friendica/src/Util/XML.php

465 lines
14 KiB
PHP
Raw Normal View History

<?php
/**
* @file src/Util/XML.php
2016-03-29 17:54:36 +02:00
*/
namespace Friendica\Util;
2016-03-29 17:54:36 +02:00
use DOMXPath;
use SimpleXMLElement;
2016-03-29 17:54:36 +02:00
/**
* @brief This class contain methods to work with XML data
*/
class XML
{
2016-03-20 16:16:15 +01:00
/**
* @brief Creates an XML structure out of a given array
*
* @param array $array The array of the XML structure that will be generated
* @param object $xml The createdXML will be returned by reference
* @param bool $remove_header Should the XML header be removed or not?
* @param array $namespaces List of namespaces
* @param bool $root interally used parameter. Mustn't be used from outside.
2016-03-20 16:16:15 +01:00
*
* @return string The created XML
*/
public static function fromArray($array, &$xml, $remove_header = false, $namespaces = [], $root = true)
{
2016-03-13 09:57:44 +01:00
if ($root) {
2016-11-27 20:42:40 +01:00
foreach ($array as $key => $value) {
foreach ($namespaces as $nskey => $nsvalue) {
2016-03-13 09:57:44 +01:00
$key .= " xmlns".($nskey == "" ? "":":").$nskey.'="'.$nsvalue.'"';
2016-11-27 20:42:40 +01:00
}
2016-03-13 09:57:44 +01:00
if (is_array($value)) {
$root = new SimpleXMLElement("<".$key."/>");
self::fromArray($value, $root, $remove_header, $namespaces, false);
2016-11-27 20:42:40 +01:00
} else {
$root = new SimpleXMLElement("<".$key.">".xmlify($value)."</".$key.">");
2016-11-27 20:42:40 +01:00
}
$dom = dom_import_simplexml($root)->ownerDocument;
$dom->formatOutput = true;
$xml = $dom;
$xml_text = $dom->saveXML();
2016-11-27 20:42:40 +01:00
if ($remove_header) {
$xml_text = trim(substr($xml_text, 21));
2016-11-27 20:42:40 +01:00
}
return $xml_text;
}
}
$element = null;
foreach ($array as $key => $value) {
if (!isset($element) && isset($xml)) {
2016-07-17 19:42:30 +02:00
$element = $xml;
2016-11-27 20:42:40 +01:00
}
2016-07-17 19:42:30 +02:00
if (is_integer($key)) {
2016-08-02 12:24:07 +02:00
if (isset($element)) {
if (is_scalar($value)) {
$element[0] = $value;
} else {
/// @todo: handle nested array values
}
}
2016-07-17 19:42:30 +02:00
continue;
}
$element_parts = explode(":", $key);
if ((count($element_parts) > 1) && isset($namespaces[$element_parts[0]])) {
$namespace = $namespaces[$element_parts[0]];
2016-11-27 20:42:40 +01:00
} elseif (isset($namespaces[""])) {
$namespace = $namespaces[""];
2016-11-27 20:42:40 +01:00
} else {
$namespace = null;
2016-11-27 20:42:40 +01:00
}
// Remove undefined namespaces from the key
if ((count($element_parts) > 1) && is_null($namespace)) {
$key = $element_parts[1];
2016-11-27 20:42:40 +01:00
}
if (substr($key, 0, 11) == "@attributes") {
if (!isset($element) || !is_array($value)) {
2016-03-13 09:57:44 +01:00
continue;
2016-11-27 20:42:40 +01:00
}
2016-03-13 09:57:44 +01:00
foreach ($value as $attr_key => $attr_value) {
$element_parts = explode(":", $attr_key);
if ((count($element_parts) > 1) && isset($namespaces[$element_parts[0]])) {
2016-03-13 09:57:44 +01:00
$namespace = $namespaces[$element_parts[0]];
2016-11-27 20:42:40 +01:00
} else {
$namespace = null;
2016-11-27 20:42:40 +01:00
}
2016-03-13 09:57:44 +01:00
2016-07-17 19:42:30 +02:00
$element->addAttribute($attr_key, $attr_value, $namespace);
2016-03-13 09:57:44 +01:00
}
continue;
}
2016-11-27 20:42:40 +01:00
if (!is_array($value)) {
2016-03-13 09:57:44 +01:00
$element = $xml->addChild($key, xmlify($value), $namespace);
2016-11-27 20:42:40 +01:00
} elseif (is_array($value)) {
$element = $xml->addChild($key, null, $namespace);
self::fromArray($value, $element, $remove_header, $namespaces, false);
2016-03-13 09:57:44 +01:00
}
}
}
2016-03-20 16:16:15 +01:00
/**
* @brief Copies an XML object
*
* @param object $source The XML source
* @param object $target The XML target
2016-03-20 16:16:15 +01:00
