Merge pull request #5 from grgk/Parser-optimization

added option of injecting a custom parser,
grgk · Oct 23, 2019 · a52ec60 · a52ec60
2 parents d2660bc + a3083dc
commit a52ec60
Show file tree

Hide file tree

Showing 15 changed files with 356 additions and 203 deletions.
diff --git a/composer.json b/composer.json
@@ -14,7 +14,8 @@
     "psr/simple-cache": "^1.0"
   },
   "require-dev": {
-    "phpunit/phpunit": "^7"
+    "phpunit/phpunit": "^7",
+    "ext-json": "*"
   },
   "autoload": {
     "psr-4": {

diff --git a/examples/analyze_file_with_custom_parser.php b/examples/analyze_file_with_custom_parser.php
@@ -0,0 +1,29 @@
+<?php
+/**
+ * File analyze example
+ *
+ * Executes file seo analyze on local html file using custom parser.
+ */
+
+require_once(__DIR__ . '/../vendor/autoload.php');
+
+use SeoAnalyzer\Analyzer;
+use SeoAnalyzer\Factor;
+use SeoAnalyzer\HttpClient\Exception\HttpException;
+use SeoAnalyzer\Page;
+use SeoAnalyzer\Parser\ExampleCustomParser;
+
+try {
+    $page = new Page('https://www.msn.com/pl-pl');
+    $parser = new ExampleCustomParser();
+    $page->setParser($parser);
+    $analyzer = new Analyzer($page);
+    $analyzer->metrics = $page->setMetrics([Factor::ALTS]);
+    $results = $analyzer->analyze();
+} catch (HttpException $e) {
+    echo "Error loading page: " . $e->getMessage();
+} catch (ReflectionException $e) {
+    echo "Error loading metric file: " . $e->getMessage();
+}
+
+print_r($results);
diff --git a/src/Metric/MetricFactory.php b/src/Metric/MetricFactory.php
@@ -2,12 +2,15 @@
 
 namespace SeoAnalyzer\Metric;
 
+use ReflectionException;
+
 class MetricFactory
 {
     /**
      * @param string $key
      * @param null $inputData
      * @return mixed
+     * @throws ReflectionException
      */
     public static function get(string $key, $inputData = null)
     {
@@ -24,6 +27,6 @@ public static function get(string $key, $inputData = null)
             }
             return $metric;
         }
-        throw new \ReflectionException('Metric class ' . $class .' not exists');
+        throw new ReflectionException('Metric class ' . $class .' not exists');
     }
 }
diff --git a/src/Metric/Page/SizeMetric.php b/src/Metric/Page/SizeMetric.php
diff --git a/src/Page.php b/src/Page.php
@@ -7,6 +7,8 @@
 use SeoAnalyzer\HttpClient\Exception\HttpException;
 use SeoAnalyzer\Metric\MetricFactory;
 use ReflectionException;
+use SeoAnalyzer\Parser\Parser;
+use SeoAnalyzer\Parser\ParserInterface;
 
 class Page
 {
@@ -52,19 +54,33 @@ class Page
      */
     public $client;
 
+    /**
+     * @var ParserInterface
+     */
+    public $parser;
+
     /**
      * Page constructor.
      *
      * @param string|null $url
      * @param string|null $locale
      * @param ClientInterface|null $client
+     * @param ParserInterface|null $parser
      */
-    public function __construct(string $url = null, string $locale = null, ClientInterface $client = null)
-    {
+    public function __construct(
+        string $url = null,
+        string $locale = null,
+        ClientInterface $client = null,
+        ParserInterface $parser = null
+    ) {
         $this->client = $client;
         if (empty($client)) {
             $this->client = new Client();
         }
+        $this->parser = $parser;
+        if (empty($parser)) {
+            $this->parser = new Parser();
+        }
         if (!empty($url)) {
             $this->url = $this->setUpUrl($url);
             $this->getContent();
@@ -74,6 +90,26 @@ public function __construct(string $url = null, string $locale = null, ClientInt
         }
     }
 
+    /**
+     * Sets custom Http Client.
+     *
+     * @param ClientInterface $client
+     */
+    public function setClient(ClientInterface $client): void
+    {
+        $this->client = $client;
+    }
+
+    /**
+     * Sets custom Html Parser.
+     *
+     * @param ParserInterface $parser
+     */
+    public function setParser(ParserInterface $parser): void
+    {
+        $this->parser = $parser;
+    }
+
     /**
      * Verifies URL and sets up some basic metrics.
      *
@@ -180,13 +216,16 @@ protected function getHttpClientOptions()
      */
     public function parse()
     {
-        $parser = new Parser($this->content);
+        if (empty($this->content)) {
+            $this->getContent();
+        }
+        $this->parser->setContent($this->content);
         $this->setFactors([
-            Factor::META_META => $parser->getMeta(),
-            Factor::HEADERS => $parser->getHeaders(),
-            Factor::META_TITLE => $parser->getTitle(),
-            Factor::TEXT => $parser->getText(),
-            Factor::ALTS => $parser->getAlts()
+            Factor::META_META => $this->parser->getMeta(),
+            Factor::HEADERS => $this->parser->getHeaders(),
+            Factor::META_TITLE => $this->parser->getTitle(),
+            Factor::TEXT => $this->parser->getText(),
+            Factor::ALTS => $this->parser->getAlts()
         ]);
     }
 
@@ -216,9 +255,6 @@ public function setMetrics(array $config)
 
     private function initializeFactors()
     {
-        if (empty($this->content)) {
-            $this->getContent();
-        }
         if (empty($this->dom)) {
             $this->parse();
         }

diff --git a/src/Parser.php b/src/Parser.php
diff --git a/src/Parser/AbstractParser.php b/src/Parser/AbstractParser.php
@@ -0,0 +1,63 @@
+<?php
+
+namespace SeoAnalyzer\Parser;
+
+use DOMDocument;
+use DOMElement;
+use DOMNodeList;
+
+abstract class AbstractParser implements ParserInterface
+{
+    /**
+     * @var DOMDocument Dom representation of HTML document
+     */
+    protected $dom;
+
+    /**
+     * @param string $html Html document to parse.
+     */
+    public function __construct(string $html = null)
+    {
+        $this->dom = new DOMDocument();
+        if (!empty($html)) {
+            $this->setContent($html);
+        }
+    }
+
+    /**
+     * @inheritDoc
+     */
+    public function setContent($html): void
+    {
+        $internalErrors = libxml_use_internal_errors(true);
+        $this->dom->loadHTML($html, LIBXML_NOWARNING);
+        libxml_use_internal_errors($internalErrors);
+    }
+
+    /**
+     * Removes specified tags with it's content from DOM.
+     *
+     * @param string $tag
+     */
+    protected function removeTags(string $tag)
+    {
+        $tagsToRemove = [];
+        foreach ($this->getDomElements($tag) as $tag) {
+            $tagsToRemove[] = $tag;
+        }
+        foreach ($tagsToRemove as $item) {
+            $item->parentNode->removeChild($item);
+        }
+    }
+
+    /**
+     * Returns DOM elements by tag name.
+     *
+     * @param string $name
+     * @return DOMNodeList|DOMElement[]
+     */
+    protected function getDomElements(string $name): DOMNodeList
+    {
+        return $this->dom->getElementsByTagName($name);
+    }
+}