mailpoet
/
vendor
/
woocommerce
/
email-editor
/
src
/
Integrations
/
Utils
/
class-html-processing-helper.php
class-dom-document-helper.php
1 year ago
class-html-processing-helper.php
6 months ago
class-social-links-helper.php
1 year ago
class-styles-helper.php
6 months ago
class-table-wrapper-helper.php
1 year ago
index.php
1 year ago
class-html-processing-helper.php
456 lines
| 1 | <?php |
| 2 | declare( strict_types = 1 ); |
| 3 | namespace Automattic\WooCommerce\EmailEditor\Integrations\Utils; |
| 4 | if (!defined('ABSPATH')) exit; |
| 5 | class Html_Processing_Helper { |
| 6 | public static function clean_css_classes( string $classes ): string { |
| 7 | // Limit input length to prevent DoS attacks. |
| 8 | if ( strlen( $classes ) > 1000 ) { |
| 9 | $classes = substr( $classes, 0, 1000 ); |
| 10 | } |
| 11 | // Remove generic background classes but keep specific color classes. |
| 12 | $result = preg_replace( '/\bhas-background\b/', '', $classes ); |
| 13 | if ( null === $result ) { |
| 14 | $classes = ''; |
| 15 | } else { |
| 16 | $classes = $result; |
| 17 | } |
| 18 | // Remove border classes. |
| 19 | $result = preg_replace( '/\bhas-[a-z-]*border[a-z-]*\b/', '', $classes ); |
| 20 | if ( null === $result ) { |
| 21 | $classes = ''; |
| 22 | } else { |
| 23 | $classes = $result; |
| 24 | } |
| 25 | $result = preg_replace( '/\b[a-z-]+-border-[a-z-]+\b/', '', $classes ); |
| 26 | if ( null === $result ) { |
| 27 | $classes = ''; |
| 28 | } else { |
| 29 | $classes = $result; |
| 30 | } |
| 31 | // Clean up multiple spaces. |
| 32 | $result = preg_replace( '/\s+/', ' ', $classes ); |
| 33 | if ( null === $result ) { |
| 34 | $classes = ''; |
| 35 | } else { |
| 36 | $classes = $result; |
| 37 | } |
| 38 | return trim( $classes ); |
| 39 | } |
| 40 | public static function sanitize_css_value( string $value ): string { |
| 41 | // Remove dangerous script injection characters (angle brackets) but preserve quotes for CSS strings. |
| 42 | $result = preg_replace( '/[<>]/', '', $value ); |
| 43 | if ( null === $result ) { |
| 44 | $value = ''; |
| 45 | } else { |
| 46 | $value = $result; |
| 47 | } |
| 48 | // Remove dangerous CSS functions and expressions. |
| 49 | $dangerous_patterns = array( |
| 50 | '/expression\s*\(/i', |
| 51 | '/url\s*\(\s*javascript\s*:/i', |
| 52 | '/url\s*\(\s*data\s*:/i', |
| 53 | '/url\s*\(\s*vbscript\s*:/i', |
| 54 | '/import\s*\(/i', |
| 55 | '/behavior\s*:/i', |
| 56 | '/binding\s*:/i', |
| 57 | '/filter\s*:/i', |
| 58 | '/progid\s*:/i', |
| 59 | ); |
| 60 | foreach ( $dangerous_patterns as $pattern ) { |
| 61 | if ( preg_match( $pattern, $value ) ) { |
| 62 | return ''; |
| 63 | } |
| 64 | } |
| 65 | return trim( $value ); |
| 66 | } |
| 67 | public static function sanitize_dimension_value( $value ): string { |
| 68 | if ( ! is_string( $value ) && ! is_numeric( $value ) ) { |
| 69 | return ''; |
| 70 | } |
| 71 | $value = (string) $value; |
| 72 | // If it's just a number, assume pixels. |
| 73 | if ( is_numeric( $value ) ) { |
| 74 | $value = $value . 'px'; |
| 75 | } |
| 76 | // Use existing CSS value sanitization for security. |
| 77 | $sanitized_value = self::sanitize_css_value( $value ); |
| 78 | // Additional validation for dimension-specific units. |
| 79 | if ( ! empty( $sanitized_value ) && preg_match( '/^(\d+(?:\.\d+)?)(px|em|rem|%|vh|vw|ex|ch|in|cm|mm|pt|pc)$/', $sanitized_value ) ) { |
