diff --git a/debug_dom.php b/debug_dom.php
new file mode 100644
index 00000000..6adde54a
--- /dev/null
+++ b/debug_dom.php
@@ -0,0 +1,15 @@
+';
+$wrapped_buffer = '' . $buffer . '';
+
+$dom = new DOMDocument();
+$encoded_buffer = mb_encode_numericentity( $wrapped_buffer, array( 0x80, 0x10FFFF, 0, 0x1FFFFF ), 'UTF-8' );
+$dom->loadHTML( $encoded_buffer, LIBXML_HTML_NOIMPLIED | LIBXML_HTML_NODEFDTD );
+
+$dummy = $dom->getElementsByTagName('dummy')->item(0);
+
+echo "Child Nodes: " . $dummy->childNodes->length . "\n";
+foreach ($dummy->childNodes as $node) {
+ echo "Node: " . $node->nodeName . "\n";
+ echo "Content: " . $dom->saveHTML($node) . "\n";
+}
diff --git a/src/lib/helper.php b/src/lib/helper.php
index 64501cfb..176b4a41 100644
--- a/src/lib/helper.php
+++ b/src/lib/helper.php
@@ -6,6 +6,8 @@
use Exception;
use InvalidArgumentException;
use RuntimeException;
+ use DOMDocument;
+ use DOMElement;
/**
* @param string $type
@@ -99,96 +101,79 @@ function cookiebot_addons_remove_class_action( $action, $class, $method, $priori
* @since 1.2.0
*/
function cookiebot_addons_manipulate_script( $buffer, $keywords ) {
- /**
- * normalize potential self-closing script tags
- */
+ if ( empty( $buffer ) ) {
+ return $buffer;
+ }
+
+ // Use DOMDocument to safely parse and modify the script tag
+ $dom = new DOMDocument();
+
+ // Suppress errors for partial HTML
+ $libxml_previous_state = libxml_use_internal_errors( true );
+
+ // Wrap buffer in a custom tag to ensure correct parsing of fragments (e.g. multiple siblings at root)
+ // This prevents DOMDocument from trying to fix structure by nesting siblings
+ $wrapped_buffer = '' . $buffer . '';
- $normalized_buffer = preg_replace( '/()/is', '', $buffer );
+ // Load HTML with UTF-8 encoding hack
+ // The mb_convert_encoding is to ensure we don't have encoding issues
+ // We use LIBXML_HTML_NOIMPLIED | LIBXML_HTML_NODEFDTD to avoid adding wrappers automatically
+ // Replacement for deprecated mb_convert_encoding(..., 'HTML-ENTITIES', 'UTF-8')
+ $encoded_buffer = mb_encode_numericentity( $wrapped_buffer, array( 0x80, 0x10FFFF, 0, 0x1FFFFF ), 'UTF-8' );
+ $dom->loadHTML( $encoded_buffer, LIBXML_HTML_NOIMPLIED | LIBXML_HTML_NODEFDTD );
- if ( $normalized_buffer !== null ) {
- $buffer = $normalized_buffer;
+ libxml_use_internal_errors( $libxml_previous_state );
+
+ $scripts = $dom->getElementsByTagName( 'script' );
+ $modified = false;
+
+ // Convert DOMNodeList to array to avoid modification issues during iteration
+ $script_nodes = array();
+ foreach ( $scripts as $script ) {
+ $script_nodes[] = $script;
}
- /**
- * Pattern to get all scripts
- *
- * @version 2.0.4
- * @since 1.2.0
- */
- $pattern = '/(]*+>)(.*?)(<\/script>)/is';
+ foreach ( $script_nodes as $script ) {
+ /** @var DOMElement $script */
+ // Get the full script HTML to check for keywords
+ // We check the outer HTML to include attributes in the search, matching the regex behavior
+ $full_script_content = $dom->saveHTML( $script );
- /**
- * Get all scripts and add cookieconsent if it does match with the criterion
- */
- $updated_scripts = preg_replace_callback(
- $pattern,
- function ( $matches ) use ( $keywords ) {
- $script = $matches[0]; // the full script html
- $script_tag_open = $matches[1]; // only the script open tag with all attributes
- $script_tag_inner = $matches[2]; // only the script's innerText
- $script_tag_close = $matches[3]; // only the script closing tag
-
- /**
- * Check if the script contains the keywords, checks keywords one by one
- *
- * If one match, then the rest of the keywords will be skipped.
