EncoderForm.php
4 months ago
Encoding.php
4 months ago
Filters.php
4 months ago
Validate.php
4 months ago
Filters.php
477 lines
| 1 | <?php |
| 2 | |
| 3 | namespace OnlineOptimisation\EmailEncoderBundle\Validate; |
| 4 | |
| 5 | use DOMDocument; |
| 6 | use OnlineOptimisation\EmailEncoderBundle\Traits\PluginHelper; |
| 7 | |
| 8 | class Filters |
| 9 | { |
| 10 | use PluginHelper; |
| 11 | |
| 12 | public function boot(): void |
| 13 | { |
| 14 | } |
| 15 | |
| 16 | |
| 17 | /** |
| 18 | * ###################### |
| 19 | * ### |
| 20 | * #### FILTERS |
| 21 | * ### |
| 22 | * ###################### |
| 23 | */ |
| 24 | |
| 25 | /** |
| 26 | * The main page filter function |
| 27 | * |
| 28 | * @param string $content - the content that needs to be filtered |
| 29 | * @param string $protect_using |
| 30 | * @return string - The filtered content |
| 31 | */ |
| 32 | public function filter_page( $content, $protect_using ) |
| 33 | { |
| 34 | |
| 35 | //Added in 2.0.6 |
| 36 | $content = apply_filters( 'eeb/validate/filter_page_content', $content, $protect_using ); |
| 37 | |
| 38 | $content = $this->filter_soft_dom_attributes( $content, 'char_encode' ); |
| 39 | |
| 40 | $htmlSplit = preg_split( '/(<body(([^>]*)>))/is', $content, -1, PREG_SPLIT_DELIM_CAPTURE ); |
| 41 | |
| 42 | if ( ! is_array( $htmlSplit) || count( $htmlSplit ) < 4 ) { |
| 43 | return $content; |
| 44 | } |
| 45 | |
| 46 | switch ( $protect_using ) { |
| 47 | case 'with_javascript': |
| 48 | case 'without_javascript': |
| 49 | case 'char_encode': |
| 50 | $head_encoding_method = 'char_encode'; |
| 51 | break; |
| 52 | default: |
| 53 | $head_encoding_method = 'default'; |
| 54 | break; |
| 55 | } |
| 56 | |
| 57 | //Filter head area |
| 58 | $filtered_head = $this->filter_plain_emails( $htmlSplit[0], null, $head_encoding_method ); |
| 59 | |
| 60 | //Filter body |
| 61 | //Soft attributes always need to be protected using only the char encode method since otherwise the logic breaks |
| 62 | $filtered_body = $this->filter_soft_attributes( $htmlSplit[4], 'char_encode' ); |
| 63 | $filtered_body = $this->filter_content( $filtered_body, $protect_using ); |
| 64 | |
| 65 | $filtered_content = $filtered_head . $htmlSplit[1] . $filtered_body; |
| 66 | |
| 67 | //Revalidate filtered emails that should not bbe encoded |
| 68 | $filtered_content = $this->tempEncodeAtSymbol( $filtered_content, true ); |
| 69 | |
| 70 | return $filtered_content; |
| 71 | } |
| 72 | |
| 73 | /** |
| 74 | * Filter content |
| 75 | * |
| 76 | * @param string $content |
| 77 | * @param string $protect_using |
| 78 | * @return string |
| 79 | */ |
| 80 | public function filter_content( $content, $protect_using ) |
| 81 | { |
| 82 | $filtered = $content; |
| 83 | $self = $this; |
| 84 | $encode_mailtos = (bool) $this->getSetting( 'encode_mailtos', true, 'filter_body' ); |
| 85 | $convert_plain_to_image = (bool) $this->getSetting( 'convert_plain_to_image', true, 'filter_body' ); |
| 86 | |
| 87 | //Added in 2.0.6 |
| 88 | $filtered = apply_filters( 'eeb/validate/filter_content_content', $filtered, $protect_using ); |
| 89 | |
| 90 | //Soft attributes always need to be protected using only the char encode method since otherwise the logic breaks |
| 91 | $filtered = $this->filter_soft_attributes( $filtered, 'char_encode' ); |
| 92 | |
| 93 | switch ( $protect_using ) { |
| 94 | case 'char_encode': |
| 95 | $filtered = $this->filter_plain_emails( $filtered, null, 'char_encode' ); |
| 96 | break; |
| 97 | case 'strong_method': |
| 98 | $filtered = $this->filter_plain_emails( $filtered ); |
| 99 | break; |
| 100 | case 'without_javascript': |
