-9] )+$|ArtofUsenet|ekiosk|(ebook|mobi).+collection|erotica|Full Video|ImwithJamie|linkoff org|Mega.+pack|^[a-z0-9]+ (?!((January|February|March|April|May|June|July|August|September|O([ck])tober|November|De([cz])ember)))[a-z]+( (ebooks?|The))?$|NY Times|(Book|Massive) Dump|Sexual/i', $releasename)) { + if ($parsed->isJunk) { if ($this->echooutput) { cli()->headerOver('Changing category to misc books: ').cli()->primary($releasename); } @@ -496,7 +476,7 @@ class BookService return false; } - if (preg_match('/^([a-z0-9ü!]+ ){1,2}(N|Vol)?\d{1,4}([abc])?$|^([a-z0-9]+ ){1,2}(Jan( |unar|$)|Feb( |ruary|$)|Mar( |ch|$)|Apr( |il|$)|May(?![a-z0-9])|Jun([ e$])|Jul([ y$])|Aug( |ust|$)|Sep( |tember|$)|O([ck])t( |ober|$)|Nov( |ember|$)|De([cz])( |ember|$))/ui', $releasename) && ! preg_match('/Part \d+/i', $releasename)) { + if ($parsed->isMagazine) { if ($this->echooutput) { cli()->headerOver('Changing category to magazines: ').cli()->primary($releasename); } @@ -506,7 +486,7 @@ class BookService return false; } if (! empty($releasename) && ! preg_match('/^[a-z0-9]+$|^([0-9]+ ){1,}$|Part \d+/i', $releasename)) { - return $releasename; + return $parsed->searchQuery(); } return false; @@ -515,7 +495,7 @@ class BookService if (! empty($releasename) && ! preg_match('/^[a-z0-9]+$|^([0-9]+ ){1,}$|Part \d+/i', $releasename)) { // We can skip category checks for audiobooks since the category is known. // As long as the release name is valid, it should still be post-processed. - return $releasename; + return $parsed->searchQuery(); } return false; @@ -524,6 +504,70 @@ class BookService return false; } + public function parseReleaseName(string $releaseName, string $releaseType = 'ebook'): BookParseResult + { + $isbn = $this->extractIsbn($releaseName); + $a = preg_replace('/\d{1,2} \d{1,2} \d{2,4}|(19|20)\d\d|anybody got .+?[a-z]\? |[ ._-](Novel|TIA)([ ._-]|$)|([ \.])HQ([-\. ])|[\(\)\.\-_ ](AVI|AZW3?|DOC|EPUB|LIT|MOBI|NFO|RETAIL|(si)?PDF|RTF|TXT)[\)\]\.\-_ ](?![a-z0-9])|compleet|DAGSTiDNiNGEN|DiRFiX|\+ extra|r?e ?Books?([\.\-_ ]English|ers)?|azw3?|ePu([bp])s?|html|mobi|^NEW[\.\-_ ]|PDF([\.\-_ ]English)?|Please post more|Post description|Proper|Repack(fix)?|[\.\-_ ](Chinese|English|French|German|Italian|Retail|Scan|Swedish|Multilingual)|^R4 |Repost|Skytwohigh|TIA!+|TruePDF|V413HAV|(would someone )?please (re)?post.+? "|with the authors name right/i', '', $releaseName); + $b = preg_replace('/^(As Req |conversion |eq |Das neue Abenteuer \d+|Fixed version( ignore previous post)?|Full |Per Req As Found|(\s+)?R4 |REQ |revised |version |\d+(\s+)?$)|(COMPLETE|INTERNAL|RELOADED| (AZW3|eB|docx|ENG?|exe|FR|Fix|gnv64|MU|NIV|R\d\s+\d{1,2} \d{1,2}|R\d|Req|TTL|UC|v(\s+)?\d))(\s+)?$/i', '', (string) $a); + $c = preg_replace('/ - \[.+\]|\[.+\]/', '', (string) $b); + $d = preg_replace('/(\(\)|\[\])/', '', (string) $c); + $normalized = trim((string) preg_replace('/\s\s+/i', ' ', (string) $d)); + $normalized = (string) preg_replace('/[._]+/', ' ', $normalized); + $normalized = (string) preg_replace('/\b97[89](?:[-\s]?\d){10}\b|\b\d(?:[-\s]?\d){8}[-\s]?[\dXx]\b/', '', $normalized); + $normalized = trim((string) preg_replace('/\s+/', ' ', $normalized)); + + $year = null; + if (preg_match('/\b(19|20)\d{2}\b/', $normalized, $yearMatch) === 1) { + $year = (int) $yearMatch[0]; + } + + $format = null; + if (preg_match('/\b(EPUB|MOBI|AZW3?|PDF|FB2|DJVU|LIT)\b/i', $releaseName, $formatMatch) === 1) { + $format = strtoupper($formatMatch[1]); + } + + $author = null; + $title = $normalized; + if (preg_match('/^(?[A-Za-z0-9&\'\.\-\s]{3,80})\s-\s(?.+)$/', $normalized, $matches) === 1) { + $author = trim($matches['author']); + $title = trim($matches['title']); + } elseif (preg_match('/^(?<author>[A-Za-z0-9&\'\.\-\s]{3,80})\s+by\s+(?<title>.+)$/i', $normalized, $matches) === 1) { + $author = trim($matches['author']); + $title = trim($matches['title']); + } elseif (preg_match('/^(?<title>.+?)\s+by\s+(?<author>[A-Za-z0-9&\'\.\-\s]{3,80})$/i', $normalized, $matches) === 1) { + $author = trim($matches['author']); + $title = trim($matches['title']); + } + + $title = trim((string) preg_replace('/\s+/', ' ', (string) preg_replace('/\((Book|Series)\s*\d+\)/i', '', $title))); + + $isJunk = false; + $junkPattern = '/^([a-z0-9] )+$|ArtofUsenet|ekiosk|(ebook|mobi).+collection|erotica|Full Video|ImwithJamie|linkoff org|Mega.+pack|^[a-z0-9]+ (?!((January|February|March|April|May|June|July|August|September|O([ck])tober|November|De([cz])ember)))[a-z]+( (ebooks?|The))?$|NY Times|(Book|Massive) Dump|Sexual/i'; + if (preg_match($junkPattern, $normalized) === 1) { + $isJunk = true; + } + if ($releaseType === 'ebook' && preg_match('/\b(v?\d+\.\d+\.\d+(?:\.\d+)?|x64|x86|portable|setup|crack(ed)?|patch)\b/i', $releaseName) === 1) { + $isJunk = true; + } + if ($releaseType === 'ebook' && preg_match('/\b(WEB[\.\-_ ]?FLAC|MP3|320kbps|FALCON|discography)\b/i', $releaseName) === 1) { + $isJunk = true; + } + + $isMagazine = preg_match('/^([a-z0-9ü!]+ ){1,2}(N|Vol)?\d{1,4}([abc])?$|^([a-z0-9]+ ){1,2}(Jan( |unar|$)|Feb( |ruary|$)|Mar( |ch|$)|Apr( |il|$)|May(?![a-z0-9])|Jun([ e$])|Jul([ y$])|Aug( |ust|$)|Sep( |tember|$)|O([ck])t( |ober|$)|Nov( |ember|$)|De([cz])( |ember|$))/ui', $normalized) === 1 + && preg_match('/Part \d+/i', $normalized) !