consertado workflow

adicionado withWebDriver no ScraPHPBuilder passado logs para classe scraphp adicionado builder para scraphp
rodrigoaramburu · Dec 17, 2023 · 555dbce · 555dbce
1 parent 9b0242a
commit 555dbce
Show file tree

Hide file tree

Showing 33 changed files with 674 additions and 551 deletions.
diff --git a/.github/workflows/ci.yml → .github/workflows/main.yml b/.github/workflows/ci.yml → .github/workflows/main.yml
@@ -20,12 +20,20 @@ jobs:
 
       - name: Download dependencies
         run: composer update --no-interaction --no-progress 
+
+
+      - uses: isbang/[email protected]
+        with:
+          compose-file: "./docker-compose.yml"
+          down-flags: "--volumes"
+          services: |
+            php
+            selenium
 
-      - name: Start test servers
+      - name: Sleep to wait for Selenium
         shell: bash
         run: |
-          cd tests/fixtures
-          php -S 0.0.0.0:8000 &
+          sleep 30 &
 
       - name: Run tests
         run: ./vendor/bin/pest
diff --git a/README.md b/README.md
@@ -1,5 +1,5 @@
 
-
+![example workflow](https://github.com/rodrigoaramburu/scraphp/actions/workflows/ci.yml/badge.svg)
 
 # Executar Selenium
 

diff --git a/composer.json b/composer.json
@@ -30,5 +30,11 @@
         "allow-plugins": {
             "pestphp/pest-plugin": true
         }
+    },
+    "scripts": {
+        "lint": ["pint"],
+        "test": ["pest"],
+        "test:lint": ["pint --test"],
+        "test:stan": ["phpstan analyse"]
     }
 }
diff --git a/tests/docker-compose.yml → docker-compose.yml b/tests/docker-compose.yml → docker-compose.yml
@@ -8,7 +8,7 @@ services:
     ports:
       - 8000:8000
     volumes:
-      - './test-pages:/application'
+      - './tests/test-pages:/application'
 
   selenium: 
     image: selenium/standalone-chrome:latest

diff --git a/tests/docker/php/Dockerfile → docker/php/Dockerfile b/tests/docker/php/Dockerfile → docker/php/Dockerfile
diff --git a/src/HttpClient/AssetFetcher.php b/src/HttpClient/AssetFetcher.php
@@ -1,25 +1,23 @@
-<?php 
+<?php
+
 declare(strict_types=1);
 
 namespace ScraPHP\HttpClient;
 
-use Psr\Log\LoggerInterface;
 use GuzzleHttp\Exception\ClientException;
 use GuzzleHttp\Exception\ConnectException;
-use ScraPHP\Exceptions\HttpClientException;
 use ScraPHP\Exceptions\AssetNotFoundException;
+use ScraPHP\Exceptions\HttpClientException;
 
 final class AssetFetcher
 {
-
     private \GuzzleHttp\Client $client;
 
-    public function __construct(
-        private LoggerInterface $logger,
-    ){
+    public function __construct()
+    {
         $this->client = new \GuzzleHttp\Client();
     }
-    
+
     /**
      * Fetches an asset from the given URL.
      *
@@ -31,18 +29,15 @@ public function __construct(
     public function fetchAsset(string $url): string
     {
         try {
-            $this->logger->info('Fetching asset '.$url);
             $response = $this->client->request('GET', $url);
-            $this->logger->info('Status: '.$response->getStatusCode().' '.$url);
         } catch (ClientException $e) {
             if ($e->getCode() === 404) {
-                $this->logger->error('404 NOT FOUND '.$url);
                 throw new AssetNotFoundException($url.' not found');
             }
-        } catch(ConnectException $e) {
+        } catch (ConnectException $e) {
             throw new HttpClientException($e->getMessage(), $e->getCode(), $e);
         }
 
         return $response->getBody()->getContents();
     }
-}
+}
diff --git a/src/HttpClient/FilteredElement.php b/src/HttpClient/FilteredElement.php
@@ -1,4 +1,5 @@
-<?php 
+<?php
+
 declare(strict_types=1);
 
 namespace ScraPHP\HttpClient;
@@ -12,5 +13,4 @@ public function attr(string $attr): ?string;
     public function filterCSS(string $cssSelector): ?FilteredElement;
 
