Merge pull request #20 from hyvor/original-url

Original url, single config object
hyvor · Oct 24, 2024 · df48ff0 · df48ff0
2 parents 64f4314 + aa82bd1
commit df48ff0
Show file tree

Hide file tree

Showing 27 changed files with 287 additions and 168 deletions.
diff --git a/src/Embed/Embed.php b/src/Embed/Embed.php
@@ -5,9 +5,9 @@
 use Hyvor\Unfold\Exception\EmbedParserException;
 use Hyvor\Unfold\Exception\EmbedUnableToResolveException;
 use Hyvor\Unfold\Exception\UnfoldException;
-use Hyvor\Unfold\UnfoldCallContext;
 use Hyvor\Unfold\UnfoldConfig;
 use Hyvor\Unfold\Unfolded\Unfolded;
+use Hyvor\Unfold\UnfoldMethod;
 
 class Embed
 {
@@ -16,33 +16,31 @@ class Embed
      */
     public static function getParsers(): array
     {
-        $namespace = __NAMESPACE__.'\\Platforms\\';
+        $namespace = __NAMESPACE__ . '\\Platforms\\';
 
         $parsers = array_map(
-            fn ($file) => $namespace.pathinfo((string) $file, PATHINFO_FILENAME),
-            (array) glob(__DIR__.'/Platforms/*.php')
+            fn($file) => $namespace . pathinfo((string)$file, PATHINFO_FILENAME),
+            (array)glob(__DIR__ . '/Platforms/*.php')
         );
 
-        usort($parsers, fn ($a, $b) => $b::PRIORITY <=> $a::PRIORITY);
+        usort($parsers, fn($a, $b) => $b::PRIORITY <=> $a::PRIORITY);
 
         return $parsers;
     }
 
     /**
      * @throws EmbedParserException
      */
-    public static function parse(
-        string $url,
-        ?UnfoldConfig $config = null,
-    ): EmbedResponseObject {
+    public static function parse(UnfoldConfig $config): EmbedResponseObject
+    {
         foreach (self::getParsers() as $parserClass) {
             /** @var EmbedParserAbstract $parser */
-            $parser = new $parserClass($url, $config);
+            $parser = new $parserClass($config);
             if ($matches = $parser->match()) {
                 return $parser->parse($matches);
             }
         }
-        throw new EmbedUnableToResolveException;
+        throw new EmbedUnableToResolveException();
     }
 
     /**
@@ -52,7 +50,7 @@ public static function getMatchingParser(string $url): ?array
     {
         foreach (self::getParsers() as $parserClass) {
             /** @var EmbedParserAbstract $parser */
-            $parser = new $parserClass($url);
+            $parser = new $parserClass(UnfoldConfig::withUrlAndMethod($url, UnfoldMethod::EMBED));
             if ($parser->match()) {
                 return [
                     'parser' => $parser,
@@ -68,14 +66,13 @@ public static function getMatchingParser(string $url): ?array
      * @throws UnfoldException
      */
     public static function getUnfoldedObject(
-        UnfoldCallContext $context,
+        UnfoldConfig $config,
     ): Unfolded {
-        $oembed = self::parse($context->url, $context->config);
+        $oembed = self::parse($config);
 
         return Unfolded::fromEmbed(
             $oembed,
-            $context->url,
-            $context
+            $config
         );
     }
 }
diff --git a/src/Embed/EmbedParserAbstract.php b/src/Embed/EmbedParserAbstract.php
@@ -19,26 +19,8 @@ abstract class EmbedParserAbstract
     // define priority relatively
     public const PRIORITY = 0;
 
-    protected UnfoldConfig $config;
-
-    public function __construct(
-        protected string $url,
-        ?UnfoldConfig $config = null,
-    ) {
-        $this->config = $config ?? new UnfoldConfig();
-    }
-
-
-    /**
-     * TODO: This is not yet used
-     * If the URL needs to be replaced before sending to the oEmbed endpoint,
-     * return the new URL here. Otherwise, return null
-     * Ex: m.facebook.com -> www.facebook.com
-     * @codeCoverageIgnore
-     */
-    public function replaceUrl(): ?string
+    public function __construct(protected UnfoldConfig $config)
     {
-        return null;
     }
 