* @param string $elementname Name of the XML element of the target
* @return void
2016-03-20 16:16:15 +01:00
*/
public static function copy(&$source, &$target, $elementname)
{
if (count($source->children()) == 0) {
2016-03-14 23:54:01 +01:00
$target->addChild($elementname, xmlify($source));
} else {
$child = $target->addChild($elementname);
foreach ($source->children() as $childfield => $childentry) {
self::copy($childentry, $child, $childfield);
2016-11-27 20:42:40 +01:00
}
}
}
/**
* @brief Create an XML element
*
* @param object $doc XML root
* @param string $element XML element name
* @param string $value XML value
* @param array $attributes array containing the attributes
*
* @return object XML element object
*/
public static function createElement($doc, $element, $value = "", $attributes = [])
{
$element = $doc->createElement($element, xmlify($value));
foreach ($attributes as $key => $value) {
$attribute = $doc->createAttribute($key);
$attribute->value = xmlify($value);
$element->appendChild($attribute);
}
return $element;
}
/**
* @brief Create an XML and append it to the parent object
*
* @param object $doc XML root
* @param object $parent parent object
* @param string $element XML element name
* @param string $value XML value
* @param array $attributes array containing the attributes
* @return void
*/
public static function addElement($doc, $parent, $element, $value = "", $attributes = [])
{
$element = self::createElement($doc, $element, $value, $attributes);
$parent->appendChild($element);
}
/**
* @brief Convert an XML document to a normalised, case-corrected array
* used by webfinger
2016-08-02 12:24:07 +02:00
*
* @param object $xml_element The XML document
2016-08-02 12:24:07 +02:00
* @param integer $recursion_depth recursion counter for internal use - default 0
* internal use, recursion counter
2016-08-02 12:24:07 +02:00
*
2017-12-17 21:27:50 +01:00
* @return array | string The array from the xml element or the string
*/
public static function elementToArray($xml_element, &$recursion_depth = 0)
{
// If we're getting too deep, bail out
if ($recursion_depth > 512) {
return(null);
}
$xml_element_copy = '';
2016-11-27 20:42:40 +01:00
if (!is_string($xml_element)
&& !is_array($xml_element)
&& (get_class($xml_element) == 'SimpleXMLElement')
) {
$xml_element_copy = $xml_element;
$xml_element = get_object_vars($xml_element);
}
if (is_array($xml_element)) {
$result_array = [];
if (count($xml_element) <= 0) {
return (trim(strval($xml_element_copy)));
}
2016-11-27 20:42:40 +01:00
foreach ($xml_element as $key => $value) {
$recursion_depth++;
$result_array[strtolower($key)] = self::elementToArray($value, $recursion_depth);
$recursion_depth--;
}
if ($recursion_depth == 0) {
$temp_array = $result_array;
$result_array = [
strtolower($xml_element_copy->getName()) => $temp_array,
];
}
return ($result_array);
} else {
return (trim(strval($xml_element)));
}
}
/**
* @brief Convert the given XML text to an array in the XML structure.
2016-08-02 12:24:07 +02:00
*
* Xml::toArray() will convert the given XML text to an array in the XML structure.
* Link: http://www.bin-co.com/php/scripts/xml2array/
* Portions significantly re-written by mike@macgirvin.com for Friendica
* (namespaces, lowercase tags, get_attribute default changed, more...)
2016-08-02 12:24:07 +02:00
*
* Examples: $array = Xml::toArray(file_get_contents('feed.xml'));
* $array = Xml::toArray(file_get_contents('feed.xml', true, 1, 'attribute'));
2016-08-02 12:24:07 +02:00
*
* @param object $contents The XML text
* @param boolean $namespaces True or false include namespace information
* in the returned array as array elements.
2016-08-02 12:24:07 +02:00
* @param integer $get_attributes 1 or 0. If this is 1 the function will get the attributes as well as the tag values -
* this results in a different array structure in the return value.