| 80 | return $sanitized_value; |
| 81 | } |
| 82 | return ''; |
| 83 | } |
| 84 | public static function sanitize_color( string $color ): string { |
| 85 | // Remove any whitespace. |
| 86 | $color = trim( $color ); |
| 87 | // Check if it's a valid hex color (#fff, #ffffff, #ffffffff). |
| 88 | if ( preg_match( '/^#([0-9a-fA-F]{3}|[0-9a-fA-F]{6}|[0-9a-fA-F]{8})$/', $color ) ) { |
| 89 | return strtolower( $color ); |
| 90 | } |
| 91 | // Check for rgb/rgba colors. |
| 92 | if ( preg_match( '/^rgba?\(\s*(25[0-5]|2[0-4]\d|1\d{2}|\d{1,2})\s*,\s*(25[0-5]|2[0-4]\d|1\d{2}|\d{1,2})\s*,\s*(25[0-5]|2[0-4]\d|1\d{2}|\d{1,2})\s*(?:,\s*(?:1(?:\.0+)?|0(?:\.\d+)?|\.\d+)\s*)?\)$/', $color ) ) { |
| 93 | return $color; |
| 94 | } |
| 95 | // Check for hsl/hsla colors. |
| 96 | if ( preg_match( '/^hsla?\(\s*(360|3[0-5]\d|[12]\d{2}|\d{1,2})\s*,\s*(100|[1-9]?\d)%\s*,\s*(100|[1-9]?\d)%\s*(?:,\s*(?:1(?:\.0+)?|0(?:\.\d+)?|\.\d+)\s*)?\)$/', $color ) ) { |
| 97 | return $color; |
| 98 | } |
| 99 | // Check for named colors and other valid CSS color values. |
| 100 | // We use a permissive approach: accept any string that doesn't contain dangerous characters |
| 101 | // and let the CSS engine handle the actual validation. |
| 102 | if ( preg_match( '/^[a-zA-Z][a-zA-Z0-9-]*$/', $color ) && ! preg_match( '/^(expression|javascript|vbscript|data|import|behavior|binding|filter|progid)/i', $color ) ) { |
| 103 | return strtolower( $color ); |
| 104 | } |
| 105 | // Check if it's a CSS variable (var(--variable-name)). |
| 106 | if ( preg_match( '/^var\(--[a-zA-Z0-9\-_]+\)$/', $color ) ) { |
| 107 | return $color; |
| 108 | } |
| 109 | // If not a valid color format, return a safe default. |
| 110 | return '#000000'; |
| 111 | } |
| 112 | private static function normalize_rel_attribute( ?string $rel_value, bool $require_security_tokens = false ): string { |
| 113 | $allowed_tokens = array( 'noopener', 'noreferrer', 'nofollow', 'external' ); |
| 114 | $required_tokens = $require_security_tokens ? array( 'noopener', 'noreferrer' ) : array(); |
| 115 | // If no rel value and no required tokens, return empty. |
| 116 | if ( null === $rel_value && empty( $required_tokens ) ) { |
| 117 | return ''; |
| 118 | } |
| 119 | // Start with required tokens. |
| 120 | $tokens = $required_tokens; |
| 121 | // If rel value exists, parse and normalize it. |
| 122 | if ( null !== $rel_value ) { |
| 123 | $existing_tokens = preg_split( '/\s+/', trim( $rel_value ) ); |
| 124 | if ( false !== $existing_tokens ) { |
| 125 | // Normalize existing tokens: lowercase, remove empty, filter allowed. |
| 126 | $normalized_existing = array_filter( |
| 127 | array_map( 'strtolower', $existing_tokens ), |
| 128 | function ( $token ) use ( $allowed_tokens ) { |
| 129 | return ! empty( $token ) && in_array( $token, $allowed_tokens, true ); |
| 130 | } |
| 131 | ); |
| 132 | // Merge with required tokens, removing duplicates. |
| 133 | $tokens = array_unique( array_merge( $tokens, $normalized_existing ) ); |
| 134 | } |
| 135 | } |
| 136 | // Return normalized rel attribute or empty string if no valid tokens. |
| 137 | return empty( $tokens ) ? '' : implode( ' ', $tokens ); |
| 138 | } |
| 139 | public static function validate_caption_attribute( \WP_HTML_Tag_Processor $html, string $attr_name ): void { |
| 140 | $attr_value = $html->get_attribute( $attr_name ); |
| 141 | if ( null === $attr_value ) { |
| 142 | return; |
