diff --git a/Blacklight/Movie.php b/Blacklight/Movie.php
index fc9305fc7..155f899ff 100755
--- a/Blacklight/Movie.php
+++ b/Blacklight/Movie.php
@@ -7,6 +7,7 @@ use App\Models\Category;
use App\Models\MovieInfo;
use App\Models\Release;
use App\Models\Settings;
+use App\Services\ImdbScraper;
use Blacklight\libraries\FanartTV;
use Blacklight\processing\tv\TraktTv;
use Blacklight\utility\Utility;
@@ -24,12 +25,9 @@ use Illuminate\Support\Facades\Cache;
use Illuminate\Support\Facades\DB;
use Illuminate\Support\Facades\File;
use Illuminate\Support\Facades\Log;
-use Illuminate\Support\Str;
-use Imdb\Config;
-use Imdb\Title;
-use Imdb\TitleSearch;
-use Tmdb\Exception\TmdbApiException;
-use Tmdb\Laravel\Facades\Tmdb;
+use Illuminate\Support\Str; // ensure scraper imported
+use Tmdb; // add TMDB facade
+use Tmdb\Exception\TmdbApiException; // add TMDB exception
/**
* Class Movie.
@@ -101,8 +99,6 @@ class Movie
public ?OMDbAPI $omdbApi;
- private Config $config;
-
/**
* @var null|string
*/
@@ -133,14 +129,11 @@ class Movie
}
$this->lookuplanguage = Settings::settingValue('imdblanguage') !== '' ? (string) Settings::settingValue('imdblanguage') : 'en';
- $this->config = new Config;
- $this->config->language = $this->lookuplanguage;
- $this->config->throwHttpExceptions = false;
$cacheDir = storage_path('framework/cache/imdb_cache');
if (! File::isDirectory($cacheDir)) {
File::makeDirectory($cacheDir, 0777, false, true);
}
- $this->config->cachedir = $cacheDir;
+ // $this->config->cachedir = $cacheDir; // removed imdbphp config
$this->imdburl = (int) Settings::settingValue('imdburl') !== 0;
$this->movieqty = Settings::settingValue('maximdbprocessed') !== '' ? (int) Settings::settingValue('maximdbprocessed') : 100;
@@ -169,40 +162,23 @@ class Movie
{
$page = max(1, $page);
$start = max(0, $start);
-
$catsrch = '';
- if (\count($cat) > 0 && $cat[0] !== -1) {
+ if (count($cat) > 0 && $cat[0] !== -1) {
$catsrch = Category::getCategorySearch($cat);
}
$order = $this->getMovieOrder($orderBy);
$expiresAt = now()->addMinutes(config('nntmux.cache_expiry_medium'));
- $moviesSql =
- sprintf(
- "
- SELECT SQL_CALC_FOUND_ROWS
- m.imdbid,
- GROUP_CONCAT(r.id ORDER BY r.postdate DESC SEPARATOR ',') AS grp_release_id
- FROM movieinfo m
- LEFT JOIN releases r USING (imdbid)
- WHERE m.title != ''
- AND m.imdbid != '0000000'
- AND r.passwordstatus %s
- %s %s %s %s
- GROUP BY m.imdbid
- ORDER BY %s %s %s",
- $this->showPasswords,
- $this->getBrowseBy(),
- (! empty($catsrch) ? $catsrch : ''),
- (
- $maxAge > 0
- ? 'AND r.postdate > NOW() - INTERVAL '.$maxAge.'DAY '
- : ''
- ),
- \count($excludedCats) > 0 ? ' AND r.categories_id NOT IN ('.implode(',', $excludedCats).')' : '',
- $order[0],
- $order[1],
- $start === false ? '' : ' LIMIT '.$num.' OFFSET '.$start
- );
+ $whereAge = $maxAge > 0 ? 'AND r.postdate > NOW() - INTERVAL '.$maxAge.' DAY ' : '';
+ $whereExcluded = count($excludedCats) > 0 ? ' AND r.categories_id NOT IN ('.implode(',', $excludedCats).')' : '';
+ $limitClause = $start === false ? '' : ' LIMIT '.$num.' OFFSET '.$start;
+ $moviesSql = "SELECT SQL_CALC_FOUND_ROWS m.imdbid, GROUP_CONCAT(r.id ORDER BY r.postdate DESC SEPARATOR ',' ) AS grp_release_id "
+ ."FROM movieinfo m LEFT JOIN releases r USING (imdbid) WHERE m.title != '' AND m.imdbid != '0000000' "
+ ."AND r.passwordstatus {$this->showPasswords} "
+ .$this->getBrowseBy().' '
+ .(! empty($catsrch) ? $catsrch.' ' : '')
+ .$whereAge
+ .$whereExcluded.' '
+ ."GROUP BY m.imdbid ORDER BY {$order[0]} {$order[1]} {$limitClause}";
$movieCache = Cache::get(md5($moviesSql.$page));
if ($movieCache !== null) {
$movies = $movieCache;
@@ -213,57 +189,46 @@ class Movie
}
$movieIDs = $releaseIDs = [];
if (! empty($movies['result'])) {
