Fix styling

Author: cloudstudio

src/HtmlCrawler.php

@@ -10,50 +10,36 @@ class HtmlCrawler
 {
     /**
      * The original HTML content.
-     *
-     * @var string
      */
     protected string $html;
 
     /**
      * Array of allowed HTML tags (e.g. ['p', 'a', 'h1']).
-     *
-     * @var array
      */
     protected array $allowedTags = [];
 
     /**
      * A custom regex pattern to remove parts of the HTML.
-     *
-     * @var string|null
      */
     protected ?string $customPattern = null;
 
     /**
      * Whether to preserve newlines in the output.
-     *
-     * @var bool
      */
     protected bool $preserveNewlines;
 
     /**
      * Whether to convert the cleaned HTML to Markdown.
-     *
-     * @var bool
      */
     protected bool $convertToMarkdown;
 
     /**
      * Whether to remove script blocks by default.
-     *
-     * @var bool
      */
     protected bool $removeScripts;
 
     /**
      * Whether to remove style blocks by default.
-     *
-     * @var bool
      */
     protected bool $removeStyles;
 
@@ -64,38 +50,37 @@ class HtmlCrawler
      */
     public function __construct()
     {
-        $this->preserveNewlines   = config('htmlcrawler.preserve_newlines', true);
-        $this->allowedTags        = config('htmlcrawler.allowed_tags', []);
-        $this->convertToMarkdown  = config('htmlcrawler.convert_to_markdown', false);
-        $this->removeScripts      = config('htmlcrawler.remove_scripts', true);
-        $this->removeStyles       = config('htmlcrawler.remove_styles', true);
+        $this->preserveNewlines = config('htmlcrawler.preserve_newlines', true);
+        $this->allowedTags = config('htmlcrawler.allowed_tags', []);
+        $this->convertToMarkdown = config('htmlcrawler.convert_to_markdown', false);
+        $this->removeScripts = config('htmlcrawler.remove_scripts', true);
+        $this->removeStyles = config('htmlcrawler.remove_styles', true);
     }
 
     /**
      * Initialize the crawler with an HTML string.
      *
-     * @param string $html
      * @return static
      */
     public static function fromHtml(string $html): self
     {
-        $instance = new self();
+        $instance = new self;
         $instance->html = $html;
+
         return $instance;
     }
 
     /**
      * Initialize the crawler with HTML loaded from a URL.
      *
-     * @param string $url
      * @return static
      *
      * @throws InvalidUrlException
      * @throws \RuntimeException
      */
     public static function fromUrl(string $url): self
     {
-        if (!filter_var($url, FILTER_VALIDATE_URL)) {
+        if (! filter_var($url, FILTER_VALIDATE_URL)) {
             throw new InvalidUrlException("Invalid URL provided: {$url}");
         }
 
@@ -110,25 +95,26 @@ public static function fromUrl(string $url): self
     /**
      * Specify which HTML tags should be preserved.
      *
-     * @param string|array $tags A tag or an array of tags (e.g., 'p' or ['p', 'a']).
+     * @param  string|array  $tags  A tag or an array of tags (e.g., 'p' or ['p', 'a']).
      * @return $this
      */
     public function keepTags(string|array $tags): self
     {
         $tags = is_array($tags) ? $tags : [$tags];
         $this->allowedTags = array_unique(array_merge($this->allowedTags, $tags));
+
         return $this;
     }
 
     /**
      * Replace the allowed tags list with a new set.
      *
-     * @param array $tags
      * @return $this
      */
     public function setAllowedTags(array $tags): self
     {
         $this->allowedTags = $tags;
+
         return $this;
     }
 
@@ -201,9 +187,10 @@ public function keepLinks(): self
     public function keepScripts(): self
     {
         $this->removeScripts = false;
-        if (!in_array('script', $this->allowedTags)) {
+        if (! in_array('script', $this->allowedTags)) {
             $this->allowedTags[] = 'script';
         }
+
         return $this;
     }
 
