src/Util/XML.php

   1 <?php
   2 /**
   3  * @file src/Util/XML.php
   4  */
   5 namespace Friendica\Util;
   6
   7 use DOMXPath;
   8 use SimpleXMLElement;
   9
  10 /**
  11  * @brief This class contain methods to work with XML data
  12  */
  13 class XML
  14 {
  15         /**
  16          * @brief Creates an XML structure out of a given array
  17          *
  18          * @param array  $array         The array of the XML structure that will be generated
  19          * @param object $xml           The createdXML will be returned by reference
  20          * @param bool   $remove_header Should the XML header be removed or not?
  21          * @param array  $namespaces    List of namespaces
  22          * @param bool   $root          interally used parameter. Mustn't be used from outside.
  23          *
  24          * @return string The created XML
  25          */
  26         public static function fromArray($array, &$xml, $remove_header = false, $namespaces = [], $root = true)
  27         {
  28                 if ($root) {
  29                         foreach ($array as $key => $value) {
  30                                 foreach ($namespaces as $nskey => $nsvalue) {
  31                                         $key .= " xmlns".($nskey == "" ? "":":").$nskey.'="'.$nsvalue.'"';
  32                                 }
  33
  34                                 if (is_array($value)) {
  35                                         $root = new SimpleXMLElement("<".$key."/>");
  36                                         self::fromArray($value, $root, $remove_header, $namespaces, false);
  37                                 } else {
  38                                         $root = new SimpleXMLElement("<".$key.">".xmlify($value)."</".$key.">");
  39                                 }
  40
  41                                 $dom = dom_import_simplexml($root)->ownerDocument;
  42                                 $dom->formatOutput = true;
  43                                 $xml = $dom;
  44
  45                                 $xml_text = $dom->saveXML();
  46
  47                                 if ($remove_header) {
  48                                         $xml_text = trim(substr($xml_text, 21));
  49                                 }
  50
  51                                 return $xml_text;
  52                         }
  53                 }
  54
  55                 $element = null;
  56                 foreach ($array as $key => $value) {
  57                         if (!isset($element) && isset($xml)) {
  58                                 $element = $xml;
  59                         }
  60
  61                         if (is_integer($key)) {
  62                                 if (isset($element)) {
  63                                         if (is_scalar($value)) {
  64                                                 $element[0] = $value;
  65                                         } else {
  66                                                 /// @todo: handle nested array values
  67                                         }
  68                                 }
  69                                 continue;
  70                         }
  71
  72                         $element_parts = explode(":", $key);
  73                         if ((count($element_parts) > 1) && isset($namespaces[$element_parts[0]])) {
  74                                 $namespace = $namespaces[$element_parts[0]];
  75                         } elseif (isset($namespaces[""])) {
  76                                 $namespace = $namespaces[""];
  77                         } else {
  78                                 $namespace = null;
  79                         }
  80
  81                         // Remove undefined namespaces from the key
  82                         if ((count($element_parts) > 1) && is_null($namespace)) {
  83                                 $key = $element_parts[1];
  84                         }
  85
  86                         if (substr($key, 0, 11) == "@attributes") {
  87                                 if (!isset($element) || !is_array($value)) {
  88                                         continue;
  89                                 }
  90
  91                                 foreach ($value as $attr_key => $attr_value) {
  92                                         $element_parts = explode(":", $attr_key);
  93                                         if ((count($element_parts) > 1) && isset($namespaces[$element_parts[0]])) {
  94                                                 $namespace = $namespaces[$element_parts[0]];
  95                                         } else {
  96                                                 $namespace = null;
  97                                         }
  98
  99                                         $element->addAttribute($attr_key, $attr_value, $namespace);
 100                                 }
 101
 102                                 continue;
 103                         }
 104
 105                         if (!is_array($value)) {
 106                                 $element = $xml->addChild($key, xmlify($value), $namespace);
 107                         } elseif (is_array($value)) {
 108                                 $element = $xml->addChild($key, null, $namespace);
 109                                 self::fromArray($value, $element, $remove_header, $namespaces, false);