- */
- foreach ( $keywords as $needle => $cookie_type ) {
- /**
- * The script contains the needle
- */
- if ( strpos( $script, $needle ) !== false ) {
- /**
- * replace all single quotes with double quotes in the open tag
- * remove previously set data-cookieconsent attribute
- * remove type attribute
- */
- $script_tag_open = str_replace( '\'', '"', $script_tag_open );
- $script_tag_open = preg_replace( '/\sdata-cookieconsent=\"[^"]*+\"/', '', $script_tag_open );
- $script_tag_open = preg_replace( '/\stype=\"[^"]*+\"/', '', $script_tag_open );
-
- /**
- * set the type attribute to text/plain to prevent javascript execution
- * add data-cookieconsent attribute
- */
- $cookie_types = cookiebot_addons_output_cookie_types( $cookie_type );
-
- $script_tag_open = str_replace(
- ' $cookie_type ) {
+ if ( strpos( $full_script_content, $needle ) !== false ) {
+ // Match found
- /**
- * return the reconstructed script
- */
- return $script;
- },
- $buffer
- );
+ // Remove existing attributes
+ $script->removeAttribute( 'type' );
+ $script->removeAttribute( 'data-cookieconsent' );
- /**
- * Fallback when the regex fails to work due to PCRE_ERROR_JIT_STACKLIMIT
- *
- * @version 2.0.4
- * @since 2.0.4
- */
- if ( $updated_scripts === null ) {
- $updated_scripts = $buffer;
+ // Add new attributes
+ $script->setAttribute( 'type', 'text/plain' );
+ $script->setAttribute( 'data-cookieconsent', cookiebot_addons_output_cookie_types( $cookie_type ) );
- if ( get_option( 'cookiebot_regex_stacklimit' ) === false ) {
- update_option( 'cookiebot_regex_stacklimit', 1 );
+ $modified = true;
+ // Break inner loop (keywords) as we found a match
+ break;
+ }
+ }
+ }
+
+ if ( $modified ) {
+ // Save HTML
+ // We extract children of our wrapper
+ $wrapper = $dom->getElementsByTagName( 'cookiebot-wrapper' )->item( 0 );
+ if ( $wrapper ) {
+ $output = '';
+ foreach ( $wrapper->childNodes as $node ) {
+ $output .= $dom->saveHTML( $node );
+ }
+ return $output;
}
+ // Fallback if wrapper is missing (should not happen)
+ return $dom->saveHTML();
}
- return $updated_scripts;
+ return $buffer;
}
/**
diff --git a/src/lib/script_loader_tag/Script_Loader_Tag.php b/src/lib/script_loader_tag/Script_Loader_Tag.php
index 876f9b73..a5b222b2 100644
--- a/src/lib/script_loader_tag/Script_Loader_Tag.php
+++ b/src/lib/script_loader_tag/Script_Loader_Tag.php
@@ -2,6 +2,9 @@
namespace cybot\cookiebot\lib\script_loader_tag;
+use DOMDocument;
+use DOMElement;
+
class Script_Loader_Tag implements Script_Loader_Tag_Interface {
@@ -74,27 +77,51 @@ public function ignore_script( $script ) {
* @since 1.2.0
*/
public function cookiebot_add_consent_attribute_to_tag( $tag, $handle, $src ) {
+ // Check if the handle is in our list of tags to modify
if ( array_key_exists( $handle, $this->tags ) && ! empty( $this->tags[ $handle ] ) ) {
+ // If we have a match, we completely replace the tag with our own constructed one
+ // This is safer than parsing for this specific case as we know exactly what we want
//phpcs:ignore WordPress.WP.EnqueuedResources.NonEnqueuedScript
- return '';
+ return '';
}
+ // Check if the script should be ignored
if ( $this->check_ignore_script( $src ) ) {
- return preg_replace_callback(
- '/[^>]*)>/',
- function ( $tag ) use ( $handle ) {
- // Prevent modification of the script tags inside the other script tag by validating the ID of the
- // script and checking if we already set the consent status for the script. This will fix the issue
- // on Gutenberg editor pages.
