se/SE/se-lib/XML.php

<?php

Lib::loadClass('Core_XmlWriter');
// $zxo = new XMLDiff\Memory;// @require xmldiff pecl package installed

// TODO:TYPE_RECURCE: if type has local prefix - choose:
//	 - find recurse typeName and restrictions
//	 - find types and save in SystemObjectField table (if simpleType)
//	 - ? find types and save in SystemObject table + SystemObjectField table (if complexType)

class XML {

	public static function xmlToArray($xml) {

	}

	public static function readXmlFileToArray($filePath) {
		$z = new XMLReader();
		if (!$z) throw new HttpException("Class XMLReader not installed", 500);
		$fileName = basename($filePath);
		if (!$z->open($filePath)) throw new Exception("Failed to open ant schema file '{$fileName}'", 500);
		return self::xmlReadToArray($z);
	}
	public static function xmlReadToArray($z) {
		while ($z->read()) {
			if (XMLReader::ELEMENT !== $z->nodeType) continue;
			return self::xmlReadRecurse($z);
		}
	}
	public static function xmlReadRecurse($z) {
		$node = [ $z->name, [], [] ];// name, attrs, childrens
		$depth = $z->depth;
		$isEmpty = $z->isEmptyElement;
		if ($z->hasAttributes) {
			while ($z->moveToNextAttribute()) {
				$node[1][$z->name] = $z->value;
			}
		}
		if ($isEmpty) {
			$node[2] = null;
			return $node;
		}
		// switch ($z->name) {
		// 	case 'xsd:complexType': $node[1]['name'] = $z->getAttribute('name'); break;
		// 	case 'xsd:simpleType': $node[1]['name'] = $z->getAttribute('name'); break;
		// 	case 'xsd:restriction': $node[1]['base'] = $z->getAttribute('base'); break;
		// 	case 'xsd:enumeration': $node[1]['value'] = $z->getAttribute('value'); break;
		// 	case 'xsd:element': $node[1] = [
		// 		'type' => $z->getAttribute('type')
		// 	]; break;
		// 	case 'xsd:complexContent': break;
		// 	case 'xsd:sequence': break;
		// 	case 'xsd:attribute': $node[1]['base'] = $z->getAttribute('base'); break;
		// 	case 'xsd:extension': $node[1]['base'] = $z->getAttribute('base'); break;
		// 	default: UI::alert('warning', "TODO: read attributes from: d({$z->depth}) name($z->name)");
		// }
		while ($z->read() && $z->depth > $depth) {
			if (XMLReader::ELEMENT !== $z->nodeType) continue;
			// if ($z->depth == $depth + 2)
			$node[2][] = self::xmlReadRecurse($z);
			// else $node[2][] = "d({$z->depth}) name($z->name)";
		}
		return $node;
	}

	public static function printXmlFromArray($xml) {
		$xmlWriter = new Core_XmlWriter();
		$xmlWriter->openUri('php://output');
		$xmlWriter->setIndent(true);
		$xmlWriter->startDocument('1.0','UTF-8');
		$xmlWriter->h($xml[0], $xml[1], $xml[2]);
		$xmlWriter->endDocument();
	}

	public static function findTargetNamespace($docArray) {
		return V::get('targetNamespace', '', $docArray[1]);
	}

	public static function findSimpleTypeNode($docArray, $name) {
		foreach ($docArray[2] as $child) {
			if ('simpleType' === self::getTagName($child[0])) {
				if ($name === $child[1]['name']) {
					return $child;
				}
			}
		}
	}

	public static function findTargetNamespacePrefix($docArray) {
		$tns = self::findTargetNamespace($docArray);
		if (!$tns) throw new Exception("targetNamespace not defined");
		foreach ($docArray[1] as $attr => $val) {
			if ('xmlns:' !== substr($attr, 0, 6)) continue;
			if ($tns === $val) return substr($attr, 6);
		}
		throw new Exception("Missing targetNamespace xmlns:...");
	}

