| 1 |
<?php |
| 2 |
|
| 3 |
namespace Linguise\WordPress; |
| 4 |
|
| 5 |
use Linguise\WordPress\FragmentHandler; |
| 6 |
use Linguise\WordPress\HTMLHelper; |
| 7 |
use Linguise\Vendor\JsonPath\JsonObject; |
| 8 |
use Linguise\Vendor\Linguise\Script\Core\Debug; |
| 9 |
|
| 10 |
defined('ABSPATH') || die(''); |
| 11 |
|
| 12 |
/** |
| 13 |
* AttributeHandler, derived from FragmentHandler |
| 14 |
* |
| 15 |
* This class handle parsing and injecting |
| 16 |
* JSON data from HTML attributes. |
| 17 |
*/ |
| 18 |
class AttributeHandler extends FragmentHandler |
| 19 |
{ |
| 20 |
/** |
| 21 |
* Get attribute JSON fragment matching |
| 22 |
* |
| 23 |
* Similar to `getJSONOverrideMatcher()` but we use DOMDocument to do the matching |
| 24 |
* - name: The name of the plugin, e.g. 'wpca', must be unique |
| 25 |
* - key: the data attribute to match, e.g. 'data-wpca' |
| 26 |
* - encode: If the JSON data is URL encoded or not, default to false (optional) |
| 27 |
* - strict: If the JSON data should be strict or not, default to false (optional) |
| 28 |
* - mode: The mode of the matcher, default to 'json', can be a 'string' or 'json' (optional) |
| 29 |
* - entity: The entity name to match, e.g. 'sc-wpca', default to nil (optional) |
| 30 |
* |
| 31 |
* @param string $html_data The HTML data input |
| 32 |
* |
| 33 |
* @return array The array of JSON to match with fragment |
| 34 |
*/ |
| 35 |
private static function getMatcher($html_data) |
| 36 |
{ |
| 37 |
$current_list = []; |
| 38 |
|
| 39 |
if (defined('LINGUISE_WP_PLUGIN_TEST_MODE')) { |
| 40 |
$current_list[] = [ |
| 41 |
'name' => 'linguise-demo-test', |
| 42 |
'key' => 'data-linguise-demo', |
| 43 |
]; |
| 44 |
|
| 45 |
$current_list[] = [ |
| 46 |
'name' => 'linguise-demo-string', |
| 47 |
'key' => 'data-label', |
| 48 |
'mode' => 'string', |
| 49 |
'matchers' => [ |
| 50 |
[ |
| 51 |
'key' => 'sc-linguise-demo', |
| 52 |
'type' => 'tag' |
| 53 |
] |
| 54 |
], |
| 55 |
]; |
| 56 |
|
| 57 |
$current_list[] = [ |
| 58 |
'name' => 'linguise-demo-string-encoded', |
| 59 |
'key' => 'data-label-encoded', |
| 60 |
'mode' => 'string-encoded', |
| 61 |
'matchers' => [ |
| 62 |
[ |
| 63 |
'key' => 'sc-linguise-demo-encoded', |
| 64 |
'type' => 'tag' |
| 65 |
] |
| 66 |
], |
| 67 |
]; |
| 68 |
} |
| 69 |
|
| 70 |
$current_list = apply_filters('linguise_fragment_attributes', $current_list, $html_data); |
| 71 |
|
| 72 |
// loop through the list and add extra field, use reference |
| 73 |
foreach ($current_list as &$matcher) { |
| 74 |
$name_sanitized = str_replace(' ', '-', $matcher['name']); |
| 75 |
$name_sanitized = sanitize_key($name_sanitized); |
| 76 |
$matcher['attr_ids'] = 'data-linguise-attribute-' . $name_sanitized; |
| 77 |
} |
| 78 |
|
| 79 |
return $current_list; |
| 80 |
} |
| 81 |
|
| 82 |
/** |
| 83 |
* Find the matcher by name. |
| 84 |
* |
| 85 |
* @param string $matcher_name The name of the matcher to find |
| 86 |
* @param array $matchers The list of matchers to search in |
| 87 |
* |
| 88 |
* @return array|null The matcher if found, null otherwise |
| 89 |
*/ |
| 90 |
private static function findMatcher($matcher_name, $matchers) |