- foreach ($movies['result'] as $movie => $id) {
+ foreach ($movies['result'] as $id) {
$movieIDs[] = $id->imdbid;
$releaseIDs[] = $id->grp_release_id;
}
}
-
- $sql = sprintf(
- "
- SELECT
- GROUP_CONCAT(r.id ORDER BY r.postdate DESC SEPARATOR ',') AS grp_release_id,
- GROUP_CONCAT(r.rarinnerfilecount ORDER BY r.postdate DESC SEPARATOR ',') AS grp_rarinnerfilecount,
- GROUP_CONCAT(r.haspreview ORDER BY r.postdate DESC SEPARATOR ',') AS grp_haspreview,
- GROUP_CONCAT(r.passwordstatus ORDER BY r.postdate DESC SEPARATOR ',') AS grp_release_password,
- GROUP_CONCAT(r.guid ORDER BY r.postdate DESC SEPARATOR ',') AS grp_release_guid,
- GROUP_CONCAT(rn.releases_id ORDER BY r.postdate DESC SEPARATOR ',') AS grp_release_nfoid,
- GROUP_CONCAT(g.name ORDER BY r.postdate DESC SEPARATOR ',') AS grp_release_grpname,
- GROUP_CONCAT(r.searchname ORDER BY r.postdate DESC SEPARATOR '#') AS grp_release_name,
- GROUP_CONCAT(r.postdate ORDER BY r.postdate DESC SEPARATOR ',') AS grp_release_postdate,
- GROUP_CONCAT(r.adddate ORDER BY r.postdate DESC SEPARATOR ',') AS grp_release_adddate,
- GROUP_CONCAT(r.size ORDER BY r.postdate DESC SEPARATOR ',') AS grp_release_size,
- GROUP_CONCAT(r.totalpart ORDER BY r.postdate DESC SEPARATOR ',') AS grp_release_totalparts,
- GROUP_CONCAT(r.comments ORDER BY r.postdate DESC SEPARATOR ',') AS grp_release_comments,
- GROUP_CONCAT(r.grabs ORDER BY r.postdate DESC SEPARATOR ',') AS grp_release_grabs,
- GROUP_CONCAT(df.failed ORDER BY r.postdate DESC SEPARATOR ',') AS grp_release_failed,
- GROUP_CONCAT(cp.title, ' > ', c.title ORDER BY r.postdate DESC SEPARATOR ',') AS grp_release_catname,
- m.*,
- g.name AS group_name,
- rn.releases_id AS nfoid
- FROM releases r
- LEFT OUTER JOIN usenet_groups g ON g.id = r.groups_id
- LEFT OUTER JOIN release_nfos rn ON rn.releases_id = r.id
- LEFT OUTER JOIN dnzb_failures df ON df.release_id = r.id
- LEFT OUTER JOIN categories c ON c.id = r.categories_id
- LEFT OUTER JOIN root_categories cp ON cp.id = c.root_categories_id
- INNER JOIN movieinfo m ON m.imdbid = r.imdbid
- WHERE m.imdbid IN (%s)
- AND r.id IN (%s) %s
- GROUP BY m.imdbid
- ORDER BY %s %s",
- (\is_array($movieIDs) && ! empty($movieIDs) ? implode(',', $movieIDs) : -1),
- (\is_array($releaseIDs) && ! empty($releaseIDs) ? implode(',', $releaseIDs) : -1),
- (! empty($catsrch) ? $catsrch : ''),
- $order[0],
- $order[1]
- );
+ $inMovieIds = (is_array($movieIDs) && ! empty($movieIDs)) ? implode(',', $movieIDs) : -1;
+ $inReleaseIds = (is_array($releaseIDs) && ! empty($releaseIDs)) ? implode(',', $releaseIDs) : -1;
+ $sql = 'SELECT '
+ ."GROUP_CONCAT(r.id ORDER BY r.postdate DESC SEPARATOR ',' ) AS grp_release_id, "
+ ."GROUP_CONCAT(r.rarinnerfilecount ORDER BY r.postdate DESC SEPARATOR ',' ) AS grp_rarinnerfilecount, "
+ ."GROUP_CONCAT(r.haspreview ORDER BY r.postdate DESC SEPARATOR ',' ) AS grp_haspreview, "
+ ."GROUP_CONCAT(r.passwordstatus ORDER BY r.postdate DESC SEPARATOR ',' ) AS grp_release_password, "
+ ."GROUP_CONCAT(r.guid ORDER BY r.postdate DESC SEPARATOR ',' ) AS grp_release_guid, "
+ ."GROUP_CONCAT(rn.releases_id ORDER BY r.postdate DESC SEPARATOR ',' ) AS grp_release_nfoid, "
+ ."GROUP_CONCAT(g.name ORDER BY r.postdate DESC SEPARATOR ',' ) AS grp_release_grpname, "
+ ."GROUP_CONCAT(r.searchname ORDER BY r.postdate DESC SEPARATOR '#') AS grp_release_name, "
+ ."GROUP_CONCAT(r.postdate ORDER BY r.postdate DESC SEPARATOR ',' ) AS grp_release_postdate, "
+ ."GROUP_CONCAT(r.adddate ORDER BY r.postdate DESC SEPARATOR ',' ) AS grp_release_adddate, "
+ ."GROUP_CONCAT(r.size ORDER BY r.postdate DESC SEPARATOR ',' ) AS grp_release_size, "
+ ."GROUP_CONCAT(r.totalpart ORDER BY r.postdate DESC SEPARATOR ',' ) AS grp_release_totalparts, "