	public static function findElementName($docArray, $nodeArray) {
		if (!empty($nodeArray[1]['name'])) return $nodeArray[1]['name'];
		if (!empty($nodeArray[1]['ref'])) return $nodeArray[1]['ref'];
		throw new Exception("Missing xsd:element name");
	}

	public static function findElementType($docArray, $nodeArray) {
		$fieldName = self::findElementName($docArray, $nodeArray);
		if (!empty($nodeArray[1]['type'])) {
			// TODO:TYPE_RECURCE: if local ns prefix then find correct typeName?
			// TODO:TYPE_RECURCE:the same for restrictions
			$type = $nodeArray[1]['type'];
			if ('xs:' == substr($type, 0, 3)) return "xsd:" . substr($type, 3);
			if ('xsd:' == substr($type, 0, 4)) return $type;
			list($prefix, $name) = explode(':', $type);
			if ($prefix === self::findTargetNamespacePrefix($docArray)) {
				$simpleTypeNode = self::findSimpleTypeNode($docArray, $name);
				DBG::log($simpleTypeNode, 'array', "\$simpleTypeNode \$fieldName='{$fieldName}' type='{$type}'");
				if (!empty($simpleTypeNode[1]['type'])) {
					throw new Exception("TODO: findElementType node/@type => 'xsd:simpleType/@type' = '{$simpleTypeNode[1]['type']}'");
				}
				if (!empty($simpleTypeNode[2][0]) && self::isXsdTag($simpleTypeNode[2][0][0], 'restriction')) {
					$restrictionNode = $simpleTypeNode[2][0];
					if (empty($restrictionNode[1]['base'])) throw new Exception("Missing xsd:restriction/@base (node/@type => xsd:simpleType/[@type='{$simpleTypeNode[1]['type']}']/xsd:restriction')");
					$type = $restrictionNode[1]['base'];
					DBG::log($type, 'array', "findElementType \$fieldName='{$fieldName}' type='{$nodeArray[1]['type']}' => type='{$type}'");
					// check restrictions - if has enumeration then return 'p5:enum'
					$isEnum = false;
					foreach ($restrictionNode[2] as $restr) {
						if ('enumeration' === self::getTagName($restr[0])) {
							$isEnum = true;
							break;
						}
					}
					if ($isEnum) return 'p5:enum';

					// TODO: recurse with limit
					if ('xs:' == substr($type, 0, 3)) return "xsd:" . substr($type, 3);
					if ('xsd:' == substr($type, 0, 4)) return $type;
					list($prefix, $name) = explode(':', $type);
					$simpleTypeNode = self::findSimpleTypeNode($docArray, $name);
					DBG::log($simpleTypeNode, 'array', "\$simpleTypeNode \$fieldName='{$fieldName}' ... type='{$type}'");
					if (!empty($simpleTypeNode[1]['type'])) {
						throw new Exception("TODO: findElementType node/@type => 'xsd:simpleType/@type' = '{$simpleTypeNode[1]['type']}'");
					}
					if (!empty($simpleTypeNode[2][0]) && self::isXsdTag($simpleTypeNode[2][0][0], 'restriction')) {
						$restrictionNode = $simpleTypeNode[2][0];
						if (empty($restrictionNode[1]['base'])) throw new Exception("Missing xsd:restriction/@base (node/@type => xsd:simpleType/[@type='{$simpleTypeNode[1]['type']}']/xsd:restriction')");
						$type = $restrictionNode[1]['base'];
						DBG::log($type, 'array', "findElementType \$fieldName='{$fieldName}' ... type='{$type}'");
						if ('xs:' == substr($type, 0, 3)) return "xsd:" . substr($type, 3);
						if ('xsd:' == substr($type, 0, 4)) return $type;
					}
					// TODO: throw...
				}
				// TODO: throw...
				//	 0 => 'xsd:simpleType',
				//	 1 => [
				//		 'name' => 'PROCES_INIT_Simple',
				//	 ],
				//	 2 => [
				//		 0 => [
				//			 0 => 'xsd:restriction',
				//			 1 => [
				//				 'base' => 'default_db__x3A__CRM_PROCES:TYPE_Simple',
				//			 ],
				//			 2 => [
				//				 0 => [
				//					 0 => 'xsd:enumeration',
				//					 1 => [
				//						 'value' => 'PROCES_INIT',
				//					 ],
				//					 2 => NULL,
			} else {
				return $nodeArray[1]['type'];
			}
		}
		if (!empty($nodeArray[1]['ref'])) return 'ref:' . $nodeArray[1]['ref'];
		if (empty($nodeArray[2])) throw new Exception("Missing xsd:element childrens - cannot find type");
		// TODO: find in loop (xsd:annotation may accure)
		if (empty($nodeArray[2][0][0]) || !self::isXsdTag($nodeArray[2][0][0], 'simpleType')) throw new Exception("Missing 'xsd:simpleType' for field '{$fieldName}'");
		if (empty($nodeArray[2][0][2][0]) || !self::isXsdTag($nodeArray[2][0][2][0][0], 'restriction')) throw new Exception("Missing 'xsd:restriction' for field '{$fieldName}'");
		if (empty($nodeArray[2][0][2][0][1]['base'])) throw new Exception("Missing 'xsd:restriction/@base' for field '{$fieldName}'");
		return $nodeArray[2][0][2][0][1]['base'];
	}

