Merge pull request #465 from shikorism/feature/metadata-error

メタデータ取得エラーの記録とリトライ制限の適用

app/Exceptions/Handler.php

@@ -23,6 +23,7 @@ class Handler extends ExceptionHandler
         \Illuminate\Database\Eloquent\ModelNotFoundException::class,
         \Illuminate\Session\TokenMismatchException::class,
         \Illuminate\Validation\ValidationException::class,
+        \App\MetadataResolver\ResolverCircuitBreakException::class,
     ];
 
     /**

app/Metadata.php

@@ -2,6 +2,8 @@
 
 namespace App;
 
+use Carbon\CarbonInterface;
+use GuzzleHttp\Exception\RequestException;
 use Illuminate\Database\Eloquent\Model;
 
 class Metadata extends Model
@@ -13,10 +15,66 @@ class Metadata extends Model
     protected $fillable = ['url', 'title', 'description', 'image', 'expires_at'];
     protected $visible = ['url', 'title', 'description', 'image', 'expires_at', 'tags'];
 
-    protected $dates = ['created_at', 'updated_at', 'expires_at'];
+    protected $dates = ['created_at', 'updated_at', 'expires_at', 'error_at'];
 
     public function tags()
     {
         return $this->belongsToMany(Tag::class)->withTimestamps();
     }
+
+    public function needRefresh(): bool
+    {
+        return $this->isExpired() || $this->error_at !== null;
+    }
+
+    public function isExpired(): bool
+    {
+        return $this->expires_at !== null && $this->expires_at < now();
+    }
+
+    public function storeException(CarbonInterface $error_at, \Exception $exception): self
+    {
+        $this->prepareFieldsOnError();
+        $this->error_at = $error_at;
+        $this->error_exception_class = get_class($exception);
+        $this->error_body = $exception->getMessage();
+        if ($exception instanceof RequestException) {
+            $this->error_http_code = $exception->getCode();
+        } else {
+            $this->error_http_code = null;
+        }
+        $this->error_count++;
+
+        return $this;
+    }
+
+    public function storeError(CarbonInterface $error_at, string $body, ?int $httpCode = null): self
+    {
+        $this->prepareFieldsOnError();
+        $this->error_at = $error_at;
+        $this->error_exception_class = null;
+        $this->error_body = $body;
+        $this->error_http_code = $httpCode;
+        $this->error_count++;
+
+        return $this;
+    }
+
+    public function clearError(): self
+    {
+        $this->error_at = null;
+        $this->error_exception_class = null;
+        $this->error_body = null;
+        $this->error_http_code = null;
+        $this->error_count = 0;
+
+        return $this;
+    }
+
+    private function prepareFieldsOnError()
+    {
+        $this->title = $this->title ?? '';
+        $this->description = $this->description ?? '';
+        $this->image = $this->image ?? '';
+    }
 }

app/MetadataResolver/ResolverCircuitBreakException.php

@@ -0,0 +1,16 @@
+<?php
+
+namespace App\MetadataResolver;
+
+use Throwable;
+
+/**
+ * 規定回数以上の解決失敗により、メタデータの取得が不能となっている場合にスローされます。
+ */
+class ResolverCircuitBreakException extends \RuntimeException
+{
+    public function __construct(int $errorCount, string $url, Throwable $previous = null)
+    {
+        parent::__construct("{$errorCount}回失敗しているためメタデータの取得を中断しました: {$url}", 0, $previous);
+    }
+}

app/MetadataResolver/UncaughtResolverException.php

@@ -0,0 +1,10 @@
+<?php
+
+namespace App\MetadataResolver;
+
+/**
+ * MetadataResolver内で未キャッチの例外が発生した場合にスローされます。
+ */
+class UncaughtResolverException extends \RuntimeException
+{
+}

app/Services/MetadataResolveService.php

@@ -5,14 +5,17 @@ namespace App\Services;
 use App\Metadata;
 use App\MetadataResolver\DeniedHostException;
 use App\MetadataResolver\MetadataResolver;
+use App\MetadataResolver\ResolverCircuitBreakException;
+use App\MetadataResolver\UncaughtResolverException;
 use App\Tag;
 use App\Utilities\Formatter;
-use GuzzleHttp\Exception\TransferException;
 use Illuminate\Support\Facades\DB;
-use Illuminate\Support\Facades\Log;
 
 class MetadataResolveService
 {
+    /** @var int メタデータの解決を中断するエラー回数。この回数以上エラーしていたら処理は行わない。 */
+    const CIRCUIT_BREAK_COUNT = 5;
+
     /** @var MetadataResolver */
     private $resolver;
     /** @var Formatter */
@@ -24,6 +27,13 @@ class MetadataResolveService
         $this->formatter = $formatter;
     }
 
