src/Util/XML.php

   1 <?php
   2
   3 /**
   4  * @file src/Util/XML.php
   5  */
   6 namespace Friendica\Util;
   7
   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   $root          interally used parameter. Mustn't be used from outside.
  24          *
  25          * @return string The created XML
  26          */
  27         public static function from_array($array, &$xml, $remove_header = false, $namespaces = array(), $root = true)
  28         {
  29                 if ($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::from_array($value, $root, $remove_header, $namespaces, false);
  38                                 } else {
  39                                         $root = new SimpleXMLElement("<".$key.">".xmlify($value)."</".$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 $xml_text;
  53                         }
  54                 }
  55
  56                 foreach ($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, $namespace);
 109                                 self::from_array($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 $elementname Name of the XML element of the target
 120          */
 121         public static function copy(&$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  $attributes array containing the attributes
 140          *
 141          * @return object XML element object
 142          */
 143         public static function create_element($doc, $element, $value = "", $attributes = array())
 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  $attributes array containing the attributes
 163          */
 164         public static function add_element($doc, $parent, $element, $value = "", $attributes = array())
 165         {
 166                 $element = self::create_element($doc, $element, $value, $attributes);
 167                 $parent->appendChild($element);
 168         }
 169
 170         /**
 171          * @brief Convert an XML document to a normalised, case-corrected array
 172          *   used by webfinger
 173          *
 174          * @param object  $xml_element     The XML document
 175          * @param integer $recursion_depth recursion counter for internal use - default 0
 176          *                                 internal use, recursion counter
 177          *
 178          * @return array | sring The array from the xml element or the string
 179          */
 180         public static function element_to_array($xml_element, &$recursion_depth=0)
 181         {
 182                 // If we're getting too deep, bail out
 183                 if ($recursion_depth > 512) {
 184                         return(null);
 185                 }
 186
 187                 if (!is_string($xml_element)
 188                         && !is_array($xml_element)
 189                         && (get_class($xml_element) == 'SimpleXMLElement')
 190                 ) {
 191                                 $xml_element_copy = $xml_element;
 192                                 $xml_element = get_object_vars($xml_element);
 193                 }
 194
 195                 if (is_array($xml_element)) {
 196                         $result_array = array();
 197                         if (count($xml_element) <= 0) {
 198                                 return (trim(strval($xml_element_copy)));
 199                         }
 200
 201                         foreach ($xml_element as $key => $value) {
 202                                 $recursion_depth++;
 203                                 $result_array[strtolower($key)] = self::element_to_array($value, $recursion_depth);
 204                                 $recursion_depth--;
 205                         }
 206
 207                         if ($recursion_depth == 0) {
 208                                 $temp_array = $result_array;
 209                                 $result_array = array(
 210                                         strtolower($xml_element_copy->getName()) => $temp_array,
 211                                 );
 212                         }
 213
 214                         return ($result_array);
 215                 } else {
 216                         return (trim(strval($xml_element)));
 217                 }
 218         }
 219
 220         /**
 221          * @brief Convert the given XML text to an array in the XML structure.
 222          *
 223          * Xml::to_array() will convert the given XML text to an array in the XML structure.
 224          * Link: http://www.bin-co.com/php/scripts/xml2array/
 225          * Portions significantly re-written by mike@macgirvin.com for Friendica
 226          * (namespaces, lowercase tags, get_attribute default changed, more...)
 227          *
 228          * Examples: $array =  Xml::to_array(file_get_contents('feed.xml'));
 229          *              $array =  Xml::to_array(file_get_contents('feed.xml', true, 1, 'attribute'));
 230          *
 231          * @param object  $contents       The XML text
 232          * @param boolean $namespaces     True or false include namespace information
 233          *                                    in the returned array as array elements.
 234          * @param integer $get_attributes 1 or 0. If this is 1 the function will get the attributes as well as the tag values -
 235          *                                    this results in a different array structure in the return value.
 236          * @param string  $priority       Can be 'tag' or 'attribute'. This will change the way the resulting
 237          *                                    array sturcture. For 'tag', the tags are given more importance.