| 143 | } |
| 144 | // Block all event handler attributes (on*) - Critical security fix. |
| 145 | if ( str_starts_with( $attr_name, 'on' ) ) { |
| 146 | $html->remove_attribute( $attr_name ); |
| 147 | return; |
| 148 | } |
| 149 | switch ( $attr_name ) { |
| 150 | case 'href': |
| 151 | // Only allow http, https, mailto, and tel protocols. |
| 152 | if ( ! preg_match( '/^(https?:\/\/|mailto:|tel:)/i', (string) $attr_value ) ) { |
| 153 | $html->remove_attribute( $attr_name ); |
| 154 | break; |
| 155 | } |
| 156 | // Sanitize and normalize the URL using WordPress's esc_url_raw. |
| 157 | $sanitized_url = esc_url_raw( (string) $attr_value ); |
| 158 | if ( empty( $sanitized_url ) ) { |
| 159 | // If esc_url_raw returns empty, the URL was invalid - remove the attribute. |
| 160 | $html->remove_attribute( $attr_name ); |
| 161 | } else { |
| 162 | // Set the attribute to the sanitized/normalized value. |
| 163 | $html->set_attribute( $attr_name, $sanitized_url ); |
| 164 | } |
| 165 | break; |
| 166 | case 'target': |
| 167 | // Allow only common safe targets. |
| 168 | $allowed_targets = array( '_blank', '_self' ); |
| 169 | $target_value = strtolower( (string) $attr_value ); |
| 170 | if ( ! in_array( $target_value, $allowed_targets, true ) ) { |
| 171 | $html->remove_attribute( $attr_name ); |
| 172 | } elseif ( '_blank' === $target_value ) { |
| 173 | // When target is "_blank", ensure rel attribute has noopener and noreferrer. |
| 174 | $current_rel = $html->get_attribute( 'rel' ); |
| 175 | $rel_value = is_string( $current_rel ) ? $current_rel : null; |
| 176 | $normalized_rel = self::normalize_rel_attribute( $rel_value, true ); |
| 177 | $html->set_attribute( 'rel', $normalized_rel ); |
| 178 | } |
| 179 | break; |
| 180 | case 'rel': |
| 181 | // Normalize rel attribute: lowercase, deduplicate, preserve safe tokens. |
| 182 | $rel_value = is_string( $attr_value ) ? $attr_value : null; |
| 183 | $normalized_rel = self::normalize_rel_attribute( $rel_value, false ); |
| 184 | if ( empty( $normalized_rel ) ) { |
| 185 | $html->remove_attribute( $attr_name ); |
| 186 | } else { |
| 187 | $html->set_attribute( $attr_name, $normalized_rel ); |
| 188 | } |
| 189 | break; |
| 190 | case 'style': |
| 191 | // Only allow safe CSS properties for typography and basic styling. |
| 192 | $safe_properties = self::get_safe_css_properties(); |
| 193 | $sanitized_styles = array(); |
| 194 | $style_parts = explode( ';', (string) $attr_value ); |
| 195 | foreach ( $style_parts as $style_part ) { |
| 196 | $style_part = trim( $style_part ); |
| 197 | if ( empty( $style_part ) ) { |
| 198 | continue; |
| 199 | } |
| 200 | $property_parts = explode( ':', $style_part, 2 ); |
| 201 | if ( count( $property_parts ) !== 2 ) { |
| 202 | continue; |
| 203 | } |
| 204 | $property = trim( strtolower( $property_parts[0] ) ); |
| 205 | $value = trim( $property_parts[1] ); |
| 206 | // Only allow safe properties. |
| 207 | if ( in_array( $property, $safe_properties, true ) ) { |
| 208 | // Use centralized CSS value sanitization. |
| 209 | $sanitized_value = self::sanitize_css_value( $value ); |
| 210 | if ( ! empty( $sanitized_value ) ) { |
| 211 | $sanitized_styles[] = $property . ': ' . $sanitized_value; |
| 212 | } |
| 213 | } |
| 214 | } |
| 215 | if ( empty( $sanitized_styles ) ) { |
| 216 | $html->remove_attribute( $attr_name ); |
| 217 | } else { |
