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 $xml_text;
  52                         }
  53                 }
  54
  55                 foreach ($array as $key => $value) {
  56                         if (!isset($element) && isset($xml)) {
  57                                 $element = $xml;
  58                         }
  59
  60                         if (is_integer($key)) {
  61                                 if (isset($element)) {
  62                                         if (is_scalar($value)) {
  63                                                 $element[0] = $value;
  64                                         } else {
  65                                                 /// @todo: handle nested array values
  66                                         }
  67                                 }
  68                                 continue;
  69                         }
  70
  71                         $element_parts = explode(":", $key);
  72                         if ((count($element_parts) > 1) && isset($namespaces[$element_parts[0]])) {
  73                                 $namespace = $namespaces[$element_parts[0]];
  74                         } elseif (isset($namespaces[""])) {
  75                                 $namespace = $namespaces[""];
  76                         } else {
  77                                 $namespace = null;
  78                         }
  79
  80                         // Remove undefined namespaces from the key
  81                         if ((count($element_parts) > 1) && is_null($namespace)) {
  82                                 $key = $element_parts[1];
  83                         }
  84
  85                         if (substr($key, 0, 11) == "@attributes") {
  86                                 if (!isset($element) || !is_array($value)) {
  87                                         continue;
  88                                 }
  89
  90                                 foreach ($value as $attr_key => $attr_value) {
  91                                         $element_parts = explode(":", $attr_key);
  92                                         if ((count($element_parts) > 1) && isset($namespaces[$element_parts[0]])) {
  93                                                 $namespace = $namespaces[$element_parts[0]];
  94                                         } else {
  95                                                 $namespace = null;
  96                                         }
  97
  98                                         $element->addAttribute($attr_key, $attr_value, $namespace);
  99                                 }
 100
 101                                 continue;
 102                         }
 103
 104                         if (!is_array($value)) {
 105                                 $element = $xml->addChild($key, xmlify($value), $namespace);
 106                         } elseif (is_array($value)) {
 107                                 $element = $xml->addChild($key, null, $namespace);
 108                                 self::fromArray($value, $element, $remove_header, $namespaces, false);
 109                         }
 110                 }
 111         }
 112
 113         /**
 114          * @brief Copies an XML object
 115          *
 116          * @param object $source      The XML source
 117          * @param object $target      The XML target
 118          * @param string $elementname Name of the XML element of the target
 119          * @return void
 120          */
 121         public static function copy(&$source, &$target, $elementname)
 122         {
 123                 if (count($source->children()) == 0) {
 124                         $target->addChild($elementname, xmlify($source));
 125                 } else {
 126                         $child = $target->addChild($elementname);
 127                         foreach ($source->children() as $childfield => $childentry) {
 128                                 self::copy($childentry, $child, $childfield);
 129                         }
 130                 }
 131         }
 132
 133         /**
 134          * @brief Create an XML element
 135          *
 136          * @param object $doc        XML root
 137          * @param string $element    XML element name
 138          * @param string $value      XML value
 139          * @param array  $attributes array containing the attributes
 140          *
 141          * @return object XML element object
 142          */
 143         public static function createElement($doc, $element, $value = "", $attributes = [])
 144         {
 145                 $element = $doc->createElement($element, xmlify($value));
 146
 147                 foreach ($attributes as $key => $value) {
 148                         $attribute = $doc->createAttribute($key);
 149                         $attribute->value = xmlify($value);
 150                         $element->appendChild($attribute);
 151                 }
 152                 return $element;
 153         }
 154
 155         /**
 156          * @brief Create an XML and append it to the parent object
 157          *
 158          * @param object $doc        XML root
 159          * @param object $parent     parent object
 160          * @param string $element    XML element name
 161          * @param string $value      XML value
 162          * @param array  $attributes array containing the attributes
 163          * @return void
 164          */
 165         public static function addElement($doc, $parent, $element, $value = "", $attributes = [])
 166         {
 167                 $element = self::createElement($doc, $element, $value, $attributes);
 168                 $parent->appendChild($element);
 169         }
 170
 171         /**
 172          * @brief Convert an XML document to a normalised, case-corrected array
 173          *   used by webfinger
 174          *
 175          * @param object  $xml_element     The XML document
 176          * @param integer $recursion_depth recursion counter for internal use - default 0
 177          *                                 internal use, recursion counter
 178          *
 179          * @return array | string The array from the xml element or the string
 180          */
 181         public static function elementToArray($xml_element, &$recursion_depth = 0)
 182         {
 183                 // If we're getting too deep, bail out
 184                 if ($recursion_depth > 512) {
 185                         return(null);
 186                 }
 187
 188                 if (!is_string($xml_element)
 189                         && !is_array($xml_element)
 190                         && (get_class($xml_element) == 'SimpleXMLElement')
 191                 ) {
 192                                 $xml_element_copy = $xml_element;
 193                                 $xml_element = get_object_vars($xml_element);
 194                 }
 195
 196                 if (is_array($xml_element)) {
 197                         $result_array = [];
 198                         if (count($xml_element) <= 0) {
 199                                 return (trim(strval($xml_element_copy)));
 200                         }
 201
 202                         foreach ($xml_element as $key => $value) {
 203                                 $recursion_depth++;
 204                                 $result_array[strtolower($key)] = self::elementToArray($value, $recursion_depth);
 205                                 $recursion_depth--;
 206                         }
 207
 208                         if ($recursion_depth == 0) {
 209                                 $temp_array = $result_array;
 210                                 $result_array = [
 211                                         strtolower($xml_element_copy->getName()) => $temp_array,
 212                                 ];
 213                         }
 214
 215                         return ($result_array);
 216                 } else {
 217                         return (trim(strval($xml_element)));
 218                 }
 219         }
 220
 221         /**
 222          * @brief Convert the given XML text to an array in the XML structure.