 110                         }
 111                 }
 112         }
 113
 114         /**
 115          * @brief Copies an XML object
 116          *
 117          * @param object $source      The XML source
 118          * @param object $target      The XML target
 119          * @param string $elementname Name of the XML element of the target
 120          * @return void
 121          */
 122         public static function copy(&$source, &$target, $elementname)
 123         {
 124                 if (count($source->children()) == 0) {
 125                         $target->addChild($elementname, xmlify($source));
 126                 } else {
 127                         $child = $target->addChild($elementname);
 128                         foreach ($source->children() as $childfield => $childentry) {
 129                                 self::copy($childentry, $child, $childfield);
 130                         }
 131                 }
 132         }
 133
 134         /**
 135          * @brief Create an XML element
 136          *
 137          * @param object $doc        XML root
 138          * @param string $element    XML element name
 139          * @param string $value      XML value
 140          * @param array  $attributes array containing the attributes
 141          *
 142          * @return object XML element object
 143          */
 144         public static function createElement($doc, $element, $value = "", $attributes = [])
 145         {
 146                 $element = $doc->createElement($element, xmlify($value));
 147
 148                 foreach ($attributes as $key => $value) {
 149                         $attribute = $doc->createAttribute($key);
 150                         $attribute->value = xmlify($value);
 151                         $element->appendChild($attribute);
 152                 }
 153                 return $element;
 154         }
 155
 156         /**
 157          * @brief Create an XML and append it to the parent object
 158          *
 159          * @param object $doc        XML root
 160          * @param object $parent     parent object
 161          * @param string $element    XML element name
 162          * @param string $value      XML value
 163          * @param array  $attributes array containing the attributes
 164          * @return void
 165          */
 166         public static function addElement($doc, $parent, $element, $value = "", $attributes = [])
 167         {
 168                 $element = self::createElement($doc, $element, $value, $attributes);
 169                 $parent->appendChild($element);
 170         }
 171
 172         /**
 173          * @brief Convert an XML document to a normalised, case-corrected array
 174          *   used by webfinger
 175          *
 176          * @param object  $xml_element     The XML document
 177          * @param integer $recursion_depth recursion counter for internal use - default 0
 178          *                                 internal use, recursion counter
 179          *
 180          * @return array | string The array from the xml element or the string
 181          */
 182         public static function elementToArray($xml_element, &$recursion_depth = 0)
 183         {
 184                 // If we're getting too deep, bail out
 185                 if ($recursion_depth > 512) {
 186                         return(null);
 187                 }
 188
 189                 $xml_element_copy = '';
 190                 if (!is_string($xml_element)
 191                         && !is_array($xml_element)
 192                         && (get_class($xml_element) == 'SimpleXMLElement')
 193                 ) {
 194                         $xml_element_copy = $xml_element;
 195                         $xml_element = get_object_vars($xml_element);
 196                 }
 197
 198                 if (is_array($xml_element)) {
 199                         $result_array = [];
 200                         if (count($xml_element) <= 0) {
 201                                 return (trim(strval($xml_element_copy)));
 202                         }
 203
 204                         foreach ($xml_element as $key => $value) {
 205                                 $recursion_depth++;
 206                                 $result_array[strtolower($key)] = self::elementToArray($value, $recursion_depth);
 207                                 $recursion_depth--;
 208                         }
 209
 210                         if ($recursion_depth == 0) {
 211                                 $temp_array = $result_array;
 212                                 $result_array = [
 213                                         strtolower($xml_element_copy->getName()) => $temp_array,
 214                                 ];
 215                         }
 216
 217                         return ($result_array);
 218                 } else {
 219                         return (trim(strval($xml_element)));
 220                 }
 221         }
 222
 223         /**
 224          * @brief Convert the given XML text to an array in the XML structure.
 225          *
 226          * Xml::toArray() will convert the given XML text to an array in the XML structure.
 227          * Link: http://www.bin-co.com/php/scripts/xml2array/
 228          * Portions significantly re-written by mike@macgirvin.com for Friendica
 229          * (namespaces, lowercase tags, get_attribute default changed, more...)
