src/Util/XML.php

   1 <?php
   2 /**
   3  * @file src/Util/XML.php
   4  */
   5 namespace Friendica\Util;
   6
   7 use DOMXPath;
   8 use Friendica\Core\Logger;
   9 use Friendica\Core\System;
  10 use SimpleXMLElement;
  11
  12 /**
  13  * @brief This class contain methods to work with XML data
  14  */
  15 class XML
  16 {
  17         /**
  18          * @brief Creates an XML structure out of a given array
  19          *
  20          * @param array  $array         The array of the XML structure that will be generated
  21          * @param object $xml           The createdXML will be returned by reference
  22          * @param bool   $remove_header Should the XML header be removed or not?
  23          * @param array  $namespaces    List of namespaces
  24          * @param bool   $root          interally used parameter. Mustn't be used from outside.
  25          *
  26          * @return string The created XML
  27          */
  28         public static function fromArray($array, &$xml, $remove_header = false, $namespaces = [], $root = true)
  29         {
  30                 if ($root) {
  31                         foreach ($array as $key => $value) {
  32                                 foreach ($namespaces as $nskey => $nsvalue) {
  33                                         $key .= " xmlns".($nskey == "" ? "":":").$nskey.'="'.$nsvalue.'"';
  34                                 }
  35
  36                                 if (is_array($value)) {
  37                                         $root = new SimpleXMLElement("<".$key."/>");
  38                                         self::fromArray($value, $root, $remove_header, $namespaces, false);
  39                                 } else {
  40                                         $root = new SimpleXMLElement("<".$key.">".self::escape($value)."</".$key.">");
  41                                 }
  42
  43                                 $dom = dom_import_simplexml($root)->ownerDocument;
  44                                 $dom->formatOutput = true;
  45                                 $xml = $dom;
  46
  47                                 $xml_text = $dom->saveXML();
  48
  49                                 if ($remove_header) {
  50                                         $xml_text = trim(substr($xml_text, 21));
  51                                 }
  52
  53                                 return $xml_text;
  54                         }
  55                 }
  56
  57                 $element = null;
  58                 foreach ($array as $key => $value) {
  59                         if (!isset($element) && isset($xml)) {
  60                                 $element = $xml;
  61                         }
  62
  63                         if (is_integer($key)) {
  64                                 if (isset($element)) {
  65                                         if (is_scalar($value)) {
  66                                                 $element[0] = $value;
  67                                         } else {
  68                                                 /// @todo: handle nested array values
  69                                         }
  70                                 }
  71                                 continue;
  72                         }
  73
  74                         $element_parts = explode(":", $key);
  75                         if ((count($element_parts) > 1) && isset($namespaces[$element_parts[0]])) {
  76                                 $namespace = $namespaces[$element_parts[0]];
  77                         } elseif (isset($namespaces[""])) {
  78                                 $namespace = $namespaces[""];
  79                         } else {
  80                                 $namespace = null;
  81                         }
  82
  83                         // Remove undefined namespaces from the key
  84                         if ((count($element_parts) > 1) && is_null($namespace)) {
  85                                 $key = $element_parts[1];
  86                         }
  87
  88                         if (substr($key, 0, 11) == "@attributes") {
  89                                 if (!isset($element) || !is_array($value)) {
  90                                         continue;
  91                                 }
  92
  93                                 foreach ($value as $attr_key => $attr_value) {
  94                                         $element_parts = explode(":", $attr_key);
  95                                         if ((count($element_parts) > 1) && isset($namespaces[$element_parts[0]])) {
  96                                                 $namespace = $namespaces[$element_parts[0]];
  97                                         } else {
  98                                                 $namespace = null;
  99                                         }
 100
 101                                         $element->addAttribute($attr_key, $attr_value, $namespace);
 102                                 }
 103
 104                                 continue;
 105                         }
 106
 107                         if (!is_array($value)) {
 108                                 $element = $xml->addChild($key, self::escape($value), $namespace);
 109                         } elseif (is_array($value)) {