| 91 |
{ |
| 92 |
foreach ($matchers as $matcher) { |
| 93 |
if ($matcher['name'] === $matcher_name) { |
| 94 |
return $matcher; |
| 95 |
} |
| 96 |
} |
| 97 |
|
| 98 |
return null; |
| 99 |
} |
| 100 |
|
| 101 |
/** |
| 102 |
* Process a list of matchers and see if it would match the entity. |
| 103 |
* |
| 104 |
* @param array $matchers The list of matchers to process |
| 105 |
* @param \DomElement $entity The entity to match against |
| 106 |
* @param string $match_mode The match mode, can be 'any' or 'all', default to 'any' |
| 107 |
* |
| 108 |
* @return boolean True if the entity matches any of the matchers, false otherwise |
| 109 |
*/ |
| 110 |
private static function processMatcher($matchers, $entity, $match_mode = 'any') |
| 111 |
{ |
| 112 |
if (empty($matchers)) { |
| 113 |
return true; |
| 114 |
} |
| 115 |
|
| 116 |
$matchers_results = []; |
| 117 |
foreach ($matchers as $matcher) { |
| 118 |
$kind = isset($matcher['type']) ? $matcher['type'] : 'attribute'; |
| 119 |
if ($kind === 'attribute') { |
| 120 |
$key = isset($matcher['key']) ? $matcher['key'] : ''; |
| 121 |
if (!empty($key) && $entity->hasAttribute($key)) { |
| 122 |
$value_orig = $entity->getAttribute($key); |
| 123 |
if (isset($matcher['value'])) { |
| 124 |
$matchers_results[] = $value_orig === $matcher['value']; |
| 125 |
} else { |
| 126 |
$matchers_results[] = true; |
| 127 |
} |
| 128 |
} |
| 129 |
} elseif ($kind === 'tag') { |
| 130 |
// Check with tag name |
| 131 |
if (isset($matcher['key']) && $matcher['key'] === $entity->tagName) { |
| 132 |
$matchers_results[] = true; |
| 133 |
} else { |
| 134 |
$matchers_results[] = false; |
| 135 |
} |
| 136 |
} elseif ($kind === 'class') { |
| 137 |
// Values |
| 138 |
$values = isset($matcher['value']) ? $matcher['value'] : []; |
| 139 |
$class_list = explode(' ', $entity->getAttribute('class')); |
| 140 |
$mode = isset($matcher['mode']) ? $matcher['mode'] : 'any'; |
| 141 |
if ($mode === 'all') { |
| 142 |
// If all of the class matches, we return true |
| 143 |
$matchers_results[] = empty(array_diff($values, $class_list)); |
| 144 |
} else { |
| 145 |
// If any of the class matches, we return true |
| 146 |
$matchers_results[] = !empty(array_intersect($values, $class_list)); |
| 147 |
} |
| 148 |
} |
| 149 |
} |
| 150 |
|
| 151 |
if ($match_mode === 'all') { |
| 152 |
// If all of the matchers matched, we return true |
| 153 |
return !in_array(false, $matchers_results, true); |
| 154 |
} else { |
| 155 |
// If any of the matchers matched, we return true |
| 156 |
return in_array(true, $matchers_results, true); |
| 157 |
} |
| 158 |
} |
| 159 |
|
| 160 |
/** |
| 161 |
* Parse the HTML input or data into the fragments. |
| 162 |
* |
| 163 |
* @param string $html_data The HTML data to be parsed |
| 164 |
* |
| 165 |
* @return array |
| 166 |
*/ |
| 167 |
public static function findWPFragments(&$html_data) |
| 168 |
{ |
| 169 |
$html_dom = HTMLHelper::loadHTML($html_data); |
| 170 |
if (empty($html_dom)) { |
| 171 |
return []; |
| 172 |
} |
| 173 |
|
| 174 |
$matchers = self::getMatcher($html_data); |
| 175 |
$internal_counter = 0; |
| 176 |
|
| 177 |
$all_fragments = []; |