+ ."GROUP_CONCAT(r.comments ORDER BY r.postdate DESC SEPARATOR ',' ) AS grp_release_comments, "
+ ."GROUP_CONCAT(r.grabs ORDER BY r.postdate DESC SEPARATOR ',' ) AS grp_release_grabs, "
+ ."GROUP_CONCAT(df.failed ORDER BY r.postdate DESC SEPARATOR ',' ) AS grp_release_failed, "
+ ."GROUP_CONCAT(cp.title, ' > ', c.title ORDER BY r.postdate DESC SEPARATOR ',' ) AS grp_release_catname, "
+ .'m.*, g.name AS group_name, rn.releases_id AS nfoid FROM releases r '
+ .'LEFT OUTER JOIN usenet_groups g ON g.id = r.groups_id '
+ .'LEFT OUTER JOIN release_nfos rn ON rn.releases_id = r.id '
+ .'LEFT OUTER JOIN dnzb_failures df ON df.release_id = r.id '
+ .'LEFT OUTER JOIN categories c ON c.id = r.categories_id '
+ .'LEFT OUTER JOIN root_categories cp ON cp.id = c.root_categories_id '
+ .'INNER JOIN movieinfo m ON m.imdbid = r.imdbid '
+ ."WHERE m.imdbid IN ($inMovieIds) AND r.id IN ($inReleaseIds) "
+ .(! empty($catsrch) ? $catsrch.' ' : '')
+ ."GROUP BY m.imdbid ORDER BY {$order[0]} {$order[1]}";
$return = Cache::get(md5($sql.$page));
if ($return !== null) {
return $return;
}
$return = Release::fromQuery($sql);
- if (\count($return) > 0) {
+ if (count($return) > 0) {
$return[0]->_totalcount = $movies['total'][0]->total ?? 0;
}
Cache::put(md5($sql.$page), $return, $expiresAt);
@@ -624,7 +589,7 @@ class Movie
'title' => $mov['title'],
'tmdbid' => $mov['tmdbid'],
'traktid' => $mov['traktid'],
- 'type' => html_entity_decode(ucwords(preg_replace('/[\.\_]/', ' ', $mov['type'])), ENT_QUOTES, 'UTF-8'),
+ 'type' => html_entity_decode(ucwords(preg_replace('/[._]/', ' ', $mov['type'])), ENT_QUOTES, 'UTF-8'),
'year' => $mov['year'],
]);
@@ -838,76 +803,43 @@ class Movie
*/
public function fetchIMDBProperties(string $imdbId): array|false
{
- // Create a cache key for this request
$cacheKey = 'imdb_movie_'.md5($imdbId);
- $expiresAt = now()->addDays(7); // Cache for 7 days
-
- // Check if we have this cached
+ $expiresAt = now()->addDays(7);
if (Cache::has($cacheKey)) {
return Cache::get($cacheKey);
}
-
try {
- // Get real ID and title information
- $realId = (new Title($imdbId, $this->config))->real_id();
- $result = new Title($realId, $this->config);
- $title = ! empty($result->orig_title()) ? $result->orig_title() : $result->title();
-
- // If no title was found, cache the failure and return false
- if (empty($title)) {
+ $scraper = app(ImdbScraper::class);
+ $scraped = $scraper->fetchById($imdbId);
+ if ($scraped === false || empty($scraped['title'])) {
Cache::put($cacheKey, false, now()->addHours(6));
return false;
}
-
- // Title similarity check when we have a title to compare against
if (! empty($this->currentTitle)) {
- similar_text($this->currentTitle, $title, $percent);
+ similar_text($this->currentTitle, $scraped['title'], $percent);
if ($percent < self::MATCH_PERCENT) {
Cache::put($cacheKey, false, now()->addHours(6));
return false;
}
+ if (! empty($this->currentYear) && ! empty($scraped['year'])) {
+ similar_text($this->currentYear, $scraped['year'], $yearPercent);
+ if ($yearPercent < self::YEAR_MATCH_PERCENT) {
+ Cache::put($cacheKey, false, now()->addHours(6));
- // Year similarity check
- similar_text($this->currentYear, $result->year(), $percent);
- if ($percent < self::YEAR_MATCH_PERCENT) {
- Cache::put($cacheKey, false, now()->addHours(6));
-
- return false;
+ return false;
+ }
}
}
-
- // Build the movie data array
- $movieData = [
- 'title' => $title,
- 'tagline' => $result->tagline() ?? '',
- 'plot' => Arr::get($result->plot_split(), '0.plot', ''),
- 'rating' => $result->rating() ?? '',
- 'year' => $result->year() ?? '',
- 'cover' => $result->photo() ?? '',
- 'genre' => $result->genre() ?? '',
- 'language' => $result->language() ?? '',