 110                                 $element = $xml->addChild($key, null, $namespace);
 111                                 self::fromArray($value, $element, $remove_header, $namespaces, false);
 112                         }
 113                 }
 114         }
 115
 116         /**
 117          * @brief Copies an XML object
 118          *
 119          * @param object $source      The XML source
 120          * @param object $target      The XML target
 121          * @param string $elementname Name of the XML element of the target
 122          * @return void
 123          */
 124         public static function copy(&$source, &$target, $elementname)
 125         {
 126                 if (count($source->children()) == 0) {
 127                         $target->addChild($elementname, self::escape($source));
 128                 } else {
 129                         $child = $target->addChild($elementname);
 130                         foreach ($source->children() as $childfield => $childentry) {
 131                                 self::copy($childentry, $child, $childfield);
 132                         }
 133                 }
 134         }
 135
 136         /**
 137          * @brief Create an XML element
 138          *
 139          * @param \DOMDocument $doc        XML root
 140          * @param string       $element    XML element name
 141          * @param string       $value      XML value
 142          * @param array        $attributes array containing the attributes
 143          *
 144          * @return \DOMElement XML element object
 145          */
 146         public static function createElement(\DOMDocument $doc, $element, $value = "", $attributes = [])
 147         {
 148                 $element = $doc->createElement($element, self::escape($value));
 149
 150                 foreach ($attributes as $key => $value) {
 151                         $attribute = $doc->createAttribute($key);
 152                         $attribute->value = self::escape($value);
 153                         $element->appendChild($attribute);
 154                 }
 155                 return $element;
 156         }
 157
 158         /**
 159          * @brief Create an XML and append it to the parent object
 160          *
 161          * @param \DOMDocument $doc        XML root
 162          * @param object $parent     parent object
 163          * @param string $element    XML element name
 164          * @param string $value      XML value
 165          * @param array  $attributes array containing the attributes
 166          * @return void
 167          */
 168         public static function addElement(\DOMDocument $doc, $parent, $element, $value = "", $attributes = [])
 169         {
 170                 $element = self::createElement($doc, $element, $value, $attributes);
 171                 $parent->appendChild($element);
 172         }
 173
 174         /**
 175          * @brief Convert an XML document to a normalised, case-corrected array
 176          *   used by webfinger
 177          *
 178          * @param object  $xml_element     The XML document
 179          * @param integer $recursion_depth recursion counter for internal use - default 0
 180          *                                 internal use, recursion counter
 181          *
 182          * @return array | string The array from the xml element or the string
 183          */
 184         public static function elementToArray($xml_element, &$recursion_depth = 0)
 185         {
 186                 // If we're getting too deep, bail out
 187                 if ($recursion_depth > 512) {
 188                         return(null);
 189                 }
 190
 191                 $xml_element_copy = '';
 192                 if (!is_string($xml_element)
 193                         && !is_array($xml_element)
 194                         && (get_class($xml_element) == 'SimpleXMLElement')
 195                 ) {
 196                         $xml_element_copy = $xml_element;
 197                         $xml_element = get_object_vars($xml_element);
 198                 }
 199
 200                 if (is_array($xml_element)) {
 201                         $result_array = [];
 202                         if (count($xml_element) <= 0) {
 203                                 return (trim(strval($xml_element_copy)));
 204                         }
 205
 206                         foreach ($xml_element as $key => $value) {
 207                                 $recursion_depth++;
 208                                 $result_array[strtolower($key)] = self::elementToArray($value, $recursion_depth);
 209                                 $recursion_depth--;
 210                         }
 211
 212                         if ($recursion_depth == 0) {
 213                                 $temp_array = $result_array;
 214                                 $result_array = [
 215                                         strtolower($xml_element_copy->getName()) => $temp_array,
 216                                 ];
 217                         }
 218
 219                         return ($result_array);
 220                 } else {
 221                         return (trim(strval($xml_element)));
 222                 }
 223         }
 224
 225         /**
 226          * @brief Convert the given XML text to an array in the XML structure.
