X-Git-Url: https://scripts.mit.edu/gitweb/autoinstallsdev/mediawiki.git/blobdiff_plain/19e297c21b10b1b8a3acad5e73fc71dcb35db44a..6932310fd58ebef145fa01eb76edf7150284d8ea:/includes/api/ApiFormatXml.php

diff --git a/includes/api/ApiFormatXml.php b/includes/api/ApiFormatXml.php
index 29d40d08..e4dfda0f 100644
--- a/includes/api/ApiFormatXml.php
+++ b/includes/api/ApiFormatXml.php
@@ -1,10 +1,10 @@
 <?php
 /**
- * API for MediaWiki 1.8+
+ *
  *
  * Created on Sep 19, 2006
  *
- * Copyright © 2006 Yuri Astrakhan <Firstname><Lastname>@gmail.com
+ * Copyright © 2006 Yuri Astrakhan "<Firstname><Lastname>@gmail.com"
  *
  * This program is free software; you can redistribute it and/or modify
  * it under the terms of the GNU General Public License as published by
@@ -24,11 +24,6 @@
  * @file
  */
 
-if ( !defined( 'MEDIAWIKI' ) ) {
-	// Eclipse helper - will be ignored in production
-	require_once( 'ApiFormatBase.php' );
-}
-
 /**
  * API XML output formatter
  * @ingroup API
@@ -36,178 +31,271 @@ if ( !defined( 'MEDIAWIKI' ) ) {
 class ApiFormatXml extends ApiFormatBase {
 
 	private $mRootElemName = 'api';
-	private $mDoubleQuote = false;
+	public static $namespace = 'http://www.mediawiki.org/xml/api/';
+	private $mIncludeNamespace = false;
 	private $mXslt = null;
 
-	public function __construct( $main, $format ) {
-		parent::__construct( $main, $format );
-	}
-
 	public function getMimeType() {
 		return 'text/xml';
 	}
 
-	public function getNeedsRawData() {
-		return true;
-	}
-
 	public function setRootElement( $rootElemName ) {
 		$this->mRootElemName = $rootElemName;
 	}
 
 	public function execute() {
 		$params = $this->extractRequestParams();
-		$this->mDoubleQuote = $params['xmldoublequote'];
+		$this->mIncludeNamespace = $params['includexmlnamespace'];
 		$this->mXslt = $params['xslt'];
 
 		$this->printText( '<?xml version="1.0"?>' );
 		if ( !is_null( $this->mXslt ) ) {
 			$this->addXslt();
 		}
+
+		$result = $this->getResult();
+		if ( $this->mIncludeNamespace && $result->getResultData( 'xmlns' ) === null ) {
+			// If the result data already contains an 'xmlns' namespace added
+			// for custom XML output types, it will override the one for the
+			// generic API results.
+			// This allows API output of other XML types like Atom, RSS, RSD.
+			$result->addValue( null, 'xmlns', self::$namespace, ApiResult::NO_SIZE_CHECK );
+		}
+		$data = $result->getResultData( null, [
+			'Custom' => function ( &$data, &$metadata ) {
+				if ( isset( $metadata[ApiResult::META_TYPE] ) ) {
+					// We want to use non-BC for BCassoc to force outputting of _idx.
+					switch ( $metadata[ApiResult::META_TYPE] ) {
+						case 'BCassoc':
+							$metadata[ApiResult::META_TYPE] = 'assoc';
+							break;
+					}
+				}
+			},
+			'BC' => [ 'nobool', 'no*', 'nosub' ],
+			'Types' => [ 'ArmorKVP' => '_name' ],
+		] );
+
 		$this->printText(
-			self::recXmlPrint( $this->mRootElemName,
-				$this->getResultData(),
-				$this->getIsHtml() ? - 2 : null,
-				$this->mDoubleQuote
+			static::recXmlPrint( $this->mRootElemName,
+				$data,
+				$this->getIsHtml() ? -2 : null
 			)
 		);
 	}
 
