src/Util/XML.php

   1 <?php
   2 /**
   3  * @copyright Copyright (C) 2020, Friendica
   4  *
   5  * @license GNU AGPL version 3 or any later version
   6  *
   7  * This program is free software: you can redistribute it and/or modify
   8  * it under the terms of the GNU Affero General Public License as
   9  * published by the Free Software Foundation, either version 3 of the
  10  * License, or (at your option) any later version.
  11  *
  12  * This program is distributed in the hope that it will be useful,
  13  * but WITHOUT ANY WARRANTY; without even the implied warranty of
  14  * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
  15  * GNU Affero General Public License for more details.
  16  *
  17  * You should have received a copy of the GNU Affero General Public License
  18  * along with this program.  If not, see <https://www.gnu.org/licenses/>.
  19  *
  20  */
  21
  22 namespace Friendica\Util;
  23
  24 use DOMXPath;
  25 use Friendica\Core\Logger;
  26 use Friendica\Core\System;
  27 use SimpleXMLElement;
  28
  29 /**
  30  * This class contain methods to work with XML data
  31  */
  32 class XML
  33 {
  34         /**
  35          * Creates an XML structure out of a given array
  36          *
  37          * @param array  $array         The array of the XML structure that will be generated
  38          * @param object $xml           The createdXML will be returned by reference
  39          * @param bool   $remove_header Should the XML header be removed or not?
  40          * @param array  $namespaces    List of namespaces
  41          * @param bool   $root          interally used parameter. Mustn't be used from outside.
  42          *
  43          * @return string The created XML
  44          */
  45         public static function fromArray($array, &$xml, $remove_header = false, $namespaces = [], $root = true)
  46         {
  47                 if ($root) {
  48                         foreach ($array as $key => $value) {
  49                                 foreach ($namespaces as $nskey => $nsvalue) {
  50                                         $key .= " xmlns".($nskey == "" ? "":":").$nskey.'="'.$nsvalue.'"';
  51                                 }
  52
  53                                 if (is_array($value)) {
  54                                         $root = new SimpleXMLElement("<".$key."/>");
  55                                         self::fromArray($value, $root, $remove_header, $namespaces, false);
  56                                 } else {
  57                                         $root = new SimpleXMLElement("<".$key.">".self::escape($value)."</".$key.">");
  58                                 }
  59
  60                                 $dom = dom_import_simplexml($root)->ownerDocument;
  61                                 $dom->formatOutput = true;
  62                                 $xml = $dom;
  63
  64                                 $xml_text = $dom->saveXML();
  65
  66                                 if ($remove_header) {
  67                                         $xml_text = trim(substr($xml_text, 21));
  68                                 }
  69
  70                                 return $xml_text;
  71                         }
  72                 }
  73
  74                 $element = null;
  75                 foreach ($array as $key => $value) {
  76                         if (!isset($element) && isset($xml)) {
  77                                 $element = $xml;
  78                         }
  79
  80                         if (is_integer($key)) {
  81                                 if (isset($element)) {
  82                                         if (is_scalar($value)) {
  83                                                 $element[0] = $value;
  84                                         } else {
  85                                                 /// @todo: handle nested array values
  86                                         }
  87                                 }
  88                                 continue;
  89                         }
  90
  91                         $element_parts = explode(":", $key);
  92                         if ((count($element_parts) > 1) && isset($namespaces[$element_parts[0]])) {
  93                                 $namespace = $namespaces[$element_parts[0]];
  94                         } elseif (isset($namespaces[""])) {
  95                                 $namespace = $namespaces[""];
  96                         } else {
  97                                 $namespace = null;
  98                         }
  99
 100                         // Remove undefined namespaces from the key
 101                         if ((count($element_parts) > 1) && is_null($namespace)) {
 102                                 $key = $element_parts[1];
 103                         }
 104
 105                         if (substr($key, 0, 11) == "@attributes") {
 106                                 if (!isset($element) || !is_array($value)) {
 107                                         continue;
 108                                 }
 109
 110                                 foreach ($value as $attr_key => $attr_value) {
 111                                         $element_parts = explode(":", $attr_key);
 112                                         if ((count($element_parts) > 1) && isset($namespaces[$element_parts[0]])) {
 113                                                 $namespace = $namespaces[$element_parts[0]];
 114                                         } else {
 115                                                 $namespace = null;
 116                                         }
 117