| 101 | $filtered = $this->filter_input_fields( $filtered, $protect_using ); |
| 102 | $filtered = $this->filter_mailto_links( $filtered, 'without_javascript' ); |
| 103 | $filtered = $this->filter_custom_links( $filtered, 'without_javascript' ); |
| 104 | |
| 105 | if ( $convert_plain_to_image ) { |
| 106 | $replace_by = 'convert_image'; |
| 107 | } else { |
| 108 | $replace_by = 'use_css'; |
| 109 | } |
| 110 | |
| 111 | if ( $encode_mailtos ) { |
| 112 | if ( ! ( function_exists( 'et_fb_enabled' ) && et_fb_enabled() ) ) { |
| 113 | $filtered = $this->filter_plain_emails( $filtered, function ( $match ) use ( $self ) { |
| 114 | return $self->createProtectedMailto( $match[0], array( 'href' => 'mailto:' . $match[0] ), 'without_javascript' ); |
| 115 | }, $replace_by); |
| 116 | } else { |
| 117 | $filtered = $this->filter_plain_emails( $filtered, null, $replace_by ); |
| 118 | } |
| 119 | } else { |
| 120 | $filtered = $this->filter_plain_emails( $filtered, null, $replace_by ); |
| 121 | } |
| 122 | |
| 123 | break; |
| 124 | case 'with_javascript': |
| 125 | $filtered = $this->filter_input_fields( $filtered, $protect_using ); |
| 126 | $filtered = $this->filter_mailto_links( $filtered ); |
| 127 | $filtered = $this->filter_custom_links( $filtered ); |
| 128 | |
| 129 | if ( $convert_plain_to_image ) { |
| 130 | $replace_by = 'convert_image'; |
| 131 | } else { |
| 132 | $replace_by = 'use_javascript'; |
| 133 | } |
| 134 | |
| 135 | if ( $encode_mailtos ) { |
| 136 | if ( ! ( function_exists( 'et_fb_enabled' ) && et_fb_enabled() ) ) { |
| 137 | $filtered = $this->filter_plain_emails( $filtered, function ( $match ) use ( $self ) { |
| 138 | return $self->createProtectedMailto( $match[0], array( 'href' => 'mailto:' . $match[0] ), 'with_javascript' ); |
| 139 | }, $replace_by); |
| 140 | } else { |
| 141 | $filtered = $this->filter_plain_emails( $filtered, null, $replace_by ); |
| 142 | } |
| 143 | } else { |
| 144 | $filtered = $this->filter_plain_emails( $filtered, null, $replace_by ); |
| 145 | } |
| 146 | |
| 147 | break; |
| 148 | } |
| 149 | |
| 150 | //Revalidate filtered emails that should not be encoded |
| 151 | $filtered = $this->tempEncodeAtSymbol( $filtered, true ); |
| 152 | |
| 153 | return $filtered; |
| 154 | } |
| 155 | |
| 156 | /** |
| 157 | * Emails will be replaced by '*protected email*' |
| 158 | * @param string $content |
| 159 | * @param string|callable $replace_by Optional |
| 160 | * @param string $protection_method Optional |
| 161 | * @param mixed $show_encoded_check Optional |
| 162 | * @return string |
| 163 | */ |
| 164 | public function filter_plain_emails( $content, $replace_by = null, $protection_method = 'default', $show_encoded_check = 'default' ) |
| 165 | { |
| 166 | |
| 167 | if ( $show_encoded_check === 'default' ) { |
| 168 | $show_encoded_check = (bool) $this->getSetting( 'show_encoded_check', true ); |
| 169 | } |
| 170 | |
| 171 | if ( $replace_by === null ) { |
| 172 | $replace_by = __( $this->getSetting( 'protection_text', true ), 'email-encoder-bundle' ); |
| 173 | } |
| 174 | |
| 175 | $self = $this; |
| 176 | |
| 177 | return preg_replace_callback( $this->settings()->get_email_regex(), function ( $matches ) use ( $replace_by, $protection_method, $show_encoded_check, $self ) { |
| 178 | // workaround to skip responsive image names containing @ |
| 179 | $extention = strtolower( $matches[4] ); |
| 180 | $excludedList = array( |
| 181 | '.jpg', |
| 182 | '.jpeg', |
| 183 | '.png', |
| 184 | '.gif', |
| 185 | '.svg', |
| 186 | '.webp', |
| 187 | '.bmp', |
| 188 | '.tiff', |
| 189 | '.avif', |
| 190 | ); |
| 191 | |
| 192 | //Added in 2.1.1 |