 223          *
 224          * Xml::toArray() will convert the given XML text to an array in the XML structure.
 225          * Link: http://www.bin-co.com/php/scripts/xml2array/
 226          * Portions significantly re-written by mike@macgirvin.com for Friendica
 227          * (namespaces, lowercase tags, get_attribute default changed, more...)
 228          *
 229          * Examples: $array =  Xml::toArray(file_get_contents('feed.xml'));
 230          *              $array =  Xml::toArray(file_get_contents('feed.xml', true, 1, 'attribute'));
 231          *
 232          * @param object  $contents       The XML text
 233          * @param boolean $namespaces     True or false include namespace information
 234          *                                    in the returned array as array elements.
 235          * @param integer $get_attributes 1 or 0. If this is 1 the function will get the attributes as well as the tag values -
 236          *                                    this results in a different array structure in the return value.
 237          * @param string  $priority       Can be 'tag' or 'attribute'. This will change the way the resulting
 238          *                                    array sturcture. For 'tag', the tags are given more importance.
 239          *
 240          * @return array The parsed XML in an array form. Use print_r() to see the resulting array structure.
 241          */
 242         public static function toArray($contents, $namespaces = true, $get_attributes = 1, $priority = 'attribute')
 243         {
 244                 if (!$contents) {
 245                         return [];
 246                 }
 247
 248                 if (!function_exists('xml_parser_create')) {
 249                         logger('Xml::toArray: parser function missing');
 250                         return [];
 251                 }
 252
 253
 254                 libxml_use_internal_errors(true);
 255                 libxml_clear_errors();
 256
 257                 if ($namespaces) {
 258                         $parser = @xml_parser_create_ns("UTF-8", ':');
 259                 } else {
 260                         $parser = @xml_parser_create();
 261                 }
 262
 263                 if (! $parser) {
 264                         logger('Xml::toArray: xml_parser_create: no resource');
 265                         return [];
 266                 }
 267
 268                 xml_parser_set_option($parser, XML_OPTION_TARGET_ENCODING, "UTF-8");
 269                 // http://minutillo.com/steve/weblog/2004/6/17/php-xml-and-character-encodings-a-tale-of-sadness-rage-and-data-loss
 270                 xml_parser_set_option($parser, XML_OPTION_CASE_FOLDING, 0);
 271                 xml_parser_set_option($parser, XML_OPTION_SKIP_WHITE, 1);
 272                 @xml_parse_into_struct($parser, trim($contents), $xml_values);
 273                 @xml_parser_free($parser);
 274
 275                 if (! $xml_values) {
 276                         logger('Xml::toArray: libxml: parse error: ' . $contents, LOGGER_DATA);
 277                         foreach (libxml_get_errors() as $err) {
 278                                 logger('libxml: parse: ' . $err->code . " at " . $err->line . ":" . $err->column . " : " . $err->message, LOGGER_DATA);
 279                         }
 280                         libxml_clear_errors();
 281                         return;
 282                 }
 283
 284                 //Initializations
 285                 $xml_array = [];
 286                 $parents = [];
 287                 $opened_tags = [];
 288                 $arr = [];
 289
 290                 $current = &$xml_array; // Reference
 291
 292                 // Go through the tags.
 293                 $repeated_tag_index = []; // Multiple tags with same name will be turned into an array
 294                 foreach ($xml_values as $data) {
 295                         $tag        = $data['tag'];
 296                         $type       = $data['type'];
 297                         $level      = $data['level'];
 298                         $attributes = isset($data['attributes']) ? $data['attributes'] : null;
 299                         $value      = isset($data['value']) ? $data['value'] : null;
 300
 301                         $result = [];
 302                         $attributes_data = [];
 303
 304                         if (isset($value)) {
 305                                 if ($priority == 'tag') {
 306                                         $result = $value;
 307                                 } else {
 308                                         $result['value'] = $value; // Put the value in a assoc array if we are in the 'Attribute' mode
 309                                 }
 310                         }
 311
 312                         //Set the attributes too.