== 1; + + return new BookParseResult( + rawName: $releaseName, + title: $title, + author: $author, + isbn: $isbn, + year: $year, + format: $format, + isJunk: $isJunk, + isMagazine: $isMagazine, + ); + } + public function extractIsbn(string $releaseName): ?string { if (preg_match('/\b(97[89](?:[-\s]?\d){10})\b/', $releaseName, $matches) === 1) { @@ -554,27 +598,63 @@ class BookService { $ri = new ReleaseImageService; $isbnDb = new IsbnDbService; + $googleBooks = new GoogleBooksService; + $openLibrary = new OpenLibraryService; + $itunes = new ItunesService; + $scorer = new BookMatchScorer; $bookId = -2; - - $book = false; + $book = null; if ($bookInfo !== '') { + $parsed = $this->parsedBookResult ?? $this->parseReleaseName($bookInfo); + $candidates = []; if ($isbnDb->isConfigured()) { if ($isbn !== null && $isbn !== '') { cli()->info('Fetching data from ISBNdb by ISBN '.$isbn); - $book = $isbnDb->findByIsbn($isbn); + $candidate = $isbnDb->findByIsbn($isbn); + if (is_array($candidate)) { + $candidates[] = $candidate; + } } - - if ($book === false || $book === null) { + if ($candidates === []) { cli()->info('Fetching data from ISBNdb for '.$bookInfo); - $book = $isbnDb->searchBook($bookInfo); + $candidates = array_merge($candidates, $isbnDb->searchBooks($bookInfo)); } } - if ($book === false || $book === null) { - cli()->info('Fetching data from iTunes for '.$bookInfo); - $book = $this->fetchItunesBookProperties($bookInfo); + if ($candidates === []) { + cli()->info('Fetching data from Google Books for '.$bookInfo); + $candidates = array_merge( + $candidates, + $googleBooks->searchBooks( + $bookInfo, + $parsed->title, + $parsed->author, + $isbn + ) + ); } + + if ($candidates === []) { + cli()->info('Fetching data from iTunes for '.$bookInfo); + $candidates = array_merge($candidates, $this->fetchItunesBookProperties($bookInfo, $itunes)); + } + + if ($candidates === []) { + cli()->info('Fetching data from Open Library for '.$bookInfo); + if ($isbn !== null && $isbn !== '') { + $candidate = $openLibrary->findByIsbn($isbn); + if (is_array($candidate)) { + $candidates[] = $candidate; + } + } + + if ($candidates === []) { + $candidates = array_merge($candidates, $openLibrary->searchBooks($bookInfo)); + } + } + + $book = $this->pickBestCandidate($candidates, $parsed, $scorer); } if (empty($book)) { @@ -652,43 +732,94 @@ class BookService /** * Fetch book properties from iTunes. * - * @return array<string, mixed>|bool + * @return array<int, array<string, mixed>> */ - public function fetchItunesBookProperties(string $bookInfo) + public function fetchItunesBookProperties(string $bookInfo, ?ItunesService $itunesService = null): array { - $itunes = new ItunesService; - $iTunesBook = $itunes->findEbook($bookInfo); + $itunes = $itunesService ?? new ItunesService; + $iTunesBooks = $itunes->findEbooks($bookInfo); - if ($iTunesBook === null) { + if ($iTunesBooks === []) { cli()->notice('Could not find a match on iTunes!'); - return false; + return []; } - cli()->info('Found matching title: '.$iTunesBook['name']); - - $book = [ - 'title' => $iTunesBook['name'], - 'author' => $iTunesBook['author'], - 'asin' => $iTunesBook['id'], - 'isbn' => null, - 'ean' => null, - 'url' => $iTunesBook['store_url'], - 'salesrank' => '', - 'publisher' => '', - 'pages' => '', - 'coverurl' => ! empty($iTunesBook['cover']) ? $iTunesBook['cover'] : '', - 'genre' => is_array($iTunesBook['genres']) ? implode(', ', $iTunesBook['genres']) : $iTunesBook['genre'], - 'overview' => strip_tags($iTunesBook['description'] ?? ''), - 'publishdate' => $iTunesBook['release_date'], - ]; - - if (! empty($book['coverurl'])) { - $book['cover'] = 1; - } else { - $book['cover'] = 0; + $books = []; + foreach ($iTunesBooks as $iTunesBook) { + $books[] = [ + 'title' => $iTunesBook['name'], + 'author' => $iTunesBook['author'], + 'asin' => (string) $iTunesBook['id'], + 'isbn' => null, + 'ean' => null, + 'url' => $iTunesBook['store_url'], + 'salesrank' => '', + 'publisher' => '', + 'pages' => '', + 'coverurl' => ! empty($iTunesBook['cover']) ? $iTunesBook['cover'] : '', + 'genre' => is_array($iTunesBook['genres']) ? implode(', ', $iTunesBook['genres']) : $iTunesBook['genre'], + 'overview' => strip_tags($iTunesBook['description'] ?? ''), + 'publishdate' => $iTunesBook['release_date'], + 'cover' => ! empty($iTunesBook['cover']) ? 1 : 0, + ]; } - return $book; + cli()->info('Found '.\count($books).' matching title(s) on iTunes'); + + return $books; + } + + /** + * @param array<int, BookInfo> $books + */ + private function pickBestExistingBook(array $books, ?BookParseResult $parsed): ?BookInfo + { + if ($books === []) { + return null; + } + if ($parsed === null) { + return $books[0]; + } + + $scorer = new BookMatchScorer; + $best = null; + $bestScore = 0.0; + foreach ($books as $book) { + $score = $scorer->scoreBookInfo($book, $parsed); + if ($score > $bestScore) { + $best = $book; + $bestScore = $score; + } + } + + return $bestScore >= 0.4 ? $best : null; + } + + /** + * @param array<int, array<string, mixed>> $candidates + * @return array<string, mixed>|null + */ + private function pickBestCandidate(array $candidates, BookParseResult $parsed, BookMatchScorer $scorer): ?array + { + if ($candidates === []) { + return null; + } + + $best = null; + $bestScore = 0.0; + foreach ($candidates as $candidate) { + $score = $scorer->score($candidate, $parsed); + if ($score > $bestScore) { + $bestScore = $score; + $best = $candidate; + } + } + + if ($best === null || $bestScore < 0.4) { + return null; + } + + return $best; } } diff --git a/app/Services/Categorization/Categorizers/BookCategorizer.php b/app/Services/Categorization/Categorizers/BookCategorizer.php index 92b8180cb..033ebf9c8 100644 --- a/app/Services/Categorization/Categorizers/BookCategorizer.php +++ b/app/Services/Categorization/Categorizers/BookCategorizer.php @@ -28,6 +28,14 @@ class BookCategorizer extends AbstractCategorizer if (preg_match('/\b(?:PS[1-5]|PlayStation|Xbox|Switch|Nintendo|Wii|3DS|GameCube)\b/i', $context->releaseName)) { return true; } + // Skip software and utility release naming patterns. + if (preg_match('/\bv?