- 'type' => $result->movietype() ?? '',
- 'actors' => $result->cast() ?? [],
- 'director' => $result->director() ?? [],
- ];
-
- // Log success
+ Cache::put($cacheKey, $scraped, $expiresAt);
if ($this->echooutput) {
- $this->colorCli->info('IMDb found '.$title);
+ $this->colorCli->info('IMDb scraped '.$scraped['title']);
}
- // Cache the successful result
- Cache::put($cacheKey, $movieData, $expiresAt);
-
- return $movieData;
-
- } catch (\Exception $e) {
- // Log the error
- Log::error('IMDB API error for '.$imdbId.': '.$e->getMessage());
-
- // Cache the failure but for shorter time
+ return $scraped;
+ } catch (\Throwable $e) {
+ Log::error('IMDb scrape error for '.$imdbId.': '.$e->getMessage());
Cache::put($cacheKey, false, now()->addHours(6));
return false;
@@ -1441,35 +1373,32 @@ class Movie
*/
private function searchIMDb(int $releaseId): bool
{
+ // Simplified scraper-only search
try {
- $imdbSearch = new TitleSearch($this->config);
- foreach ($imdbSearch->search($this->currentTitle, [TitleSearch::MOVIE]) as $imdbTitle) {
- if (empty($imdbTitle->title())) {
+ $scraper = app(ImdbScraper::class);
+ $matches = $scraper->search($this->currentTitle);
+ foreach ($matches as $match) {
+ $title = $match['title'] ?? '';
+ if ($title === '') {
continue;
}
-
- // Compare title similarity
- similar_text($imdbTitle->title(), $this->currentTitle, $percent);
+ similar_text($title, $this->currentTitle, $percent);
if ($percent < self::MATCH_PERCENT) {
continue;
}
-
- // Compare year similarity if year is available
- similar_text($this->currentYear, $imdbTitle->year(), $yearPercent);
- if ($yearPercent < self::YEAR_MATCH_PERCENT) {
- continue;
+ if (! empty($this->currentYear) && ! empty($match['year'])) {
+ similar_text($this->currentYear, $match['year'], $yearPercent);
+ if ($yearPercent < self::YEAR_MATCH_PERCENT) {
+ continue;
+ }
}
-
- // Found a good match, update the release
- $getIMDBid = $imdbTitle->imdbid();
- $imdbId = $this->doMovieUpdate('tt'.$getIMDBid, 'IMDb', $releaseId);
+ $imdbId = $this->doMovieUpdate('tt'.$match['imdbid'], 'IMDb(scrape)', $releaseId);
if ($imdbId !== false) {
return true;
}
}
- } catch (\ErrorException $e) {
- $this->colorCli->error('Error fetching data from IMDb: '.$e->getMessage(), true);
- Log::debug($e->getMessage());
+ } catch (\Throwable $e) {
+ Log::debug('IMDb scraper search failed: '.$e->getMessage());
}
return false;
@@ -1714,8 +1643,13 @@ class Movie
// 1. Remove any common movie flags like 1080p, BluRay, etc.
$name = preg_replace('/'.$followingList.'/i', ' ', $name);
- // 2. Remove content in parentheses, brackets, and periods/underscores
- $name = preg_replace('/\(.*?\)|\[.*?\]|[._]/i', ' ', $name);
+ // 2. Remove content in parentheses, brackets, and periods/underscores without complex alternation
+ $name = preg_replace('/\([^)]*\)/i', ' ', $name);
+ // Remove [ ... ] blocks without regex escapes to satisfy linter
+ while (($openPos = strpos($name, '[')) !== false && ($closePos = strpos($name, ']', $openPos)) !== false) {
+ $name = substr($name, 0, $openPos).' '.substr($name, $closePos + 1);
+ }
+ $name = str_replace(['.', '_'], ' ', $name);
// 3. Remove scene group names (typically after a hyphen)
$name = preg_replace('/-[A-Z0-9].*$/i', '', $name);
diff --git a/app/Services/ImdbScraper.php b/app/Services/ImdbScraper.php
new file mode 100644
index 000000000..6b80296d6
--- /dev/null
+++ b/app/Services/ImdbScraper.php
@@ -0,0 +1,225 @@
+client = new Client([
+ 'timeout' => 10,
+ 'headers' => [
+ 'User-Agent' => 'Mozilla/5.0 (compatible; NNTmuxBot/1.0; +https://github.com/NNTmux)',
+ ],
+ ]);
+ }
+
+ /**
+ * Fetch a movie by IMDB numeric ID.
+ *
+ * @param string $id Numeric part without 'tt'.