 313                         if (isset($attributes) and $get_attributes) {
 314                                 foreach ($attributes as $attr => $val) {
 315                                         if ($priority == 'tag') {
 316                                                 $attributes_data[$attr] = $val;
 317                                         } else {
 318                                                 $result['@attributes'][$attr] = $val; // Set all the attributes in a array called 'attr'
 319                                         }
 320                                 }
 321                         }
 322
 323                         // See tag status and do the needed.
 324                         if ($namespaces && strpos($tag, ':')) {
 325                                 $namespc = substr($tag, 0, strrpos($tag, ':'));
 326                                 $tag = strtolower(substr($tag, strlen($namespc)+1));
 327                                 $result['@namespace'] = $namespc;
 328                         }
 329                         $tag = strtolower($tag);
 330
 331                         if ($type == "open") {   // The starting of the tag '<tag>'
 332                                 $parent[$level-1] = &$current;
 333                                 if (!is_array($current) || (!in_array($tag, array_keys($current)))) { // Insert New tag
 334                                         $current[$tag] = $result;
 335                                         if ($attributes_data) {
 336                                                 $current[$tag. '_attr'] = $attributes_data;
 337                                         }
 338                                         $repeated_tag_index[$tag.'_'.$level] = 1;
 339
 340                                         $current = &$current[$tag];
 341                                 } else { // There was another element with the same tag name
 342
 343                                         if (isset($current[$tag][0])) { // If there is a 0th element it is already an array
 344                                                 $current[$tag][$repeated_tag_index[$tag.'_'.$level]] = $result;
 345                                                 $repeated_tag_index[$tag.'_'.$level]++;
 346                                         } else { // This section will make the value an array if multiple tags with the same name appear together
 347                                                 $current[$tag] = [$current[$tag], $result]; // This will combine the existing item and the new item together to make an array
 348                                                 $repeated_tag_index[$tag.'_'.$level] = 2;
 349
 350                                                 if (isset($current[$tag.'_attr'])) { // The attribute of the last(0th) tag must be moved as well
 351                                                         $current[$tag]['0_attr'] = $current[$tag.'_attr'];
 352                                                         unset($current[$tag.'_attr']);
 353                                                 }
 354                                         }
 355                                         $last_item_index = $repeated_tag_index[$tag.'_'.$level]-1;
 356                                         $current = &$current[$tag][$last_item_index];
 357                                 }
 358                         } elseif ($type == "complete") { // Tags that ends in 1 line '<tag />'
 359                                 //See if the key is already taken.
 360                                 if (!isset($current[$tag])) { //New Key
 361                                         $current[$tag] = $result;
 362                                         $repeated_tag_index[$tag.'_'.$level] = 1;
 363                                         if ($priority == 'tag' and $attributes_data) {
 364                                                 $current[$tag. '_attr'] = $attributes_data;
 365                                         }
 366                                 } else { // If taken, put all things inside a list(array)
 367                                         if (isset($current[$tag][0]) and is_array($current[$tag])) { // If it is already an array...
 368
 369                                                 // ...push the new element into that array.
 370                                                 $current[$tag][$repeated_tag_index[$tag.'_'.$level]] = $result;
 371
 372                                                 if ($priority == 'tag' and $get_attributes and $attributes_data) {
 373                                                         $current[$tag][$repeated_tag_index[$tag.'_'.$level] . '_attr'] = $attributes_data;
 374                                                 }
 375                                                 $repeated_tag_index[$tag.'_'.$level]++;
 376                                         } else { // If it is not an array...
 377                                                 $current[$tag] = [$current[$tag], $result]; //...Make it an array using using the existing value and the new value
 378                                                 $repeated_tag_index[$tag.'_'.$level] = 1;
 379                                                 if ($priority == 'tag' and $get_attributes) {
 380                                                         if (isset($current[$tag.'_attr'])) { // The attribute of the last(0th) tag must be moved as well
 381
 382                                                                 $current[$tag]['0_attr'] = $current[$tag.'_attr'];
 383                                                                 unset($current[$tag.'_attr']);
 384                                                         }
 385
 386                                                         if ($attributes_data) {
 387                                                                 $current[$tag][$repeated_tag_index[$tag.'_'.$level] . '_attr'] = $attributes_data;
 388                                                         }
 389                                                 }
 390                                                 $repeated_tag_index[$tag.'_'.$level]++; // 0 and 1 indexes are already taken
 391                                         }
 392                                 }
 393                         } elseif ($type == 'close') { // End of tag '</tag>'
 394                                 $current = &$parent[$level-1];
 395                         }
 396                 }
 397
 398                 return($xml_array);
 399         }
 400
 401         /**
 402          * @brief Delete a node in a XML object
 403          *
 404          * @param object $doc  XML document
 405          * @param string $node Node name
 406          * @return void
 407          */
 408         public static function deleteNode(&$doc, $node)
 409         {
 410                 $xpath = new DOMXPath($doc);
 411                 $list = $xpath->query("//".$node);
 412                 foreach ($list as $child) {
 413                         $child->parentNode->removeChild($child);
 414                 }
 415         }
 416 }