From d27fb225c2fadb454a4bb19394798a096a72aa1b Mon Sep 17 00:00:00 2001
From: --system <info@online-gamer.nl>
Date: Mon, 6 Jan 2025 22:25:55 +0100
Subject: [PATCH] [TASK] Split up ContentParser to multiple classes and
 interfaces

---
 .../ContentExtractors/BaseUrlParser.php       |  19 +++
 .../BaseUrlParserInterface.php                |  10 ++
 .../ContentExtractors/BodyProcessor.php       |  52 ++++++
 .../BodyProcessorInterface.php                |  10 ++
 .../ContentMetadataExtractor.php              |  48 ++++++
 .../ContentMetadataExtractorInterface.php     |  12 ++
 .../ContentExtractors/FaviconExtractor.php    |  22 +++
 .../FaviconExtractorInterface.php             |  10 ++
 .../TitleConfigurationExtractor.php           |  24 +++
 .../TitleConfigurationExtractorInterface.php  |  13 ++
 Classes/Service/Preview/ContentParser.php     | 153 +++---------------
 Configuration/Services.yaml                   |   9 ++
 12 files changed, 250 insertions(+), 132 deletions(-)
 create mode 100644 Classes/Service/Preview/ContentExtractors/BaseUrlParser.php
 create mode 100644 Classes/Service/Preview/ContentExtractors/BaseUrlParserInterface.php
 create mode 100644 Classes/Service/Preview/ContentExtractors/BodyProcessor.php
 create mode 100644 Classes/Service/Preview/ContentExtractors/BodyProcessorInterface.php
 create mode 100644 Classes/Service/Preview/ContentExtractors/ContentMetadataExtractor.php
 create mode 100644 Classes/Service/Preview/ContentExtractors/ContentMetadataExtractorInterface.php
 create mode 100644 Classes/Service/Preview/ContentExtractors/FaviconExtractor.php
 create mode 100644 Classes/Service/Preview/ContentExtractors/FaviconExtractorInterface.php
 create mode 100644 Classes/Service/Preview/ContentExtractors/TitleConfigurationExtractor.php
 create mode 100644 Classes/Service/Preview/ContentExtractors/TitleConfigurationExtractorInterface.php

diff --git a/Classes/Service/Preview/ContentExtractors/BaseUrlParser.php b/Classes/Service/Preview/ContentExtractors/BaseUrlParser.php
new file mode 100644
index 00000000..983548f0
--- /dev/null
+++ b/Classes/Service/Preview/ContentExtractors/BaseUrlParser.php
@@ -0,0 +1,19 @@
+<?php
+
+declare(strict_types=1);
+
+namespace YoastSeoForTypo3\YoastSeo\Service\Preview\ContentExtractors;
+
+class BaseUrlParser implements BaseUrlParserInterface
+{
+    public function getBaseUrl(mixed $urlParts): string
+    {
+        if (!is_array($urlParts)) {
+            return '://';
+        }
+        if ($urlParts['port'] ?? false) {
+            return (isset($urlParts['scheme']) ? $urlParts['scheme'] . ':' : '') . '//' . ($urlParts['host'] ?? '') . ':' . $urlParts['port'];
+        }
+        return (isset($urlParts['scheme']) ? $urlParts['scheme'] . ':' : '') . '//' . ($urlParts['host'] ?? '');
+    }
+}
diff --git a/Classes/Service/Preview/ContentExtractors/BaseUrlParserInterface.php b/Classes/Service/Preview/ContentExtractors/BaseUrlParserInterface.php
new file mode 100644
index 00000000..7615e754
--- /dev/null
+++ b/Classes/Service/Preview/ContentExtractors/BaseUrlParserInterface.php
@@ -0,0 +1,10 @@
+<?php
+
+declare(strict_types=1);
+
+namespace YoastSeoForTypo3\YoastSeo\Service\Preview\ContentExtractors;
+
+interface BaseUrlParserInterface
+{
+    public function getBaseUrl(mixed $urlParts): string;
+}
diff --git a/Classes/Service/Preview/ContentExtractors/BodyProcessor.php b/Classes/Service/Preview/ContentExtractors/BodyProcessor.php
new file mode 100644
index 00000000..706edef6
--- /dev/null
+++ b/Classes/Service/Preview/ContentExtractors/BodyProcessor.php
@@ -0,0 +1,52 @@
+<?php
+
+declare(strict_types=1);
+
+namespace YoastSeoForTypo3\YoastSeo\Service\Preview\ContentExtractors;
+
+class BodyProcessor implements BodyProcessorInterface
+{
+    public function getBody(string $content): string
+    {
+        $body = '';
+
+        $bodyFound = preg_match("/<body[^>]*>(.*)<\/body>/is", $content, $matchesBody);
+