\d+\.\d+\.\d+(?:\.\d+)?\b.*\b(?:Multilingual|x64|x86|Portable|Setup|Patch)\b/i', $context->releaseName)) { + return true; + } + // Skip music-style release names that are often misfiled as books. + if (preg_match('/\b(?:WEB[\.\-_ ]?FLAC|MP3|320kbps|discography|FALCON)\b/i', $context->releaseName)) { + return true; + } // Skip TV shows (season patterns) if (preg_match('/[._ -]S\d{1,3}[._ -]?(E\d|Complete|Full|1080|720|480|2160|WEB|HDTV|BluRay)/i', $context->releaseName)) { return true; diff --git a/app/Services/GoogleBooksService.php b/app/Services/GoogleBooksService.php new file mode 100644 index 000000000..e24a5e208 --- /dev/null +++ b/app/Services/GoogleBooksService.php @@ -0,0 +1,224 @@ +<?php + +declare(strict_types=1); + +namespace App\Services; + +use Carbon\Carbon; +use GuzzleHttp\Client; +use GuzzleHttp\Exception\GuzzleException; +use Illuminate\Support\Facades\Cache; +use Illuminate\Support\Facades\Log; + +class GoogleBooksService +{ + protected const API_URL = 'https://www.googleapis.com/books/v1/volumes'; + + protected Client $client; + + protected ?string $apiKey; + + protected int $maxResults = 5; + + public function __construct(?Client $client = null, ?string $apiKey = null) + { + $this->apiKey = trim($apiKey ?? (string) config('nntmux_api.google_books_api_key', '')) ?: null; + + $this->client = $client ?? new Client([ + 'timeout' => 15, + 'connect_timeout' => 10, + 'http_errors' => false, + 'headers' => [ + 'Accept' => 'application/json', + 'User-Agent' => 'NNTmux/1.0', + ], + ]); + } + + public function hasApiKey(): bool + { + return $this->apiKey !== null; + } + + public function isConfigured(): bool + { + return true; + } + + /** + * @return array<string, mixed>|null + */ + public function findByIsbn(string $isbn): ?array + { + $books = $this->searchBooks('', null, null, $isbn); + + return $books[0] ?? null; + } + + /** + * @return array<string, mixed>|null + */ + public function searchBook(string $query, ?string $title = null, ?string $author = null): ?array + { + $books = $this->searchBooks($query, $title, $author); + + return $books[0] ?? null; + } + + /** + * @return array<int, array<string, mixed>> + */ + public function searchBooks(string $query, ?string $title = null, ?string $author = null, ?string $isbn = null): array + { + $query = trim($query); + $title = $title !== null ? trim($title) : null; + $author = $author !== null ? trim($author) : null; + $isbn = $isbn !== null ? strtoupper((string) preg_replace('/[^0-9X]/i', '', $isbn)) : null; + + $searchQuery = $this->buildSearchQuery($query, $title, $author, $isbn); + if ($searchQuery === '') { + return []; + } + + $cacheKey = 'google_books_search_'.md5($searchQuery.'_'.$this->maxResults.'_'.$this->apiKey); + $cached = Cache::get($cacheKey); + if (is_array($cached)) { + return $cached; + } + + $response = $this->request([ + 'q' => $searchQuery, + 'maxResults' => $this->maxResults, + 'printType' => 'books', + ]); + if (! is_array($response['items'] ?? null)) { + return []; + } + + $books = []; + foreach ($response['items'] as $item) { + if (is_array($item)) { + $books[] = $this->normalizeBookResult($item); + } + } + + Cache::put($cacheKey, $books, now()->addHours(12)); + + return $books; + } + + /** + * @param array<string, mixed> $query + * @return array<string, mixed>|null + */ + protected function request(array $query): ?array + { + try { + if ($this->apiKey !== null) { + $query['key'] = $this->apiKey; + } + + $response = $this->client->get(self::API_URL, ['query' => $query]); + if ($response->getStatusCode() !== 200) { + Log::warning('Google Books API request failed', ['status' => $response->getStatusCode()]); + + return null; + } + + $decoded = json_decode($response->getBody()->getContents(), true); + + return is_array($decoded) ? $decoded : null; + } catch (GuzzleException $e) { + Log::error('Google Books API request error: '.$e->getMessage()); + + return null; + } + } + + private function buildSearchQuery(string $query, ?string $title, ?string $author, ?string $isbn): string + { + if ($isbn !== null && $isbn !== '') { + return 'isbn:'.$isbn; + } + + $parts = []; + if ($title !== null && $title !== '') { + $parts[] = 'intitle:'.$title; + } + if ($author !== null && $author !== '') { + $parts[] = 'inauthor:'.$author; + } + + if ($parts !== []) { + return implode(' ', $parts); + } + + return $query; + } + + /** + * @param array<string, mixed> $item + * @return array<string, mixed> + */ + private function normalizeBookResult(array $item): array + { + $volumeInfo = is_array($item['volumeInfo'] ?? null) ? $item['volumeInfo'] : []; + $industryIdentifiers = is_array($volumeInfo['industryIdentifiers'] ?? null) ? $volumeInfo['industryIdentifiers'] : []; + + $isbn13 = ''; + $isbn10 = ''; + foreach ($industryIdentifiers as $identifier) { + if (! is_array($identifier)) { + continue; + } + $type = (string) ($identifier['type'] ?? ''); + $value = strtoupper((string) preg_replace('/[^0-9X]/i', '', (string) ($identifier['identifier'] ?? ''))); + if ($type === 'ISBN_13' && $value !