| 178 |
// Loop through all valid DOM elements |
| 179 |
$elements = $html_dom->getElementsByTagName('*'); |
| 180 |
foreach ($elements as $element) { |
| 181 |
foreach ($matchers as $matcher) { |
| 182 |
$key_data = $element->getAttribute($matcher['key']); |
| 183 |
if (empty($key_data)) { |
| 184 |
continue; |
| 185 |
} |
| 186 |
|
| 187 |
// Unroll the entity |
| 188 |
$key_data = html_entity_decode(HTMLHelper::unprotectEntity($key_data), ENT_QUOTES, 'UTF-8'); |
| 189 |
|
| 190 |
// Is the data URL encoded? |
| 191 |
$should_encode = isset($matcher['encode']) && $matcher['encode']; |
| 192 |
if ($should_encode) { |
| 193 |
$key_data = urldecode($key_data); |
| 194 |
} |
| 195 |
|
| 196 |
$additional_matchers = isset($matcher['matchers']) ? $matcher['matchers'] : []; |
| 197 |
$additional_match_mode = isset($matcher['match_mode']) ? $matcher['match_mode'] : 'any'; |
| 198 |
if (!empty($additional_matchers)) { |
| 199 |
$is_matched = self::processMatcher($additional_matchers, $element, $additional_match_mode); |
| 200 |
if (!$is_matched) { |
| 201 |
// Skip unmatched element |
| 202 |
continue; |
| 203 |
} |
| 204 |
} |
| 205 |
|
| 206 |
if (isset($matcher['entity']) && $matcher['entity'] !== $element->tagName) { |
| 207 |
// Skip unmatched entity name |
| 208 |
continue; |
| 209 |
} |
| 210 |
|
| 211 |
if (isset($matcher['mode']) && $matcher['mode'] === 'string') { |
| 212 |
// If the mode is a string we just pass the string as is |
| 213 |
$cast_data = isset($matcher['cast']) ? $matcher['cast'] : 'html-main'; |
| 214 |
$collected_temp = [ |
| 215 |
[ |
| 216 |
'key' => $matcher['key'], |
| 217 |
'value' => $key_data, |
| 218 |
'format' => $cast_data, |
| 219 |
] |
| 220 |
]; |
| 221 |
} elseif (isset($matcher['mode']) && $matcher['mode'] === 'string-encoded') { |
| 222 |
$cast_data = isset($matcher['cast']) ? $matcher['cast'] : 'html-main'; |
| 223 |
|
| 224 |
$collected_temp = [ |
| 225 |
[ |
| 226 |
'key' => $matcher['key'], |
| 227 |
'value' => html_entity_decode($key_data, ENT_QUOTES | ENT_SUBSTITUTE, 'utf-8'), |
| 228 |
'format' => $cast_data, |
| 229 |
] |
| 230 |
]; |
| 231 |
} else { |
| 232 |
$json_data = json_decode($key_data, true); |
| 233 |
if (json_last_error() !== JSON_ERROR_NONE) { |
| 234 |
continue; |
| 235 |
} |
| 236 |
|
| 237 |
$is_strict = isset($matcher['strict']) && $matcher['strict']; |
| 238 |
$collected_temp = self::collectFragmentFromJson($json_data, $is_strict); |
| 239 |
} |
| 240 |
|
| 241 |
if (!empty($collected_temp)) { |
| 242 |
$int_marker = $matcher['attr_ids'] . '-' . $internal_counter; |
| 243 |
$element->setAttribute($matcher['attr_ids'], $int_marker); |
| 244 |
$all_fragments[$matcher['name']][$int_marker] = [ |
| 245 |
'mode' => 'attribute', |
| 246 |
'attribute' => $element->tagName, |
| 247 |
'fragments' => $collected_temp, |
| 248 |
]; |
| 249 |
$internal_counter++; |
| 250 |
} |
| 251 |
} |
| 252 |
} |
| 253 |
|
| 254 |
if (!empty($all_fragments)) { |
| 255 |
$html_data = HTMLHelper::saveHTML($html_dom); |
| 256 |
} |
| 257 |
|
| 258 |
Debug::log('AttributeHandler -> Collected: ' . json_encode($all_fragments, JSON_PRETTY_PRINT)); |
| 259 |