 118                                         $element->addAttribute($attr_key, $attr_value, $namespace);
 119                                 }
 120
 121                                 continue;
 122                         }
 123
 124                         if (!is_array($value)) {
 125                                 $element = $xml->addChild($key, self::escape($value), $namespace);
 126                         } elseif (is_array($value)) {
 127                                 $element = $xml->addChild($key, null, $namespace);
 128                                 self::fromArray($value, $element, $remove_header, $namespaces, false);
 129                         }
 130                 }
 131         }
 132
 133         /**
 134          * Copies an XML object
 135          *
 136          * @param object $source      The XML source
 137          * @param object $target      The XML target
 138          * @param string $elementname Name of the XML element of the target
 139          * @return void
 140          */
 141         public static function copy(&$source, &$target, $elementname)
 142         {
 143                 if (count($source->children()) == 0) {
 144                         $target->addChild($elementname, self::escape($source));
 145                 } else {
 146                         $child = $target->addChild($elementname);
 147                         foreach ($source->children() as $childfield => $childentry) {
 148                                 self::copy($childentry, $child, $childfield);
 149                         }
 150                 }
 151         }
 152
 153         /**
 154          * Create an XML element
 155          *
 156          * @param \DOMDocument $doc        XML root
 157          * @param string       $element    XML element name
 158          * @param string       $value      XML value
 159          * @param array        $attributes array containing the attributes
 160          *
 161          * @return \DOMElement XML element object
 162          */
 163         public static function createElement(\DOMDocument $doc, $element, $value = "", $attributes = [])
 164         {
 165                 $element = $doc->createElement($element, self::escape($value));
 166
 167                 foreach ($attributes as $key => $value) {
 168                         $attribute = $doc->createAttribute($key);
 169                         $attribute->value = self::escape($value);
 170                         $element->appendChild($attribute);
 171                 }
 172                 return $element;
 173         }
 174
 175         /**
 176          * Create an XML and append it to the parent object
 177          *
 178          * @param \DOMDocument $doc        XML root
 179          * @param object $parent     parent object
 180          * @param string $element    XML element name
 181          * @param string $value      XML value
 182          * @param array  $attributes array containing the attributes
 183          * @return void
 184          */
 185         public static function addElement(\DOMDocument $doc, $parent, $element, $value = "", $attributes = [])
 186         {
 187                 $element = self::createElement($doc, $element, $value, $attributes);
 188                 $parent->appendChild($element);
 189         }
 190
 191         /**
 192          * Convert an XML document to a normalised, case-corrected array
 193          *   used by webfinger
 194          *
 195          * @param object  $xml_element     The XML document
 196          * @param integer $recursion_depth recursion counter for internal use - default 0
 197          *                                 internal use, recursion counter
 198          *
 199          * @return array | string The array from the xml element or the string
 200          */
 201         public static function elementToArray($xml_element, &$recursion_depth = 0)
 202         {
 203                 // If we're getting too deep, bail out
 204                 if ($recursion_depth > 512) {
 205                         return(null);
 206                 }
 207
 208                 $xml_element_copy = '';
 209                 if (!is_string($xml_element)
 210                         && !is_array($xml_element)
 211                         && (get_class($xml_element) == 'SimpleXMLElement')
 212                 ) {
 213                         $xml_element_copy = $xml_element;
 214                         $xml_element = get_object_vars($xml_element);
 215                 }
 216
 217                 if (is_array($xml_element)) {
 218                         $result_array = [];
 219                         if (count($xml_element) <= 0) {
 220                                 return (trim(strval($xml_element_copy)));
 221                         }
 222
 223                         foreach ($xml_element as $key => $value) {
 224                                 $recursion_depth++;
 225                                 $result_array[strtolower($key)] = self::elementToArray($value, $recursion_depth);
 226                                 $recursion_depth--;
 227                         }
 228
 229                         if ($recursion_depth == 0) {
 230                                 $temp_array = $result_array;
 231                                 $result_array = [
 232                                         strtolower($xml_element_copy->getName()) => $temp_array,
 233                                 ];
 234                         }
 235
 236                         return ($result_array);
 237                 } else {
 238                         return (trim(strval($xml_element)));
 239                 }
 240         }
 241
 242         /**
 243          * Convert the given XML text to an array in the XML structure.