- if ( ! self::validate_attributes_for_consent_ignore( $handle, $tag['atts'] ) ) {
- return $tag[0];
- }
-
- //phpcs:ignore WordPress.WP.EnqueuedResources.NonEnqueuedScript
- return str_replace( ')
+ $html = preg_replace('/]*)>\s*<\/script>/', '', $html);
+ // Remove spaces around =
+ $html = preg_replace('/\s*=\s*/', '=', $html);
+
+ // Sort attributes to handle order differences
+ // This is a naive implementation but should work for simple cases
+ $html = preg_replace_callback('/]*)>/', function($matches) {
+ $attrs = $matches[1];
+ preg_match_all('/(\w+)(?:="([^"]*)")?/', $attrs, $attr_matches, PREG_SET_ORDER);
+ $sorted_attrs = [];
+ foreach ($attr_matches as $m) {
+ $name = $m[1];
+ $value = isset($m[2]) ? $m[2] : '';
+ $sorted_attrs[$name] = $value;
+ }
+ ksort($sorted_attrs);
+ $new_attrs = '';
+ foreach ($sorted_attrs as $k => $v) {
+ if ($v === '') {
+ $new_attrs .= " $k";
+ } else {
+ $new_attrs .= " $k=\"$v\"";
+ }
+ }
+ return "";
+ }, $html);
+
+ return $html;
+}
+
+function assert_equivalent($new, $old, $test_name) {
+ global $passes, $fails;
+
+ $new_norm = normalize_html($new);
+ $old_norm = normalize_html($old);
+
+ if ($new_norm === $old_norm) {
+ echo "[PASS] $test_name\n";
+ $passes++;
+ } else {
+ echo "[FAIL] $test_name\n";
+ echo " Old (Norm): $old_norm\n";
+ echo " New (Norm): $new_norm\n";
+ echo " Old (Raw): $old\n";
+ echo " New (Raw): $new\n";
+ $fails++;
+ }
+}
+
+// --- Test Suite 1: helper.php cookiebot_addons_manipulate_script ---
+
+echo "Testing cookiebot_addons_manipulate_script (helper.php)...\n";
+
+$keywords = ['tracking.js' => 'marketing'];
+
+// Case 1: Simple Script
+$input = '';
+$new = cookiebot_addons_manipulate_script($input, $keywords);
+$old = cookiebot_addons_manipulate_script_old($input, $keywords);
+assert_equivalent($new, $old, 'Simple Script');
+
+// Case 2: Script with existing attributes
+$input = '';
+$new = cookiebot_addons_manipulate_script($input, $keywords);
+$old = cookiebot_addons_manipulate_script_old($input, $keywords);
+assert_equivalent($new, $old, 'Attributes');
+
+// Case 3: Script with existing type
+$input = '';
+$new = cookiebot_addons_manipulate_script($input, $keywords);
+$old = cookiebot_addons_manipulate_script_old($input, $keywords);
+assert_equivalent($new, $old, 'Existing Type');
+
+// Case 4: Multiple scripts in buffer
+$input = '';
+$new = cookiebot_addons_manipulate_script($input, $keywords);
+$old = cookiebot_addons_manipulate_script_old($input, $keywords);
+assert_equivalent($new, $old, 'Multiple Scripts');
+
+// Case 5: Fragment (no html/body)
+$input = '';
+$new = cookiebot_addons_manipulate_script($input, $keywords);
+$old = cookiebot_addons_manipulate_script_old($input, $keywords);
+assert_equivalent($new, $old, 'Fragment');
+
+// --- Test Suite 2: Script_Loader_Tag ---
+
+echo "\nTesting Script_Loader_Tag...\n";
+
+$loader_new = new Script_Loader_Tag();
+$loader_new->ignore_script('ignore-me.js');
+
+$loader_old = new Script_Loader_Tag_Old();
+$loader_old->ignore_script('ignore-me.js');
+
+// Case 6: Ignored Script
+$tag = '';
+$handle = 'ignore-me';
+$src = 'https://example.com/ignore-me.js';
+
+$new = $loader_new->cookiebot_add_consent_attribute_to_tag($tag, $handle, $src);