+ */
+ public function fetchById(string $id): array|false
+ {
+ $id = preg_replace('/[^0-9]/', '', $id);
+ if ($id === '') {
+ return false;
+ }
+ $cacheKey = 'imdb_scrape_id_'.$id;
+ if (Cache::has($cacheKey)) {
+ return Cache::get($cacheKey);
+ }
+ $url = 'https://www.imdb.com/title/tt'.$id.'/';
+ try {
+ $res = $this->client->get($url);
+ $html = (string) $res->getBody();
+ $dom = HtmlDomParser::str_get_html($html);
+ if (! $dom) {
+ Cache::put($cacheKey, false, now()->addHours(6));
+
+ return false;
+ }
+
+ $titleNode = $dom->findOne('h1');
+ $title = trim($titleNode?->text() ?? '');
+ if ($title === '') {
+ Cache::put($cacheKey, false, now()->addHours(6));
+
+ return false;
+ }
+
+ // Year
+ $year = '';
+ $yearNode = $dom->findOne("span[data-testid='title-details-releasedate'] a");
+ if ($yearNode) {
+ if (preg_match('/(19|20)\d{2}/', $yearNode->text(), $m)) {
+ $year = $m[0];
+ }
+ }
+ if ($year === '') {
+ $altYearNode = $dom->findOne("span[data-testid='title-details-releasedate']");
+ if ($altYearNode && preg_match('/(19|20)\d{2}/', $altYearNode->text(), $m)) {
+ $year = $m[0];
+ }
+ }
+
+ // Plot
+ $plot = '';
+ $plotNode = $dom->findOne("span[data-testid='plot-l']");
+ if (! $plotNode) {
+ $plotNode = $dom->findOne("span[data-testid='plot-xl']");
+ }
+ if ($plotNode) {
+ $plot = trim($plotNode->text());
+ }
+
+ // Rating
+ $rating = '';
+ $ratingNode = $dom->findOne("div[data-testid='hero-rating-bar__aggregate-rating__score'] span");
+ if ($ratingNode) {
+ $rating = trim($ratingNode->text());
+ }
+ $rating = preg_replace('/[^0-9\.]/', '', $rating);
+
+ // Poster
+ $cover = '';
+ $posterNode = $dom->findOne("div[data-testid='hero-media__poster'] img");
+ if ($posterNode) {
+ $cover = $posterNode->getAttribute('src');
+ }
+ if ($cover !== '' && str_contains($cover, '._V1_')) {
+ // Attempt higher res by removing size suffix
+ $cover = preg_replace('/\._V1_.*\./', '.', $cover);
+ }
+
+ // Genres
+ $genres = [];
+ foreach ($dom->find("div[data-testid='genres'] a") as $g) {
+ $gt = trim($g->text());
+ if ($gt !== '') {
+ $genres[] = $gt;
+ }
+ }
+
+ // Directors
+ $directors = [];
+ foreach ($dom->find("li[data-testid='title-pc-principal-credit']:contains(Director) a") as $d) {
+ $dt = trim($d->text());
+ if ($dt !== '') {
+ $directors[] = $dt;
+ }
+ }
+ if (empty($directors)) {
+ foreach ($dom->find("li[data-testid='title-pc-principal-credit']:contains(Creator) a") as $d) {
+ $dt = trim($d->text());
+ if ($dt !== '') {
+ $directors[] = $dt;
+ }
+ }
+ }
+
+ // Actors (top billed)
+ $actors = [];
+ foreach ($dom->find("div[data-testid='title-cast-item'] a[data-testid='title-cast-item__actor']") as $a) {
+ $at = trim($a->text());
+ if ($at !== '') {
+ $actors[] = $at;
+ }
+ if (count($actors) >= 10) {
+ break;
+ }
+ }
+
+ // Languages (from details section)
+ $language = '';
+ foreach ($dom->find("li[data-testid='title-details-languages'] a") as $ln) {
+ $lt = trim($ln->text());
+ if ($lt !== '') {
+ $language .= $lt.', ';
+ }
+ }
+ $language = rtrim($language, ', ');
+
+ $data = [
+ 'imdbid' => $id,
+ 'title' => $title,
+ 'year' => $year,
+ 'plot' => $plot,
+ 'rating' => $rating,
+ 'cover' => $cover,
+ 'genre' => $genres,
+ 'director' => $directors,
+ 'actors' => $actors,
+ 'language' => $language,
+ 'type' => 'movie',
+ ];
+ Cache::put($cacheKey, $data, now()->addDays(7));
+
+ return $data;
+ } catch (GuzzleException|\Throwable $e) {
+ Log::debug('IMDb fetch error tt'.$id.': '.$e->getMessage());
+ Cache::put($cacheKey, false, now()->addHours(6));
+
+ return false;
+ }
+ }
+
+ /**
+ * Search IMDb suggestion API for a title.
+ */
+ public function search(string $query): array
+ {
+ $query = trim($query);
+ if ($query === '') {
+ return [];
+ }
+ $norm = strtolower(preg_replace('/[^a-z0-9 ]/i', '', $query));
+ $slug = str_replace(' ', '_', $norm);
+ $prefix = substr($slug, 0, 1);
+ $cacheKey = 'imdb_search_'.md5($slug);
+ if (Cache::has($cacheKey)) {
+ return Cache::get($cacheKey);
+ }
+ $url = 'https://v2.sg.media-imdb.com/suggestion/'.urlencode($prefix).'/'.urlencode($slug).'.json';
+ try {
+ $res = $this->client->get($url);
+ $json = json_decode((string) $res->getBody(), true);
+ $results = [];
+ foreach (($json['d'] ?? []) as $row) {
+ if (! isset($row['id']) || ! str_starts_with($row['id'], 'tt')) {
+ continue;
+ }
+ $id = substr($row['id'], 2);
+ $title = $row['l'] ?? '';
+ $year = $row['y'] ?? '';
+ if ($title === '') {
+ continue;
+ }
+ $results[] = [
+ 'imdbid' => $id,
+ 'title' => $title,
+ 'year' => (string) $year,
+ ];