 227          *
 228          * Xml::toArray() will convert the given XML text to an array in the XML structure.
 229          * Link: http://www.bin-co.com/php/scripts/xml2array/
 230          * Portions significantly re-written by mike@macgirvin.com for Friendica
 231          * (namespaces, lowercase tags, get_attribute default changed, more...)
 232          *
 233          * Examples: $array =  Xml::toArray(file_get_contents('feed.xml'));
 234          *        $array =  Xml::toArray(file_get_contents('feed.xml', true, 1, 'attribute'));
 235          *
 236          * @param object  $contents         The XML text
 237          * @param boolean $namespaces       True or false include namespace information
 238          *                                  in the returned array as array elements.
 239          * @param integer $get_attributes   1 or 0. If this is 1 the function will get the attributes as well as the tag values -
 240          *                                  this results in a different array structure in the return value.
 241          * @param string  $priority         Can be 'tag' or 'attribute'. This will change the way the resulting
 242          *                                  array sturcture. For 'tag', the tags are given more importance.
 243          *
 244          * @return array The parsed XML in an array form. Use print_r() to see the resulting array structure.
 245          * @throws \Exception
 246          */
 247         public static function toArray($contents, $namespaces = true, $get_attributes = 1, $priority = 'attribute')
 248         {
 249                 if (!$contents) {
 250                         return [];
 251                 }
 252
 253                 if (!function_exists('xml_parser_create')) {
 254                         Logger::log('Xml::toArray: parser function missing');
 255                         return [];
 256                 }
 257
 258
 259                 libxml_use_internal_errors(true);
 260                 libxml_clear_errors();
 261
 262                 if ($namespaces) {
 263                         $parser = @xml_parser_create_ns("UTF-8", ':');
 264                 } else {
 265                         $parser = @xml_parser_create();
 266                 }
 267
 268                 if (! $parser) {
 269                         Logger::log('Xml::toArray: xml_parser_create: no resource');
 270                         return [];
 271                 }
 272
 273                 xml_parser_set_option($parser, XML_OPTION_TARGET_ENCODING, "UTF-8");
 274                 // http://minutillo.com/steve/weblog/2004/6/17/php-xml-and-character-encodings-a-tale-of-sadness-rage-and-data-loss
 275                 xml_parser_set_option($parser, XML_OPTION_CASE_FOLDING, 0);
 276                 xml_parser_set_option($parser, XML_OPTION_SKIP_WHITE, 1);
 277                 @xml_parse_into_struct($parser, trim($contents), $xml_values);
 278                 @xml_parser_free($parser);
 279
 280                 if (! $xml_values) {
 281                         Logger::log('Xml::toArray: libxml: parse error: ' . $contents, Logger::DATA);
 282                         foreach (libxml_get_errors() as $err) {
 283                                 Logger::log('libxml: parse: ' . $err->code . " at " . $err->line . ":" . $err->column . " : " . $err->message, Logger::DATA);
 284                         }
 285                         libxml_clear_errors();
 286                         return;
 287                 }
 288
 289                 //Initializations
 290                 $xml_array = [];
 291
 292                 $current = &$xml_array; // Reference
 293
 294                 // Go through the tags.
 295                 $repeated_tag_index = []; // Multiple tags with same name will be turned into an array
 296                 foreach ($xml_values as $data) {
 297                         $tag        = $data['tag'];
 298                         $type       = $data['type'];
 299                         $level      = $data['level'];
 300                         $attributes = isset($data['attributes']) ? $data['attributes'] : null;
 301                         $value      = isset($data['value']) ? $data['value'] : null;
 302
 303                         $result = [];
 304                         $attributes_data = [];
 305
 306                         if (isset($value)) {
 307                                 if ($priority == 'tag') {
 308                                         $result = $value;
 309                                 } else {
 310                                         $result['value'] = $value; // Put the value in a assoc array if we are in the 'Attribute' mode
 311                                 }
 312                         }
 313
 314                         //Set the attributes too.
