Merge pull request #1568 from SemanticMediaWiki/invalid

Remove/strip invalid characters/tags from property name, refs #1567

i18n/en.json

@@ -341,6 +341,7 @@
 	"smw-pa-property-predefined_errp": "\"$1\" is a predefined property to track input errors for irregular value annotations that was likely caused by type or [[Property:Allows value|allowed value]] restrictions and is provided by [https://www.semantic-mediawiki.org/wiki/Help:Special_properties Semantic MediaWiki].",
 	"smw-pa-property-predefined_pval": "[https://www.semantic-mediawiki.org/wiki/Help:Special_property_Allows_value \"$1\"] is a predefined property that can define a list of permissible values to restrict value assignments for a property and is provided by [https://www.semantic-mediawiki.org/wiki/Help:Special_properties Semantic MediaWiki].",
 	"smw-datavalue-property-restricted-use": "Property \"$1\" has been marked for restricted use.",
+	"smw-datavalue-property-invalid-name": "Property name contains invalid characters (e.g. $1) .",
 	"smw-datavalue-restricted-use": "Datavalue \"$1\" has been marked for restricted use.",
 	"smw-datavalue-invalid-number": "\"$1\" can not be interpreted as a number.",
 	"smw-query-condition-circular": "A possible circular condition has been detected in \"$1\".",

includes/datavalues/SMW_DV_Property.php

@@ -133,17 +133,13 @@ protected function parseUserValue( $value ) {
 		if ( $this->m_caption === false ) { // always use this as caption
 			$this->m_caption = $value;
 		}
-		$propertyName = smwfNormalTitleText( ltrim( rtrim( $value, ' ]' ), ' [' ) ); // slightly normalise label
-		$inverse = false;
-		if ( ( $propertyName !== '' ) && ( $propertyName { 0 } == '-' ) ) { // property refers to an inverse
-			$propertyName = smwfNormalTitleText( (string)substr( $value, 1 ) );
-			/// NOTE The cast is necessary at least in PHP 5.3.3 to get string '' instead of boolean false.
-			/// NOTE It is necessary to normalize again here, since normalization may uppercase the first letter.
-			$inverse = true;
-		}
+
+		list( $propertyName, $inverse ) = $this->doNormalizeUserValue(
+			$value
+		);
 
 		try {
-			$this->m_dataitem = DIProperty::newFromUserLabel( $propertyName, $inverse, $this->m_typeid );
+			$this->m_dataitem = DIProperty::newFromUserLabel( $propertyName, $inverse );
 		} catch ( SMWDataItemException $e ) { // happens, e.g., when trying to sort queries by property "-"
 			$this->addError( wfMessage( 'smw_noproperty', $value )->inContentLanguage()->text() );
 			$this->m_dataitem = new DIProperty( 'ERROR', false ); // just to have something
@@ -444,4 +440,28 @@ public function getText() {
 		return $this->m_dataitem->getLabel();
 	}
 
+	private function doNormalizeUserValue( $value ) {
+
+		// #1567
+		if ( strpos( $value, '#' ) !== false ) {
+			$this->addErrorMsg( array( 'smw-datavalue-property-invalid-name', '#' ) );
+			$this->m_dataitem = new DIProperty( 'ERROR', false );
+		}
+
+		$value = strip_tags( $value );
+		$inverse = false;
+
+		// slightly normalise label
+		$propertyName = smwfNormalTitleText( ltrim( rtrim( $value, ' ]' ), ' [' ) );
+
+		if ( ( $propertyName !== '' ) && ( $propertyName { 0 } == '-' ) ) { // property refers to an inverse
+			$propertyName = smwfNormalTitleText( (string)substr( $value, 1 ) );
+			/// NOTE The cast is necessary at least in PHP 5.3.3 to get string '' instead of boolean false.
+			/// NOTE It is necessary to normalize again here, since normalization may uppercase the first letter.
+			$inverse = true;
+		}
+
+		return array( $propertyName, $inverse );
+	}
+
 }

tests/phpunit/Integration/ByJsonScript/Fixtures/p-0102.json

@@ -0,0 +1,69 @@
+{
+	"description": "Test in-text annotation on properties with invalid names/charaters (#1567, `wgContLang=en`)",
+	"properties": [
+		{
+			"name": "Has #",
+			"contents": "[[Has type::Date]]"
+		}
+	],
+	"subjects": [
+		{
+			"name": "Example/P0102/1",
+			"contents": "[[| ]]</td><::123]]"
+		},
+		{
+			"name": "Example/P0102/2",
+			"contents": "[[<code>Has property a</code>::ABC]] {{#set: <code>Has property b</code>=DEF }}"
+		},
+		{
+			"name": "Example/P0102/3",
+			"contents": "[[<code>Has property e::ABC]] {{#set: <code>Has property f=DEF }}"
+		}
+	],
+	"parser-testcases": [
+		{
+			"about": "#0 do not allow #",
+			"subject": "Example/P0102/1",
+			"store": {
+				"semantic-data": {
+					"strict-mode-valuematch": false,
+					"propertyCount": 3,
+					"propertyKeys": [ "_ERRC", "_MDAT", "_SKEY" ]
+				}
+			}
+		},
+		{
+			"about": "#1 strip tags",
+			"subject": "Example/P0102/2",
+			"store": {
+				"semantic-data": {
+					"strict-mode-valuematch": false,
+					"propertyCount": 4,
+					"propertyKeys": [ "Has_property_a", "Has_property_b", "_MDAT", "_SKEY" ],
+					"propertyValues": [ "ABC", "DEF" ]
+				}
+			}
+		},
+		{
+			"about": "#2 remove broken tags",
+			"subject": "Example/P0102/3",
+			"store": {
+				"semantic-data": {
+					"strict-mode-valuematch": false,
+					"propertyCount": 4,
+					"propertyKeys": [ "Has_property_e", "Has_property_f", "_MDAT", "_SKEY" ],
+					"propertyValues": [ "ABC", "DEF" ]
+				}
+			}
+		}
+	],
+	"settings": {
+		"wgContLang": "en",
+		"smwgPageSpecialProperties": [ "_MDAT" ]
+	},
+	"meta": {
+		"version": "0.1",
+		"is-incomplete": false,
+		"debug": false
+	}
+}