| 193 | $excludedList = apply_filters( 'eeb/validate/excluded_image_urls', $excludedList ); |
| 194 | |
| 195 | if ( in_array( $extention, $excludedList ) ) { |
| 196 | return $matches[0]; |
| 197 | } |
| 198 | |
| 199 | if ( is_callable( $replace_by ) ) { |
| 200 | return call_user_func( $replace_by, $matches, $protection_method ); |
| 201 | } |
| 202 | |
| 203 | if ( $protection_method === 'char_encode' ) { |
| 204 | $protected_return = antispambot( $matches[0] ); |
| 205 | } elseif ( $protection_method === 'convert_image' ) { |
| 206 | |
| 207 | $image_link = $self->generateEmailImageUrl( $matches[0] ); |
| 208 | if ( ! empty( $image_link ) ) { |
| 209 | $protected_return = '<img src="' . $image_link . '" />'; |
| 210 | } else { |
| 211 | $protected_return = antispambot( $matches[0] ); |
| 212 | } |
| 213 | |
| 214 | } elseif ( $protection_method === 'use_javascript' ) { |
| 215 | $protection_text = __( $this->getSetting( 'protection_text', true ), 'email-encoder-bundle' ); |
| 216 | $protected_return = $this->dynamicJsEmailEncoding( $matches[0], $protection_text ); |
| 217 | } elseif ( $protection_method === 'use_css' ) { |
| 218 | $protection_text = __( $this->getSetting( 'protection_text', true ), 'email-encoder-bundle' ); |
| 219 | // $protected_return = $this->validate()->encoding->encode_email_css( $matches[0], $protection_text ); |
| 220 | $protected_return = $this->validate()->encoding->encode_email_css( $matches[0] ); |
| 221 | } elseif ( $protection_method === 'no_encoding' ) { |
| 222 | $protected_return = $matches[0]; |
| 223 | } else { |
| 224 | $protected_return = $replace_by; |
| 225 | } |
| 226 | |
| 227 | // mark link as successfully encoded (for admin users) |
| 228 | if ( current_user_can( $this->getAdminCap( 'frontend-display-security-check' ) ) && $show_encoded_check ) { |
| 229 | $protected_return .= $this->getEncodedEmailIcon(); |
| 230 | } |
| 231 | |
| 232 | return $protected_return; |
| 233 | }, $content ) ?? ''; |
| 234 | } |
| 235 | |
| 236 | /** |
| 237 | * Filter passed input fields |
| 238 | * |
| 239 | * @param string $content |
| 240 | * @return string |
| 241 | */ |
| 242 | public function filter_input_fields( string $content, string $encoding_method = 'default' ) |
| 243 | { |
| 244 | $strong_encoding = (bool) $this->getSetting( 'input_strong_protection', true, 'filter_body' ); |
| 245 | |
| 246 | $callback_encode_input_fields = function ( $match ) use ( $encoding_method, $strong_encoding ) { |
| 247 | $input = $match[0]; |
| 248 | $email = $match[2]; |
| 249 | |
| 250 | //Only allow strong encoding if javascript is supported |
| 251 | if ( $encoding_method === 'without_javascript' ) { |
| 252 | $strong_encoding = false; |
| 253 | } |
| 254 | |
| 255 | return $this->validate()->encoding->encode_input_field( $input, $email, $strong_encoding ); |
| 256 | }; |
| 257 | |
| 258 | $regexpInputField = '/<input([^>]*)value=["\'][\s+]*' . $this->settings()->get_email_regex( true ) . '[\s+]*["\']([^>]*)>/is'; |
| 259 | |
| 260 | return preg_replace_callback( $regexpInputField, $callback_encode_input_fields, $content ) ?? ''; |
| 261 | } |
| 262 | |
| 263 | /** |
| 264 | * @param string $content |
| 265 | * @param string $protection_method |
| 266 | * @return string |
| 267 | */ |
| 268 | public function filter_mailto_links( string $content, ?string $protection_method = null ) |
| 269 | { |
| 270 | $self = $this; |
| 271 | |
| 272 | $callbackEncodeMailtoLinks = function ( $match ) use ( $self, $protection_method ) { |
| 273 | $attrs = $this->helper()->parse_html_attributes( $match[1] ); |
| 274 | return $self->createProtectedMailto( $match[4], $attrs, $protection_method ); |
| 275 | }; |
| 276 | |
