diff --git a/app/MetadataResolver/ActivityPubResolver.php b/app/MetadataResolver/ActivityPubResolver.php
new file mode 100644
index 0000000..f790f02
--- /dev/null
+++ b/app/MetadataResolver/ActivityPubResolver.php
@@ -0,0 +1,71 @@
+activityClient = new \GuzzleHttp\Client([
+ 'headers' => [
+ 'Accept' => 'application/activity+json, application/ld+json; profile="https://www.w3.org/ns/activitystreams"'
+ ]
+ ]);
+ }
+
+ public function resolve(string $url): Metadata
+ {
+ $res = $this->activityClient->get($url);
+ if ($res->getStatusCode() === 200) {
+ return $this->parse($res->getBody());
+ } else {
+ throw new \RuntimeException("{$res->getStatusCode()}: $url");
+ }
+ }
+
+ public function parse(string $json): Metadata
+ {
+ $activityOrObject = json_decode($json, true);
+ $object = $activityOrObject['object'] ?? $activityOrObject;
+
+ $metadata = new Metadata();
+
+ $metadata->title = isset($object['attributedTo']) ? $this->getTitleFromActor($object['attributedTo']) : '';
+ $metadata->description = isset($object['content']) ? $this->html2text($object['content']) : '';
+ $metadata->image = $object['attachment'][0]['url'] ?? '';
+
+ return $metadata;
+ }
+
+ private function getTitleFromActor(string $url): string
+ {
+ $res = $this->activityClient->get($url);
+ if ($res->getStatusCode() !== 200) {
+ return '';
+ }
+
+ $actor = json_decode($res->getBody(), true);
+ $title = $actor['name'] ?? '';
+ if (isset($actor['preferredUsername'])) {
+ $title .= ' (@' . $actor['preferredUsername'] . '@' . parse_url($actor['id'], PHP_URL_HOST) . ')';
+ }
+
+ return $title;
+ }
+
+ private function html2text(string $html): string
+ {
+ $html = mb_convert_encoding($html, 'HTML-ENTITIES', 'UTF-8');
+ $html = preg_replace('~
|
]*>~i', "\n", $html); + $dom = new \DOMDocument(); + $dom->loadHTML($html, LIBXML_HTML_NOIMPLIED | LIBXML_HTML_NODEFDTD); + return $dom->textContent; + } +} diff --git a/app/MetadataResolver/MetadataResolver.php b/app/MetadataResolver/MetadataResolver.php index 94f2156..8312af9 100644 --- a/app/MetadataResolver/MetadataResolver.php +++ b/app/MetadataResolver/MetadataResolver.php @@ -18,7 +18,12 @@ class MetadataResolver implements Resolver '~www\.patreon\.com/~' => PatreonResolver::class, '~www\.deviantart\.com/.*/art/.*~' => DeviantArtResolver::class, '~\.syosetu\.com/n\d+[a-z]{2,}~' => NarouResolver::class, - '/.*/' => OGPResolver::class + ]; + + public $mimeTypes = [ + 'application/activity+json' => ActivityPubResolver::class, + 'application/ld+json' => ActivityPubResolver::class, + 'text/html' => OGPResolver::class ]; public function resolve(string $url): Metadata @@ -31,6 +36,27 @@ class MetadataResolver implements Resolver } } - throw new \UnexpectedValueException('URL not matched.'); + $client = new \GuzzleHttp\Client(); + $res = $client->request('GET', $url, [ + 'headers' => [ + 'Accept' => implode(', ', array_keys($this->mimeTypes)) + ] + ]); + + if ($res->getStatusCode() === 200) { + preg_match('/^[^;\s]+/', $res->getHeaderLine('Content-Type'), $matches); + $mimeType = $matches[0]; + + if (isset($this->mimeTypes[$mimeType])) { + $class = $this->mimeTypes[$mimeType]; + $parser = new $class(); + + return $parser->parse($res->getBody()); + } else { + throw new \UnexpectedValueException('URL not matched.'); + } + } else { + throw new \RuntimeException("{$res->getStatusCode()}: $url"); + } } } diff --git a/app/MetadataResolver/OGPResolver.php b/app/MetadataResolver/OGPResolver.php index 5afe83c..3002d91 100644 --- a/app/MetadataResolver/OGPResolver.php +++ b/app/MetadataResolver/OGPResolver.php @@ -2,7 +2,7 @@ namespace App\MetadataResolver; -class OGPResolver implements Resolver +class OGPResolver implements Resolver, Parser { public function resolve(string $url): Metadata { diff --git a/app/MetadataResolver/Parser.php b/app/MetadataResolver/Parser.php new file mode 100644 index 0000000..f9effde --- /dev/null +++ b/app/MetadataResolver/Parser.php @@ -0,0 +1,8 @@ +