 	/**
 	 * This method takes an array and converts it to XML.
-	 * There are several noteworthy cases:
 	 *
-	 *  If array contains a key '_element', then the code assumes that ALL other keys are not important and replaces them with the value['_element'].
-	 *	Example:	name='root',  value = array( '_element'=>'page', 'x', 'y', 'z') creates <root>  <page>x</page>  <page>y</page>  <page>z</page> </root>
-	 *
-	 *  If any of the array's element key is '*', then the code treats all other key->value pairs as attributes, and the value['*'] as the element's content.
-	 *	Example:	name='root',  value = array( '*'=>'text', 'lang'=>'en', 'id'=>10)   creates  <root lang='en' id='10'>text</root>
-	 *
-	 * If neither key is found, all keys become element names, and values become element content.
-	 * The method is recursive, so the same rules apply to any sub-arrays.
+	 * @param string|null $name Tag name
+	 * @param mixed $value Tag value (attributes/content/subelements)
+	 * @param int|null $indent Indentation
+	 * @param array $attributes Additional attributes
+	 * @return string
 	 */
-	public static function recXmlPrint( $elemName, $elemValue, $indent, $doublequote = false ) {
+	public static function recXmlPrint( $name, $value, $indent, $attributes = [] ) {
 		$retval = '';
-		if ( !is_null( $indent ) ) {
-			$indent += 2;
+		if ( $indent !== null ) {
+			if ( $name !== null ) {
+				$indent += 2;
+			}
 			$indstr = "\n" . str_repeat( ' ', $indent );
 		} else {
 			$indstr = '';
 		}
-		$elemName = str_replace( ' ', '_', $elemName );
-
-		switch ( gettype( $elemValue ) ) {
-			case 'array':
-				if ( isset( $elemValue['*'] ) ) {
-					$subElemContent = $elemValue['*'];
-					if ( $doublequote ) {
-						$subElemContent = Sanitizer::encodeAttribute( $subElemContent );
-					}
-					unset( $elemValue['*'] );
 
-					// Add xml:space="preserve" to the
-					// element so XML parsers will leave
-					// whitespace in the content alone
-					$elemValue['xml:space'] = 'preserve';
-				} else {
-					$subElemContent = null;
-				}
+		if ( is_object( $value ) ) {
+			$value = (array)$value;
+		}
+		if ( is_array( $value ) ) {
+			$contentKey = isset( $value[ApiResult::META_CONTENT] )
+				? $value[ApiResult::META_CONTENT]
+				: '*';
+			$subelementKeys = isset( $value[ApiResult::META_SUBELEMENTS] )
+				? $value[ApiResult::META_SUBELEMENTS]
+				: [];
+			if ( isset( $value[ApiResult::META_BC_SUBELEMENTS] ) ) {
+				$subelementKeys = array_merge(
+					$subelementKeys, $value[ApiResult::META_BC_SUBELEMENTS]
+				);
+			}
+			$preserveKeys = isset( $value[ApiResult::META_PRESERVE_KEYS] )
+				? $value[ApiResult::META_PRESERVE_KEYS]
+				: [];
+			$indexedTagName = isset( $value[ApiResult::META_INDEXED_TAG_NAME] )
+				? self::mangleName( $value[ApiResult::META_INDEXED_TAG_NAME], $preserveKeys )
+				: '_v';
+			$bcBools = isset( $value[ApiResult::META_BC_BOOLS] )
+				? $value[ApiResult::META_BC_BOOLS]
+				: [];
+			$indexSubelements = isset( $value[ApiResult::META_TYPE] )
+				? $value[ApiResult::META_TYPE] !== 'array'
+				: false;
 
-				if ( isset( $elemValue['_element'] ) ) {
-					$subElemIndName = $elemValue['_element'];
-					unset( $elemValue['_element'] );
-				} else {
-					$subElemIndName = null;
+			$content = null;
+			$subelements = [];
+			$indexedSubelements = [];
+			foreach ( $value as $k => $v ) {
+				if ( ApiResult::isMetadataKey( $k ) && !in_array( $k, $preserveKeys, true ) ) {
+					continue;
 				}
 
-				$indElements = array();
-				$subElements = array();
-				foreach ( $elemValue as $subElemId => & $subElemValue ) {
-					if ( is_string( $subElemValue ) && $doublequote ) {
-						$subElemValue = Sanitizer::encodeAttribute( $subElemValue );
-					}
-
-					if ( gettype( $subElemId ) === 'integer' ) {
-						$indElements[] = $subElemValue;
-						unset( $elemValue[$subElemId] );
-					} elseif ( is_array( $subElemValue ) ) {
-						$subElements[$subElemId] = $subElemValue;
-						unset ( $elemValue[$subElemId] );
-					}
+				$oldv = $v;
+				if ( is_bool( $v ) && !in_array( $k, $bcBools, true ) ) {
+					$v = $v ? 'true' : 'false';
 				}
 