 238          *
 239          * @return array The parsed XML in an array form. Use print_r() to see the resulting array structure.
 240          */
 241         public static function to_array($contents, $namespaces = true, $get_attributes = 1, $priority = 'attribute')
 242         {
 243                 if (!$contents) {
 244                         return array();
 245                 }
 246
 247                 if (!function_exists('xml_parser_create')) {
 248                         logger('Xml::to_array: parser function missing');
 249                         return array();
 250                 }
 251
 252
 253                 libxml_use_internal_errors(true);
 254                 libxml_clear_errors();
 255
 256                 if ($namespaces) {
 257                         $parser = @xml_parser_create_ns("UTF-8", ':');
 258                 } else {
 259                         $parser = @xml_parser_create();
 260                 }
 261
 262                 if (! $parser) {
 263                         logger('Xml::to_array: xml_parser_create: no resource');
 264                         return array();
 265                 }
 266
 267                 xml_parser_set_option($parser, XML_OPTION_TARGET_ENCODING, "UTF-8");
 268                 // http://minutillo.com/steve/weblog/2004/6/17/php-xml-and-character-encodings-a-tale-of-sadness-rage-and-data-loss
 269                 xml_parser_set_option($parser, XML_OPTION_CASE_FOLDING, 0);
 270                 xml_parser_set_option($parser, XML_OPTION_SKIP_WHITE, 1);
 271                 @xml_parse_into_struct($parser, trim($contents), $xml_values);
 272                 @xml_parser_free($parser);
 273
 274                 if (! $xml_values) {
 275                         logger('Xml::to_array: libxml: parse error: ' . $contents, LOGGER_DATA);
 276                         foreach (libxml_get_errors() as $err) {
 277                                 logger('libxml: parse: ' . $err->code . " at " . $err->line . ":" . $err->column . " : " . $err->message, LOGGER_DATA);
 278                         }
 279                         libxml_clear_errors();
 280                         return;
 281                 }
 282
 283                 //Initializations
 284                 $xml_array = array();
 285                 $parents = array();
 286                 $opened_tags = array();
 287                 $arr = array();
 288
 289                 $current = &$xml_array; // Reference
 290
 291                 // Go through the tags.
 292                 $repeated_tag_index = array(); // Multiple tags with same name will be turned into an array
 293                 foreach ($xml_values as $data) {
 294                         unset($attributes, $value); // Remove existing values, or there will be trouble
 295
 296                         // This command will extract these variables into the foreach scope
 297                         // tag(string), type(string), level(int), attributes(array).
 298                         extract($data); // We could use the array by itself, but this cooler.
 299
 300                         $result = array();
 301                         $attributes_data = array();
 302
 303                         if (isset($value)) {
 304                                 if ($priority == 'tag') {
 305                                         $result = $value;
 306                                 } else {
 307                                         $result['value'] = $value; // Put the value in a assoc array if we are in the 'Attribute' mode
 308                                 }
 309                         }
 310
 311                         //Set the attributes too.
 312                         if (isset($attributes) and $get_attributes) {
 313                                 foreach ($attributes as $attr => $val) {
 314                                         if ($priority == 'tag') {
 315                                                 $attributes_data[$attr] = $val;
 316                                         } else {
 317                                                 $result['@attributes'][$attr] = $val; // Set all the attributes in a array called 'attr'
 318                                         }
 319                                 }
 320                         }
 321
 322                         // See tag status and do the needed.