 315                         if (isset($attributes) and $get_attributes) {
 316                                 foreach ($attributes as $attr => $val) {
 317                                         if ($priority == 'tag') {
 318                                                 $attributes_data[$attr] = $val;
 319                                         } else {
 320                                                 $result['@attributes'][$attr] = $val; // Set all the attributes in a array called 'attr'
 321                                         }
 322                                 }
 323                         }
 324
 325                         // See tag status and do the needed.
 326                         if ($namespaces && strpos($tag, ':')) {
 327                                 $namespc = substr($tag, 0, strrpos($tag, ':'));
 328                                 $tag = strtolower(substr($tag, strlen($namespc)+1));
 329                                 $result['@namespace'] = $namespc;
 330                         }
 331                         $tag = strtolower($tag);
 332
 333                         if ($type == "open") {   // The starting of the tag '<tag>'
 334                                 $parent[$level-1] = &$current;
 335                                 if (!is_array($current) || (!in_array($tag, array_keys($current)))) { // Insert New tag
 336                                         $current[$tag] = $result;
 337                                         if ($attributes_data) {
 338                                                 $current[$tag. '_attr'] = $attributes_data;
 339                                         }
 340                                         $repeated_tag_index[$tag.'_'.$level] = 1;
 341
 342                                         $current = &$current[$tag];
 343                                 } else { // There was another element with the same tag name
 344
 345                                         if (isset($current[$tag][0])) { // If there is a 0th element it is already an array
 346                                                 $current[$tag][$repeated_tag_index[$tag.'_'.$level]] = $result;
 347                                                 $repeated_tag_index[$tag.'_'.$level]++;
 348                                         } else { // This section will make the value an array if multiple tags with the same name appear together
 349                                                 $current[$tag] = [$current[$tag], $result]; // This will combine the existing item and the new item together to make an array
 350                                                 $repeated_tag_index[$tag.'_'.$level] = 2;
 351
 352                                                 if (isset($current[$tag.'_attr'])) { // The attribute of the last(0th) tag must be moved as well
 353                                                         $current[$tag]['0_attr'] = $current[$tag.'_attr'];
 354                                                         unset($current[$tag.'_attr']);
 355                                                 }
 356                                         }
 357                                         $last_item_index = $repeated_tag_index[$tag.'_'.$level]-1;
 358                                         $current = &$current[$tag][$last_item_index];
 359                                 }
 360                         } elseif ($type == "complete") { // Tags that ends in 1 line '<tag />'
 361                                 //See if the key is already taken.
 362                                 if (!isset($current[$tag])) { //New Key
 363                                         $current[$tag] = $result;
 364                                         $repeated_tag_index[$tag.'_'.$level] = 1;
 365                                         if ($priority == 'tag' and $attributes_data) {
 366                                                 $current[$tag. '_attr'] = $attributes_data;
 367                                         }
 368                                 } else { // If taken, put all things inside a list(array)
 369                                         if (isset($current[$tag][0]) and is_array($current[$tag])) { // If it is already an array...
 370
 371                                                 // ...push the new element into that array.
 372                                                 $current[$tag][$repeated_tag_index[$tag.'_'.$level]] = $result;
 373
 374                                                 if ($priority == 'tag' and $get_attributes and $attributes_data) {
 375                                                         $current[$tag][$repeated_tag_index[$tag.'_'.$level] . '_attr'] = $attributes_data;
 376                                                 }
 377                                                 $repeated_tag_index[$tag.'_'.$level]++;
 378                                         } else { // If it is not an array...