| 277 | $regexpMailtoLink = '/<a[\s+]*(([^>]*)href=["\']mailto\:([^>]*)["\' ])>(.*?)<\/a[\s+]*>/is'; |
| 278 | |
| 279 | return preg_replace_callback( $regexpMailtoLink, $callbackEncodeMailtoLinks, $content ) ?? ''; |
| 280 | } |
| 281 | |
| 282 | /** |
| 283 | * @param string $content |
| 284 | * @param string $protection_method |
| 285 | * @return string |
| 286 | */ |
| 287 | public function filter_custom_links( string $content, ?string $protection_method = null ) |
| 288 | { |
| 289 | $self = $this; |
| 290 | $custom_href_attr = (string) $this->getSetting( 'custom_href_attr', true ); |
| 291 | |
| 292 | if ( ! empty( $custom_href_attr ) ) { |
| 293 | $custom_attr_list = explode( ',', $custom_href_attr ); |
| 294 | foreach ( $custom_attr_list as $s_attr ) { |
| 295 | $attr_name = trim( $s_attr ); |
| 296 | |
| 297 | $callbackEncodeCustomLinks = function ( $match ) use ( $self, $protection_method ) { |
| 298 | $attrs = shortcode_parse_atts( $match[1] ); |
| 299 | return $self->createProtectedHrefAtt( $match[4], $attrs, $protection_method ); |
| 300 | }; |
| 301 | |
| 302 | $regexpMailtoLink = '/<a[\s+]*(([^>]*)href=["\']' . addslashes( $attr_name ) . '\:([^>]*)["\' ])>(.*?)<\/a[\s+]*>/is'; |
| 303 | |
| 304 | $content = preg_replace_callback( $regexpMailtoLink, $callbackEncodeCustomLinks, $content ) ?? ''; |
| 305 | } |
| 306 | } |
| 307 | |
| 308 | return $content; |
| 309 | } |
| 310 | |
| 311 | /** |
| 312 | * Emails will be replaced by '*protected email*' |
| 313 | * |
| 314 | * @param string $content |
| 315 | * @param string $protection_type |
| 316 | * @return string |
| 317 | */ |
| 318 | public function filter_rss( string $content, ?string $protection_type ) |
| 319 | { |
| 320 | |
| 321 | if ( $protection_type === 'strong_method' ) { |
| 322 | $filtered = $this->filter_plain_emails( $content ); |
| 323 | } else { |
| 324 | $filtered = $this->filter_plain_emails( $content, null, 'char_encode' ); |
| 325 | } |
| 326 | |
| 327 | return $filtered; |
| 328 | } |
| 329 | |
| 330 | /** |
| 331 | * Filter plain emails using soft attributes |
| 332 | * |
| 333 | * @param string $content - the content that should be soft filtered |
| 334 | * @param string $protection_method - The method (E.g. char_encode) |
| 335 | * @return string |
| 336 | */ |
| 337 | public function filter_soft_attributes( string $content, string $protection_method ) |
| 338 | { |
| 339 | $soft_attributes = (array) $this->settings()->get_soft_attribute_regex(); |
| 340 | |
| 341 | foreach ( $soft_attributes as $ident => $regex ) { |
| 342 | |
| 343 | // $attributes = array(); |
| 344 | preg_match_all( $regex, $content, $attributes ); |
| 345 | |
| 346 | // if ( is_array( $attributes ) && isset( $attributes[0] ) ) { |
| 347 | foreach ( $attributes[0] as $single ) { |
| 348 | |
| 349 | if ( empty( $single ) ) { |
| 350 | continue; |
| 351 | } |
| 352 | |
| 353 | $content = str_replace( $single, $this->filter_plain_emails( $single, null, $protection_method, false ), $content ); |
| 354 | } |
| 355 | // } |
| 356 | |
| 357 | } |
| 358 | |
| 359 | return $content; |
| 360 | } |
| 361 | |
| 362 | /** |
| 363 | * Filter plain emails using soft dom attributes |
| 364 | * |
| 365 | * @param string $content - the content that should be soft filtered |
| 366 | * @param string $protection_method - The method (E.g. char_encode) |
| 367 | * @return string |
| 368 | */ |
| 369 | public function filter_soft_dom_attributes( $content, $protection_method ) |
| 370 | { |
| 371 | $content = (string) $content; |
| 372 | |
| 373 | $no_script_tags = (bool) $this->getSetting( 'no_script_tags', true, 'filter_body' ); |
