Improve array config validation

src/Extractor/ContentExtractor.php

@@ -1044,16 +1044,12 @@ private function getReadability(string $html, string $url, string $parser, bool
     {
         $readability = new Readability($html, $url, $parser, $enableTidy);
 
-        if (isset($this->config->getReadability()['pre_filters']) && \is_array($this->config->getReadability()['pre_filters'])) {
-            foreach ($this->config->getReadability()['pre_filters'] as $filter => $replacer) {
-                $readability->addPreFilter($filter, $replacer);
-            }
+        foreach ($this->config->getReadability()['pre_filters'] as $filter => $replacer) {
+            $readability->addPreFilter($filter, $replacer);
         }
 
-        if (isset($this->config->getReadability()['post_filters']) && \is_array($this->config->getReadability()['post_filters'])) {
-            foreach ($this->config->getReadability()['post_filters'] as $filter => $replacer) {
-                $readability->addPostFilter($filter, $replacer);
-            }
+        foreach ($this->config->getReadability()['post_filters'] as $filter => $replacer) {
+            $readability->addPostFilter($filter, $replacer);
         }
 
         return $readability;

src/Extractor/ContentExtractorConfig.php

@@ -2,13 +2,17 @@
 
 namespace Graby\Extractor;
 
+use Graby\OptionsResolver\ArrayStringOptionsTrait;
+use Symfony\Component\OptionsResolver\Options;
 use Symfony\Component\OptionsResolver\OptionsResolver;
 
 /**
  * Configuration for ContentExtractor as a Value Object.
  */
 class ContentExtractorConfig
 {
+    use ArrayStringOptionsTrait;
+
     private string $default_parser;
     /** @var array<string> */
     private array $allowed_parsers = ['libxml', 'html5lib'];
@@ -63,6 +67,18 @@ public function __construct(array $config)
             $readabilityResolver->setAllowedTypes('post_filters', 'array');
         });
 
+        $resolver->setNormalizer('readability', function (Options $options, $value) {
+            $this->validateArray($value, 'readability[pre_filters]', 'pre_filters');
+            $this->validateArray($value, 'readability[post_filters]', 'post_filters');
+
+            return $value;
+        });
+        $resolver->setNormalizer('fingerprints', function (Options $options, $value) {
+            $this->validateArray($value, 'fingerprints');
+
+            return $value;
+        });
+
         $config = $resolver->resolve($config);
 
         foreach ($config as $key => $value) {

src/Extractor/HttpClient.php

@@ -241,7 +241,7 @@ private function cleanupUrl(string $url): string
     {
         // rewrite part of urls to something more readable
         foreach ($this->config->getRewriteUrl() as $find => $action) {
-            if (false !== strpos($url, (string) $find) && \is_array($action)) {
+            if (false !== strpos($url, (string) $find)) {
                 $url = strtr($url, $action);
             }
         }

src/Extractor/HttpClientConfig.php

@@ -2,13 +2,18 @@
 
 namespace Graby\Extractor;
 
+use Graby\OptionsResolver\ArrayStringOptionsTrait;
+use Symfony\Component\OptionsResolver\Exception\InvalidOptionsException;
+use Symfony\Component\OptionsResolver\Options;
 use Symfony\Component\OptionsResolver\OptionsResolver;
 
 /**
  * Configuration for HttpClient as a Value Object.
  */
 class HttpClientConfig
 {
+    use ArrayStringOptionsTrait;
+
     private string $ua_browser;
     private string $default_referer;
     /** @var array<array<string, string>> */
@@ -69,6 +74,23 @@ public function __construct(array $config)
         $resolver->setAllowedTypes('ajax_triggers', 'string[]');
         $resolver->setAllowedTypes('max_redirect', 'int');
 
+        $resolver->setNormalizer('user_agents', function (Options $options, $value) {
+            $this->validateArray($value, 'user_agents');
+
+            return $value;
+        });
+        $resolver->setNormalizer('rewrite_url', function (Options $options, $value) {
+            foreach ($value as $url => $action) {
+                if (!\is_string($url)) {
+                    throw new InvalidOptionsException(sprintf('The option "rewrite_url" with key "%s" is expected to be of type "string", but is of type "%s".', $url, get_debug_type($url)));
+                }
+
+                $this->validateArray($action, 'rewrite_url[' . $url . ']');
+            }
+
+            return $value;
+        });
+
         $config = $resolver->resolve($config);
 
         foreach ($config as $key => $value) {

src/OptionsResolver/ArrayStringOptionsTrait.php

@@ -0,0 +1,34 @@
+<?php
+
+namespace Graby\OptionsResolver;
+
+use Symfony\Component\OptionsResolver\Exception\InvalidOptionsException;
+
+/**
+ * To be used inside a `OptionsResolver->setNormalizer` to enforce that a given value is an array on string, string.
+ */
+trait ArrayStringOptionsTrait
+{
+    public function validateArray(array $array, string $option, string $key = null): void
+    {
+        if (null === $key) {
+            foreach ($array as $arrayKey => $arrayValue) {
+                if (!\is_string($arrayKey)) {
+                    throw new InvalidOptionsException(sprintf('The option "%s" with key "%s" is expected to be of type "string", but is of type "%s".', $option, $arrayKey, get_debug_type($arrayKey)));
+                }
+                if (!\is_string($arrayValue)) {
+                    throw new InvalidOptionsException(sprintf('The option "%s" with value "%s" is expected to be of type "string", but is of type "%s".', $option, $arrayValue, get_debug_type($arrayValue)));
+                }
+            }
+        } elseif (!empty($array[$key])) {
+            foreach ($array[$key] as $arrayKey => $arrayValue) {
+                if (!\is_string($arrayKey)) {
+                    throw new InvalidOptionsException(sprintf('The option "%s" with key "%s" is expected to be of type "string", but is of type "%s".', $option, $arrayKey, get_debug_type($arrayKey)));
+                }
+                if (!\is_string($arrayValue)) {
+                    throw new InvalidOptionsException(sprintf('The option "%s" with value "%s" is expected to be of type "string", but is of type "%s".', $option, $arrayValue, get_debug_type($arrayValue)));
+                }
+            }
+        }
+    }
+}

src/SiteConfig/ConfigBuilderConfig.php

@@ -25,7 +25,7 @@ public function __construct(array $config)
 
         $resolver->setRequired('site_config');
 
-        $resolver->setAllowedTypes('site_config', 'array');
+        $resolver->setAllowedTypes('site_config', 'string[]');
         $resolver->setAllowedTypes('hostname_regex', 'string');
 
         $resolver->setNormalizer('site_config', function (Options $options, $value) {