|
| 260 |
return $all_fragments; |
| 261 |
} |
| 262 |
|
| 263 |
/** |
| 264 |
* Find the element by attribute and matcher. |
| 265 |
* |
| 266 |
* @param \DOMDocument $html_dom The HTML DOM object |
| 267 |
* @param string $tag_name The tag name to search for |
| 268 |
* @param string $attr_name The attribute name to search for |
| 269 |
* @param string $attr_value The attribute value to search for |
| 270 |
* |
| 271 |
* @return \DOMElement|null The found element or null if not found |
| 272 |
*/ |
| 273 |
private static function findElementByAttributeAndMatcher($html_dom, $tag_name, $attr_name, $attr_value) |
| 274 |
{ |
| 275 |
$elements = $html_dom->getElementsByTagName($tag_name); |
| 276 |
foreach ($elements as $element) { |
| 277 |
if ($element->nodeType !== XML_ELEMENT_NODE) { |
| 278 |
continue; |
| 279 |
} |
| 280 |
|
| 281 |
if ($element->hasAttribute($attr_name) && $element->getAttribute($attr_name) === $attr_value) { |
| 282 |
return $element; |
| 283 |
} |
| 284 |
} |
| 285 |
return null; |
| 286 |
} |
| 287 |
|
| 288 |
/** |
| 289 |
* Inject the fragments into the HTML data. |
| 290 |
* |
| 291 |
* @param string $html_data The HTML data to be injected |
| 292 |
* @param array $fragments The array of fragments to be injected, from intoJSONFragments |
| 293 |
* |
| 294 |
* @return string |
| 295 |
*/ |
| 296 |
public static function injectTranslatedWPFragments($html_data, $fragments) |
| 297 |
{ |
| 298 |
if (empty($fragments)) { |
| 299 |
return $html_data; |
| 300 |
} |
| 301 |
|
| 302 |
$html_dom = HTMLHelper::loadHTML($html_data); |
| 303 |
if (empty($html_dom)) { |
| 304 |
return $html_data; |
| 305 |
} |
| 306 |
|
| 307 |
Debug::log('AttributeHandler -> Injecting: ' . json_encode($fragments, JSON_PRETTY_PRINT)); |
| 308 |
|
| 309 |
$fragment_matchers = self::getMatcher($html_data); |
| 310 |
foreach ($fragments as $fragment_name => $fragment_jsons) { |
| 311 |
$matched = self::findMatcher($fragment_name, $fragment_matchers); |
| 312 |
foreach ($fragment_jsons as $fragment_param => $fragment_list) { |
| 313 |
self::removeFragmentMarkers($html_dom, $fragment_name, $fragment_param); |
| 314 |
|
| 315 |
if (!isset($fragment_list['attribute'])) { |
| 316 |
continue; |
| 317 |
} |
| 318 |
if (empty($matched)) { |
| 319 |
continue; |
| 320 |
} |
| 321 |
|
| 322 |
$attr_html = self::findElementByAttributeAndMatcher( |
| 323 |
$html_dom, |
| 324 |
$fragment_list['attribute'], |
| 325 |
$matched['attr_ids'], |
| 326 |
$fragment_param |
| 327 |
); |
| 328 |
|
| 329 |
if (empty($attr_html)) { |
| 330 |
continue; |
| 331 |
} |
| 332 |
|
| 333 |
$match_data = $attr_html->getAttribute($matched['key']); |
| 334 |
if (empty($match_data)) { |
| 335 |
continue; |
| 336 |
} |
| 337 |
|
| 338 |
// Since we have protection enabled around this! |
| 339 |
$match_data = html_entity_decode(HTMLHelper::unprotectEntity($match_data), ENT_QUOTES, 'UTF-8'); |
| 340 |
|
| 341 |
$should_encode = isset($matched['encode']) && $matched['encode']; |
| 342 |
if ($should_encode) { |
| 343 |
$match_data = urldecode($match_data); |
| 344 |
} |
| 345 |
|
| 346 |
// Check mode, if string mode, we just pass the string as is |