| 218 | $html->set_attribute( $attr_name, implode( '; ', $sanitized_styles ) ); |
| 219 | } |
| 220 | break; |
| 221 | case 'class': |
| 222 | // Only allow alphanumeric characters, hyphens, and underscores. |
| 223 | if ( ! preg_match( '/^[a-zA-Z0-9\s\-_]+$/', (string) $attr_value ) ) { |
| 224 | $html->remove_attribute( $attr_name ); |
| 225 | } |
| 226 | break; |
| 227 | case 'data-type': |
| 228 | case 'data-id': |
| 229 | // Only allow alphanumeric characters, hyphens, and underscores. |
| 230 | if ( ! preg_match( '/^[a-zA-Z0-9\-_]+$/', (string) $attr_value ) ) { |
| 231 | $html->remove_attribute( $attr_name ); |
| 232 | } |
| 233 | break; |
| 234 | default: |
| 235 | // Handle data-* attributes with strict validation. |
| 236 | if ( str_starts_with( $attr_name, 'data-' ) ) { |
| 237 | if ( ! preg_match( '/^[a-zA-Z0-9\-_]+$/', (string) $attr_value ) ) { |
| 238 | $html->remove_attribute( $attr_name ); |
| 239 | } |
| 240 | break; |
| 241 | } |
| 242 | // Default deny policy: Remove any attribute not explicitly allowed. |
| 243 | $html->remove_attribute( $attr_name ); |
| 244 | break; |
| 245 | } |
| 246 | } |
| 247 | public static function get_safe_css_properties(): array { |
| 248 | return array( |
| 249 | 'color', |
| 250 | 'background-color', |
| 251 | 'font-family', |
| 252 | 'font-size', |
| 253 | 'font-weight', |
| 254 | 'font-style', |
| 255 | 'text-decoration', |
| 256 | 'text-align', |
| 257 | 'line-height', |
| 258 | 'letter-spacing', |
| 259 | 'text-transform', |
| 260 | ); |
| 261 | } |
| 262 | public static function get_caption_css_properties(): array { |
| 263 | return array( |
| 264 | 'font-family', |
| 265 | 'font-size', |
| 266 | 'font-weight', |
| 267 | 'font-style', |
| 268 | 'text-decoration', |
| 269 | 'line-height', |
| 270 | 'letter-spacing', |
| 271 | 'text-transform', |
| 272 | ); |
| 273 | } |
| 274 | public static function validate_container_attributes( string $container_html ): bool { |
| 275 | // Use WP_HTML_Tag_Processor to validate container attributes. |
| 276 | $html = new \WP_HTML_Tag_Processor( $container_html ); |
| 277 | if ( ! $html->next_tag() ) { |
| 278 | return false; |
| 279 | } |
| 280 | // Get all attributes and validate each one using our existing validation logic. |
| 281 | $attributes = $html->get_attribute_names_with_prefix( '' ); |
| 282 | if ( is_array( $attributes ) ) { |
| 283 | foreach ( $attributes as $attr_name ) { |
| 284 | // Use the same validation logic as validate_caption_attribute for consistency. |
| 285 | $attr_value = $html->get_attribute( $attr_name ); |
| 286 | if ( null === $attr_value ) { |
| 287 | continue; |
| 288 | } |
| 289 | // Block event handlers immediately. |
| 290 | if ( str_starts_with( $attr_name, 'on' ) ) { |
| 291 | return false; |
| 292 | } |
| 293 | // Apply the same validation rules as caption attributes. |
| 294 | // Create a temporary processor to test validation. |
| 295 | $escaped_value = htmlspecialchars( (string) $attr_value, ENT_QUOTES, 'UTF-8' ); |
| 296 | $temp_html = new \WP_HTML_Tag_Processor( '<span ' . $attr_name . '="' . $escaped_value . '">test</span>' ); |
| 297 | if ( $temp_html->next_tag() ) { |
| 298 | $original_value = $temp_html->get_attribute( $attr_name ); |
| 299 | self::validate_caption_attribute( $temp_html, $attr_name ); |
| 300 | $validated_value = $temp_html->get_attribute( $attr_name ); |
| 301 | // If attribute was removed during validation, container is unsafe. |