	public static function findElementRestrictions($docArray, $nodeArray) {
		$restrictions = [];
		$fieldName = self::findElementName($docArray, $nodeArray);
		if (!empty($nodeArray[1]['nillable']) && 'true' === $nodeArray[1]['nillable']) $restrictions['nillable'] = true;
		if (!empty($nodeArray[2])) {
			foreach ($nodeArray[2] as $c) {
				switch (XML::getTagName($c[0])) {
					case 'annotation': break; // skip xsd:element/xsd:annotation @see findElementAppInfo
					case 'simpleType': { // xsd:element/xsd:simpleType
						if (empty($c[2][0]) || 'restriction' != XML::getTagName($c[2][0][0])) throw new Exception("Missing 'xsd:restriction' for field '{$fieldName}'");
						if (empty($c[2][0][1]['base'])) throw new Exception("Missing 'xsd:restriction/@base' for field '{$fieldName}'");
						// xsd:element/xsd:simpleType/xsd:restriction/xsd:string
						foreach ($c[2][0][2] as $tagRestriction) {
							// xsd:string/xsd:maxLength
							$val = $tagRestriction[1]['value'];
							if ('enumeration' == XML::getTagName($tagRestriction[0])) {
								$restrictions['enumeration'][$val] = $val;
							} else {
								$restrictions[XML::getTagName($tagRestriction[0])] = $val;
							}
						}
					} break;
					default: {
						DBG::log($c, 'array', "Not imeplemented element child '{$c[0]}'");
					}
				}
			}
		}
		if (empty($nodeArray[2]) && !empty($nodeArray[1]['type'])) {
			list($prefix, $name) = explode(':', $nodeArray[1]['type']);
			if ($prefix === XML::findTargetNamespacePrefix($docArray)) {
				$simpleTypeNode = self::findSimpleTypeNode($docArray, $name);
				return XML::findSimpleTypeRestrictions($docArray, $simpleTypeNode);
			}
		}
		return $restrictions;
	}
	public static function findSimpleTypeRestrictions($docArray, $nodeArray) { // $nodeArray must be 'simpleType' === XML::getTagName($nodeArray[0])
		DBG::log($nodeArray, 'array', "TODO: findElementRestrictions \$nodeArray");
		$restrictions = [];
		if (empty($nodeArray[2][0]) || 'restriction' != XML::getTagName($nodeArray[2][0][0])) throw new Exception("Missing 'xsd:restriction' for field '{$fieldName}'");
		if (empty($nodeArray[2][0][1]['base'])) throw new Exception("Missing 'xsd:restriction/@base' for field '{$fieldName}'");
		// xsd:element/xsd:simpleType/xsd:restriction/xsd:string
		foreach ($nodeArray[2][0][2] as $tagRestriction) {
			// xsd:string/xsd:maxLength
			$val = $tagRestriction[1]['value'];
			if ('enumeration' == XML::getTagName($tagRestriction[0])) {
				$restrictions['enumeration'][$val] = $val;
			} else {
				$restrictions[XML::getTagName($tagRestriction[0])] = $val;
			}
		}
		return $restrictions;
	}