+    /**
+     * メタデータをキャッシュまたはリモートに問い合わせて取得します。
+     * @param string $url メタデータを取得したいURL
+     * @return Metadata 取得できたメタデータ
+     * @throws DeniedHostException アクセス先がブラックリスト入りしているため取得できなかった場合にスロー
+     * @throws UncaughtResolverException Resolver内で例外が発生して取得できなかった場合にスロー
+     */
     public function execute(string $url): Metadata
     {
         // URLの正規化
@@ -34,19 +44,39 @@ class MetadataResolveService
             throw new DeniedHostException($url);
         }
 
-        return DB::transaction(function () use ($url) {
+        DB::beginTransaction();
+        try {
+            $metadata = Metadata::find($url);
+
             // 無かったら取得
             // TODO: ある程度古かったら再取得とかありだと思う
-            $metadata = Metadata::find($url);
+            if ($metadata == null || $metadata->needRefresh()) {
-            if ($metadata == null || ($metadata->expires_at !== null && $metadata->expires_at < now())) {
+                if ($metadata === null) {
+                    $metadata = new Metadata(['url' => $url]);
+                }
+
+                if ($metadata->error_count >= self::CIRCUIT_BREAK_COUNT) {
+                    throw new ResolverCircuitBreakException($metadata->error_count, $url);
+                }
+
                 try {
                     $resolved = $this->resolver->resolve($url);
-                    $metadata = Metadata::updateOrCreate(['url' => $url], [
+                } catch (\Exception $e) {
+                    $metadata->storeException(now(), $e);
+                    $metadata->save();
+                    throw new UncaughtResolverException(implode(': ', [
+                        $metadata->error_count . '回目のメタデータ取得失敗', get_class($e), $e->getMessage()
+                    ]), 0, $e);
+                }
+
+                $metadata->fill([
                     'title' => $resolved->title,
                     'description' => $resolved->description,
                     'image' => $resolved->image,
                     'expires_at' => $resolved->expires_at
                 ]);
+                $metadata->clearError();
+                $metadata->save();
 
                 $tagIds = [];
                 foreach ($resolved->normalizedTags() as $tagName) {
@@ -54,14 +84,18 @@ class MetadataResolveService
                     $tagIds[] = $tag->id;
                 }
                 $metadata->tags()->sync($tagIds);
-                } catch (TransferException $e) {
+            }
-                    // 何らかの通信エラーによってメタデータの取得に失敗した時、とりあえずエラーログにURLを残す
+
-                    Log::error(self::class . ': メタデータの取得に失敗 URL=' . $url);
+            DB::commit();
+
+            return $metadata;
+        } catch (UncaughtResolverException $e) {
+            // Metadataにエラー情報を記録するため
+            DB::commit();
+            throw $e;
+        } catch (\Exception $e) {
+            DB::rollBack();
             throw $e;
         }
     }
-
-            return $metadata;
-        });
-    }
 }

database/migrations/2020_08_10_114944_add_error_data_to_metadata.php

@@ -0,0 +1,36 @@
+<?php
+
+use Illuminate\Database\Migrations\Migration;
+use Illuminate\Database\Schema\Blueprint;
+use Illuminate\Support\Facades\Schema;
+
+class AddErrorDataToMetadata extends Migration
+{
+    /**
+     * Run the migrations.
+     *
+     * @return void
+     */
+    public function up()
+    {
+        Schema::table('metadata', function (Blueprint $table) {
+            $table->timestamp('error_at')->nullable();
+            $table->string('error_exception_class')->nullable();
+            $table->integer('error_http_code')->nullable();
+            $table->text('error_body')->nullable();
+            $table->integer('error_count')->default(0);
+        });
+    }
+
+    /**
+     * Reverse the migrations.
+     *
+     * @return void
+     */
+    public function down()
+    {
+        Schema::table('metadata', function (Blueprint $table) {
+            $table->dropColumn(['error_at', 'error_exception_class', 'error_http_code', 'error_body', 'error_count']);
+        });
+    }
+}