+ if (count($results) >= 25) {
+ break;
+ }
+ }
+ Cache::put($cacheKey, $results, now()->addHours(12));
+
+ return $results;
+ } catch (GuzzleException|\Throwable $e) {
+ Log::debug('IMDb search error '.$query.': '.$e->getMessage());
+ Cache::put($cacheKey, [], now()->addHours(6));
+
+ return [];
+ }
+ }
+}
diff --git a/composer.json b/composer.json
index 989f9dbd7..69fa85f27 100644
--- a/composer.json
+++ b/composer.json
@@ -60,7 +60,6 @@
"geoip2/geoip2": "^3.1",
"google/recaptcha": "^1.3",
"guzzlehttp/guzzle": "^7.9",
- "imdbphp/imdbphp": "^8.3",
"intervention/image": "^3.11",
"jrean/laravel-user-verification": "^13.0",
"laravel/framework": "^12.28",
diff --git a/composer.lock b/composer.lock
index b332f4b7a..8393fdf60 100644
--- a/composer.lock
+++ b/composer.lock
@@ -4,7 +4,7 @@
"Read more about it at https://getcomposer.org/doc/01-basic-usage.md#installing-dependencies",
"This file is @generated automatically"
],
- "content-hash": "32b40f0d4ebbdcc3116e8bfd91316751",
+ "content-hash": "51bf4d154595d4d6a5d5335d732d5be3",
"packages": [
{
"name": "aharen/omdbapi",
@@ -2552,54 +2552,6 @@
],
"time": "2025-08-22T14:27:06+00:00"
},
- {
- "name": "imdbphp/imdbphp",
- "version": "v8.3.0",
- "source": {
- "type": "git",
- "url": "https://github.com/tboothman/imdbphp.git",
- "reference": "b007c383b416e714c7212516284c41d974beba43"
- },
- "dist": {
- "type": "zip",
- "url": "https://api.github.com/repos/tboothman/imdbphp/zipball/b007c383b416e714c7212516284c41d974beba43",
- "reference": "b007c383b416e714c7212516284c41d974beba43",
- "shasum": ""
- },
- "require": {
- "ext-curl": "*",
- "ext-dom": "*",
- "ext-json": "*",
- "ext-libxml": "*",
- "ext-simplexml": "*",
- "php": ">=5.6",
- "psr/log": "^1.0 || ^2.0",
- "psr/simple-cache": "^1.0"
- },
- "require-dev": {
- "cache/array-adapter": "^1.2",
- "mockery/mockery": "^1.6",
- "php-cs-fixer/shim": "^3.9",
- "phpstan/phpstan": "^2.1",
- "phpunit/phpunit": "^9.6"
- },
- "type": "library",
- "autoload": {
- "psr-4": {
- "Imdb\\": "src/Imdb"
- }
- },
- "notification-url": "https://packagist.org/downloads/",
- "license": [
- "GPL-2.0-or-later"
- ],
- "description": "Library for retrieving film and tv information from IMDb",
- "support": {
- "issues": "https://github.com/tboothman/imdbphp/issues",
- "source": "https://github.com/tboothman/imdbphp/tree/v8.3.0"
- },
- "time": "2025-03-02T18:05:46+00:00"
- },
{
"name": "intervention/gif",
"version": "4.2.2",
@@ -7890,16 +7842,16 @@
},
{
"name": "psr/log",
- "version": "2.0.0",
+ "version": "3.0.2",
"source": {
"type": "git",
"url": "https://github.com/php-fig/log.git",
- "reference": "ef29f6d262798707a9edd554e2b82517ef3a9376"
+ "reference": "f16e1d5863e37f8d8c2a01719f5b34baa2b714d3"
},
"dist": {
"type": "zip",
- "url": "https://api.github.com/repos/php-fig/log/zipball/ef29f6d262798707a9edd554e2b82517ef3a9376",
- "reference": "ef29f6d262798707a9edd554e2b82517ef3a9376",
+ "url": "https://api.github.com/repos/php-fig/log/zipball/f16e1d5863e37f8d8c2a01719f5b34baa2b714d3",
+ "reference": "f16e1d5863e37f8d8c2a01719f5b34baa2b714d3",
"shasum": ""
},
"require": {
@@ -7908,7 +7860,7 @@
"type": "library",
"extra": {
"branch-alias": {
- "dev-master": "2.0.x-dev"
+ "dev-master": "3.x-dev"
}
},
"autoload": {
@@ -7934,31 +7886,31 @@
"psr-3"
],
"support": {
- "source": "https://github.com/php-fig/log/tree/2.0.0"
+ "source": "https://github.com/php-fig/log/tree/3.0.2"
},
- "time": "2021-07-14T16:41:46+00:00"
+ "time": "2024-09-11T13:17:53+00:00"
},
{
"name": "psr/simple-cache",
- "version": "1.0.1",
+ "version": "3.0.0",
"source": {
"type": "git",
"url": "https://github.com/php-fig/simple-cache.git",
- "reference": "408d5eafb83c57f6365a3ca330ff23aa4a5fa39b"
+ "reference": "764e0b3939f5ca87cb904f570ef9be2d78a07865"
},
"dist": {
"type": "zip",
- "url": "https://api.github.com/repos/php-fig/simple-cache/zipball/408d5eafb83c57f6365a3ca330ff23aa4a5fa39b",
- "reference": "408d5eafb83c57f6365a3ca330ff23aa4a5fa39b",
+ "url": "https://api.github.com/repos/php-fig/simple-cache/zipball/764e0b3939f5ca87cb904f570ef9be2d78a07865",
+ "reference": "764e0b3939f5ca87cb904f570ef9be2d78a07865",
"shasum": ""
},
"require": {
- "php": ">=5.3.0"
+ "php": ">=8.0.0"
},
"type": "library",
"extra": {
"branch-alias": {
- "dev-master": "1.0.x-dev"
+ "dev-master": "3.0.x-dev"
}
},
"autoload": {
@@ -7973,7 +7925,7 @@
"authors": [
{
"name": "PHP-FIG",
- "homepage": "http://www.php-fig.org/"
+ "homepage": "https://www.php-fig.org/"
}
],
"description": "Common interfaces for simple caching",
@@ -7985,9 +7937,9 @@
"simple-cache"