 244          *
 245          * Xml::toArray() will convert the given XML text to an array in the XML structure.
 246          * Link: http://www.bin-co.com/php/scripts/xml2array/
 247          * Portions significantly re-written by mike@macgirvin.com for Friendica
 248          * (namespaces, lowercase tags, get_attribute default changed, more...)
 249          *
 250          * Examples: $array =  Xml::toArray(file_get_contents('feed.xml'));
 251          *        $array =  Xml::toArray(file_get_contents('feed.xml', true, 1, 'attribute'));
 252          *
 253          * @param object  $contents         The XML text
 254          * @param boolean $namespaces       True or false include namespace information
 255          *                                  in the returned array as array elements.
 256          * @param integer $get_attributes   1 or 0. If this is 1 the function will get the attributes as well as the tag values -
 257          *                                  this results in a different array structure in the return value.
 258          * @param string  $priority         Can be 'tag' or 'attribute'. This will change the way the resulting
 259          *                                  array sturcture. For 'tag', the tags are given more importance.
 260          *
 261          * @return array The parsed XML in an array form. Use print_r() to see the resulting array structure.
 262          * @throws \Exception
 263          */
 264         public static function toArray($contents, $namespaces = true, $get_attributes = 1, $priority = 'attribute')
 265         {
 266                 if (!$contents) {
 267                         return [];
 268                 }
 269
 270                 if (!function_exists('xml_parser_create')) {
 271                         Logger::log('Xml::toArray: parser function missing');
 272                         return [];
 273                 }
 274
 275
 276                 libxml_use_internal_errors(true);
 277                 libxml_clear_errors();
 278
 279                 if ($namespaces) {
 280                         $parser = @xml_parser_create_ns("UTF-8", ':');
 281                 } else {
 282                         $parser = @xml_parser_create();
 283                 }
 284
 285                 if (! $parser) {
 286                         Logger::log('Xml::toArray: xml_parser_create: no resource');
 287                         return [];
 288                 }
 289
 290                 xml_parser_set_option($parser, XML_OPTION_TARGET_ENCODING, "UTF-8");
 291                 // http://minutillo.com/steve/weblog/2004/6/17/php-xml-and-character-encodings-a-tale-of-sadness-rage-and-data-loss
 292                 xml_parser_set_option($parser, XML_OPTION_CASE_FOLDING, 0);
 293                 xml_parser_set_option($parser, XML_OPTION_SKIP_WHITE, 1);
 294                 @xml_parse_into_struct($parser, trim($contents), $xml_values);
 295                 @xml_parser_free($parser);
 296
 297                 if (! $xml_values) {
 298                         Logger::log('Xml::toArray: libxml: parse error: ' . $contents, Logger::DATA);
 299                         foreach (libxml_get_errors() as $err) {
 300                                 Logger::log('libxml: parse: ' . $err->code . " at " . $err->line . ":" . $err->column . " : " . $err->message, Logger::DATA);
 301                         }
 302                         libxml_clear_errors();
 303                         return;
 304                 }
 305
 306                 //Initializations
 307                 $xml_array = [];
 308
 309                 $current = &$xml_array; // Reference
 310
 311                 // Go through the tags.
 312                 $repeated_tag_index = []; // Multiple tags with same name will be turned into an array
 313                 foreach ($xml_values as $data) {
 314                         $tag        = $data['tag'];
 315                         $type       = $data['type'];
 316                         $level      = $data['level'];
 317                         $attributes = isset($data['attributes']) ? $data['attributes'] : null;
 318                         $value      = isset($data['value']) ? $data['value'] : null;
 319
 320                         $result = [];
 321                         $attributes_data = [];
 322
 323                         if (isset($value)) {
 324                                 if ($priority == 'tag') {
 325                                         $result = $value;
 326                                 } else {
 327                                         $result['value'] = $value; // Put the value in a assoc array if we are in the 'Attribute' mode
 328                                 }
 329                         }
 330
 331                         //Set the attributes too.