tests/Unit/Services/MetadataResolverServiceTest.php

@@ -0,0 +1,182 @@
+<?php
+
+namespace Tests\Unit\Services;
+
+use App\MetadataResolver\MetadataResolver;
+use App\MetadataResolver\ResolverCircuitBreakException;
+use App\MetadataResolver\UncaughtResolverException;
+use App\Services\MetadataResolveService;
+use Carbon\Carbon;
+use GuzzleHttp\Client;
+use GuzzleHttp\Exception\ClientException;
+use GuzzleHttp\Exception\ServerException;
+use GuzzleHttp\Handler\MockHandler;
+use GuzzleHttp\HandlerStack;
+use GuzzleHttp\Psr7\Response;
+use Illuminate\Foundation\Testing\RefreshDatabase;
+use Mockery\MockInterface;
+use Tests\TestCase;
+
+class MetadataResolverServiceTest extends TestCase
+{
+    use RefreshDatabase;
+
+    protected function setUp(): void
+    {
+        parent::setUp();
+        $this->seed();
+        Carbon::setTestNow('2020-07-21 19:19:19');
+    }
+
+    protected function tearDown(): void
+    {
+        parent::tearDown();
+        Carbon::setTestNow();
+    }
+
+    public function testOnRuntimeException()
+    {
+        $this->mock(MetadataResolver::class, function (MockInterface $mock) {
+            $mock->shouldReceive('resolve')->andReturnUsing(function ($url) {
+                throw new \RuntimeException('Something happened!');
+            });
+        });
+
+        try {
+            $service = app()->make(MetadataResolveService::class);
+            $service->execute('http://example.com');
+        } catch (UncaughtResolverException $e) {
+            $this->assertDatabaseHas('metadata', [
+                'url' => 'http://example.com',
+                'error_at' => new Carbon('2020-07-21 19:19:19'),
+                'error_count' => 1,
+                'error_exception_class' => \RuntimeException::class,
+                'error_http_code' => null,
+                'error_body' => 'Something happened!',
+            ]);
+
+            return;
+        }
+        $this->fail();
+    }
+
+    public function testOnHttpClientError()
+    {
+        $handler = HandlerStack::create(new MockHandler([new Response(404)]));
+        $client = new Client(['handler' => $handler]);
+        $this->instance(Client::class, $client);
+
+        try {
+            $service = app()->make(MetadataResolveService::class);
+            $service->execute('http://example.com');
+        } catch (UncaughtResolverException $e) {
+            $this->assertDatabaseHas('metadata', [
+                'url' => 'http://example.com',
+                'error_at' => new Carbon('2020-07-21 19:19:19'),
+                'error_count' => 1,
+                'error_exception_class' => ClientException::class,
+                'error_http_code' => 404,
+            ]);
+
+            return;
+        }
+        $this->fail();
+    }
+
+    public function testOnHttpServerError()
+    {
+        $handler = HandlerStack::create(new MockHandler([new Response(503), new Response(503)]));
+        $client = new Client(['handler' => $handler]);
+        $this->instance(Client::class, $client);
+
+        try {
+            $service = app()->make(MetadataResolveService::class);
+            $service->execute('http://example.com');
+        } catch (UncaughtResolverException $e) {
+            $this->assertDatabaseHas('metadata', [
+                'url' => 'http://example.com',
+                'error_at' => new Carbon('2020-07-21 19:19:19'),
+                'error_count' => 1,
+                'error_exception_class' => ServerException::class,
+                'error_http_code' => 503,
+            ]);
+
+            return;
+        }
+        $this->fail();
+    }
+
+    public function testCircuitBreak()
+    {
+        $this->mock(MetadataResolver::class, function (MockInterface $mock) {
+            $mock->shouldReceive('resolve')->andReturnUsing(function ($url) {
+                throw new \RuntimeException('Something happened!');
+            });
+        });
+
+        try {
+            for ($i = 0; $i < 6; $i++) {
+                try {
+                    $service = app()->make(MetadataResolveService::class);
+                    $service->execute('http://example.com');
+                } catch (UncaughtResolverException $e) {
+                }
+            }
+        } catch (ResolverCircuitBreakException $e) {
+            $this->assertDatabaseHas('metadata', [
+                'url' => 'http://example.com',
+                'error_at' => new Carbon('2020-07-21 19:19:19'),
+                'error_count' => 5,
+                'error_exception_class' => \RuntimeException::class,
+                'error_http_code' => null,
+                'error_body' => 'Something happened!',
+            ]);
+
+            return;
+        }
+        $this->fail();
+    }
+
+    public function testOnResurrect()
+    {
+        $successBody = <<<HTML
+<!doctype html>
+<html lang="ja">
+<head>
+<meta charset="UTF-8">
+    <meta name="og:title" content="OGP Title">
+    <meta name="og:description" content="OGP Description">
+    <title>Test Document</title>
+</head>
+<body>
+</body>
+</html>
+HTML;
+        $handler = HandlerStack::create(new MockHandler([
+            new Response(404),
+            new Response(200, ['Content-Type' => 'text/html'], $successBody),
+        ]));
+        $client = new Client(['handler' => $handler]);
+        $this->instance(Client::class, $client);
+
+        for ($i = 0; $i < 2; $i++) {
+            try {
+                $service = app()->make(MetadataResolveService::class);
+                $service->execute('http://example.com');
+            } catch (UncaughtResolverException $e) {
+            }
+        }
+
+        $this->assertDatabaseHas('metadata', [
+            'url' => 'http://example.com',
+            'title' => 'OGP Title',
+            'description' => 'OGP Description',
+            'image' => '',
+            'error_at' => null,
+            'error_count' => 0,
+            'error_exception_class' => null,
+            'error_http_code' => null,
+            'error_body' => null,
+        ]);
+    }
+}