Validate language tags

rubensworks · Mar 16, 2020 · 318e3e6 · 318e3e6
1 parent 8bc5c15
commit 318e3e6
Show file tree

Hide file tree

Showing 4 changed files with 30 additions and 0 deletions.
diff --git a/lib/JsonLdParser.ts b/lib/JsonLdParser.ts
@@ -440,6 +440,14 @@ export interface IJsonLdParserOptions {
    * this must be explicitly set to true.
    */
   validateValueIndexes?: boolean;
+  /**
+   * If values should be strictly checked.
+   * If true, an error will be thrown on invalid value ranged.
+   * if false, the resulting triple/quad will be omitted.
+   *
+   * Defaults to false.
+   */
+  strictRanges?: boolean;
   /**
    * The graph to use as default graph when no explicit @graph is set.
    * Defaults to dataFactory.defaultGraph().

diff --git a/lib/ParsingContext.ts b/lib/ParsingContext.ts
@@ -28,6 +28,7 @@ export class ParsingContext {
   public readonly processingMode: string;
   public readonly errorOnInvalidProperties: boolean;
   public readonly validateValueIndexes: boolean;
+  public readonly strictRanges: boolean;
   public readonly rootContext: Promise<IJsonLdContextNormalized>;
   public readonly defaultGraph?: RDF.Term;
 
@@ -73,6 +74,7 @@ export class ParsingContext {
     this.processingMode = options.processingMode || JsonLdParser.DEFAULT_PROCESSING_MODE;
     this.errorOnInvalidProperties = options.errorOnInvalidIris;
     this.validateValueIndexes = options.validateValueIndexes;
+    this.strictRanges = options.strictRanges;
     this.defaultGraph = options.defaultGraph;
 
     // Initialize stacks

diff --git a/lib/Util.ts b/lib/Util.ts
@@ -12,6 +12,7 @@ export class Util {
   public static readonly XSD_INTEGER: string = Util.XSD + 'integer';
   public static readonly XSD_DOUBLE: string = Util.XSD + 'double';
   public static readonly RDF: string = 'http://www.w3.org/1999/02/22-rdf-syntax-ns#';
+  public static readonly REGEX_LANGUAGE_TAG: RegExp = /^[a-zA-Z]+(-[a-zA-Z0-9]+)*$/;
 
   public readonly dataFactory: RDF.DataFactory;
   public readonly rdfFirst: RDF.NamedNode;
@@ -239,6 +240,14 @@ export class Util {
           // Language tags are always normalized to lowercase.
           valueLanguage = valueLanguage.toLowerCase();
 
+          if (!Util.REGEX_LANGUAGE_TAG.test(valueLanguage)) {
+            if (this.parsingContext.strictRanges) {
+              throw new Error(`The value of an '@language' must be a valid language tag, got '${
+                JSON.stringify(valueLanguage)}'`);
+            } else {
+              return null;
+            }
+          }
           return this.dataFactory.literal(val, valueLanguage);
         } else if (valueType) { // Validate @type
           if (typeof valueType !== 'string') {

diff --git a/test/Util-test.ts b/test/Util-test.ts
@@ -303,6 +303,17 @@ describe('Util', () => {
               new Error('When an \'@language\' is set, the value of \'@value\' must be a string, got \'true\''));
         });
 
+        it('with a @value and invalid @language should return null', async () => {
+          return expect(util.valueToTerm(context, 'key', { '@value': 'abc', '@language': 'en us' }, 0))
+            .resolves.toBeNull();
+        });
+
+        it('with a @value and invalid @language should throw an error when strictRanges is true', async () => {
+          util.parsingContext.strictRanges = true;
+          return expect(util.valueToTerm(context, 'key', { '@value': 'abc', '@language': 'en us' }, 0))
+            .rejects.toThrow(new Error('The value of an \'@language\' must be a valid language tag, got \'"en us"\''));
+        });
+
         it('with a @value and boolean @type should throw an error', async () => {
           return expect(util.valueToTerm(context, 'key', { '@value': 'abc', '@type': true }, 0))
             .rejects.toThrow(new Error('The value of an \'@type\' must be a string, got \'true\''));