| 302 | if ( null !== $original_value && null === $validated_value ) { |
| 303 | return false; |
| 304 | } |
| 305 | } |
| 306 | } |
| 307 | } |
| 308 | return true; |
| 309 | } |
| 310 | public static function sanitize_caption_html( string $caption_html ): string { |
| 311 | // If no HTML tags, return as-is. |
| 312 | if ( false === strpos( $caption_html, '<' ) ) { |
| 313 | return $caption_html; |
| 314 | } |
| 315 | // Remove dangerous content: script, style, and other executable elements. |
| 316 | $result = preg_replace( '/<(script|style|iframe|object|embed|form|input|button)\b[^>]*>.*?<\/\1>/is', '', $caption_html ); |
| 317 | if ( null === $result ) { |
| 318 | $caption_html = ''; |
| 319 | } else { |
| 320 | $caption_html = $result; |
| 321 | } |
| 322 | // Use a more conservative approach - only validate attributes, don't modify tags. |
| 323 | $allowed_tags = array( 'strong', 'em', 'a', 'mark', 'kbd', 's', 'sub', 'sup', 'span', 'br' ); |
| 324 | $html = new \WP_HTML_Tag_Processor( $caption_html ); |
| 325 | // First pass: Process attributes for allowed tags only. |
| 326 | while ( $html->next_tag() ) { |
| 327 | $tag_name = $html->get_tag(); |
| 328 | // Skip processing for disallowed tags. |
| 329 | if ( ! in_array( $tag_name, $allowed_tags, true ) ) { |
| 330 | continue; |
| 331 | } |
| 332 | // Only process attributes for allowed tags. |
| 333 | $attributes = $html->get_attribute_names_with_prefix( '' ); |
| 334 | if ( is_array( $attributes ) ) { |
| 335 | foreach ( $attributes as $attr_name ) { |
| 336 | // Validate and sanitize each attribute individually. |
| 337 | self::validate_caption_attribute( $html, $attr_name ); |
| 338 | } |
| 339 | } |
| 340 | } |
| 341 | // Second pass: Remove disallowed tags using a simple regex approach. |
| 342 | $final_html = $html->get_updated_html(); |
| 343 | // Create a regex pattern to match disallowed tags. |
| 344 | $allowed_tags_pattern = implode( '|', array_map( 'preg_quote', $allowed_tags ) ); |
| 345 | // Remove disallowed opening and closing tags, keeping only their content. |
| 346 | $result = preg_replace( '/<(?!(?:' . $allowed_tags_pattern . ')\b)[^>]*>(.*?)<\/(?!(?:' . $allowed_tags_pattern . ')\b)[^>]*>/s', '$1', $final_html ); |
| 347 | if ( null === $result ) { |
| 348 | $final_html = ''; |
| 349 | } else { |
| 350 | $final_html = $result; |
| 351 | } |
| 352 | // Remove disallowed self-closing tags. |
| 353 | $result = preg_replace( '/<(?!(?:' . $allowed_tags_pattern . ')\b)[^>]*\/>/s', '', $final_html ); |
| 354 | if ( null === $result ) { |
| 355 | $final_html = ''; |
| 356 | } else { |
| 357 | $final_html = $result; |
| 358 | } |
| 359 | return $final_html; |
| 360 | } |
| 361 | public static function sanitize_image_html( string $image_html ): string { |
| 362 | // If no HTML tags, return as-is. |
| 363 | if ( false === strpos( $image_html, '<' ) ) { |
| 364 | return $image_html; |
| 365 | } |
| 366 | // Extract img tag using regex for reliable processing. |
| 367 | if ( ! preg_match( '/<img[^>]*>/i', $image_html, $matches ) ) { |
| 368 | return $image_html; |
| 369 | } |
| 370 | $img_tag = $matches[0]; |
| 371 | $sanitized_attributes = array(); |
| 372 | $has_src = false; |
| 373 | // Extract and sanitize individual attributes using WP_HTML_Tag_Processor for attribute processing. |
| 374 | $html = new \WP_HTML_Tag_Processor( $img_tag ); |
