src/Util/XML.php

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