     /**
@@ -65,7 +47,7 @@ public function match(): false|array
         $regex = $this->regex();
 
         foreach ($regex as $reg) {
-            if (preg_match("~$reg~", $this->url, $matches)) {
+            if (preg_match("~$reg~", $this->config->url, $matches)) {
                 return $matches;
             }
         }
@@ -108,7 +90,7 @@ private function parseOEmbed(): EmbedResponseObject
         $uri = Uri::withQueryValues(
             new Uri($oEmbedUrl),
             [
-                'url' => $this->url,
+                'url' => $this->config->url,
                 'format' => 'json'
             ]
         );

diff --git a/src/Embed/Platforms/FacebookPage.php b/src/Embed/Platforms/FacebookPage.php
@@ -23,7 +23,7 @@ public function regex()
 
     public function getEmbedHtml(array $matches): string
     {
-        $url = $this->url;
+        $url = $this->config->url;
 
         $name = explode('/', $url)[3] ?? '';
         $sdk = FacebookHelper::sdkScript();

diff --git a/src/Embed/Platforms/FacebookPost.php b/src/Embed/Platforms/FacebookPost.php
@@ -39,7 +39,7 @@ public function regex()
 
     public function getEmbedHtml(array $matches): string
     {
-        $url = $this->url;
+        $url = $this->config->url;
         $sdk = FacebookHelper::sdkScript();
 
         return <<<HTML

diff --git a/src/Embed/Platforms/FacebookVideo.php b/src/Embed/Platforms/FacebookVideo.php
@@ -37,7 +37,7 @@ public function regex()
     public function getEmbedHtml(array $matches): string
     {
         $sdk = FacebookHelper::sdkScript();
-        $url = $this->url;
+        $url = $this->config->url;
         return <<<HTML
 $sdk
 <div class="fb-video" data-href="$url" data-width="500" data-show-text="true"></div>

diff --git a/src/Embed/Platforms/GithubGist.php b/src/Embed/Platforms/GithubGist.php
@@ -16,6 +16,6 @@ public function regex()
 
     public function getEmbedHtml($matches): string
     {
-        return "<script src=\"$this->url.js\"></script>";
+        return "<script src=\"{$this->config->url}.js\"></script>";
     }
 }
diff --git a/src/Embed/Platforms/Youtube.php b/src/Embed/Platforms/Youtube.php
@@ -57,9 +57,9 @@ public function getEmbedHtml($matches): string
     {
         $id = $matches[1] ?? '';
 
-        $isShort = str_contains($this->url, '/shorts/');
-        $isPlaylist = str_contains($this->url, '/playlist?list=');
-        $isUser = str_contains($this->url, '/user/');
+        $isShort = str_contains($this->config->url, '/shorts/');
+        $isPlaylist = str_contains($this->config->url, '/playlist?list=');
+        $isUser = str_contains($this->config->url, '/user/');
 
         $padding = round(100 * ($isShort ? 16 / 9 : 9 / 16), 2);
 

diff --git a/src/Link/Link.php b/src/Link/Link.php
@@ -3,31 +3,39 @@
 namespace Hyvor\Unfold\Link;
 
 use GuzzleHttp\Psr7\Request;
+use Http\Client\Common\Exception\LoopException;
 use Hyvor\Unfold\Exception\LinkScrapeException;
 use Hyvor\Unfold\Link\Metadata\MetadataParser;
-use Hyvor\Unfold\UnfoldCallContext;
 use Hyvor\Unfold\UnfoldConfig;
 use Hyvor\Unfold\Unfolded\Unfolded;
 use Psr\Http\Client\ClientExceptionInterface;
+use Psr\Http\Message\RequestInterface;
 
 class Link
 {
+
+    private ?RequestInterface $lastRequest = null;
+
     public function __construct(
-        private string $url,
         private UnfoldConfig $config,
-    ) {}
+    ) {
+    }
 
     public function scrape(): string
     {
         $request = new Request(
             'GET',
-            $this->url
+            $this->config->url
         );
 
         try {
             $response = $this->config->httpClient->sendRequest($request);
         } catch (ClientExceptionInterface $e) {
-            throw new LinkScrapeException($e->getMessage());
+            $message = match (true) {
+                $e instanceof LoopException => 'Too many redirects',
+                default => $e->getMessage(),
+            };
+            throw new LinkScrapeException($message);
         }
 
         $status = $response->getStatusCode();
@@ -36,19 +44,24 @@ public function scrape(): string
             throw new LinkScrapeException("Unable to scrape link. HTTP status code: $status");
         }
 