     public function filterCSSEach(string $cssSelector, callable $callback): array;
-
 }
diff --git a/src/HttpClient/Guzzle/GuzzleFilteredElement.php b/src/HttpClient/Guzzle/GuzzleFilteredElement.php
@@ -5,12 +5,10 @@
 namespace ScraPHP\HttpClient\Guzzle;
 
 use ScraPHP\HttpClient\FilteredElement;
-use ScraPHP\HttpClient\HtmlElement;
 use Symfony\Component\DomCrawler\Crawler;
 
 final class GuzzleFilteredElement implements FilteredElement
 {
-
     public function __construct(private Crawler $crawler)
     {
     }
@@ -36,7 +34,8 @@ public function attr(string $attr): ?string
         return $this->crawler->attr($attr);
     }
 
-    public function filterCSS(string $cssSelector): ?FilteredElement{
+    public function filterCSS(string $cssSelector): ?FilteredElement
+    {
         $crawler = $this->crawler->filter($cssSelector);
         if ($crawler->count() === 0) {
             return null;
@@ -46,7 +45,7 @@ public function filterCSS(string $cssSelector): ?FilteredElement{
     }
 
     public function filterCSSEach(string $cssSelector, callable $callback): array
-    {     
+    {
         $filter = $this->crawler->filter($cssSelector);
 
         return $filter->each(static function (Crawler $crawler, int $i) use ($callback) {

diff --git a/src/HttpClient/Guzzle/GuzzleHttpClient.php b/src/HttpClient/Guzzle/GuzzleHttpClient.php
@@ -4,33 +4,31 @@
 
 namespace ScraPHP\HttpClient\Guzzle;
 
-use Psr\Log\LoggerInterface;
-use ScraPHP\HttpClient\Page;
-use ScraPHP\HttpClient\HttpClient;
-use ScraPHP\HttpClient\AssetFetcher;
 use GuzzleHttp\Exception\ClientException;
 use GuzzleHttp\Exception\ConnectException;
+use Psr\Log\LoggerInterface;
+use ScraPHP\Exceptions\AssetNotFoundException;
 use ScraPHP\Exceptions\HttpClientException;
 use ScraPHP\Exceptions\UrlNotFoundException;
-use ScraPHP\Exceptions\AssetNotFoundException;
+use ScraPHP\HttpClient\AssetFetcher;
+use ScraPHP\HttpClient\HttpClient;
+use ScraPHP\HttpClient\Page;
 
 final class GuzzleHttpClient implements HttpClient
 {
     private \GuzzleHttp\Client $client;
+
     private AssetFetcher $assetFetcher;
-    
+
     /**
      * Constructor for the class.
      *
-     * @param LoggerInterface $logger The logger instance.
+     * @param  LoggerInterface  $logger The logger instance.
      */
-    public function __construct(
-        private LoggerInterface $logger,
-
-    )
+    public function __construct()
     {
         $this->client = new \GuzzleHttp\Client();
-        $this->assetFetcher = new AssetFetcher($this->logger);
+        $this->assetFetcher = new AssetFetcher();
     }
 
     /**
@@ -45,15 +43,12 @@ public function __construct(
     public function get(string $url): Page
     {
         try {
-            $this->logger->info('Accessing '.$url);
             $response = $this->client->request('GET', $url);
-            $this->logger->info('Status: '.$response->getStatusCode().' '.$url);
         } catch (ClientException $e) {
             if ($e->getCode() === 404) {
-                $this->logger->error('404 NOT FOUND '.$url);
                 throw new UrlNotFoundException($url.' not found');
             }
-        } catch(ConnectException $e) {
+        } catch (ConnectException $e) {
             throw new HttpClientException($e->getMessage(), $e->getCode(), $e);
         }
 
@@ -77,5 +72,4 @@ public function fetchAsset(string $url): string
     {
         return $this->assetFetcher->fetchAsset($url);
     }
-
 }
diff --git a/src/HttpClient/Guzzle/GuzzlePage.php b/src/HttpClient/Guzzle/GuzzlePage.php
@@ -1,11 +1,11 @@
-<?php 
+<?php
 
 declare(strict_types=1);
 
 namespace ScraPHP\HttpClient\Guzzle;
 