== '') { + $isbn13 = $value; + } + if ($type === 'ISBN_10' && $value !== '') { + $isbn10 = $value; + } + } + + $identifier = $isbn13 !== '' ? $isbn13 : ($isbn10 !== '' ? $isbn10 : (string) ($item['id'] ?? md5(json_encode($item) ?: ''))); + $authors = is_array($volumeInfo['authors'] ?? null) + ? array_values(array_filter(array_map('strval', $volumeInfo['authors']))) + : []; + $categories = is_array($volumeInfo['categories'] ?? null) + ? array_values(array_filter(array_map('strval', $volumeInfo['categories']))) + : []; + $imageLinks = is_array($volumeInfo['imageLinks'] ?? null) ? $volumeInfo['imageLinks'] : []; + $coverUrl = (string) ($imageLinks['thumbnail'] ?? $imageLinks['smallThumbnail'] ?? ''); + + return [ + 'title' => (string) ($volumeInfo['title'] ?? ''), + 'author' => implode(', ', $authors), + 'asin' => 'googlebooks:'.$identifier, + 'isbn' => $isbn13 !== '' ? $isbn13 : null, + 'ean' => $isbn10 !== '' ? $isbn10 : null, + 'url' => (string) ($volumeInfo['infoLink'] ?? ''), + 'salesrank' => '', + 'publisher' => (string) ($volumeInfo['publisher'] ?? ''), + 'publishdate' => $this->normalizePublishDate($volumeInfo['publishedDate'] ?? null), + 'pages' => isset($volumeInfo['pageCount']) && is_numeric($volumeInfo['pageCount']) ? (string) ((int) $volumeInfo['pageCount']) : '', + 'overview' => strip_tags((string) ($volumeInfo['description'] ?? '')), + 'genre' => implode(', ', $categories), + 'coverurl' => $coverUrl, + 'cover' => $coverUrl !== '' ? 1 : 0, + ]; + } + + private function normalizePublishDate(mixed $date): ?string + { + if (! is_string($date) || trim($date) === '') { + return null; + } + + try { + return Carbon::parse($date)->format('Y-m-d'); + } catch (\Exception) { + return null; + } + } +} diff --git a/app/Services/IsbnDbService.php b/app/Services/IsbnDbService.php index 09305cedb..51210076b 100644 --- a/app/Services/IsbnDbService.php +++ b/app/Services/IsbnDbService.php @@ -46,10 +46,20 @@ class IsbnDbService * @return array<string, mixed>|null */ public function searchBook(string $query): ?array + { + $books = $this->searchBooks($query); + + return $books[0] ?? null; + } + + /** + * @return array<int, array<string, mixed>> + */ + public function searchBooks(string $query): array { $query = trim($query); if ($query === '' || ! $this->isConfigured()) { - return null; + return []; } $cacheKey = 'isbndb_search_'.md5($query.'_'.$this->pageSize); @@ -64,14 +74,20 @@ class IsbnDbService ]); $books = $response['data'] ?? null; - if (! is_array($books) || $books === [] || ! isset($books[0]) || ! is_array($books[0])) { - return null; + if (! is_array($books) || $books === []) { + return []; } - $book = $this->normalizeBookResult($books[0]); - Cache::put($cacheKey, $book, now()->addHours(24)); + $normalized = []; + foreach ($books as $book) { + if (is_array($book)) { + $normalized[] = $this->normalizeBookResult($book); + } + } - return $book; + Cache::put($cacheKey, $normalized, now()->addHours(24)); + + return $normalized; } /** diff --git a/app/Services/ItunesService.php b/app/Services/ItunesService.php index 29b5fb85c..926735b25 100644 --- a/app/Services/ItunesService.php +++ b/app/Services/ItunesService.php @@ -206,13 +206,31 @@ class ItunesService */ public function findEbook(string $term): ?array { - $results = $this->searchEbooks($term); + $results = $this->findEbooks($term); + return $results[0] ?? null; + } + + /** + * Get normalized ebook results for a search term. + * + * @return array<int, array<string, mixed>> + */ + public function findEbooks(string $term): array + { + $results = $this->searchEbooks($term); if (empty($results)) { - return null; + return []; } - return $this->normalizeEbookResult($results[0]); // @phpstan-ignore offsetAccess.notFound + $normalized = []; + foreach ($results as $result) { + if (is_array($result)) { + $normalized[] = $this->normalizeEbookResult($result); + } + } + + return $normalized; } /** diff --git a/app/Services/OpenLibraryService.php b/app/Services/OpenLibraryService.php new file mode 100644 index 000000000..aac38e4d0 --- /dev/null +++ b/app/Services/OpenLibraryService.php @@ -0,0 +1,234 @@ +<?php + +declare(strict_types=1); + +namespace App\Services; + +use Carbon\Carbon; +use GuzzleHttp\Client; +use GuzzleHttp\Exception\GuzzleException; +use Illuminate\Support\Facades\Cache; +use Illuminate\Support\Facades\Log; + +class OpenLibraryService +{ + protected const SEARCH_URL = 'https://openlibrary.org/search.json'; + + protected const ISBN_URL = 'https://openlibrary.org/isbn/'; + + protected Client $client; + + protected int $limit = 5; + + public function __construct(?Client $client = null) + { + $this->client = $client ?? new Client([ + 'timeout' => 15, + 'connect_timeout' => 10, + 'http_errors' => false, + 'headers' => [ + 'Accept' => 'application/json', + 'User-Agent' => 'NNTmux/1.0', + ], + ]); + } + + /** + * @return array<string, mixed>|null + */ + public function findByIsbn(string $isbn): ?array + { + $isbn = strtoupper((string) preg_replace('/[^0-9X]/i', '', $isbn)); + if ($isbn === '') { + return null; + } + + $cacheKey = 'openlibrary_isbn_'.