+$old = $loader_old->cookiebot_add_consent_attribute_to_tag($tag, $handle, $src);
+assert_equivalent($new, $old, 'Ignored Script');
+
+// Case 7: Non-ignored script
+$tag = '';
+$handle = 'normal';
+$src = 'https://example.com/normal.js';
+
+$new = $loader_new->cookiebot_add_consent_attribute_to_tag($tag, $handle, $src);
+$old = $loader_old->cookiebot_add_consent_attribute_to_tag($tag, $handle, $src);
+assert_equivalent($new, $old, 'Non-ignored script');
+
+echo "\nSummary: $passes Passed, $fails Failed.\n";
+if ($fails > 0) {
+ exit(1);
+}
diff --git a/tests/mock_wp_functions.php b/tests/mock_wp_functions.php
new file mode 100644
index 00000000..983a8770
--- /dev/null
+++ b/tests/mock_wp_functions.php
@@ -0,0 +1,59 @@
+ []]; }
+}
+if (!function_exists('is_multisite')) {
+ function is_multisite() { return false; }
+}
+if (!function_exists('get_site_url')) {
+ function get_site_url() { return 'http://example.com'; }
+}
+if (!function_exists('get_site_option')) {
+ function get_site_option($option, $default = false) { return $default; }
+}
+if (!function_exists('get_locale')) {
+ function get_locale() { return 'en_US'; }
+}
+if (!function_exists('plugin_dir_url')) {
+ function plugin_dir_url($file) { return 'http://example.com/wp-content/plugins/cookiebot/'; }
+}
+if (!function_exists('plugin_dir_path')) {
+ function plugin_dir_path($file) { return '/path/to/plugin/'; }
+}
+
+// Constants
+if (!defined('ABSPATH')) define('ABSPATH', '/tmp/');
+if (!defined('CYBOT_COOKIEBOT_PLUGIN_DIR')) define('CYBOT_COOKIEBOT_PLUGIN_DIR', __DIR__ . '/../');
+if (!defined('CYBOT_COOKIEBOT_PLUGIN_ASSETS_DIR')) define('CYBOT_COOKIEBOT_PLUGIN_ASSETS_DIR', 'assets/');
diff --git a/tests/old_logic.php b/tests/old_logic.php
new file mode 100644
index 00000000..d1b397c5
--- /dev/null
+++ b/tests/old_logic.php
@@ -0,0 +1,101 @@
+)/is', '', $buffer );
+ if ( $normalized_buffer !== null ) {
+ $buffer = $normalized_buffer;
+ }
+ $pattern = '/(]*+>)(.*?)(<\/script>)/is';
+ $updated_scripts = preg_replace_callback(
+ $pattern,
+ function ( $matches ) use ( $keywords ) {
+ $script = $matches[0];
+ $script_tag_open = $matches[1];
+ $script_tag_inner = $matches[2];
+ $script_tag_close = $matches[3];
+
+ foreach ( $keywords as $needle => $cookie_type ) {
+ if ( strpos( $script, $needle ) !== false ) {
+ $script_tag_open = str_replace( '\'', '"', $script_tag_open );
+ $script_tag_open = preg_replace( '/\sdata-cookieconsent=\"[^"]*+\"/', '', $script_tag_open );
+ $script_tag_open = preg_replace( '/\stype=\"[^"]*+\"/', '', $script_tag_open );
+
+ $cookie_types = cookiebot_addons_output_cookie_types( $cookie_type );
+
+ $script_tag_open = str_replace(
+ 'tags[ $tag ] = $type;
+ }
+
+ public function ignore_script( $script ) {
+ array_push( $this->ignore_scripts, $script );
+ }
+
+ public function cookiebot_add_consent_attribute_to_tag( $tag, $handle, $src ) {
+ if ( array_key_exists( $handle, $this->tags ) && ! empty( $this->tags[ $handle ] ) ) {
+ return '';
+ }
+
+ if ( $this->check_ignore_script( $src ) ) {
+ return preg_replace_callback(
+ '/[^>]*)>/',
+ function ( $tag ) use ( $handle ) {
+ if ( ! self::validate_attributes_for_consent_ignore( $handle, $tag['atts'] ) ) {
+ return $tag[0];
+ }
+ return str_replace( '