-use ScraPHP\HttpClient\Page;
 use ScraPHP\HttpClient\FilteredElement;
+use ScraPHP\HttpClient\Page;
 use Symfony\Component\DomCrawler\Crawler;
 
 final class GuzzlePage implements Page
@@ -15,26 +15,29 @@ public function __construct(
         private int $statusCode,
         private string $content,
         private array $headers
-    )
-    {        
+    ) {
     }
 
     public function statusCode(): int
     {
         return $this->statusCode;
     }
+
     public function url(): string
     {
         return $this->url;
     }
+
     public function htmlBody(): string
     {
         return $this->content;
     }
+
     public function headers(): array
     {
         return $this->headers;
     }
+
     public function header(string $key): array
     {
         return $this->headers[$key] ?? [];
@@ -50,14 +53,15 @@ public function filterCSS(string $cssSelector): ?FilteredElement
 
         return new GuzzleFilteredElement(crawler: $crawler);
     }
+
     public function filterCSSEach(string $cssSelector, callable $callback): array
     {
         $crawler = new Crawler($this->content);
-     
+
         $filter = $crawler->filter($cssSelector);
 
         return $filter->each(static function (Crawler $crawler, int $i) use ($callback) {
             return $callback(new GuzzleFilteredElement(crawler: $crawler), $i);
         });
     }
-}
+}
diff --git a/src/HttpClient/HttpClient.php b/src/HttpClient/HttpClient.php
@@ -4,12 +4,9 @@
 
 namespace ScraPHP\HttpClient;
 
-use ScraPHP\HttpClient\Page;
-
 interface HttpClient
 {
     public function get(string $url): Page;
 
     public function fetchAsset(string $url): string;
-
 }
diff --git a/src/HttpClient/Page.php b/src/HttpClient/Page.php
@@ -1,19 +1,22 @@
-<?php 
+<?php
+
 declare(strict_types=1);
 
 namespace ScraPHP\HttpClient;
 
-use ScraPHP\HttpClient\FilteredElement;
-
 interface Page
 {
     public function statusCode(): int;
+
     public function url(): string;
+
     public function htmlBody(): string;
+
     public function headers(): array;
+
     public function header(string $key): array;
 
     public function filterCSS(string $cssSelector): ?FilteredElement;
-    public function filterCSSEach(string $cssSelector, callable $callback): array;
 
+    public function filterCSSEach(string $cssSelector, callable $callback): array;
 }
diff --git a/src/HttpClient/WebDriver/WebDriverFilteredElement.php b/src/HttpClient/WebDriver/WebDriverFilteredElement.php
@@ -1,20 +1,20 @@
-<?php 
+<?php
 
 declare(strict_types=1);
 
 namespace ScraPHP\HttpClient\WebDriver;
 
+use Facebook\WebDriver\Exception\NoSuchElementException;
+use Facebook\WebDriver\Remote\RemoteWebElement;
 use Facebook\WebDriver\WebDriverBy;
 use ScraPHP\HttpClient\FilteredElement;
-use Facebook\WebDriver\Remote\RemoteWebElement;
-use Facebook\WebDriver\Exception\NoSuchElementException;
 
 final class WebDriverFilteredElement implements FilteredElement
 {
-
     public function __construct(
         private RemoteWebElement $remoteWebElement
-    ){}
+    ) {
+    }
 
     public function text(): string
     {
@@ -28,13 +28,14 @@ public function attr(string $attr): ?string
 
     public function filterCSS(string $cssSelector): ?FilteredElement
     {
-        try{
+        try {
             $remoteWebElement = $this->remoteWebElement->findElement(
                 WebDriverBy::cssSelector($cssSelector)
             );
-        }catch (NoSuchElementException $exception){
+        } catch (NoSuchElementException $exception) {
             return null;
         }
+
         return new WebDriverFilteredElement(
             remoteWebElement: $remoteWebElement
         );
@@ -48,6 +49,7 @@ public function filterCSSEach(string $cssSelector, callable $callback): array
         foreach ($elements as $key => $element) {
             $data[] = $callback(new WebDriverFilteredElement(remoteWebElement: $element), $key);
         }
+
         return $data;
     }
-}
+}