$isbn; + $cached = Cache::get($cacheKey); + if (is_array($cached)) { + return $cached; + } + + try { + $response = $this->client->get(self::ISBN_URL.rawurlencode($isbn).'.json'); + if ($response->getStatusCode() !== 200) { + return null; + } + + $data = json_decode($response->getBody()->getContents(), true); + if (! is_array($data)) { + return null; + } + + $book = $this->normalizeIsbnResult($data, $isbn); + Cache::put($cacheKey, $book, now()->addHours(24)); + + return $book; + } catch (GuzzleException $e) { + Log::error('OpenLibrary ISBN lookup error: '.$e->getMessage()); + + return null; + } + } + + /** + * @return array<string, mixed>|null + */ + public function searchBook(string $query): ?array + { + $results = $this->searchBooks($query); + + return $results[0] ?? null; + } + + /** + * @return array<int, array<string, mixed>> + */ + public function searchBooks(string $query): array + { + $query = trim($query); + if ($query === '') { + return []; + } + + $cacheKey = 'openlibrary_search_'.md5($query.'_'.$this->limit); + $cached = Cache::get($cacheKey); + if (is_array($cached)) { + return $cached; + } + + try { + $response = $this->client->get(self::SEARCH_URL, [ + 'query' => [ + 'q' => $query, + 'limit' => $this->limit, + ], + ]); + if ($response->getStatusCode() !== 200) { + return []; + } + + $data = json_decode($response->getBody()->getContents(), true); + if (! is_array($data['docs'] ?? null)) { + return []; + } + + $results = []; + foreach ($data['docs'] as $doc) { + if (is_array($doc)) { + $results[] = $this->normalizeSearchResult($doc); + } + } + + Cache::put($cacheKey, $results, now()->addHours(12)); + + return $results; + } catch (GuzzleException $e) { + Log::error('OpenLibrary search error: '.$e->getMessage()); + + return []; + } + } + + /** + * @param array<string, mixed> $book + * @return array<string, mixed> + */ + private function normalizeIsbnResult(array $book, string $isbn): array + { + $title = (string) ($book['title'] ?? ''); + $authors = []; + if (is_array($book['authors'] ?? null)) { + foreach ($book['authors'] as $author) { + if (is_array($author) && isset($author['name'])) { + $authors[] = (string) $author['name']; + } + } + } + + $publishDate = $this->normalizePublishDate($book['publish_date'] ?? null); + $coverUrl = ''; + if (is_array($book['covers'] ?? null) && isset($book['covers'][0])) { + $coverUrl = 'https://covers.openlibrary.org/b/id/'.(int) $book['covers'][0].'-L.jpg'; + } + + return [ + 'title' => $title, + 'author' => implode(', ', $authors), + 'asin' => 'openlibrary:'.$isbn, + 'isbn' => strlen($isbn) === 13 ? $isbn : null, + 'ean' => strlen($isbn) === 10 ? $isbn : null, + 'url' => (string) ($book['url'] ?? ''), + 'salesrank' => '', + 'publisher' => is_array($book['publishers'] ?? null) ? (string) ($book['publishers'][0] ?? '') : '', + 'publishdate' => $publishDate, + 'pages' => isset($book['number_of_pages']) && is_numeric($book['number_of_pages']) ? (string) ((int) $book['number_of_pages']) : '', + 'overview' => '', + 'genre' => '', + 'coverurl' => $coverUrl, + 'cover' => $coverUrl !== '' ? 1 : 0, + ]; + } + + /** + * @param array<string, mixed> $doc + * @return array<string, mixed> + */ + private function normalizeSearchResult(array $doc): array + { + $isbn13 = null; + $isbn10 = null; + if (is_array($doc['isbn'] ?? null)) { + foreach ($doc['isbn'] as $isbn) { + $normalized = strtoupper((string) preg_replace('/[^0-9X]/i', '', (string) $isbn)); + if ($normalized === '') { + continue; + } + if (strlen($normalized) === 13 && $isbn13 === null) { + $isbn13 = $normalized; + } elseif (strlen($normalized) === 10 && $isbn10 === null) { + $isbn10 = $normalized; + } + } + } + + $identifier = $isbn13 ?? $isbn10 ?? (string) ($doc['key'] ?? md5((string) json_encode($doc))); + $authorNames = is_array($doc['author_name'] ?? null) + ? array_values(array_filter(array_map('strval', $doc['author_name']))) + : []; + + $coverUrl = isset($doc['cover_i']) && is_numeric($doc['cover_i']) + ? 'https://covers.openlibrary.org/b/id/'.(int) $doc['cover_i'].'-L.jpg' + : ''; + + return [ + 'title' => (string) ($doc['title'] ?? ''), + 'author' => implode(', ', $authorNames), + 'asin' => 'openlibrary:'.$identifier, + 'isbn' => $isbn13, + 'ean' => $isbn10, + 'url' => isset($doc['key']) ? 'https://openlibrary.org'.$doc['key'] : '', + 'salesrank' => '', + 'publisher' => is_array($doc['publisher'] ?? null) ? (string) ($doc['publisher'][0] ?? '') : '', + 'publishdate' => isset($doc['first_publish_year']) ? ((int) $doc['first_publish_year']).'-01-01' : null, + 'pages' => '', + 'overview' => '', + 'genre' => '', + 'coverurl' => $coverUrl, + 'cover' => $coverUrl !== '' ? 1 : 0, + ]; + } + + private function normalizePublishDate(mixed $date): ?string + { + if (! is_string($date) || trim($date) === '') { + return null; + } + + try { + return Carbon::parse($date)->format('Y-m-d'); + } catch (\Exception) { + return null; + } + } +} diff --git a/app/Services/Tmux/Tmux.php b/app/Services/Tmux/Tmux.php index 9bd1ffbe0..f7b433d2e 100644 --- a/app/Services/Tmux/Tmux.php +++ b/app/Services/Tmux/Tmux.php @@ -294,7 +294,7 @@ class Tmux /** * @throws \Exception */ - public function proc_query(mixed $qry, mixed $bookreqids, string $db_name, ?string $ppmax = '', ?string $ppmin = ''): bool|string + public function proc_query(mixed $qry, string $db_name, ?string $ppmax = '', ?string $ppmin = ''): bool|string { // Convert null to empty string for backward compatibility $ppmax = $ppmax ?? ''; @@ -319,7 +319,7 @@ class Tmux SUM(IF(categories_id BETWEEN %d AND %d AND '.