| 347 |
if (isset($matched['mode']) && $matched['mode'] === 'string') { |
| 348 |
// Get first item |
| 349 |
$first_fragment = isset($fragment_list['fragments'][0]) ? $fragment_list['fragments'][0] : null; |
| 350 |
if (empty($first_fragment)) { |
| 351 |
continue; |
| 352 |
} |
| 353 |
|
| 354 |
$first_value = isset($first_fragment['value']) ? $first_fragment['value'] : null; |
| 355 |
if (empty($first_value)) { |
| 356 |
continue; |
| 357 |
} |
| 358 |
|
| 359 |
// Replace the data |
| 360 |
$replaced_text = $first_value; |
| 361 |
if ($should_encode) { |
| 362 |
$replaced_text = rawurlencode($replaced_text); |
| 363 |
} |
| 364 |
|
| 365 |
$protected_json = HTMLHelper::protectEntity($replaced_text); |
| 366 |
} elseif (isset($matched['mode']) && $matched['mode'] === 'string-encoded') { |
| 367 |
// Get first item |
| 368 |
$first_fragment = isset($fragment_list['fragments'][0]) ? $fragment_list['fragments'][0] : null; |
| 369 |
if (empty($first_fragment)) { |
| 370 |
continue; |
| 371 |
} |
| 372 |
|
| 373 |
$first_value = isset($first_fragment['value']) ? $first_fragment['value'] : null; |
| 374 |
if (empty($first_value)) { |
| 375 |
continue; |
| 376 |
} |
| 377 |
|
| 378 |
// Replace the data |
| 379 |
$replaced_text = htmlspecialchars($first_value, ENT_QUOTES, 'UTF-8', false); |
| 380 |
if ($should_encode) { |
| 381 |
$replaced_text = rawurlencode($replaced_text); |
| 382 |
} |
| 383 |
|
| 384 |
$protected_json = HTMLHelper::protectEntity($replaced_text); |
| 385 |
} else { |
| 386 |
// JSON mode, we need to decode the JSON data |
| 387 |
$json_decoded = json_decode($match_data, true); |
| 388 |
if (json_last_error() !== JSON_ERROR_NONE) { |
| 389 |
continue; |
| 390 |
} |
| 391 |
|
| 392 |
$json_data = new JsonObject($json_decoded); |
| 393 |
foreach ($fragment_list['fragments'] as $fragment) { |
| 394 |
$dec_key = self::unwrapKey($fragment['key']); |
| 395 |
try { |
| 396 |
$json_data->set('$.' . $dec_key, $fragment['value']); |
| 397 |
} catch (\Linguise\Vendor\JsonPath\InvalidJsonPathException $e) { |
| 398 |
Debug::log('Failed to set key in attributes: ' . $dec_key . ' -> ' . $e->getMessage()); |
| 399 |
} |
| 400 |
} |
| 401 |
|
| 402 |
$replaced_json = $json_data->getJson(); |
| 403 |
if ($should_encode) { |
| 404 |
$replaced_json = rawurlencode($replaced_json); |
| 405 |
} |
| 406 |
|
| 407 |
// Protect the entity back |
| 408 |
$protected_json = HTMLHelper::protectEntity(htmlspecialchars($replaced_json, ENT_QUOTES, 'UTF-8', false)); |
| 409 |
} |
| 410 |
|
| 411 |
$attr_html->setAttribute($matched['key'], $protected_json); |
| 412 |
$attr_html->removeAttribute($matched['attr_ids']); |
| 413 |
} |
| 414 |
} |
| 415 |
|
| 416 |
$html_data = HTMLHelper::saveHTML($html_dom); |
| 417 |
|
| 418 |
// Unmangle stuff like &#xE5; |
| 419 |
$html_data = preg_replace('/&#x([0-9A-Fa-f]+);/', '&#x$1;', $html_data); |
| 420 |
|
| 421 |
$mod_html_data = apply_filters('linguise_after_attribute_translation', $html_data); |
| 422 |
if (!empty($mod_html_data)) { |
| 423 |
$html_data = $mod_html_data; |
| 424 |
} |
| 425 |
|
| 426 |
return $html_data; |
| 427 |
} |
| 428 |
} |
| 429 |
|