@@ -216,37 +203,38 @@ public function keepScripts(): self
     public function keepCss(): self
     {
         $this->removeStyles = false;
-        if (!in_array('style', $this->allowedTags)) {
+        if (! in_array('style', $this->allowedTags)) {
             $this->allowedTags[] = 'style';
         }
-        if (!in_array('link', $this->allowedTags)) {
+        if (! in_array('link', $this->allowedTags)) {
             $this->allowedTags[] = 'link';
         }
+
         return $this;
     }
 
     /**
      * Set a custom regex pattern to remove parts of the HTML.
      * This pattern takes precedence over allowed tags.
      *
-     * @param string $pattern
      * @return $this
      */
     public function useCustomPattern(string $pattern): self
     {
         $this->customPattern = $pattern;
+
         return $this;
     }
 
     /**
      * Set whether to preserve newlines.
      *
-     * @param bool $preserve
      * @return $this
      */
     public function preserveNewlines(bool $preserve = true): self
     {
         $this->preserveNewlines = $preserve;
+
         return $this;
     }
 
@@ -258,6 +246,7 @@ public function preserveNewlines(bool $preserve = true): self
     public function withMarkdown(): self
     {
         $this->convertToMarkdown = true;
+
         return $this;
     }
 
@@ -269,7 +258,7 @@ public function withMarkdown(): self
      */
     public function clean(): string
     {
-        $cleaner = new HtmlCleaner();
+        $cleaner = new HtmlCleaner;
         $result = $cleaner->clean(
             $this->html,
             $this->allowedTags,
@@ -288,8 +277,6 @@ public function clean(): string
 
     /**
      * Magic method to return the cleaned content when the object is treated as a string.
-     *
-     * @return string
      */
     public function __toString(): string
     {

src/HtmlCrawlerServiceProvider.php

@@ -8,17 +8,16 @@
 /**
  * Class HtmlCrawlerServiceProvider
  *
- * Service provider for the HtmlCrawler package. 
- * Responsible for configuring the package, including 
+ * Service provider for the HtmlCrawler package.
+ * Responsible for configuring the package, including
  * setting up configuration files and views.
  */
 class HtmlCrawlerServiceProvider extends PackageServiceProvider
 {
     /**
      * Configure the package.
      *
-     * @param Package $package The package instance to configure.
-     * @return void
+     * @param  Package  $package  The package instance to configure.
      */
     public function configurePackage(Package $package): void
     {
@@ -32,13 +31,11 @@ public function configurePackage(Package $package): void
      *
      * This method binds the HtmlCrawler class to the service container
      * and creates an alias for easier access.
-     *
-     * @return void
      */
     public function packageRegistered(): void
     {
         $this->app->singleton(HtmlCrawler::class, function ($app) {
-            return new HtmlCrawler();
+            return new HtmlCrawler;
         });
 
         $this->app->alias(HtmlCrawler::class, 'html-crawler');

src/Services/Cleaner/CleanerInterface.php

@@ -6,9 +6,6 @@ interface CleanerInterface
 {
     /**
      * Clean the given HTML and return the modified content.
-     *
-     * @param string $html
-     * @return string
      */
     public function clean(string $html): string;
 }

src/Services/Cleaner/CustomPatternCleaner.php

@@ -18,7 +18,7 @@ class CustomPatternCleaner implements CleanerInterface
     /**
      * CustomPatternCleaner constructor.
      *
-     * @param string|null $pattern The regex pattern to apply for cleaning.
+     * @param  string|null  $pattern  The regex pattern to apply for cleaning.
      */
     public function __construct(?string $pattern)
     {
@@ -28,14 +28,15 @@ public function __construct(?string $pattern)
     /**
      * Clean the given HTML content by applying the custom regex pattern.
      *
-     * @param string $html The HTML content to clean.
+     * @param  string  $html  The HTML content to clean.
      * @return string The cleaned HTML content.
      */
     public function clean(string $html): string
     {
         if ($this->pattern) {
             return preg_replace($this->pattern, '', $html);
         }
+
         return $html;
     }
 }