+        if ($bodyFound) {
+            $body = $matchesBody[1];
+
+            preg_match_all(
+                '/<!--\s*?TYPO3SEARCH_begin\s*?-->.*?<!--\s*?TYPO3SEARCH_end\s*?-->/mis',
+                $body,
+                $indexableContents
+            );
+
+            if (is_array($indexableContents[0]) && !empty($indexableContents[0])) {
+                $body = implode('', $indexableContents[0]);
+            }
+        }
+
+        return $this->prepareBody($body);
+    }
+
+    protected function prepareBody(string $body): string
+    {
+        $body = $this->stripTagsContent($body, '<script><noscript>');
+        $body = preg_replace(['/\s?\n\s?/', '/\s{2,}/'], [' ', ' '], $body);
+        $body = strip_tags((string)$body, '<h1><h2><h3><h4><h5><p><a><img>');
+
+        return trim($body);
+    }
+
+    protected function stripTagsContent(string $text, string $tags = ''): string
+    {
+        preg_match_all('/<(.+?)[\s]*\/?[\s]*>/si', trim($tags), $foundTags);
+        $tagsArray = array_unique($foundTags[1]);
+
+        if (count($tagsArray) > 0) {
+            return (string)preg_replace('@<(' . implode('|', $tagsArray) . ')\b.*?>.*?</\1>@si', '', $text);
+        }
+
+        return $text;
+    }
+}
diff --git a/Classes/Service/Preview/ContentExtractors/BodyProcessorInterface.php b/Classes/Service/Preview/ContentExtractors/BodyProcessorInterface.php
new file mode 100644
index 00000000..46d456eb
--- /dev/null
+++ b/Classes/Service/Preview/ContentExtractors/BodyProcessorInterface.php
@@ -0,0 +1,10 @@
+<?php
+
+declare(strict_types=1);
+
+namespace YoastSeoForTypo3\YoastSeo\Service\Preview\ContentExtractors;
+
+interface BodyProcessorInterface
+{
+    public function getBody(string $content): string;
+}
diff --git a/Classes/Service/Preview/ContentExtractors/ContentMetadataExtractor.php b/Classes/Service/Preview/ContentExtractors/ContentMetadataExtractor.php
new file mode 100644
index 00000000..3872f00f
--- /dev/null
+++ b/Classes/Service/Preview/ContentExtractors/ContentMetadataExtractor.php
@@ -0,0 +1,48 @@
+<?php
+
+declare(strict_types=1);
+
+namespace YoastSeoForTypo3\YoastSeo\Service\Preview\ContentExtractors;
+
+class ContentMetadataExtractor implements ContentMetadataExtractorInterface
+{
+    public function getTitle(string $content): string
+    {
+        $title = '';
+        $titleFound = preg_match("/<title[^>]*>(.*?)<\/title>/is", $content, $matchesTitle);
+
+        if ($titleFound) {
+            $title = $matchesTitle[1];
+        }
+
+        return strip_tags(html_entity_decode($title));
+    }
+
+    public function getDescription(string $content): string
+    {
+        $metaDescription = '';
+        $descriptionFound = preg_match(
+            "/<meta[^>]*name=[\" | \']description[\"|\'][^>]*content=[\"]([^\"]*)[\"][^>]*>/i",
+            $content,
+            $matchesDescription
+        );
+
+        if ($descriptionFound) {
+            $metaDescription = $matchesDescription[1];
+        }
+
+        return strip_tags(html_entity_decode($metaDescription));
+    }
+
+    public function getLocale(string $content): string
+    {
+        $locale = 'en';
+        $localeFound = preg_match('/<html[^>]*lang="([a-z\-A-Z]*)"/is', $content, $matchesLocale);
+
+        if ($localeFound) {
+            [$locale] = explode('-', trim($matchesLocale[1]));
+        }
+
+        return $locale;
+    }
+}
diff --git a/Classes/Service/Preview/ContentExtractors/ContentMetadataExtractorInterface.php b/Classes/Service/Preview/ContentExtractors/ContentMetadataExtractorInterface.php
new file mode 100644
index 00000000..990842fb
--- /dev/null
+++ b/Classes/Service/Preview/ContentExtractors/ContentMetadataExtractorInterface.php
@@ -0,0 +1,12 @@
+<?php
+
+declare(strict_types=1);
+
+namespace YoastSeoForTypo3\YoastSeo\Service\Preview\ContentExtractors;
+
+interface ContentMetadataExtractorInterface
+{
+    public function getTitle(string $content): string;
