+ prime speed scraper

src/Exceptions/AppraiserException.php

@@ -6,6 +6,6 @@
  * Class AppraiserException
  * @package Vantoozz\ProxyScraper\Exceptions
  */
-class AppraiserException extends ProxyScraperException
+final class AppraiserException extends ProxyScraperException
 {
 }

src/Exceptions/HttpClientException.php

@@ -6,6 +6,6 @@
  * Class HttpClientException
  * @package Vantoozz\ProxyScraper\Exceptions
  */
-class HttpClientException extends ProxyScraperException
+final class HttpClientException extends ProxyScraperException
 {
 }

src/Exceptions/InvalidArgumentException.php

@@ -6,6 +6,6 @@
  * Class InvalidArgumentException
  * @package Vantoozz\ProxyScraper\Exceptions
  */
-class InvalidArgumentException extends ProxyScraperException
+final class InvalidArgumentException extends ProxyScraperException
 {
 }

src/Exceptions/ProxyScraperException.php

@@ -6,6 +6,6 @@
  * Class ProxyScraperException
  * @package Vantoozz\ProxyScraper\Exceptions
  */
-class ProxyScraperException extends \Exception
+abstract class ProxyScraperException extends \Exception
 {
 }

src/Exceptions/RuntimeException.php

@@ -6,6 +6,6 @@
  * Class RuntimeException
  * @package Vantoozz\ProxyScraper\Exceptions
  */
-class RuntimeException extends ProxyScraperException
+final class RuntimeException extends ProxyScraperException
 {
 }

src/Exceptions/ScraperException.php

@@ -6,6 +6,6 @@
  * Class ScraperException
  * @package Vantoozz\ProxyScraper\Exceptions
  */
-class ScraperException extends ProxyScraperException
+final class ScraperException extends ProxyScraperException
 {
 }

src/Exceptions/ValidationException.php

@@ -6,6 +6,6 @@
  * Class ValidationException
  * @package Vantoozz\ProxyScraper\Exceptions
  */
-class ValidationException extends ProxyScraperException
+final class ValidationException extends ProxyScraperException
 {
 }

src/Scrapers/PrimeSpeedScraper.php

@@ -0,0 +1,70 @@
+<?php declare(strict_types = 1);
+
+namespace Vantoozz\ProxyScraper\Scrapers;
+
+use Vantoozz\ProxyScraper\Exceptions\HttpClientException;
+use Vantoozz\ProxyScraper\Exceptions\ScraperException;
+use Vantoozz\ProxyScraper\HttpClient\HttpClientInterface;
+use Vantoozz\ProxyScraper\Proxy;
+
+/**
+ * Class PrimeSpeedScraper
+ * @package Vantoozz\ProxyScraper\Scrapers
+ */
+final class PrimeSpeedScraper implements ScraperInterface
+{
+    private const URL = 'http://www.prime-speed.ru/proxy/free-proxy-list/all-working-proxies.php';
+
+    /**
+     * @var HttpClientInterface
+     */
+    private $httpClient;
+
+    /**
+     * FreeProxyListScraper constructor.
+     * @param HttpClientInterface $httpClient
+     */
+    public function __construct(HttpClientInterface $httpClient)
+    {
+        $this->httpClient = $httpClient;
+    }
+
+    /**
+     * @return \Generator|Proxy[]
+     * @throws \Vantoozz\ProxyScraper\Exceptions\ScraperException
+     */
+    public function get(): \Generator
+    {
+        try {
+            $html = $this->httpClient->get(static::URL);
+        } catch (HttpClientException $e) {
+            throw new ScraperException($e->getMessage(), $e->getCode(), $e);
+        }
+
+        $list = $this->extractList($html);
+
+        yield from (new TextScraper($list))->get();
+    }
+
+    /**
+     * @param string $html
+     * @return string
+     * @throws \Vantoozz\ProxyScraper\Exceptions\ScraperException
+     */
+    private function extractList(string $html): string
+    {
+        $expectedPartsCount = 2;
+
+        $parts = explode("&lt;proxy_server_name&gt; : &lt;proxy_port_number&gt;\n\n0.0.0.0:80\n", $html);
+        if ($expectedPartsCount !== count($parts)) {
+            throw new ScraperException('Unexpected markup');
+        }
+
+        $parts = explode("\n\n\n\n</pre>", $parts[1]);
+        if ($expectedPartsCount !== count($parts)) {
+            throw new ScraperException('Unexpected markup');
+        }
+
+        return $parts[0];
+    }
+}

src/Scrapers/ProxyDbScraper.php

@@ -65,7 +65,7 @@ private function getPage(int $offset, int $pageSize): \Generator
         }
 
         if (!(new Text($html))->isHtml()) {
-            throw new ScraperException($html);
+            throw new ScraperException('Unexpected markup');
         }
 