],
"support": {
- "source": "https://github.com/php-fig/simple-cache/tree/master"
+ "source": "https://github.com/php-fig/simple-cache/tree/3.0.0"
},
- "time": "2017-10-23T01:57:42+00:00"
+ "time": "2021-10-29T13:26:27+00:00"
},
{
"name": "psy/psysh",
diff --git a/tests/Fixtures/imdb/example_title.html b/tests/Fixtures/imdb/example_title.html
new file mode 100644
index 000000000..6495fefa1
--- /dev/null
+++ b/tests/Fixtures/imdb/example_title.html
@@ -0,0 +1,23 @@
+Example Movie - IMDb
+ "director": [{"@type":"Person","name":"Famous Director"}]
+ ],
+ {"@type":"Person","name":"Second Actor"}
+ {"@type":"Person","name":"First Actor"},
+ "actor": [
+ "genre": ["Action", "Adventure"],
+ "image": "https://example.com/poster_from_jsonld.jpg",
+ "datePublished": "2024-05-17",
+ "aggregateRating": {"@type":"AggregateRating","ratingValue":"7.3","ratingCount":12345},
+ "description": "An example plot goes here.",
+ "name": "Example Movie",
+ "@type": "Movie",
+ "@context": "https://schema.org",
+
diff --git a/tests/Integration/ImdbScraperIntegrationTest.php b/tests/Integration/ImdbScraperIntegrationTest.php
new file mode 100644
index 000000000..3b69463c5
--- /dev/null
+++ b/tests/Integration/ImdbScraperIntegrationTest.php
@@ -0,0 +1,121 @@
+markTestSkipped('Integration tests disabled by default. Remove this line to run live tests.');
+ }
+
+ /**
+ * Test fetching real data from IMDb for The Shawshank Redemption.
+ *
+ * @group integration
+ * @group slow
+ */
+ public function test_fetch_shawshank_redemption_from_live_site(): void
+ {
+ $scraper = new ImdbScraper;
+ $data = $scraper->fetchById('0111161');
+
+ $this->assertIsArray($data);
+ $this->assertSame('The Shawshank Redemption', $data['title']);
+ $this->assertNotEmpty($data['plot']);
+ $this->assertGreaterThan(9.0, (float) $data['rating']);
+ $this->assertSame('1994', $data['year']);
+ $this->assertNotEmpty($data['cover']);
+ $this->assertStringContainsString('Drama', $data['genre']);
+ $this->assertContains('Tim Robbins', $data['actors']);
+ $this->assertContains('Morgan Freeman', $data['actors']);
+ $this->assertContains('Frank Darabont', $data['director']);
+ }
+
+ /**
+ * Test searching for movies on live IMDb site.
+ *
+ * @group integration
+ * @group slow
+ */
+ public function test_search_inception_on_live_site(): void
+ {
+ $scraper = new ImdbScraper;
+ $results = $scraper->search('Inception');
+
+ $this->assertIsArray($results);
+ $this->assertNotEmpty($results);
+
+ // Find the 2010 Inception movie in results
+ $found = false;
+ foreach ($results as $result) {
+ if ($result['title'] === 'Inception' && $result['year'] === '2010') {
+ $found = true;
+ $this->assertSame('1375666', $result['imdbid']);
+ break;
+ }
+ }
+
+ $this->assertTrue($found, 'Inception (2010) should be found in search results');
+ }
+
+ /**
+ * Test that invalid IMDb IDs return false.
+ *
+ * @group integration
+ * @group slow
+ */
+ public function test_fetch_invalid_id_returns_false(): void
+ {
+ $scraper = new ImdbScraper;
+ $data = $scraper->fetchById('9999999');
+
+ $this->assertFalse($data);
+ }
+
+ /**
+ * Test caching works correctly on live requests.
+ *
+ * @group integration
+ * @group slow
+ */
+ public function test_caching_reduces_network_calls(): void
+ {
+ $scraper = new ImdbScraper;
+
+ // First call - should hit the network
+ $start1 = microtime(true);
+ $data1 = $scraper->fetchById('0111161');
+ $time1 = microtime(true) - $start1;
+
+ $this->assertIsArray($data1);
+
+ // Second call - should be cached (much faster)
+ $start2 = microtime(true);
+ $data2 = $scraper->fetchById('0111161');
+ $time2 = microtime(true) - $start2;
+
+ $this->assertIsArray($data2);
+ $this->assertSame($data1['title'], $data2['title']);
+
+ // Cached call should be at least 10x faster
+ $this->assertLessThan($time1 / 10, $time2, 'Cached call should be significantly faster');
+ }
+}
diff --git a/tests/Unit/ImdbScraperRealDataTest.php b/tests/Unit/ImdbScraperRealDataTest.php
new file mode 100644
index 000000000..1b7640af8
--- /dev/null
+++ b/tests/Unit/ImdbScraperRealDataTest.php
@@ -0,0 +1,169 @@
+ 'array']);
+ }
+
+ public function test_parse_shawshank_redemption_real_data(): void
+ {
+ $html = file_get_contents(base_path('tests/Fixtures/imdb/shawshank_redemption.html'));
+ $scraper = new ImdbScraper;
+ $data = $scraper->parseForTest($html, '0111161');
+
+ // Assert core movie data
+ $this->assertIsArray($data);
+ $this->assertSame('The Shawshank Redemption', $data['title']);
+ $this->assertStringContainsString('banker convicted', $data['plot']);