 323                         if ($namespaces && strpos($tag, ':')) {
 324                                 $namespc = substr($tag, 0, strrpos($tag, ':'));
 325                                 $tag = strtolower(substr($tag, strlen($namespc)+1));
 326                                 $result['@namespace'] = $namespc;
 327                         }
 328                         $tag = strtolower($tag);
 329
 330                         if ($type == "open") {   // The starting of the tag '<tag>'
 331                                 $parent[$level-1] = &$current;
 332                                 if (!is_array($current) || (!in_array($tag, array_keys($current)))) { // Insert New tag
 333                                         $current[$tag] = $result;
 334                                         if ($attributes_data) {
 335                                                 $current[$tag. '_attr'] = $attributes_data;
 336                                         }
 337                                         $repeated_tag_index[$tag.'_'.$level] = 1;
 338
 339                                         $current = &$current[$tag];
 340                                 } else { // There was another element with the same tag name
 341
 342                                         if (isset($current[$tag][0])) { // If there is a 0th element it is already an array
 343                                                 $current[$tag][$repeated_tag_index[$tag.'_'.$level]] = $result;
 344                                                 $repeated_tag_index[$tag.'_'.$level]++;
 345                                         } else { // This section will make the value an array if multiple tags with the same name appear together
 346                                                 $current[$tag] = array($current[$tag], $result); // This will combine the existing item and the new item together to make an array
 347                                                 $repeated_tag_index[$tag.'_'.$level] = 2;
 348
 349                                                 if (isset($current[$tag.'_attr'])) { // The attribute of the last(0th) tag must be moved as well
 350                                                         $current[$tag]['0_attr'] = $current[$tag.'_attr'];
 351                                                         unset($current[$tag.'_attr']);
 352                                                 }
 353                                         }
 354                                         $last_item_index = $repeated_tag_index[$tag.'_'.$level]-1;
 355                                         $current = &$current[$tag][$last_item_index];
 356                                 }
 357                         } elseif ($type == "complete") { // Tags that ends in 1 line '<tag />'
 358                                 //See if the key is already taken.
 359                                 if (!isset($current[$tag])) { //New Key
 360                                         $current[$tag] = $result;
 361                                         $repeated_tag_index[$tag.'_'.$level] = 1;
 362                                         if ($priority == 'tag' and $attributes_data) {
 363                                                 $current[$tag. '_attr'] = $attributes_data;
 364                                         }
 365                                 } else { // If taken, put all things inside a list(array)
 366                                         if (isset($current[$tag][0]) and is_array($current[$tag])) { // If it is already an array...
 367
 368                                                 // ...push the new element into that array.
 369                                                 $current[$tag][$repeated_tag_index[$tag.'_'.$level]] = $result;
 370
 371                                                 if ($priority == 'tag' and $get_attributes and $attributes_data) {
 372                                                         $current[$tag][$repeated_tag_index[$tag.'_'.$level] . '_attr'] = $attributes_data;
 373                                                 }
 374                                                 $repeated_tag_index[$tag.'_'.$level]++;
 375                                         } else { // If it is not an array...
 376                                                 $current[$tag] = array($current[$tag], $result); //...Make it an array using using the existing value and the new value
 377                                                 $repeated_tag_index[$tag.'_'.$level] = 1;
 378                                                 if ($priority == 'tag' and $get_attributes) {
 379                                                         if (isset($current[$tag.'_attr'])) { // The attribute of the last(0th) tag must be moved as well
 380
 381                                                                 $current[$tag]['0_attr'] = $current[$tag.'_attr'];
 382                                                                 unset($current[$tag.'_attr']);
 383                                                         }
 384
 385                                                         if ($attributes_data) {
 386                                                                 $current[$tag][$repeated_tag_index[$tag.'_'.$level] . '_attr'] = $attributes_data;
 387                                                         }
 388                                                 }
 389                                                 $repeated_tag_index[$tag.'_'.$level]++; // 0 and 1 indexes are already taken
 390                                         }
 391                                 }
 392                         } elseif ($type == 'close') { // End of tag '</tag>'
 393                                 $current = &$parent[$level-1];
 394                         }
 395                 }
 396
 397                 return($xml_array);
 398         }
 399
 400         /**
 401          * @brief Delete a node in a XML object
 402          *
 403          * @param object $doc  XML document
 404          * @param string $node Node name
 405          */
 406         public static function deleteNode(&$doc, $node)
 407         {
 408                 $xpath = new DomXPath($doc);
 409                 $list = $xpath->query("//".$node);
 410                 foreach ($list as $child) {
 411                         $child->parentNode->removeChild($child);
 412                 }
 413         }
 414 }