+    public function getDescription(string $content): string;
+    public function getLocale(string $content): string;
+}
diff --git a/Classes/Service/Preview/ContentExtractors/FaviconExtractor.php b/Classes/Service/Preview/ContentExtractors/FaviconExtractor.php
new file mode 100644
index 00000000..12617a92
--- /dev/null
+++ b/Classes/Service/Preview/ContentExtractors/FaviconExtractor.php
@@ -0,0 +1,22 @@
+<?php
+
+declare(strict_types=1);
+
+namespace YoastSeoForTypo3\YoastSeo\Service\Preview\ContentExtractors;
+
+class FaviconExtractor implements FaviconExtractorInterface
+{
+    public function getFaviconSrc(string $baseUrl, string $content): string
+    {
+        $faviconSrc = $baseUrl . '/favicon.ico';
+        $favIconFound = preg_match('/<link rel=\"shortcut icon\" href=\"([^"]*)\"/i', $content, $matchesFavIcon);
+        if ($favIconFound) {
+            $faviconSrc = str_contains($matchesFavIcon[1], '://') ? $matchesFavIcon[1] : $baseUrl . $matchesFavIcon[1];
+        }
+        $favIconHeader = @get_headers($faviconSrc);
+        if (($favIconHeader[0] ?? '') === 'HTTP/1.1 404 Not Found') {
+            $faviconSrc = '';
+        }
+        return $faviconSrc;
+    }
+}
diff --git a/Classes/Service/Preview/ContentExtractors/FaviconExtractorInterface.php b/Classes/Service/Preview/ContentExtractors/FaviconExtractorInterface.php
new file mode 100644
index 00000000..aa9a9063
--- /dev/null
+++ b/Classes/Service/Preview/ContentExtractors/FaviconExtractorInterface.php
@@ -0,0 +1,10 @@
+<?php
+
+declare(strict_types=1);
+
+namespace YoastSeoForTypo3\YoastSeo\Service\Preview\ContentExtractors;
+
+interface FaviconExtractorInterface
+{
+    public function getFaviconSrc(string $baseUrl, string $content): string;
+}
diff --git a/Classes/Service/Preview/ContentExtractors/TitleConfigurationExtractor.php b/Classes/Service/Preview/ContentExtractors/TitleConfigurationExtractor.php
new file mode 100644
index 00000000..2674e4a9
--- /dev/null
+++ b/Classes/Service/Preview/ContentExtractors/TitleConfigurationExtractor.php
@@ -0,0 +1,24 @@
+<?php
+
+declare(strict_types=1);
+
+namespace YoastSeoForTypo3\YoastSeo\Service\Preview\ContentExtractors;
+
+class TitleConfigurationExtractor implements TitleConfigurationExtractorInterface
+{
+    /**
+     * @return array{titlePrepend: string, titleAppend: string}
+     */
+    public function getTitleConfiguration(string $content): array
+    {
+        $prepend = $append = '';
+        preg_match('/<meta name=\"x-yoast-title-config\" value=\"([^"]*)\"/i', $content, $matchesTitleConfig);
+        if (count($matchesTitleConfig) > 1) {
+            [$prepend, $append] = explode('|||', (string)$matchesTitleConfig[1]);
+        }
+        return [
+            'titlePrepend' => $prepend,
+            'titleAppend' => $append,
+        ];
+    }
+}
diff --git a/Classes/Service/Preview/ContentExtractors/TitleConfigurationExtractorInterface.php b/Classes/Service/Preview/ContentExtractors/TitleConfigurationExtractorInterface.php
new file mode 100644
index 00000000..3e0cb522
--- /dev/null
+++ b/Classes/Service/Preview/ContentExtractors/TitleConfigurationExtractorInterface.php
@@ -0,0 +1,13 @@
+<?php
+
+declare(strict_types=1);
+
+namespace YoastSeoForTypo3\YoastSeo\Service\Preview\ContentExtractors;
+
+interface TitleConfigurationExtractorInterface
+{
+    /**
+     * @return array{titlePrepend: string, titleAppend: string}
+     */
+    public function getTitleConfiguration(string $content): array;
+}
diff --git a/Classes/Service/Preview/ContentParser.php b/Classes/Service/Preview/ContentParser.php
index 72093063..7a7679e7 100644
--- a/Classes/Service/Preview/ContentParser.php
+++ b/Classes/Service/Preview/ContentParser.php
@@ -4,156 +4,45 @@
 
 namespace YoastSeoForTypo3\YoastSeo\Service\Preview;
 
+use YoastSeoForTypo3\YoastSeo\Service\Preview\ContentExtractors\BaseUrlParserInterface;