| 375 | if ( $html->next_tag() ) { |
| 376 | $attributes = $html->get_attribute_names_with_prefix( '' ); |
| 377 | if ( is_array( $attributes ) ) { |
| 378 | foreach ( $attributes as $attr_name ) { |
| 379 | $attr_value = $html->get_attribute( $attr_name ); |
| 380 | // Sanitize specific attributes. |
| 381 | switch ( $attr_name ) { |
| 382 | case 'src': |
| 383 | // Sanitize image source URL. |
| 384 | $sanitized_src = esc_url( (string) $attr_value ); |
| 385 | if ( ! empty( $sanitized_src ) ) { |
| 386 | $sanitized_attributes[] = $attr_name . '="' . $sanitized_src . '"'; |
| 387 | $has_src = true; |
| 388 | } |
| 389 | break; |
| 390 | case 'alt': |
| 391 | case 'width': |
| 392 | case 'height': |
| 393 | // Sanitize text attributes. |
| 394 | $sanitized_attributes[] = $attr_name . '="' . esc_attr( (string) $attr_value ) . '"'; |
| 395 | break; |
| 396 | case 'class': |
| 397 | // Clean CSS classes. |
| 398 | $cleaned_classes = self::clean_css_classes( (string) $attr_value ); |
| 399 | if ( ! empty( $cleaned_classes ) ) { |
| 400 | $sanitized_attributes[] = $attr_name . '="' . esc_attr( $cleaned_classes ) . '"'; |
| 401 | } |
| 402 | break; |
| 403 | case 'style': |
| 404 | // Sanitize inline styles - only allow safe properties for email rendering. |
| 405 | $sanitized_styles = self::sanitize_image_styles( (string) $attr_value ); |
| 406 | if ( ! empty( $sanitized_styles ) ) { |
| 407 | $sanitized_attributes[] = $attr_name . '="' . esc_attr( $sanitized_styles ) . '"'; |
| 408 | } |
| 409 | break; |
| 410 | } |
| 411 | } |
| 412 | } |
| 413 | } |
| 414 | // If no valid src attribute, return empty string. |
| 415 | if ( ! $has_src ) { |
| 416 | return ''; |
| 417 | } |
| 418 | // Rebuild the img tag with sanitized attributes. |
| 419 | if ( empty( $sanitized_attributes ) ) { |
| 420 | return ''; |
| 421 | } |
| 422 | return '<img ' . implode( ' ', $sanitized_attributes ) . '>'; |
| 423 | } |
| 424 | public static function extract_url_from_text( string $text ): string { |
| 425 | if ( preg_match( '/(?<![a-zA-Z0-9.-])https?:\/\/[a-zA-Z0-9.-]+\.[a-zA-Z]{2,}[a-zA-Z0-9\/?=&%_.~+#-]*(?![a-zA-Z0-9._~+#-])/', $text, $matches ) ) { |
| 426 | return $matches[0]; |
| 427 | } |
| 428 | return ''; |
| 429 | } |
| 430 | private static function sanitize_image_styles( string $style_value ): string { |
| 431 | $sanitized_styles = array(); |
| 432 | $style_parts = explode( ';', $style_value ); |
| 433 | foreach ( $style_parts as $style_part ) { |
| 434 | $style_part = trim( $style_part ); |
| 435 | if ( empty( $style_part ) ) { |
| 436 | continue; |
| 437 | } |
| 438 | $property_parts = explode( ':', $style_part, 2 ); |
| 439 | if ( count( $property_parts ) !== 2 ) { |
| 440 | continue; |
| 441 | } |
| 442 | $property = trim( strtolower( $property_parts[0] ) ); |
| 443 | $value = trim( $property_parts[1] ); |
| 444 | // Allow safe CSS properties for images in email rendering. |
| 445 | $safe_properties = array( 'width', 'height', 'max-width', 'max-height', 'display', 'margin', 'padding', 'border', 'border-radius' ); |
| 446 | if ( in_array( $property, $safe_properties, true ) ) { |
| 447 | $sanitized_value = self::sanitize_css_value( $value ); |
| 448 | if ( ! empty( $sanitized_value ) ) { |
| 449 | $sanitized_styles[] = $property . ': ' . $sanitized_value; |
| 450 | } |
| 451 | } |
| 452 | } |
| 453 | return implode( '; ', $sanitized_styles ); |
| 454 | } |
| 455 | } |
| 456 |