+        $this->lastRequest = $this->config->httpRequestRecorder->getLastRequest();
+
         return $response->getBody()->getContents();
     }
 
     public static function getUnfoldedObject(
-        UnfoldCallContext $context,
+        UnfoldConfig $config,
     ): Unfolded {
-        $content = (new Link($context->url, $context->config))->scrape();
-        $metadata = (new MetadataParser($content, $context))->parse();
+        $link = new Link($config);
+        $content = $link->scrape();
+        $lastRequest = $link->lastRequest;
+
+        $metadata = (new MetadataParser($content, $config))->parse();
 
         return Unfolded::fromMetadata(
-            $context->url,
+            $config,
             $metadata,
-            $context,
+            $lastRequest,
         );
     }
 }
diff --git a/src/Link/Metadata/MetadataParser.php b/src/Link/Metadata/MetadataParser.php
@@ -12,7 +12,7 @@
 use Hyvor\Unfold\Link\Metadata\Parsers\OgParser;
 use Hyvor\Unfold\Link\Metadata\Parsers\TitleParser;
 use Hyvor\Unfold\Link\Metadata\Parsers\TwitterParser;
-use Hyvor\Unfold\UnfoldCallContext;
+use Hyvor\Unfold\UnfoldConfig;
 use Symfony\Component\DomCrawler\Crawler;
 
 class MetadataParser
@@ -28,7 +28,7 @@ class MetadataParser
 
     public function __construct(
         private string $content,
-        public UnfoldCallContext $context,
+        public UnfoldConfig $config,
     ) {
         $this->crawler = new Crawler($this->content);
         $this->crawlerMeta = $this->crawler->filterXPath('//meta');

diff --git a/src/Link/Metadata/Parsers/LinkParser.php b/src/Link/Metadata/Parsers/LinkParser.php
@@ -15,11 +15,15 @@ public function add(): void
             $href = $node->attr('href');
 
             if ($rel === 'canonical' && is_string($href)) {
-                $this->parser->addMetadata(new MetadataObject(MetadataKeyType::CANONICAL_URL, $this->fixIfRelativeUrl($href)));
+                $this->parser->addMetadata(
+                    new MetadataObject(MetadataKeyType::CANONICAL_URL, $this->fixIfRelativeUrl($href))
+                );
             }
 
             if ($rel === 'icon' && is_string($href)) {
-                $this->parser->addMetadata(new MetadataObject(MetadataKeyType::FAVICON_URL, $this->fixIfRelativeUrl($href)));
+                $this->parser->addMetadata(
+                    new MetadataObject(MetadataKeyType::FAVICON_URL, $this->fixIfRelativeUrl($href))
+                );
             }
         });
     }
@@ -30,7 +34,7 @@ public function fixIfRelativeUrl(string $url): string
         if (isset($parsedUrl['host'])) {
             return $url;
         } else {
-            return Uri::fromBaseUri($url, $this->parser->context->url)->toString();
+            return Uri::fromBaseUri($url, $this->parser->config->url)->toString();
         }
     }
 }
diff --git a/src/Link/RequestRecorder.php b/src/Link/RequestRecorder.php
@@ -0,0 +1,30 @@
+<?php
+
+namespace Hyvor\Unfold\Link;
+
+use Http\Client\Common\Plugin\Journal;
+use Psr\Http\Client\ClientExceptionInterface;
+use Psr\Http\Message\RequestInterface;
+use Psr\Http\Message\ResponseInterface;
+
+class RequestRecorder implements Journal
+{
+    /** @var RequestInterface[] */
+    private array $requests = [];
+
+    public function addSuccess(RequestInterface $request, ResponseInterface $response): void
+    {
+        $this->requests[] = $request;
+    }
+
+    public function addFailure(RequestInterface $request, ClientExceptionInterface $exception): void
+    {
+    }
+
+    public function getLastRequest(): ?RequestInterface
+    {
+        return count($this->requests) > 0 ?
+            $this->requests[count($this->requests) - 1] :
+            null;
+    }
+}
diff --git a/src/Unfold.php b/src/Unfold.php
@@ -18,16 +18,12 @@ public static function unfold(
         UnfoldConfig $config = null,
     ): Unfolded {
         $config ??= new UnfoldConfig();
-        $context = new UnfoldCallContext(
-            $url,
-            $method,
-            $config,
-        );
+        $config->start($url, $method);
 
         if ($method === UnfoldMethod::LINK) {
-            return Link::getUnfoldedObject($context);
+            return Link::getUnfoldedObject($config);
         } elseif ($method === UnfoldMethod::EMBED) {
-            return Embed::getUnfoldedObject($context);
+            return Embed::getUnfoldedObject($config);
         } else {
             // both
             // TODO:

diff --git a/src/UnfoldCallContext.php b/src/UnfoldCallContext.php