+use YoastSeoForTypo3\YoastSeo\Service\Preview\ContentExtractors\BodyProcessorInterface;
+use YoastSeoForTypo3\YoastSeo\Service\Preview\ContentExtractors\ContentMetadataExtractorInterface;
+use YoastSeoForTypo3\YoastSeo\Service\Preview\ContentExtractors\FaviconExtractorInterface;
+use YoastSeoForTypo3\YoastSeo\Service\Preview\ContentExtractors\TitleConfigurationExtractorInterface;
+
 class ContentParser
 {
+    public function __construct(
+        protected BaseUrlParserInterface $baseUrlParser,
+        protected BodyProcessorInterface $bodyProcessor,
+        protected ContentMetadataExtractorInterface $contentMetadataExtractor,
+        protected FaviconExtractorInterface $faviconExtractor,
+        protected TitleConfigurationExtractorInterface $titleConfigurationExtractor,
+    ) {}
+
     /**
      * @return array<string, int|string>
      */
     public function parse(string $content, string $uriToCheck, int $pageId): array
     {
         $urlParts = parse_url((string)preg_replace('/\/$/', '', $uriToCheck));
-        $baseUrl = $this->getBaseUrl($urlParts);
+        $baseUrl = $this->baseUrlParser->getBaseUrl($urlParts);
         $url = $baseUrl . ($urlParts['path'] ?? '');
 
-        $titleConfiguration = $this->getTitleConfiguration($content);
+        $titleConfiguration = $this->titleConfigurationExtractor->getTitleConfiguration($content);
 
         return [
             'id' => $pageId,
             'url' => $url,
             'baseUrl' => $baseUrl,
             'slug' => '/',
-            'title' => $this->getTitle($content),
-            'description' => $this->getDescription($content),
-            'locale' => $this->getLocale($content),
-            'body' => $this->getBody($content),
-            'faviconSrc' => $this->getFaviconSrc($baseUrl, $content),
+            'title' => $this->contentMetadataExtractor->getTitle($content),
+            'description' => $this->contentMetadataExtractor->getDescription($content),
+            'locale' => $this->contentMetadataExtractor->getLocale($content),
+            'body' => $this->bodyProcessor->getBody($content),
+            'faviconSrc' => $this->faviconExtractor->getFaviconSrc($baseUrl, $content),
             'pageTitlePrepend' => $titleConfiguration['titlePrepend'],
             'pageTitleAppend' => $titleConfiguration['titleAppend'],
         ];
     }
-
-    protected function getBaseUrl(mixed $urlParts): string
-    {
-        if (!is_array($urlParts)) {
-            return '://';
-        }
-        if ($urlParts['port'] ?? false) {
-            return (isset($urlParts['scheme']) ? $urlParts['scheme'] . ':' : '') . '//' . ($urlParts['host'] ?? '') . ':' . $urlParts['port'];
-        }
-        return (isset($urlParts['scheme']) ? $urlParts['scheme'] . ':' : '') . '//' . ($urlParts['host'] ?? '');
-    }
-
-    protected function getTitle(string $content): string
-    {
-        $title = '';
-        $titleFound = preg_match("/<title[^>]*>(.*?)<\/title>/is", $content, $matchesTitle);
-
-        if ($titleFound) {
-            $title = $matchesTitle[1];
-        }
-
-        return strip_tags(html_entity_decode($title));
-    }
-
-    protected function getDescription(string $content): string
-    {
-        $metaDescription = '';
-        $descriptionFound = preg_match(
-            "/<meta[^>]*name=[\" | \']description[\"|\'][^>]*content=[\"]([^\"]*)[\"][^>]*>/i",
-            $content,
-            $matchesDescription
-        );
-
-        if ($descriptionFound) {
-            $metaDescription = $matchesDescription[1];
-        }
-
-        return strip_tags(html_entity_decode($metaDescription));
-    }
-
-    protected function getLocale(string $content): string
-    {
-        $locale = 'en';
-        $localeFound = preg_match('/<html[^>]*lang="([a-z\-A-Z]*)"/is', $content, $matchesLocale);
-
-        if ($localeFound) {
-            [$locale] = explode('-', trim($matchesLocale[1]));
-        }
-
-        return $locale;
-    }
-