	public static function findElementAppInfo($docArray, $nodeArray) {
		$appInfo = [];
		$fieldName = self::findElementName($docArray, $nodeArray);
		if (!empty($nodeArray[2])) {
			foreach ($nodeArray[2] as $c) {
				switch (XML::getTagName($c[0])) {
					case 'annotation': { // skip xsd:element/xsd:annotation
						DBG::log($c, 'array', "xsd:annotation/xsd:appinfo '{$fieldName}'");
						// <xsd:annotation>
						//		<xsd:appinfo>
						//			 <system_cache__appinfo:flat_relation_cache>
						//					<system_cache__appinfo:source system_cache__appinfo:name="ID"
						//																				system_cache__appinfo:xpath="default_db__x3A__CRM_WSKAZNIK:CRM_WSKAZNIK/ID_PROCES"/>
						$prefix = 'system_cache__appinfo';
						foreach ($c[2] as $cc) {
							if ('appinfo' == XML::getTagName($cc[0])) {
								foreach ($cc[2] as $appTag) {
									$appInfo[XML::getTagName($appTag[0])] = XML::readAppInfoRecurse($docArray, $appTag);
								}
							}
						}
					} break;
					case 'simpleType': break; // skip xsd:element/xsd:simpleType @see findElementRestrictions
					default: {
						DBG::log($c, 'array', "Not imeplemented element child '{$c[0]}'");
					}
				}
			}
		}
		DBG::log($fieldName, 'array', "findElementAppInfo fieldName='{$fieldName}'");
		return $appInfo;
	}

	public static function getTagName($xsdName) {
		return (false !== ($pos = strpos($xsdName, ':')))
			? substr($xsdName, $pos + 1)
			: $xsdName;
	}

	public static function isXsdTag($xsdName, $expectedTagName) {
		list($xsdPrefix, $tagName) = explode(':', $xsdName);
		switch ($xsdPrefix) {
			case 'xs':
			case 'xsd': return ($tagName === $expectedTagName);
		}
		return false;
	}

	public static function readAppInfoRecurse($docArray, $nodeArray) {
		$appInfo = [];
		if (!empty($nodeArray[1])) foreach ($nodeArray[1] as $attrName => $attrVal) {
			$appInfo['@' . XML::getTagName($attrName)] = $attrVal;
		}
		if (!empty($nodeArray[2])) foreach ($nodeArray[2] as $appTag) {
			$appInfo[XML::getTagName($appTag[0])] = XML::readAppInfoRecurse($docArray, $appTag);
		}
		// TODO: text nodes
		return $appInfo;
	}

	public static function findFieldsFromSequence($docArray, $nodeArray) {
		if (!self::isXsdTag($nodeArray[0], 'sequence')) throw new Exception("Error Parsing Schema - expected 'sequence'");
		$fields = [];
		foreach ($nodeArray[2] as $f) {
			if (!self::isXsdTag($f[0], 'element')) {
				DBG::log($n, 'array', "Schema xsd parse error - Not implemented node type '{$f[0]}'");
				continue;
			}
			$fieldName = XML::findElementName($docArray, $f); // V::get('name', '', $f[1]);
			if (!$fieldName) throw new Exception("Error Parsing Schema - expected 'element[@name]'");
			if ('__' === substr($fieldName, 0, 2)) continue;
			if (!V::get('type', '', $f[1]) && !V::get('ref', '', $f[1]) && empty($f[2])) {
				UI::alert('danger', "Skipping not implemented field structure '{$fieldName}'");
				DBG::log($f, 'array', "Skipping not implemented field structure '{$fieldName}'");
				continue;
			}
			$fields[$fieldName] = [
				'type' => XML::findElementType($docArray, $f),
				'minOccurs' => V::get('minOccurs', 0, $f[1], 'int'),
				'maxOccurs' => V::get('maxOccurs', '1', $f[1]),
				'restrictions' => XML::findElementRestrictions($docArray, $f),
				'appInfo' => XML::findElementAppInfo($docArray, $f),
			];
		}
		return $fields;
	}