 379                                                 $current[$tag] = [$current[$tag], $result]; //...Make it an array using using the existing value and the new value
 380                                                 $repeated_tag_index[$tag.'_'.$level] = 1;
 381                                                 if ($priority == 'tag' and $get_attributes) {
 382                                                         if (isset($current[$tag.'_attr'])) { // The attribute of the last(0th) tag must be moved as well
 383
 384                                                                 $current[$tag]['0_attr'] = $current[$tag.'_attr'];
 385                                                                 unset($current[$tag.'_attr']);
 386                                                         }
 387
 388                                                         if ($attributes_data) {
 389                                                                 $current[$tag][$repeated_tag_index[$tag.'_'.$level] . '_attr'] = $attributes_data;
 390                                                         }
 391                                                 }
 392                                                 $repeated_tag_index[$tag.'_'.$level]++; // 0 and 1 indexes are already taken
 393                                         }
 394                                 }
 395                         } elseif ($type == 'close') { // End of tag '</tag>'
 396                                 $current = &$parent[$level-1];
 397                         }
 398                 }
 399
 400                 return($xml_array);
 401         }
 402
 403         /**
 404          * @brief Delete a node in a XML object
 405          *
 406          * @param \DOMDocument $doc  XML document
 407          * @param string $node Node name
 408          * @return void
 409          */
 410         public static function deleteNode(\DOMDocument $doc, $node)
 411         {
 412                 $xpath = new DOMXPath($doc);
 413                 $list = $xpath->query("//".$node);
 414                 foreach ($list as $child) {
 415                         $child->parentNode->removeChild($child);
 416                 }
 417         }
 418
 419         public static function parseString($s, $strict = true)
 420         {
 421                 // the "strict" parameter is deactivated
 422                 libxml_use_internal_errors(true);
 423
 424                 $x = @simplexml_load_string($s);
 425                 if (!$x) {
 426                         Logger::error('Error(s) while parsing XML string.', ['callstack' => System::callstack()]);
 427                         foreach (libxml_get_errors() as $err) {
 428                                 Logger::info('libxml error', ['code' => $err->code, 'position' => $err->line . ":" . $err->column, 'message' => $err->message]);
 429                         }
 430                         Logger::debug('Erroring XML string', ['xml' => $s]);
 431                         libxml_clear_errors();
 432                 }
 433                 return $x;
 434         }
 435
 436         public static function getFirstNodeValue(DOMXPath $xpath, $element, $context = null)
 437         {
 438                 $result = $xpath->evaluate($element, $context);
 439                 if (!is_object($result)) {
 440                         return '';
 441                 }
 442
 443                 $first_item = $result->item(0);
 444                 if (!is_object($first_item)) {
 445                         return '';
 446                 }
 447
 448                 return $first_item->nodeValue;
 449         }
 450
 451         public static function getFirstAttributes(DOMXPath $xpath, $element, $context = null)
 452         {
 453                 $result = $xpath->query($element, $context);
 454                 if (!is_object($result)) {
 455                         return false;
 456                 }
 457
 458                 $first_item = $result->item(0);
 459                 if (!is_object($first_item)) {
 460                         return false;
 461                 }
 462
 463                 return $first_item->attributes;
 464         }
 465
 466         /**
 467          * escape text ($str) for XML transport
 468          *
 469          * @param string $str
 470          * @return string Escaped text.
 471          */
 472         public static function escape($str)
 473         {
 474                 $buffer = htmlspecialchars($str, ENT_QUOTES, 'UTF-8');
 475                 $buffer = trim($buffer);
 476
 477                 return $buffer;
 478         }
 479
 480         /**
 481          * undo an escape
 482          *
 483          * @param string $s xml escaped text
 484          * @return string unescaped text
 485          */
 486         public static function unescape($s)
 487         {
 488                 $ret = htmlspecialchars_decode($s, ENT_QUOTES);
 489                 return $ret;
 490         }
 491
 492         /**
 493          * apply escape() to all values of array $val, recursively
 494          *
 495          * @param array $val
 496          * @return array|string
 497          */
 498         public static function arrayEscape($val)
 499         {
 500                 if (is_bool($val)) {
 501                         return $val ? 'true' : 'false';
 502                 } elseif (is_array($val)) {
 503                         return array_map('XML::arrayEscape', $val);
 504                 }
 505
 506                 return self::escape((string) $val);
 507         }
 508 }