$movieLookupSql.',1,0)) AS processmovies, SUM(IF(categories_id IN (%d, %d, %d) AND musicinfo_id IS NULL,1,0)) AS processmusic, SUM(IF(categories_id BETWEEN %d AND %d AND consoleinfo_id IS NULL,1,0)) AS processconsole, - SUM(IF(categories_id IN (%s) AND bookinfo_id IS NULL,1,0)) AS processbooks, + SUM(IF(categories_id BETWEEN %d AND %d AND bookinfo_id IS NULL,1,0)) AS processbooks, SUM(IF(categories_id = %d AND gamesinfo_id = 0,1,0)) AS processgames, SUM(IF(1=1 %s,1,0)) AS processnfo, SUM(IF(isrenamed = %d AND predb_id = 0 AND passwordstatus >= 0 AND nfostatus > %d @@ -340,7 +340,8 @@ class Tmux Category::MUSIC_OTHER, Category::GAME_ROOT, Category::GAME_OTHER, - $bookreqids, + Category::BOOKS_ROOT, + Category::BOOKS_UNKNOWN, Category::PC_GAMES, NfoService::NfoQueryString(), NameFixingService::IS_RENAMED_NONE, diff --git a/app/Services/Tmux/TmuxMonitorService.php b/app/Services/Tmux/TmuxMonitorService.php index 6334c33b8..e2623370b 100644 --- a/app/Services/Tmux/TmuxMonitorService.php +++ b/app/Services/Tmux/TmuxMonitorService.php @@ -215,10 +215,9 @@ class TmuxMonitorService $timer = time(); try { - $bookReqIds = $this->runVar['settings']['book_reqids'] ?? Category::BOOKS_ROOT; $dbName = config('nntmux.db_name'); - $proc1Query = $this->tmux->proc_query(1, $bookReqIds, $dbName); + $proc1Query = $this->tmux->proc_query(1, $dbName, ''); $proc1Result = DB::selectOne($proc1Query); if ($proc1Result) { @@ -234,7 +233,7 @@ class TmuxMonitorService $maxSize = $this->runVar['settings']['maxsize_pp'] ?? ''; $minSize = $this->runVar['settings']['minsize_pp'] ?? ''; - $proc2Query = $this->tmux->proc_query(2, $bookReqIds, $dbName, (string) $maxSize, (string) $minSize); + $proc2Query = $this->tmux->proc_query(2, $dbName, (string) $maxSize, (string) $minSize); $proc2Result = DB::selectOne($proc2Query); if ($proc2Result) { @@ -285,8 +284,7 @@ class TmuxMonitorService // Get additional table counts (query 4) $timer4 = time(); - $bookReqIds = $this->runVar['settings']['book_reqids'] ?? Category::BOOKS_ROOT; - $proc4Query = $this->tmux->proc_query(4, $bookReqIds, $dbName); + $proc4Query = $this->tmux->proc_query(4, $dbName, ''); $proc4Result = DB::selectOne($proc4Query); if ($proc4Result) { @@ -297,7 +295,7 @@ class TmuxMonitorService // Get newest/oldest data (query 6) $timer6 = time(); - $proc6Query = $this->tmux->proc_query(6, $bookReqIds, $dbName); + $proc6Query = $this->tmux->proc_query(6, $dbName, ''); $proc6Result = DB::selectOne($proc6Query); if ($proc6Result) { diff --git a/app/Support/BookMatchScorer.php b/app/Support/BookMatchScorer.php new file mode 100644 index 000000000..df82364cd --- /dev/null +++ b/app/Support/BookMatchScorer.php @@ -0,0 +1,99 @@ +<?php + +declare(strict_types=1); + +namespace App\Support; + +use App\Models\BookInfo; +use App\Support\DTOs\BookParseResult; + +class BookMatchScorer +{ + /** + * @param array<string, mixed> $candidate + */ + public function score(array $candidate, BookParseResult $parsed): float + { + $candidateIsbn = $this->normalizeIsbn((string) ($candidate['isbn'] ?? '')); + $parsedIsbn = $this->normalizeIsbn((string) ($parsed->isbn ?? '')); + if ($parsedIsbn !== '' && $candidateIsbn !== '' && $parsedIsbn === $candidateIsbn) { + return 1.0; + } + + $titleScore = $this->similarity($parsed->title, (string) ($candidate['title'] ?? '')); + $authorScore = $parsed->hasAuthor() + ? $this->similarity((string) $parsed->author, (string) ($candidate['author'] ?? '')) + : 0.5; + $yearScore = $this->yearScore($parsed->year, (string) ($candidate['publishdate'] ?? '')); + $publisherScore = ! empty($candidate['publisher']) ? 1.0 : 0.0; + $coverScore = (int) ($candidate['cover'] ?? 0) === 1 ? 1.0 : 0.0; + + return (0.4 * $titleScore) + + (0.3 * $authorScore) + + (0.1 * $yearScore) + + (0.1 * $publisherScore) + + (0.1 * $coverScore); + } + + public function scoreBookInfo(BookInfo $book, BookParseResult $parsed): float + { + return $this->score([ + 'title' => $book->title, + 'author' => $book->author, + 'isbn' => $book->isbn, + 'publishdate' => $book->publishdate, + 'publisher' => $book->publisher, + 'cover' => $book->cover ? 1 : 0, + ], $parsed); + } + + private function similarity(string $left, string $right): float + { + $left = $this->normalizeText($left); + $right = $this->normalizeText($right); + if ($left === '' || $right === '') { + return 0.0; + } + + similar_text($left, $right, $percent); + + return max(0.0, min(1.0, $percent / 100)); + } + + private function yearScore(?int $parsedYear, string $candidatePublishDate): float + { + if ($parsedYear === null) { + return 0.5; + } + + if (preg_match('/\b(19|20)\d{2}\b/', $candidatePublishDate, $matches) !