 230          *
 231          * Examples: $array =  Xml::toArray(file_get_contents('feed.xml'));
 232          *              $array =  Xml::toArray(file_get_contents('feed.xml', true, 1, 'attribute'));
 233          *
 234          * @param object  $contents       The XML text
 235          * @param boolean $namespaces     True or false include namespace information
 236          *                                    in the returned array as array elements.
 237          * @param integer $get_attributes 1 or 0. If this is 1 the function will get the attributes as well as the tag values -
 238          *                                    this results in a different array structure in the return value.
 239          * @param string  $priority       Can be 'tag' or 'attribute'. This will change the way the resulting
 240          *                                    array sturcture. For 'tag', the tags are given more importance.
 241          *
 242          * @return array The parsed XML in an array form. Use print_r() to see the resulting array structure.
 243          */
 244         public static function toArray($contents, $namespaces = true, $get_attributes = 1, $priority = 'attribute')
 245         {
 246                 if (!$contents) {
 247                         return [];
 248                 }
 249
 250                 if (!function_exists('xml_parser_create')) {
 251                         logger('Xml::toArray: parser function missing');
 252                         return [];
 253                 }
 254
 255
 256                 libxml_use_internal_errors(true);
 257                 libxml_clear_errors();
 258
 259                 if ($namespaces) {
 260                         $parser = @xml_parser_create_ns("UTF-8", ':');
 261                 } else {
 262                         $parser = @xml_parser_create();
 263                 }
 264
 265                 if (! $parser) {
 266                         logger('Xml::toArray: xml_parser_create: no resource');
 267                         return [];
 268                 }
 269
 270                 xml_parser_set_option($parser, XML_OPTION_TARGET_ENCODING, "UTF-8");
 271                 // http://minutillo.com/steve/weblog/2004/6/17/php-xml-and-character-encodings-a-tale-of-sadness-rage-and-data-loss
 272                 xml_parser_set_option($parser, XML_OPTION_CASE_FOLDING, 0);
 273                 xml_parser_set_option($parser, XML_OPTION_SKIP_WHITE, 1);
 274                 @xml_parse_into_struct($parser, trim($contents), $xml_values);
 275                 @xml_parser_free($parser);
 276
 277                 if (! $xml_values) {
 278                         logger('Xml::toArray: libxml: parse error: ' . $contents, LOGGER_DATA);
 279                         foreach (libxml_get_errors() as $err) {
 280                                 logger('libxml: parse: ' . $err->code . " at " . $err->line . ":" . $err->column . " : " . $err->message, LOGGER_DATA);
 281                         }
 282                         libxml_clear_errors();
 283                         return;
 284                 }
 285
 286                 //Initializations
 287                 $xml_array = [];
 288                 $parents = [];
 289                 $opened_tags = [];
 290                 $arr = [];
 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 object $doc  XML document
 407          * @param string $node Node name
 408          * @return void
 409          */
 410         public static function deleteNode(&$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('libxml: parse: error: ' . $s, LOGGER_DATA);
 427                         foreach (libxml_get_errors() as $err) {
 428                                 logger('libxml: parse: ' . $err->code." at ".$err->line.":".$err->column." : ".$err->message, LOGGER_DATA);
 429                         }
 430                         libxml_clear_errors();
 431                 }
 432                 return $x;
 433         }
 434
 435         public static function getFirstNodeValue($xpath, $element, $context = null)
 436         {
 437                 $result = $xpath->evaluate($element, $context);
 438                 if (!is_object($result)) {
 439                         return '';
 440                 }
 441
 442                 $first_item = $result->item(0);
 443                 if (!is_object($first_item)) {
 444                         return '';
 445                 }
 446
 447                 return $first_item->nodeValue;
 448         }
 449
 450         public static function getFirstAttributes($xpath, $element, $context = null)
 451         {
 452                 $result = $xpath->query($element, $context);
 453                 if (!is_object($result)) {
 454                         return false;
 455                 }
 456
 457                 $first_item = $result->item(0);
 458                 if (!is_object($first_item)) {
 459                         return false;
 460                 }
 461
 462                 return $first_item->attributes;
 463         }
 464 }