* @param string $priority Can be 'tag' or 'attribute'. This will change the way the resulting
* array sturcture. For 'tag', the tags are given more importance.
2016-08-02 12:24:07 +02:00
*
* @return array The parsed XML in an array form. Use print_r() to see the resulting array structure.
*/
public static function toArray($contents, $namespaces = true, $get_attributes = 1, $priority = 'attribute')
{
2016-11-27 20:42:40 +01:00
if (!$contents) {
return [];
2016-11-27 20:42:40 +01:00
}
2016-11-27 20:42:40 +01:00
if (!function_exists('xml_parser_create')) {
logger('Xml::toArray: parser function missing');
return [];
}
libxml_use_internal_errors(true);
libxml_clear_errors();
2016-11-27 20:42:40 +01:00
if ($namespaces) {
$parser = @xml_parser_create_ns("UTF-8", ':');
2016-11-27 20:42:40 +01:00
} else {
$parser = @xml_parser_create();
2016-11-27 20:42:40 +01:00
}
2016-11-27 20:42:40 +01:00
if (! $parser) {
logger('Xml::toArray: xml_parser_create: no resource');
return [];
}
xml_parser_set_option($parser, XML_OPTION_TARGET_ENCODING, "UTF-8");
// http://minutillo.com/steve/weblog/2004/6/17/php-xml-and-character-encodings-a-tale-of-sadness-rage-and-data-loss
xml_parser_set_option($parser, XML_OPTION_CASE_FOLDING, 0);
xml_parser_set_option($parser, XML_OPTION_SKIP_WHITE, 1);
@xml_parse_into_struct($parser, trim($contents), $xml_values);
@xml_parser_free($parser);
2016-11-27 20:42:40 +01:00
if (! $xml_values) {
logger('Xml::toArray: libxml: parse error: ' . $contents, LOGGER_DATA);
2016-11-27 20:42:40 +01:00
foreach (libxml_get_errors() as $err) {
logger('libxml: parse: ' . $err->code . " at " . $err->line . ":" . $err->column . " : " . $err->message, LOGGER_DATA);
2016-11-27 20:42:40 +01:00
}
libxml_clear_errors();
return;
}
//Initializations
$xml_array = [];
$parents = [];
$opened_tags = [];
$arr = [];
$current = &$xml_array; // Reference
// Go through the tags.
$repeated_tag_index = []; // Multiple tags with same name will be turned into an array
2016-11-27 20:42:40 +01:00
foreach ($xml_values as $data) {
$tag = $data['tag'];
$type = $data['type'];
$level = $data['level'];
$attributes = isset($data['attributes']) ? $data['attributes'] : null;
$value = isset($data['value']) ? $data['value'] : null;
$result = [];
$attributes_data = [];
2016-11-27 20:42:40 +01:00
if (isset($value)) {
if ($priority == 'tag') {
$result = $value;
} else {
$result['value'] = $value; // Put the value in a assoc array if we are in the 'Attribute' mode
}
}
//Set the attributes too.
2016-11-27 20:42:40 +01:00
if (isset($attributes) and $get_attributes) {
foreach ($attributes as $attr => $val) {
if ($priority == 'tag') {
2016-11-27 20:42:40 +01:00
$attributes_data[$attr] = $val;
} else {
$result['@attributes'][$attr] = $val; // Set all the attributes in a array called 'attr'
}
}
}
// See tag status and do the needed.