	public static function findFieldsFromComplexContent($docArray, $nodeArray) {
		// xsd:complexType / xsd:complexContent / xsd:restriction [ @base = "default_db__x3A__CRM_PROCES:CRM_PROCES" ]
		// xsd:complexType / xsd:complexContent / xsd:extension [ @base = "default_db__x3A__CRM_PROCES:CRM_PROCES" ]
		switch (XML::getTagName($nodeArray[2][0][0])) {
			case 'extension': return XML::findFieldsFromExtension($docArray, $nodeArray[2][0]);
			case 'restriction': return XML::findFieldsFromRestriction($docArray, $nodeArray[2][0]);
		}
		// TODO:? $xsdType['extensionBase'] = V::get('base', '', $nodeArray[1]);
		// TODO:? $xsdType['restrictionBase'] = V::get('base', '', $nodeArray[1]);
	}
	public static function findFieldsFromExtension($docArray, $nodeArray) {
		$fields = [];
		if (!self::isXsdTag($nodeArray[2][0][0], 'sequence')) throw new Exception("Error Parsing Schema - expected 'complexType/complexContent/extension/sequence'");
		foreach ($nodeArray[2][0][2] as $f) {
			if (!self::isXsdTag($f[0], 'element')) {
				DBG::log($n, 'array', "Schema xsd parse error - Not implemented node type '{$f[0]}'");
				continue;
			}
			$fieldName = XML::findElementName($docArray, $f); // V::get('name', '', $f[1]);
			if (!$fieldName) throw new Exception("Error Parsing Schema - expected 'element[@name]'");
			if ('__' === substr($fieldName, 0, 2)) continue;
			if (!V::get('type', '', $f[1]) && !V::get('ref', '', $f[1]) && empty($f[2])) {
				UI::alert('danger', "Skipping not implemented field structure '{$fieldName}'");
				continue;
			}
			$fields[$fieldName] = [
				'type' => XML::findElementType($docArray, $f),
				'minOccurs' => V::get('minOccurs', 0, $f[1], 'int'),
				'maxOccurs' => V::get('maxOccurs', '1', $f[1]),
				'restrictions' => XML::findElementRestrictions($docArray, $f),
				'appInfo' => XML::findElementAppInfo($docArray, $f),
			];
		}
		return $fields;
	}
	public static function findFieldsFromRestriction($docArray, $nodeArray) {
		$fields = [];
		if (!self::isXsdTag($nodeArray[2][0][0], 'sequence')) throw new Exception("Error Parsing Schema - expected 'complexType/complexContent/restriction/sequence'");
		foreach ($nodeArray[2][0][2] as $f) {
			if (!self::isXsdTag($f[0], 'element')) {
				DBG::log($n, 'array', "Schema xsd parse error - Not implemented node type '{$f[0]}'");
				continue;
			}
			$fieldName = XML::findElementName($docArray, $f); // V::get('name', '', $f[1]);
			if (!$fieldName) throw new Exception("Error Parsing Schema - expected 'element[@name]'");
			if ('__' === substr($fieldName, 0, 2)) continue;
			if (!V::get('type', '', $f[1]) && !V::get('ref', '', $f[1]) && empty($f[2])) {
				UI::alert('danger', "Skipping not implemented field structure '{$fieldName}'");
				continue;
			}
			$fields[$fieldName] = [
				'type' => XML::findElementType($docArray, $f),
				'minOccurs' => V::get('minOccurs', 0, $f[1], 'int'),
				'maxOccurs' => V::get('maxOccurs', '1', $f[1]),
				'restrictions' => XML::findElementRestrictions($docArray, $f),
				'appInfo' => XML::findElementAppInfo($docArray, $f),
			];
		}
		return $fields;
	}