| 374 | $no_attribute_validation = (bool) $this->getSetting( 'no_attribute_validation', true, 'filter_body' ); |
| 375 | |
| 376 | if ( $content !== '' ) { |
| 377 | |
| 378 | if ( class_exists( 'DOMDocument' ) ) { |
| 379 | $dom = new DOMDocument(); |
| 380 | @$dom->loadHTML($content); |
| 381 | |
| 382 | //Filter html attributes |
| 383 | if ( ! $no_attribute_validation ) { |
| 384 | $allNodes = $dom->getElementsByTagName('*'); |
| 385 | foreach ( $allNodes as $snote ) { |
| 386 | if ( $snote->hasAttributes() ) { |
| 387 | foreach ( $snote->attributes as $attr ) { |
| 388 | if ( $attr->nodeName == 'href' || $attr->nodeName == 'src' ) { |
| 389 | continue; |
| 390 | } |
| 391 | |
| 392 | if ( $attr->nodeValue !== null && strpos( $attr->nodeValue, '@' ) !== false ) { |
| 393 | // $single_tags = array(); |
| 394 | preg_match_all( '/' . $attr->nodeName . '=["\']([^"]*)["\']/i', $content, $single_tags ); |
| 395 | |
| 396 | // if ( is_array( $single_tags ) && isset( $single_tags[0] ) ) { |
| 397 | foreach ( $single_tags[0] as $single ) { |
| 398 | |
| 399 | if ( empty( $single ) ) { |
| 400 | continue; |
| 401 | } |
| 402 | |
| 403 | $content = str_replace( $single, $this->filter_plain_emails( $single, null, $protection_method, false ), $content ); |
| 404 | } |
| 405 | // } |
| 406 | |
| 407 | } |
| 408 | } |
| 409 | } |
| 410 | } |
| 411 | } |
| 412 | |
| 413 | //Keep for now |
| 414 | //Soft-encode scripts |
| 415 | // $script = $dom->getElementsByTagName('script'); |
| 416 | // if ( ! empty( $script ) ) { |
| 417 | // $scripts_encoded = true; |
| 418 | |
| 419 | // if ( ! $no_script_tags ) { |
| 420 | // foreach( $script as $item ) { |
| 421 | // $content = str_replace( $item->nodeValue, $this->filter_plain_emails( $item->nodeValue, null, $protection_method, false ), $content ); |
| 422 | // } |
| 423 | // } else { |
| 424 | // foreach( $script as $item ) { |
| 425 | // $content = str_replace( $item->nodeValue, $this->temp_encode_at_symbol( $item->nodeValue ), $content ); |
| 426 | // } |
| 427 | // } |
| 428 | // } |
| 429 | |
| 430 | } |
| 431 | |
| 432 | //Validate script tags for better encoding |
| 433 | $pattern = '/<script\b[^>]*>(.*?)<\/script>/is'; |
| 434 | |
| 435 | preg_match_all($pattern, $content, $matches); |
| 436 | if ( ! empty( $matches[1] ) ) { |
| 437 | if ( ! $no_script_tags ) { |
| 438 | foreach ( $matches[1] as $key => $item ) { |
| 439 | |
| 440 | //Don't do anything if something doesn't add up |
| 441 | if ( ! isset( $matches[0][ $key ] ) ) { |
| 442 | continue; |
| 443 | } |
| 444 | |
| 445 | $org_script = $matches[0][ $key ]; |
| 446 | |
| 447 | //Only encode emails when a CDATA is given to not cause any break within the scripts |
| 448 | if ( strpos( $item, '<![CDATA' ) !== false ) { |
| 449 | $validated_script = str_replace( $item, $this->filter_plain_emails( $item, null, $protection_method, false ), $org_script ); |
| 450 | } else { |
| 451 | $validated_script = str_replace( $item, $this->tempEncodeAtSymbol( $item ), $org_script ); |
| 452 | } |
| 453 | |
| 454 | $content = str_replace( $org_script, $validated_script, $content ); |
| 455 | } |
| 456 | } else { |
| 457 | foreach ( $matches[1] as $key => $item ) { |
| 458 | |
| 459 | //Don't do anything if something doesn't add up |
| 460 | if ( ! isset( $matches[0][ $key ] ) ) { |
| 461 | continue; |
| 462 | } |
| 463 | |
| 464 | $org_script = $matches[0][ $key ]; |
| 465 | $validated_script = str_replace( $item, $this->tempEncodeAtSymbol( $item ), $org_script ); |
| 466 | |
| 467 | $content = str_replace( $org_script, $validated_script, $content ); |
| 468 | } |
| 469 | } |
| 470 | } |
| 471 | |
| 472 | } |
| 473 | |
| 474 | return $content; |
| 475 | } |
| 476 | } |
| 477 |