src/Services/Cleaner/ScriptRemover.php

@@ -23,8 +23,8 @@ class ScriptRemover implements CleanerInterface
     /**
      * ScriptRemover constructor.
      *
-     * @param bool $remove Indicates whether to remove <script> tags.
-     * @param array $allowedTags List of tags that are allowed in the HTML.
+     * @param  bool  $remove  Indicates whether to remove <script> tags.
+     * @param  array  $allowedTags  List of tags that are allowed in the HTML.
      */
     public function __construct(bool $remove, array $allowedTags)
     {
@@ -35,14 +35,15 @@ public function __construct(bool $remove, array $allowedTags)
     /**
      * Clean the given HTML content by removing <script> tags if specified.
      *
-     * @param string $html The HTML content to clean.
+     * @param  string  $html  The HTML content to clean.
      * @return string The cleaned HTML content.
      */
     public function clean(string $html): string
     {
-        if ($this->remove && !in_array('script', $this->allowedTags)) {
+        if ($this->remove && ! in_array('script', $this->allowedTags)) {
             return preg_replace('#<script.*?</script>#is', '', $html);
         }
+
         return $html;
     }
 }

src/Services/Cleaner/StyleRemover.php

@@ -23,8 +23,8 @@ class StyleRemover implements CleanerInterface
     /**
      * StyleRemover constructor.
      *
-     * @param bool $remove Indicates whether to remove <style> tags.
-     * @param array $allowedTags List of tags that are allowed in the HTML.
+     * @param  bool  $remove  Indicates whether to remove <style> tags.
+     * @param  array  $allowedTags  List of tags that are allowed in the HTML.
      */
     public function __construct(bool $remove, array $allowedTags)
     {
@@ -35,14 +35,15 @@ public function __construct(bool $remove, array $allowedTags)
     /**
      * Clean the given HTML content by removing <style> tags if specified.
      *
-     * @param string $html The HTML content to clean.
+     * @param  string  $html  The HTML content to clean.
      * @return string The cleaned HTML content.
      */
     public function clean(string $html): string
     {
-        if ($this->remove && !in_array('style', $this->allowedTags)) {
+        if ($this->remove && ! in_array('style', $this->allowedTags)) {
             return preg_replace('#<style.*?</style>#is', '', $html);
         }
+
         return $html;
     }
 }

src/Services/Cleaner/TagCleaner.php

@@ -18,7 +18,7 @@ class TagCleaner implements CleanerInterface
     /**
      * TagCleaner constructor.
      *
-     * @param array $allowedTags List of tags that are allowed in the HTML.
+     * @param  array  $allowedTags  List of tags that are allowed in the HTML.
      */
     public function __construct(array $allowedTags)
     {
@@ -28,7 +28,7 @@ public function __construct(array $allowedTags)
     /**
      * Clean the given HTML content by stripping disallowed tags.
      *
-     * @param string $html The HTML content to clean.
+     * @param  string  $html  The HTML content to clean.
      * @return string The cleaned HTML content.
      */
     public function clean(string $html): string
@@ -41,6 +41,7 @@ public function clean(string $html): string
         foreach ($this->allowedTags as $tag) {
             $allowed .= "<{$tag}>";
         }
+
         return strip_tags($html, $allowed);
     }
 }

src/Services/Cleaner/WhitespaceNormalizer.php

@@ -19,7 +19,7 @@ class WhitespaceNormalizer implements CleanerInterface
     /**
      * WhitespaceNormalizer constructor.
      *
-     * @param bool $preserveNewlines Indicates whether to preserve newlines.
+     * @param  bool  $preserveNewlines  Indicates whether to preserve newlines.
      */
     public function __construct(bool $preserveNewlines)
     {
@@ -32,7 +32,7 @@ public function __construct(bool $preserveNewlines)
      * This method collapses multiple spaces into a single space and
      * preserves newlines if specified.
      *
-     * @param string $html The HTML content to clean.
+     * @param  string  $html  The HTML content to clean.
      * @return string The cleaned HTML content with normalized whitespace.
      */
     public function clean(string $html): string