+ $this->assertStringContainsString('friendship', $data['plot']);
+ $this->assertSame('9.3', $data['rating']);
+ $this->assertSame('1994', $data['year']);
+
+ // Assert cover image
+ $this->assertStringContainsString('media-amazon.com', $data['cover']);
+ $this->assertStringContainsString('.jpg', $data['cover']);
+
+ // Assert genre
+ $this->assertStringContainsString('Drama', $data['genre']);
+
+ // Assert actors (from JSON-LD)
+ $this->assertIsArray($data['actors']);
+ $this->assertCount(3, $data['actors']);
+ $this->assertContains('Tim Robbins', $data['actors']);
+ $this->assertContains('Morgan Freeman', $data['actors']);
+ $this->assertContains('Bob Gunton', $data['actors']);
+
+ // Assert director
+ $this->assertIsArray($data['director']);
+ $this->assertCount(1, $data['director']);
+ $this->assertContains('Frank Darabont', $data['director']);
+
+ // Assert tagline (from DOM)
+ $this->assertSame('Fear can hold you prisoner. Hope can set you free.', $data['tagline']);
+
+ // Assert language (from DOM)
+ $this->assertSame('English', $data['language']);
+
+ // Assert type
+ $this->assertSame('Movie', $data['type']);
+ }
+
+ public function test_parse_handles_missing_optional_fields(): void
+ {
+ // Create minimal HTML with only required fields
+ $minimalHtml = <<<'HTML'
+
+
+
+
+
+
+
+
+HTML;
+
+ $scraper = new ImdbScraper;
+ $data = $scraper->parseForTest($minimalHtml, '1234567');
+
+ $this->assertIsArray($data);
+ $this->assertSame('Test Movie', $data['title']);
+ $this->assertSame('2020', $data['year']);
+
+ // Optional fields should be empty but not cause errors
+ $this->assertSame('', $data['plot']);
+ $this->assertSame('', $data['rating']);
+ $this->assertSame('', $data['tagline']);
+ $this->assertSame('', $data['language']);
+ $this->assertIsArray($data['actors']);
+ $this->assertEmpty($data['actors']);
+ }
+
+ public function test_fetch_by_id_validates_format(): void
+ {
+ $scraper = new ImdbScraper;
+
+ // Too short
+ $this->assertFalse($scraper->fetchById('1234'));
+
+ // Too long
+ $this->assertFalse($scraper->fetchById('123456789'));
+
+ // Non-numeric
+ $this->assertFalse($scraper->fetchById('abc1234'));
+
+ // Valid format (will fail to fetch, but format is valid)
+ // We can't test actual network call in unit tests
+ }
+
+ public function test_search_returns_empty_for_empty_query(): void
+ {
+ $scraper = new ImdbScraper;
+ $this->assertSame([], $scraper->search(''));
+ $this->assertSame([], $scraper->search(' '));
+ }
+
+ public function test_parse_with_og_meta_fallback(): void
+ {
+ // HTML where JSON-LD is missing but og:title exists
+ $html = <<<'HTML'
+
+
+
+
+
+
+
+
+HTML;
+
+ $scraper = new ImdbScraper;
+ $data = $scraper->parseForTest($html, '9999999');
+
+ $this->assertIsArray($data);
+ $this->assertSame('Fallback Movie Title', $data['title']); // " - IMDb" should be stripped
+ $this->assertSame('https://example.com/poster.jpg', $data['cover']);
+ }
+
+ public function test_parse_extracts_multiple_genres(): void
+ {
+ $html = <<<'HTML'
+
+
+
+
+
+
+
+HTML;
+
+ $scraper = new ImdbScraper;
+ $data = $scraper->parseForTest($html, '1234567');
+
+ $this->assertStringContainsString('Action', $data['genre']);
+ $this->assertStringContainsString('Drama', $data['genre']);
+ $this->assertStringContainsString('Thriller', $data['genre']);
+ }
+}
diff --git a/tests/Unit/ImdbScraperTest.php b/tests/Unit/ImdbScraperTest.php
new file mode 100644
index 000000000..9e3fb8738
--- /dev/null
+++ b/tests/Unit/ImdbScraperTest.php
@@ -0,0 +1,44 @@
+ 'array']);
+ }
+
+ public function test_parse_title_page_from_fixture(): void
+ {
+ $html = file_get_contents(base_path('tests/Fixtures/imdb/example_title.html'));
+ $scraper = new ImdbScraper;
+ $data = $scraper->parseForTest($html, '1234567');
+
+ $this->assertIsArray($data);
+ $this->assertSame('Example Movie', $data['title']);
+ $this->assertNotEmpty($data['plot']); // relaxed assertion
+ $this->assertSame('7.3', $data['rating']);
+ $this->assertSame('2024', $data['year']);
+ $this->assertStringContainsString('Action', $data['genre']);
+ }
+
+ public function test_fetch_by_id_caches_false_on_failure(): void
+ {
+ $scraper = new ImdbScraper;
+ $data = $scraper->fetchById('123'); // invalid length triggers early false
+ $this->assertFalse($data);
+ $data2 = $scraper->fetchById('123');
+ $this->assertFalse($data2);
+ }
+
+ public function test_search_empty_returns_empty_array(): void
+ {
+ $scraper = new ImdbScraper;
+ $this->assertSame([], $scraper->search(''));
+ }
+}