 332                         if (isset($attributes) and $get_attributes) {
 333                                 foreach ($attributes as $attr => $val) {
 334                                         if ($priority == 'tag') {
 335                                                 $attributes_data[$attr] = $val;
 336                                         } else {
 337                                                 $result['@attributes'][$attr] = $val; // Set all the attributes in a array called 'attr'
 338                                         }
 339                                 }
 340                         }
 341
 342                         // See tag status and do the needed.
 343                         if ($namespaces && strpos($tag, ':')) {
 344                                 $namespc = substr($tag, 0, strrpos($tag, ':'));
 345                                 $tag = strtolower(substr($tag, strlen($namespc)+1));
 346                                 $result['@namespace'] = $namespc;
 347                         }
 348                         $tag = strtolower($tag);
 349
 350                         if ($type == "open") {   // The starting of the tag '<tag>'
 351                                 $parent[$level-1] = &$current;
 352                                 if (!is_array($current) || (!in_array($tag, array_keys($current)))) { // Insert New tag
 353                                         $current[$tag] = $result;
 354                                         if ($attributes_data) {
 355                                                 $current[$tag. '_attr'] = $attributes_data;
 356                                         }
 357                                         $repeated_tag_index[$tag.'_'.$level] = 1;
 358
 359                                         $current = &$current[$tag];
 360                                 } else { // There was another element with the same tag name
 361
 362                                         if (isset($current[$tag][0])) { // If there is a 0th element it is already an array
 363                                                 $current[$tag][$repeated_tag_index[$tag.'_'.$level]] = $result;
 364                                                 $repeated_tag_index[$tag.'_'.$level]++;
 365                                         } else { // This section will make the value an array if multiple tags with the same name appear together
 366                                                 $current[$tag] = [$current[$tag], $result]; // This will combine the existing item and the new item together to make an array
 367                                                 $repeated_tag_index[$tag.'_'.$level] = 2;
 368
 369                                                 if (isset($current[$tag.'_attr'])) { // The attribute of the last(0th) tag must be moved as well
 370                                                         $current[$tag]['0_attr'] = $current[$tag.'_attr'];
 371                                                         unset($current[$tag.'_attr']);
 372                                                 }
 373                                         }
 374                                         $last_item_index = $repeated_tag_index[$tag.'_'.$level]-1;
 375                                         $current = &$current[$tag][$last_item_index];
 376                                 }
 377                         } elseif ($type == "complete") { // Tags that ends in 1 line '<tag />'
 378                                 //See if the key is already taken.
 379                                 if (!isset($current[$tag])) { //New Key
 380                                         $current[$tag] = $result;
 381                                         $repeated_tag_index[$tag.'_'.$level] = 1;
 382                                         if ($priority == 'tag' and $attributes_data) {
 383                                                 $current[$tag. '_attr'] = $attributes_data;
 384                                         }
 385                                 } else { // If taken, put all things inside a list(array)
 386                                         if (isset($current[$tag][0]) and is_array($current[$tag])) { // If it is already an array...
 387
 388                                                 // ...push the new element into that array.
 389                                                 $current[$tag][$repeated_tag_index[$tag.'_'.$level]] = $result;
 390
 391                                                 if ($priority == 'tag' and $get_attributes and $attributes_data) {
 392                                                         $current[$tag][$repeated_tag_index[$tag.'_'.$level] . '_attr'] = $attributes_data;
 393                                                 }
 394                                                 $repeated_tag_index[$tag.'_'.$level]++;
 395                                         } else { // If it is not an array...