2016-11-27 20:42:40 +01:00
if ($namespaces && strpos($tag, ':')) {
$namespc = substr($tag, 0, strrpos($tag, ':'));
$tag = strtolower(substr($tag, strlen($namespc)+1));
$result['@namespace'] = $namespc;
}
$tag = strtolower($tag);
2016-11-27 20:42:40 +01:00
if ($type == "open") { // The starting of the tag '<tag>'
$parent[$level-1] = &$current;
2017-06-09 03:03:44 +02:00
if (!is_array($current) || (!in_array($tag, array_keys($current)))) { // Insert New tag
$current[$tag] = $result;
2016-11-27 20:42:40 +01:00
if ($attributes_data) {
$current[$tag. '_attr'] = $attributes_data;
}
$repeated_tag_index[$tag.'_'.$level] = 1;
$current = &$current[$tag];
} else { // There was another element with the same tag name
2016-11-27 20:42:40 +01:00
if (isset($current[$tag][0])) { // If there is a 0th element it is already an array
$current[$tag][$repeated_tag_index[$tag.'_'.$level]] = $result;
$repeated_tag_index[$tag.'_'.$level]++;
} else { // This section will make the value an array if multiple tags with the same name appear together
$current[$tag] = [$current[$tag], $result]; // This will combine the existing item and the new item together to make an array
$repeated_tag_index[$tag.'_'.$level] = 2;
2016-11-27 20:42:40 +01:00
if (isset($current[$tag.'_attr'])) { // The attribute of the last(0th) tag must be moved as well
$current[$tag]['0_attr'] = $current[$tag.'_attr'];
unset($current[$tag.'_attr']);
}
}
$last_item_index = $repeated_tag_index[$tag.'_'.$level]-1;
$current = &$current[$tag][$last_item_index];
}
2016-11-27 20:42:40 +01:00
} elseif ($type == "complete") { // Tags that ends in 1 line '<tag />'
//See if the key is already taken.
2016-11-27 20:42:40 +01:00
if (!isset($current[$tag])) { //New Key
$current[$tag] = $result;
$repeated_tag_index[$tag.'_'.$level] = 1;
2016-11-27 20:42:40 +01:00
if ($priority == 'tag' and $attributes_data) {
$current[$tag. '_attr'] = $attributes_data;
}
} else { // If taken, put all things inside a list(array)
2016-11-27 20:42:40 +01:00
if (isset($current[$tag][0]) and is_array($current[$tag])) { // If it is already an array...
// ...push the new element into that array.
$current[$tag][$repeated_tag_index[$tag.'_'.$level]] = $result;
2016-11-27 20:42:40 +01:00
if ($priority == 'tag' and $get_attributes and $attributes_data) {
$current[$tag][$repeated_tag_index[$tag.'_'.$level] . '_attr'] = $attributes_data;
}
$repeated_tag_index[$tag.'_'.$level]++;
} else { // If it is not an array...
$current[$tag] = [$current[$tag], $result]; //...Make it an array using using the existing value and the new value
$repeated_tag_index[$tag.'_'.$level] = 1;
2016-11-27 20:42:40 +01:00
if ($priority == 'tag' and $get_attributes) {
if (isset($current[$tag.'_attr'])) { // The attribute of the last(0th) tag must be moved as well
$current[$tag]['0_attr'] = $current[$tag.'_attr'];
unset($current[$tag.'_attr']);
}
2016-11-27 20:42:40 +01:00
if ($attributes_data) {
$current[$tag][$repeated_tag_index[$tag.'_'.$level] . '_attr'] = $attributes_data;
}
}
$repeated_tag_index[$tag.'_'.$level]++; // 0 and 1 indexes are already taken
}
}
2016-11-27 20:42:40 +01:00
} elseif ($type == 'close') { // End of tag '</tag>'
$current = &$parent[$level-1];
}
}
return($xml_array);
}
/**
* @brief Delete a node in a XML object
*
* @param object $doc XML document
* @param string $node Node name
* @return void
*/
public static function deleteNode(&$doc, $node)
{
$xpath = new DOMXPath($doc);
$list = $xpath->query("//".$node);
foreach ($list as $child) {
$child->parentNode->removeChild($child);
}
}
public static function parseString($s, $strict = true)
{
// the "strict" parameter is deactivated
libxml_use_internal_errors(true);
$x = @simplexml_load_string($s);
if (!$x) {
logger('libxml: parse: error: ' . $s, LOGGER_DATA);
foreach (libxml_get_errors() as $err) {
logger('libxml: parse: ' . $err->code." at ".$err->line.":".$err->column." : ".$err->message, LOGGER_DATA);
}
libxml_clear_errors();
}
return $x;
}
public static function getFirstNodeValue($xpath, $element, $context = null)
{
$result = $xpath->evaluate($element, $context);
if (!is_object($result)) {
return '';
}
$first_item = $result->item(0);
if (!is_object($first_item)) {
return '';
}
return $first_item->nodeValue;
}
public static function getFirstAttributes($xpath, $element, $context = null)
{
$result = $xpath->query($element, $context);
if (!is_object($result)) {
return false;
}
$first_item = $result->item(0);
if (!is_object($first_item)) {
return false;
}
return $first_item->attributes;
}
}