== 1) { + return 0.0; + } + + $candidateYear = (int) $matches[0]; + if ($parsedYear === $candidateYear) { + return 1.0; + } + + if (abs($parsedYear - $candidateYear) <= 1) { + return 0.6; + } + + return 0.0; + } + + private function normalizeText(string $value): string + { + $value = strtolower($value); + $value = (string) preg_replace('/[._-]+/', ' ', $value); + $value = (string) preg_replace('/[^\p{L}\p{N}\s]+/u', ' ', $value); + $value = (string) preg_replace('/\s+/', ' ', $value); + + return trim($value); + } + + private function normalizeIsbn(string $value): string + { + return strtoupper((string) preg_replace('/[^0-9X]/i', '', $value)); + } +} diff --git a/app/Support/DTOs/BookParseResult.php b/app/Support/DTOs/BookParseResult.php new file mode 100644 index 000000000..03024f5ce --- /dev/null +++ b/app/Support/DTOs/BookParseResult.php @@ -0,0 +1,29 @@ +<?php + +declare(strict_types=1); + +namespace App\Support\DTOs; + +final readonly class BookParseResult +{ + public function __construct( + public string $rawName, + public string $title, + public ?string $author = null, + public ?string $isbn = null, + public ?int $year = null, + public ?string $format = null, + public bool $isJunk = false, + public bool $isMagazine = false, + ) {} + + public function hasAuthor(): bool + { + return $this->author !== null && $this->author !== ''; + } + + public function searchQuery(): string + { + return trim(($this->author ?? '').' '.$this->title); + } +} diff --git a/config/nntmux_api.php b/config/nntmux_api.php index 9695a2996..93898cbe3 100644 --- a/config/nntmux_api.php +++ b/config/nntmux_api.php @@ -3,6 +3,7 @@ return [ 'anidb_api_key' => env('ANIDB_APIKEY', ''), 'fanarttv_api_key' => env('FANARTTV_APIKEY', ''), + 'google_books_api_key' => env('GOOGLE_BOOKS_API_KEY', ''), 'isbndb_api_key' => env('ISBNDB_API_KEY', ''), 'imdbapi_dev_enabled' => env('IMDBAPI_DEV_ENABLED', true), 'imdbapi_dev_base_url' => env('IMDBAPI_DEV_BASE_URL', 'https://api.imdbapi.dev'), diff --git a/database/seeders/SettingsTableSeeder.php b/database/seeders/SettingsTableSeeder.php index d39096708..0d49fa302 100644 --- a/database/seeders/SettingsTableSeeder.php +++ b/database/seeders/SettingsTableSeeder.php @@ -31,10 +31,6 @@ class SettingsTableSeeder extends Seeder 'name' => 'binarythreads', 'value' => '1', ], - 6 => [ - 'name' => 'book_reqids', - 'value' => '7010', - ], 7 => [ 'name' => 'checkpasswordedrar', 'value' => '0', diff --git a/resources/views/admin/site/edit.blade.php b/resources/views/admin/site/edit.blade.php index 257893d99..59afd3b66 100644 --- a/resources/views/admin/site/edit.blade.php +++ b/resources/views/admin/site/edit.blade.php @@ -326,20 +326,6 @@ <p class="mt-1 text-sm text-gray-500">Whether to attempt to lookup book information from ISBNdb, with iTunes fallback.</p> </div> - <div> - <label for="book_reqids" class="block text-sm font-medium text-gray-700 dark:text-gray-300 mb-1"> - <i class="fas fa-list mr-1"></i>Type of Books to Look Up - </label> - <select id="book_reqids" name="book_reqids[]" multiple class="w-full px-3 py-2 border border-gray-300 dark:border-gray-600 rounded-md focus:ring-blue-500 focus:border-blue-500 bg-white dark:bg-gray-700 text-gray-900 dark:text-gray-100" size="4"> - @foreach($book_reqids['ids'] as $index => $bookReqId) - <option value="{{ $bookReqId }}" {{ in_array($bookReqId, $book_reqids['selected'] ?? []) ? 'selected' : '' }}> - {{ $book_reqids['names'][$index] }} - </option> - @endforeach - </select> - <p class="mt-1 text-sm text-gray-500">Categories of Books to lookup information for (only work if Lookup Books is set to yes).</p> - </div> - <div> <label for="lookupimdb" class="block text-sm font-medium text-gray-700 dark:text-gray-300 mb-1"> <i class="fas fa-film mr-1"></i>Lookup Movies diff --git a/tests/Unit/Services/BookServiceTest.php b/tests/Unit/Services/BookServiceTest.php new file mode 100644 index 000000000..1ad4e62d2 --- /dev/null +++ b/tests/Unit/Services/BookServiceTest.php @@ -0,0 +1,33 @@ +<?php + +declare(strict_types=1); + +namespace Tests\Unit\Services; + +use App\Services\BookService; +use Tests\TestCase; + +class BookServiceTest extends TestCase +{ + public function test_parse_release_name_extracts_author_title_and_isbn(): void + { + /** @var BookService $service */ + $service = (new \ReflectionClass(BookService::class))->newInstanceWithoutConstructor(); + + $parsed = $service->parseReleaseName('Eric.Evans - Domain.Driven.Design 978-0321125217 RETAIL EPUB'); + + $this->assertSame('Domain Driven Design', $parsed->title); + $this->assertSame('Eric Evans', $parsed->author); + $this->assertSame('9780321125217', $parsed->isbn); + } + + public function test_parse_release_name_flags_software_as_junk(): void + { + /** @var BookService $service */ + $service = (new \ReflectionClass(BookService::class))->newInstanceWithoutConstructor(); + + $parsed = $service->parseReleaseName('Foxit PDF Editor Pro 14.0.4.33508 Multilingual x64', 'ebook'); + + $this->assertTrue($parsed->isJunk); + } +} diff --git a/tests/Unit/Services/GoogleBooksServiceTest.php b/tests/Unit/Services/GoogleBooksServiceTest.php new file mode 100644 index 000000000..f07d551b5 --- /dev/null +++ b/tests/Unit/Services/GoogleBooksServiceTest.php @@ -0,0 +1,64 @@ +<?php + +declare(strict_types=1); + +namespace Tests\Unit\Services; + +use App\Services\GoogleBooksService; +use GuzzleHttp\Client; +use GuzzleHttp\Handler\MockHandler; +use GuzzleHttp\HandlerStack; +use GuzzleHttp\Psr7\Response; +use Tests\TestCase; + +class GoogleBooksServiceTest extends TestCase +{ + public function test_search_books_maps_google_books_payload(): void + { + $mock = new MockHandler([ + new Response(200, [], json_encode([ + 'items' => [[ + 'id' => 'gbook1', + 'volumeInfo' => [ + 'title' => 'The Pragmatic Programmer', + 'authors' => ['Andrew Hunt', 'David Thomas'], + 'publisher' => 'Addison-Wesley', + 'publishedDate' => '1999-10-20', + 'pageCount' => 352, + 'description' => '<p>Classic software engineering guidance.