 396                                                 $current[$tag] = [$current[$tag], $result]; //...Make it an array using using the existing value and the new value
 397                                                 $repeated_tag_index[$tag.'_'.$level] = 1;
 398                                                 if ($priority == 'tag' and $get_attributes) {
 399                                                         if (isset($current[$tag.'_attr'])) { // The attribute of the last(0th) tag must be moved as well
 400
 401                                                                 $current[$tag]['0_attr'] = $current[$tag.'_attr'];
 402                                                                 unset($current[$tag.'_attr']);
 403                                                         }
 404
 405                                                         if ($attributes_data) {
 406                                                                 $current[$tag][$repeated_tag_index[$tag.'_'.$level] . '_attr'] = $attributes_data;
 407                                                         }
 408                                                 }
 409                                                 $repeated_tag_index[$tag.'_'.$level]++; // 0 and 1 indexes are already taken
 410                                         }
 411                                 }
 412                         } elseif ($type == 'close') { // End of tag '</tag>'
 413                                 $current = &$parent[$level-1];
 414                         }
 415                 }
 416
 417                 return($xml_array);
 418         }
 419
 420         /**
 421          * Delete a node in a XML object
 422          *
 423          * @param \DOMDocument $doc  XML document
 424          * @param string $node Node name
 425          * @return void
 426          */
 427         public static function deleteNode(\DOMDocument $doc, $node)
 428         {
 429                 $xpath = new DOMXPath($doc);
 430                 $list = $xpath->query("//".$node);
 431                 foreach ($list as $child) {
 432                         $child->parentNode->removeChild($child);
 433                 }
 434         }
 435
 436         /**
 437          * Parse XML string
 438          *
 439          * @param string  $s
 440          * @param boolean $suppress_log
 441          * @return Object
 442          */
 443         public static function parseString(string $s, bool $suppress_log = false)
 444         {
 445                 libxml_use_internal_errors(true);
 446
 447                 $x = @simplexml_load_string($s);
 448                 if (!$x) {
 449                         if (!$suppress_log) {
 450                                 Logger::error('Error(s) while parsing XML string.', ['callstack' => System::callstack()]);
 451                                 foreach (libxml_get_errors() as $err) {
 452                                         Logger::info('libxml error', ['code' => $err->code, 'position' => $err->line . ":" . $err->column, 'message' => $err->message]);
 453                                 }
 454                                 Logger::debug('Erroring XML string', ['xml' => $s]);
 455                         }
 456                         libxml_clear_errors();
 457                 }
 458                 return $x;
 459         }
 460
 461         public static function getFirstNodeValue(DOMXPath $xpath, $element, $context = null)
 462         {
 463                 $result = $xpath->evaluate($element, $context);
 464                 if (!is_object($result)) {
 465                         return '';
 466                 }
 467
 468                 $first_item = $result->item(0);
 469                 if (!is_object($first_item)) {
 470                         return '';
 471                 }
 472
 473                 return $first_item->nodeValue;
 474         }
 475
 476         public static function getFirstAttributes(DOMXPath $xpath, $element, $context = null)
 477         {
 478                 $result = $xpath->query($element, $context);
 479                 if (!is_object($result)) {
 480                         return false;
 481                 }
 482
 483                 $first_item = $result->item(0);
 484                 if (!is_object($first_item)) {
 485                         return false;
 486                 }
 487
 488                 return $first_item->attributes;
 489         }
 490
 491         public static function queryValue($xpath, $search, $context)
 492         {
 493                 $result = $xpath->query($search, $context)->item(0)->nodeValue;
 494                 if (!is_object($result)) {
 495                         return '';
 496                 }
 497                 return $result->nodeValue;
 498         }
 499
 500         /**
 501          * escape text ($str) for XML transport
 502          *
 503          * @param string $str
 504          * @return string Escaped text.
 505          */
 506         public static function escape($str)
 507         {
 508                 $buffer = htmlspecialchars($str, ENT_QUOTES, 'UTF-8');
 509                 $buffer = trim($buffer);
 510
 511                 return $buffer;
 512         }
 513
 514         /**
 515          * undo an escape
 516          *
 517          * @param string $s xml escaped text
 518          * @return string unescaped text
 519          */
 520         public static function unescape($s)
 521         {
 522                 $ret = htmlspecialchars_decode($s, ENT_QUOTES);
 523                 return $ret;
 524         }
 525
 526         /**
 527          * apply escape() to all values of array $val, recursively
 528          *
 529          * @param array $val
 530          * @return array|string
 531          */
 532         public static function arrayEscape($val)
 533         {
 534                 if (is_bool($val)) {
 535                         return $val ? 'true' : 'false';
 536                 } elseif (is_array($val)) {
 537                         return array_map('XML::arrayEscape', $val);
 538                 }
 539
 540                 return self::escape((string) $val);
 541         }
 542 }