	public static function getXsdTypeFromXsdSchema($xsdFilePath, $namespace, $name) {
		$schema = XML::readXmlFileToArray($xsdFilePath);
		if (empty($schema)) throw new Exception("Missing schema file for '{$namespace}'");
		$xsdType = [ // find xsd:element with @name = $name
			'nsPrefix' => null,
			'name' => null,
			'nsUri' => null,
			'primaryKey' => null,
			'targetNsUri' => V::get('targetNamespace', '', $schema[1])
		];
		if (!$xsdType['targetNsUri']) throw new Exception("Missing schema target namespace declaration '{$name}'");
		foreach ($schema[2] as $n) {
			if (!XML::isXsdTag($n[0], 'element')) continue;
			if ($name != V::get('name', '', $n[1])) continue;
			list($xsdType['nsPrefix'], $xsdType['name']) = explode(':', V::get('type', '', $n[1]));
		}
		if (!$xsdType['nsPrefix'] || !$xsdType['name']) throw new Exception("Missing schema root element name = '{$name}'");

		$xsdType['nsUri'] = V::get("xmlns:{$xsdType['nsPrefix']}", '', $schema[1]);// find xmlns:default_objects = "https://biuro.biall-net.pl/wfs/default_objects"
		if (!$xsdType['nsUri']) throw new Exception("Missing schema root element namespace declaration = '{$name}'");
		if ($xsdType['nsUri'] != $xsdType['targetNsUri']) throw new Exception("TODO: type ns is not the same as targetNamespace '{$name}'");// TODO

		foreach ($schema[2] as $n) {
			if (!XML::isXsdTag($n[0], 'complexType')) continue;
			if ($xsdType['name'] != V::get('name', '', $n[1])) continue;
			DBG::log($n, 'array', "find system_cache__appinfo:primaryKey from complexType");
			if (array_key_exists('system_cache__appinfo:primaryKey', $n[1])) {
				$xsdType['primaryKey'] = V::get('system_cache__appinfo:primaryKey', '', $n[1]);
			}
			// complexType/sequence/element
			// complexType/complexContent/extension[base=...]/sequence/element
			switch ($n[2][0][0]) {
				case 'xs:sequence':
				case 'xsd:sequence': $xsdType['struct'] = XML::findFieldsFromSequence($schema, $n[2][0]); break;
				case 'xs:complexContent':
				case 'xsd:complexContent': $xsdType['struct'] = XML::findFieldsFromComplexContent($schema, $n[2][0]); break;
				case 'xs:annotation':
				case 'xsd:annotation': {
					switch ($n[2][1][0]) {
						case 'xs:sequence':
						case 'xsd:sequence': $xsdType['struct'] = XML::findFieldsFromSequence($schema, $n[2][1]); break;
						case 'xs:complexContent':
						case 'xsd:complexContent': $xsdType['struct'] = XML::findFieldsFromComplexContent($schema, $n[2][1]); break;
					}
				} break;
			}
			// if ($n[2][0][0] != 'xsd:complexContent') throw new Exception("Error Parsing Schema - expected 'complexType/complexContent'");
		}
		if (empty($xsdType['primaryKey'])) {
			foreach ($xsdType['struct'] as $fieldName => $field) {
				if ('ID' === strtoupper($fieldName)) {
					$xsdType['primaryKey'] = $fieldName;
					break;
				}
			}
		}
		if (empty($xsdType['primaryKey'])) throw new Exception("Missing primaryKey for schema '{$namespace}'");
		return $xsdType;
	}

}