</p>', + 'categories' => ['Computers'], + 'industryIdentifiers' => [ + ['type' => 'ISBN_13', 'identifier' => '9780201616224'], + ['type' => 'ISBN_10', 'identifier' => '020161622X'], + ], + 'imageLinks' => [ + 'thumbnail' => 'https://example.com/cover.jpg', + ], + ], + ]], + ])), + ]); + + $service = new GoogleBooksService( + new Client(['handler' => HandlerStack::create($mock)]), + null + ); + + $books = $service->searchBooks('pragmatic programmer'); + + $this->assertCount(1, $books); + $this->assertSame('The Pragmatic Programmer', $books[0]['title']); + $this->assertSame('Andrew Hunt, David Thomas', $books[0]['author']); + $this->assertSame('9780201616224', $books[0]['isbn']); + $this->assertSame('020161622X', $books[0]['ean']); + $this->assertSame('googlebooks:9780201616224', $books[0]['asin']); + } + + public function test_has_api_key_is_false_when_missing(): void + { + $service = new GoogleBooksService(null, ''); + + $this->assertTrue($service->isConfigured()); + $this->assertFalse($service->hasApiKey()); + } +} diff --git a/tests/Unit/Services/IsbnDbServiceTest.php b/tests/Unit/Services/IsbnDbServiceTest.php index 96e5c8ffc..ca22208a7 100644 --- a/tests/Unit/Services/IsbnDbServiceTest.php +++ b/tests/Unit/Services/IsbnDbServiceTest.php @@ -50,8 +50,10 @@ class IsbnDbServiceTest extends TestCase ); $book = $service->searchBook('domain driven design'); + $books = $service->searchBooks('domain driven design'); $this->assertNotNull($book); + $this->assertCount(1, $books); $this->assertSame('Domain-Driven Design', $book['title']); $this->assertSame('Eric Evans', $book['author']); $this->assertSame('9780321125217', $book['isbn']); diff --git a/tests/Unit/Services/OpenLibraryServiceTest.php b/tests/Unit/Services/OpenLibraryServiceTest.php new file mode 100644 index 000000000..829d2e19a --- /dev/null +++ b/tests/Unit/Services/OpenLibraryServiceTest.php @@ -0,0 +1,69 @@ +<?php + +declare(strict_types=1); + +namespace Tests\Unit\Services; + +use App\Services\OpenLibraryService; +use GuzzleHttp\Client; +use GuzzleHttp\Handler\MockHandler; +use GuzzleHttp\HandlerStack; +use GuzzleHttp\Psr7\Response; +use Tests\TestCase; + +class OpenLibraryServiceTest extends TestCase +{ + public function test_search_books_normalizes_docs_payload(): void + { + $mock = new MockHandler([ + new Response(200, [], json_encode([ + 'docs' => [[ + 'key' => '/works/OL123W', + 'title' => 'Refactoring', + 'author_name' => ['Martin Fowler'], + 'isbn' => ['9780201485677', '0201485672'], + 'publisher' => ['Addison-Wesley'], + 'first_publish_year' => 1999, + 'cover_i' => 54321, + ]], + ])), + ]); + + $service = new OpenLibraryService( + new Client(['handler' => HandlerStack::create($mock)]) + ); + + $books = $service->searchBooks('refactoring'); + + $this->assertCount(1, $books); + $this->assertSame('Refactoring', $books[0]['title']); + $this->assertSame('Martin Fowler', $books[0]['author']); + $this->assertSame('9780201485677', $books[0]['isbn']); + $this->assertSame('0201485672', $books[0]['ean']); + $this->assertSame('openlibrary:9780201485677', $books[0]['asin']); + } + + public function test_find_by_isbn_normalizes_payload(): void + { + $mock = new MockHandler([ + new Response(200, [], json_encode([ + 'title' => 'Clean Architecture', + 'authors' => [['name' => 'Robert C. Martin']], + 'publish_date' => '2017-09-20', + 'publishers' => ['Prentice Hall'], + 'covers' => [12345], + ])), + ]); + + $service = new OpenLibraryService( + new Client(['handler' => HandlerStack::create($mock)]) + ); + + $book = $service->findByIsbn('9780134494166'); + + $this->assertNotNull($book); + $this->assertSame('Clean Architecture', $book['title']); + $this->assertSame('Robert C. Martin', $book['author']); + $this->assertSame('9780134494166', $book['isbn']); + } +} diff --git a/tests/Unit/Support/BookMatchScorerTest.php b/tests/Unit/Support/BookMatchScorerTest.php new file mode 100644 index 000000000..c5e712604 --- /dev/null +++ b/tests/Unit/Support/BookMatchScorerTest.php @@ -0,0 +1,63 @@ +<?php + +declare(strict_types=1); + +namespace Tests\Unit\Support; + +use App\Support\BookMatchScorer; +use App\Support\DTOs\BookParseResult; +use Tests\TestCase; + +class BookMatchScorerTest extends TestCase +{ + public function test_returns_perfect_score_for_isbn_match(): void + { + $parsed = new BookParseResult( + rawName: 'Domain Driven Design', + title: 'Domain-Driven Design', + author: 'Eric Evans', + isbn: '9780321125217' + ); + + $score = (new BookMatchScorer)->score([ + 'title' => 'Some Other Title', + 'author' => 'Unknown', + 'isbn' => '9780321125217', + 'publishdate' => '2020-01-01', + 'publisher' => '', + 'cover' => 0, + ], $parsed); + + $this->assertSame(1.0, $score); + } + + public function test_higher_score_for_better_title_author_match(): void + { + $parsed = new BookParseResult( + rawName: 'Refactoring', + title: 'Refactoring', + author: 'Martin Fowler', + year: 1999 + ); + + $scorer = new BookMatchScorer; + + $good = $scorer->score([ + 'title' => 'Refactoring: Improving the Design of Existing Code', + 'author' => 'Martin Fowler', + 'publishdate' => '1999-07-08', + 'publisher' => 'Addison-Wesley', + 'cover' => 1, + ], $parsed); + + $bad = $scorer->score([ + 'title' => 'Cooking for Beginners', + 'author' => 'Random Author', + 'publishdate' => '2018-01-01', + 'publisher' => '', + 'cover' => 0, + ], $parsed); + + $this->assertGreaterThan($bad, $good); + } +}