-				if ( is_null( $subElemIndName ) && count( $indElements ) ) {
-					ApiBase::dieDebug( __METHOD__, "($elemName, ...) has integer keys without _element value. Use ApiResult::setIndexedTagName()." );
+				if ( $name !== null && $k === $contentKey ) {
+					$content = $v;
+				} elseif ( is_int( $k ) ) {
+					$indexedSubelements[$k] = $v;
+				} elseif ( is_array( $v ) || is_object( $v ) ) {
+					$subelements[self::mangleName( $k, $preserveKeys )] = $v;
+				} elseif ( in_array( $k, $subelementKeys, true ) || $name === null ) {
+					$subelements[self::mangleName( $k, $preserveKeys )] = [
+						'content' => $v,
+						ApiResult::META_CONTENT => 'content',
+						ApiResult::META_TYPE => 'assoc',
+					];
+				} elseif ( is_bool( $oldv ) ) {
+					if ( $oldv ) {
+						$attributes[self::mangleName( $k, $preserveKeys )] = '';
+					}
+				} elseif ( $v !== null ) {
+					$attributes[self::mangleName( $k, $preserveKeys )] = $v;
 				}
+			}
 
-				if ( count( $subElements ) && count( $indElements ) && !is_null( $subElemContent ) ) {
-					ApiBase::dieDebug( __METHOD__, "($elemName, ...) has content and subelements" );
+			if ( $content !== null ) {
+				if ( $subelements || $indexedSubelements ) {
+					$subelements[self::mangleName( $contentKey, $preserveKeys )] = [
+						'content' => $content,
+						ApiResult::META_CONTENT => 'content',
+						ApiResult::META_TYPE => 'assoc',
+					];
+					$content = null;
+				} elseif ( is_scalar( $content ) ) {
+					// Add xml:space="preserve" to the element so XML parsers
+					// will leave whitespace in the content alone
+					$attributes += [ 'xml:space' => 'preserve' ];
 				}
+			}
 
-				if ( !is_null( $subElemContent ) ) {
-					$retval .= $indstr . Xml::element( $elemName, $elemValue, $subElemContent );
-				} elseif ( !count( $indElements ) && !count( $subElements ) ) {
-					$retval .= $indstr . Xml::element( $elemName, $elemValue );
+			if ( $content !== null ) {
+				if ( is_scalar( $content ) ) {
+					$retval .= $indstr . Xml::element( $name, $attributes, $content );
 				} else {
-					$retval .= $indstr . Xml::element( $elemName, $elemValue, null );
-
-					foreach ( $subElements as $subElemId => & $subElemValue ) {
-						$retval .= self::recXmlPrint( $subElemId, $subElemValue, $indent );
+					if ( $name !== null ) {
+						$retval .= $indstr . Xml::element( $name, $attributes, null );
 					}
-
-					foreach ( $indElements as &$subElemValue ) {
-						$retval .= self::recXmlPrint( $subElemIndName, $subElemValue, $indent );
+					$retval .= static::recXmlPrint( null, $content, $indent );
+					if ( $name !== null ) {
+						$retval .= $indstr . Xml::closeElement( $name );
 					}
-
-					$retval .= $indstr . Xml::closeElement( $elemName );
 				}
-				break;
-			case 'object':
-				// ignore
-				break;
-			default:
-				$retval .= $indstr . Xml::element( $elemName, null, $elemValue );
-				break;
+			} elseif ( !$indexedSubelements && !$subelements ) {
+				if ( $name !== null ) {
+					$retval .= $indstr . Xml::element( $name, $attributes );
+				}
+			} else {
+				if ( $name !== null ) {
+					$retval .= $indstr . Xml::element( $name, $attributes, null );
+				}
+				foreach ( $subelements as $k => $v ) {
+					$retval .= static::recXmlPrint( $k, $v, $indent );
+				}
+				foreach ( $indexedSubelements as $k => $v ) {
+					$retval .= static::recXmlPrint( $indexedTagName, $v, $indent,
+						$indexSubelements ? [ '_idx' => $k ] : []
+					);
+				}
+				if ( $name !== null ) {
+					$retval .= $indstr . Xml::closeElement( $name );
+				}
+			}
+		} else {
+			// to make sure null value doesn't produce unclosed element,
+			// which is what Xml::element( $name, null, null ) returns
+			if ( $value === null ) {
+				$retval .= $indstr . Xml::element( $name, $attributes );
+			} else {
+				$retval .= $indstr . Xml::element( $name, $attributes, $value );
+			}
 		}
+
 		return $retval;
 	}
 