-    protected function getBody(string $content): string
-    {
-        $body = '';
-
-        $bodyFound = preg_match("/<body[^>]*>(.*)<\/body>/is", $content, $matchesBody);
-
-        if ($bodyFound) {
-            $body = $matchesBody[1];
-
-            preg_match_all(
-                '/<!--\s*?TYPO3SEARCH_begin\s*?-->.*?<!--\s*?TYPO3SEARCH_end\s*?-->/mis',
-                $body,
-                $indexableContents
-            );
-
-            if (is_array($indexableContents[0]) && !empty($indexableContents[0])) {
-                $body = implode('', $indexableContents[0]);
-            }
-        }
-
-        return $this->prepareBody($body);
-    }
-
-    protected function prepareBody(string $body): string
-    {
-        $body = $this->stripTagsContent($body, '<script><noscript>');
-        $body = preg_replace(['/\s?\n\s?/', '/\s{2,}/'], [' ', ' '], $body);
-        $body = strip_tags((string)$body, '<h1><h2><h3><h4><h5><p><a><img>');
-
-        return trim($body);
-    }
-
-    protected function stripTagsContent(string $text, string $tags = ''): string
-    {
-        preg_match_all('/<(.+?)[\s]*\/?[\s]*>/si', trim($tags), $foundTags);
-        $tagsArray = array_unique($foundTags[1]);
-
-        if (is_array($tagsArray) && count($tagsArray) > 0) {
-            return (string)preg_replace('@<(' . implode('|', $tagsArray) . ')\b.*?>.*?</\1>@si', '', $text);
-        }
-
-        return $text;
-    }
-
-    protected function getFaviconSrc(string $baseUrl, string $content): string
-    {
-        $faviconSrc = $baseUrl . '/favicon.ico';
-        $favIconFound = preg_match('/<link rel=\"shortcut icon\" href=\"([^"]*)\"/i', $content, $matchesFavIcon);
-        if ($favIconFound) {
-            $faviconSrc = str_contains($matchesFavIcon[1], '://') ? $matchesFavIcon[1] : $baseUrl . $matchesFavIcon[1];
-        }
-        $favIconHeader = @get_headers($faviconSrc);
-        if (($favIconHeader[0] ?? '') === 'HTTP/1.1 404 Not Found') {
-            $faviconSrc = '';
-        }
-        return $faviconSrc;
-    }
-
-    /**
-     * @return array{titlePrepend: string, titleAppend: string}
-     */
-    protected function getTitleConfiguration(string $content): array
-    {
-        $prepend = $append = '';
-        preg_match('/<meta name=\"x-yoast-title-config\" value=\"([^"]*)\"/i', $content, $matchesTitleConfig);
-        if (count($matchesTitleConfig) > 1) {
-            [$prepend, $append] = explode('|||', (string)$matchesTitleConfig[1]);
-        }
-        return [
-            'titlePrepend' => $prepend,
-            'titleAppend' => $append,
-        ];
-    }
 }
diff --git a/Configuration/Services.yaml b/Configuration/Services.yaml
index d96e4c71..78acf27f 100644
--- a/Configuration/Services.yaml
+++ b/Configuration/Services.yaml
@@ -16,6 +16,15 @@ services:
   YoastSeoForTypo3\YoastSeo\Service\SnippetPreviewService:
     public: true
 
+  YoastSeoForTypo3\YoastSeo\Service\Preview\ContentParser:
+    public: true
+
+  YoastSeoForTypo3\YoastSeo\Service\Preview\ContentExtractors\BaseUrlParserInterface: '@YoastSeoForTypo3\YoastSeo\Service\Preview\ContentExtractors\BaseUrlParser'
+  YoastSeoForTypo3\YoastSeo\Service\Preview\ContentExtractors\BodyProcessorInterface: '@YoastSeoForTypo3\YoastSeo\Service\Preview\ContentExtractors\BodyProcessor'
+  YoastSeoForTypo3\YoastSeo\Service\Preview\ContentExtractors\ContentMetadataExtractorInterface: '@YoastSeoForTypo3\YoastSeo\Service\Preview\ContentExtractors\ContentMetadataExtractor'
+  YoastSeoForTypo3\YoastSeo\Service\Preview\ContentExtractors\FaviconExtractorInterface: '@YoastSeoForTypo3\YoastSeo\Service\Preview\ContentExtractors\FaviconExtractor'
+  YoastSeoForTypo3\YoastSeo\Service\Preview\ContentExtractors\TitleConfigurationExtractorInterface: '@YoastSeoForTypo3\YoastSeo\Service\Preview\ContentExtractors\TitleConfigurationExtractor'
+
   YoastSeoForTypo3\YoastSeo\Controller\AjaxController:
     public: true