         $rows = (new Dom($html))->filter('table tbody tr');

tests/integration/Scrapers/PrimeSpeedScraperTest.php

@@ -0,0 +1,25 @@
+<?php declare(strict_types = 1);
+
+namespace Vantoozz\ProxyScraper\IntegrationTests\Scrapers;
+
+use Vantoozz\ProxyScraper\IntegrationTests\IntegrationTest;
+use Vantoozz\ProxyScraper\Scrapers\PrimeSpeedScraper;
+use Vantoozz\ProxyScraper\Scrapers\UsProxyScraper;
+
+/**
+ * Class PrimeSpeedScraperTest
+ * @package Vantoozz\ProxyScraper\Scrapers
+ */
+final class PrimeSpeedScraperTest extends IntegrationTest
+{
+    /**
+     * @test
+     */
+    public function it_works(): void
+    {
+        $scrapper = new PrimeSpeedScraper($this->httpClient());
+
+        $proxies = iterator_to_array($scrapper->get());
+        $this->assertGreaterThanOrEqual(100, count($proxies));
+    }
+}

tests/systemTests.php

@@ -31,6 +31,7 @@
              Scrapers\FreeProxyListScraper::class,
              Scrapers\HideMyIpScraper::class,
              Scrapers\MultiproxyScraper::class,
+             Scrapers\PrimeSpeedScraper::class,
              Scrapers\ProxyDbScraper::class,
              Scrapers\SocksProxyScraper::class,
              Scrapers\SpysMeScraper::class,

tests/unit/Scrapers/PrimeSpeedScraperTest.php

@@ -0,0 +1,128 @@
+<?php declare(strict_types = 1);
+
+namespace Vantoozz\ProxyScraper\UnitTests\Scrapers;
+
+use PHPUnit\Framework\TestCase;
+use Vantoozz\ProxyScraper\Exceptions\HttpClientException;
+use Vantoozz\ProxyScraper\HttpClient\HttpClientInterface;
+use Vantoozz\ProxyScraper\Proxy;
+use Vantoozz\ProxyScraper\Scrapers\FoxToolsScraper;
+use Vantoozz\ProxyScraper\Scrapers\PrimeSpeedScraper;
+
+/**
+ * Class PrimeSpeedScraperTest
+ * @package Vantoozz\ProxyScraper\Scrapers
+ */
+final class PrimeSpeedScraperTest extends TestCase
+{
+    /**
+     * @test
+     * @expectedException \Vantoozz\ProxyScraper\Exceptions\ScraperException
+     * @expectedExceptionMessage error message
+     */
+    public function it_throws_an_exception_on_http_client_error(): void
+    {
+        /** @var HttpClientInterface|\PHPUnit_Framework_MockObject_MockObject $httpClient */
+        $httpClient = $this->createMock(HttpClientInterface::class);
+        $httpClient
+            ->expects(static::once())
+            ->method('get')
+            ->willThrowException(new HttpClientException('error message'));
+
+        $scraper = new PrimeSpeedScraper($httpClient);
+        $scraper->get()->current();
+    }
+
+    /**
+     * @test
+     */
+    public function it_returns_a_proxy(): void
+    {
+
+        $html  = <<<HTML
+<pre>
+format:
+&lt;proxy_server_name&gt; : &lt;proxy_port_number&gt;
+
+0.0.0.0:80
+222.111.222.111:8118
+222.111.222.122:8118
+
+
+
+</pre>
+
+HTML;
+
+
+        /** @var HttpClientInterface|\PHPUnit_Framework_MockObject_MockObject $httpClient */
+        $httpClient = $this->createMock(HttpClientInterface::class);
+        $httpClient
+            ->expects(static::once())
+            ->method('get')
+            ->willReturn($html);
+
+        $scraper = new PrimeSpeedScraper($httpClient);
+        $proxies = iterator_to_array($scraper->get(), false);
+
+        $this->assertInstanceOf(Proxy::class, $proxies[0]);
+        $this->assertSame('222.111.222.111:8118', (string)$proxies[0]);
+    }
+
+
+    /**
+     * @test
+     * @expectedException \Vantoozz\ProxyScraper\Exceptions\ScraperException
+     * @expectedExceptionMessage Unexpected markup
+     */
+    public function it_throws_an_exception_on_unexpected_markup(): void
+    {
+        $html  = <<<HTML
+<pre>
+</pre>
+HTML;
+
+
+        /** @var HttpClientInterface|\PHPUnit_Framework_MockObject_MockObject $httpClient */
+        $httpClient = $this->createMock(HttpClientInterface::class);
+        $httpClient
+            ->expects(static::once())
+            ->method('get')
+            ->willReturn($html);
+
+        $scraper = new PrimeSpeedScraper($httpClient);
+        $scraper->get()->current();
+    }
+
+
+    /**
+     * @test
+     * @expectedException \Vantoozz\ProxyScraper\Exceptions\ScraperException
+     * @expectedExceptionMessage Unexpected markup
+     */
+    public function it_throws_more_exceptions_on_unexpected_markup(): void
+    {
+        $html  = <<<HTML
+<pre>
+format:
+&lt;proxy_server_name&gt; : &lt;proxy_port_number&gt;
+
+0.0.0.0:80
+222.111.222.111:8118
+222.111.222.122:8118
+
+HTML;
+
+
+        /** @var HttpClientInterface|\PHPUnit_Framework_MockObject_MockObject $httpClient */
+        $httpClient = $this->createMock(HttpClientInterface::class);
+        $httpClient
+            ->expects(static::once())
+            ->method('get')
+            ->willReturn($html);
+
+        $scraper = new PrimeSpeedScraper($httpClient);
+        $scraper->get()->current();
+    }
+
+}

tests/unit/Scrapers/ProxyDbScraperTest.php

@@ -34,7 +34,7 @@ public function it_throws_an_exception_on_http_client_error(): void
     /**
      * @test
      * @expectedException \Vantoozz\ProxyScraper\Exceptions\ScraperException
-     * @expectedExceptionMessage some text
+     * @expectedExceptionMessage Unexpected markup
      */
     public function it_throws_an_exception_on_non_html_response(): void
     {