-	function addXslt() {
+	/**
+	 * Mangle XML-invalid names to be valid in XML
+	 * @param string $name
+	 * @param array $preserveKeys Names to not mangle
+	 * @return string Mangled name
+	 */
+	private static function mangleName( $name, $preserveKeys = [] ) {
+		static $nsc = null, $nc = null;
+
+		if ( in_array( $name, $preserveKeys, true ) ) {
+			return $name;
+		}
+
+		if ( $name === '' ) {
+			return '_';
+		}
+
+		if ( $nsc === null ) {
+			// Note we omit ':' from $nsc and $nc because it's reserved for XML
+			// namespacing, and we omit '_' from $nsc (but not $nc) because we
+			// reserve it.
+			$nsc = 'A-Za-z\x{C0}-\x{D6}\x{D8}-\x{F6}\x{F8}-\x{2FF}\x{370}-\x{37D}\x{37F}-\x{1FFF}' .
+				'\x{200C}-\x{200D}\x{2070}-\x{218F}\x{2C00}-\x{2FEF}\x{3001}-\x{D7FF}' .
+				'\x{F900}-\x{FDCF}\x{FDF0}-\x{FFFD}\x{10000}-\x{EFFFF}';
+			$nc = $nsc . '_\-.0-9\x{B7}\x{300}-\x{36F}\x{203F}-\x{2040}';
+		}
+
+		if ( preg_match( "/^[$nsc][$nc]*$/uS", $name ) ) {
+			return $name;
+		}
+
+		return '_' . preg_replace_callback(
+			"/[^$nc]/uS",
+			function ( $m ) {
+				return sprintf( '.%X.', UtfNormal\Utils::utf8ToCodepoint( $m[0] ) );
+			},
+			str_replace( '.', '.2E.', $name )
+		);
+	}
+
+	protected function addXslt() {
 		$nt = Title::newFromText( $this->mXslt );
 		if ( is_null( $nt ) || !$nt->exists() ) {
-			$this->setWarning( 'Invalid or non-existent stylesheet specified' );
+			$this->addWarning( 'apiwarn-invalidxmlstylesheet' );
+
 			return;
 		}
 		if ( $nt->getNamespace() != NS_MEDIAWIKI ) {
-			$this->setWarning( 'Stylesheet should be in the MediaWiki namespace.' );
+			$this->addWarning( 'apiwarn-invalidxmlstylesheetns' );
+
 			return;
 		}
-		if ( substr( $nt->getText(), - 4 ) !== '.xsl' ) {
-			$this->setWarning( 'Stylesheet should have .xsl extension.' );
+		if ( substr( $nt->getText(), -4 ) !== '.xsl' ) {
+			$this->addWarning( 'apiwarn-invalidxmlstylesheetext' );
+
 			return;
 		}
-		$this->printText( '<?xml-stylesheet href="' . $nt->escapeLocalURL( 'action=raw' ) . '" type="text/xsl" ?>' );
+		$this->printText( '<?xml-stylesheet href="' .
+			htmlspecialchars( $nt->getLocalURL( 'action=raw' ) ) . '" type="text/xsl" ?>' );
 	}
 
 	public function getAllowedParams() {
-		return array(
-			'xmldoublequote' => false,
-			'xslt' => null,
-		);
-	}
-
-	public function getParamDescription() {
-		return array(
-			'xmldoublequote' => 'If specified, double quotes all attributes and content',
-			'xslt' => 'If specified, adds <xslt> as stylesheet',
-		);
-	}
-
-	public function getDescription() {
-		return 'Output data in XML format' . parent::getDescription();
-	}
-
-	public function getVersion() {
-		return __CLASS__ . ': $Id$';
+		return parent::getAllowedParams() + [
+			'xslt' => [
+				ApiBase::PARAM_HELP_MSG => 'apihelp-xml-param-xslt',
+			],
+			'includexmlnamespace' => [
+				ApiBase::PARAM_DFLT => false,
+				ApiBase::PARAM_HELP_MSG => 'apihelp-xml-param-includexmlnamespace',
+			],
+		];
 	}
 }