Merge pull request #7 from wdwf/enhancement/cache-and-retryRequest-fu…

…nction

Enhancement/cache and retry request function

.env.dist

@@ -1 +1,2 @@
 GITHUB_TOKEN=
+CACHE_PATH=/tmp/cache

build.php

@@ -29,24 +29,31 @@
         ->addOrganization('laminas', 'laminas')
 );
 
+$curlExec = new \App\CurlExec(
+    cachePath: \getenv('CACHE_PATH') ?: null,
+);
+
 [$latestPhpVersion, $latestPhpVersionReleaseDate] = $run(
     'Última versão do PHP',
     new App\Crawler\LatestPhpVersion(
         'https://www.php.net/releases/active.php',
+        $curlExec,
     )
 );
 
 $phpUsagePercentage = $run(
     'Uso do PHP no w3techs.com',
     new App\Crawler\PhpUsagePercentage(
         'https://w3techs.com/technologies/history_overview/programming_language',
+        $curlExec,
     )
 );
 
 [$wordpressCmsUsagePercentage, $wordpressTotalUsagePercentage] = $run(
     'Uso do WordPress no w3techs.com',
     new App\Crawler\WordpressUsagePercentage(
         'https://w3techs.com/technologies/details/cm-wordpress',
+        $curlExec,
     )
 );
 

composer.json

@@ -17,7 +17,10 @@
     ],
     "minimum-stability": "dev",
     "require": {
-        "php": "^8.3"
+        "php": "^8.3",
+        "ext-curl": "*",
+        "ext-dom": "*",
+        "ext-libxml": "*"
     },
     "require-dev": {
         "squizlabs/php_codesniffer": "4.0.x-dev",

src/Crawler/LatestPhpVersion.php

@@ -7,15 +7,17 @@
 use App\CurlExec;
 use DateTimeImmutable;
 
-class LatestPhpVersion
+readonly class LatestPhpVersion
 {
-    public function __construct(private readonly string $url)
-    {
+    public function __construct(
+        private string $url,
+        private CurlExec $curlExec,
+    ) {
     }
 
     public function __invoke(): ?array
     {
-        $json = (new CurlExec())->fetchAsString($this->url);
+        $json = $this->curlExec->fetchAsString($this->url);
         $json = \json_decode($json, true);
         $latestMajor = end($json);
         $latestMinor = end($latestMajor);

src/Crawler/PhpUsagePercentage.php

@@ -8,15 +8,17 @@
 use DOMDocument;
 use DOMXPath;
 
-class PhpUsagePercentage
+readonly class PhpUsagePercentage
 {
-    public function __construct(private readonly string $url)
-    {
+    public function __construct(
+        private string $url,
+        private CurlExec $curlExec
+    ) {
     }
 
     public function __invoke(): ?float
     {
-        $xpath = (new CurlExec())->fetchAsXpath($this->url);
+        $xpath = $this->curlExec->fetchAsXpath($this->url);
 
         $languages = $xpath->query('//table[@class="hist"]/tr');
         foreach ($languages as $language) {

src/Crawler/WordpressUsagePercentage.php

@@ -8,15 +8,17 @@
 use DOMDocument;
 use DOMXPath;
 
-class WordpressUsagePercentage
+readonly class WordpressUsagePercentage
 {
-    public function __construct(private readonly string $url)
-    {
+    public function __construct(
+        private string $url,
+        private CurlExec $curlExec
+    ) {
     }
 
     public function __invoke(): ?array
     {
-        $xpath = (new CurlExec())->fetchAsXpath($this->url);
+        $xpath = $this->curlExec->fetchAsXpath($this->url);
 
         $languages = $xpath->query('//p[@class="surv"]');
         foreach ($languages as $language) {

src/CurlExec.php

@@ -8,9 +8,58 @@
 use DOMXPath;
 use RuntimeException;
 
-class CurlExec
+readonly class CurlExec
 {
+    public function __construct(
+        private ?string $cachePath = null,
+        private int $maxRetries = 3,
+        private int $retryDelay = 5,
+    ) {
+        if (($cachePath) && (!\is_dir($cachePath))) {
+            \mkdir($cachePath, 0755, true);
+        }
+    }
+
     public function fetchAsString(string $url): string
+    {
+        if ($this->cachePath) {
+            $cacheFile = $this->getCacheFilePath($url);
+            if (\file_exists($cacheFile)) {
+                return \file_get_contents($cacheFile);
+            }
+        }
+
+        $attempt = 0;
+        while ($attempt < $this->maxRetries) {
+            $response = $this->makeRequest($url);
+
+            // Timeout
+            if ($response === false) {
+                \sleep($this->retryDelay);
+                $attempt++;
+                continue;
+            }
+
+            // Resposta verdadeira
+            if (isset($cacheFile)) {
+                \file_put_contents($cacheFile, $response);
+            }
+            return $response;
+        }
+
+        throw new RuntimeException("Falha ao realizar a requisição para URL: {$url}");
+    }
+
+    public function fetchAsXpath(string $url): DOMXPath
+    {
+        $htmlString = $this->fetchAsString($url);
+        \libxml_use_internal_errors(true);
+        $doc = new DOMDocument();
+        $doc->loadHTML($htmlString);
+        return new DOMXPath($doc);
+    }
+
+    protected function makeRequest(string $url): string|false
     {
         $curlHandle = \curl_init();
         \curl_setopt_array($curlHandle, [
@@ -20,19 +69,21 @@ public function fetchAsString(string $url): string
         ]);
 
         $response = \curl_exec($curlHandle);
+
         if ($response === false) {
+            $errorCode = \curl_errno($curlHandle);
+            if ($errorCode === CURLE_OPERATION_TIMEDOUT) {
+                return false;
+            }
             throw new RuntimeException(\curl_error($curlHandle));
         }
 
         return $response;
     }
 
-    public function fetchAsXpath(string $url): DOMXPath
+    private function getCacheFilePath(string $url): string
     {
-        $htmlString = $this->fetchAsString($url);
-        \libxml_use_internal_errors(true);
-        $doc = new DOMDocument();
-        $doc->loadHTML($htmlString);
-        return new DOMXPath($doc);
+        $fileName = \sha1($url) . '.html';
+        return $this->cachePath . DIRECTORY_SEPARATOR . $fileName;
     }
 }