n3 2.12.2 → 2.13.1

package/lib/N3Lexer.js

@@ -135,6 +135,8 @@ class N3Lexer {
       for (const key in this) {
         if (!(key in lineModeRegExps) && this[key] instanceof RegExp) this[key] = invalidRegExp;
       }
+      // The only keyword in N-Triples and N-Quads is VERSION, which is case-sensitive
+      this._keyword = /VERSION(?=[\s#<])/y;
     }
     // When not in line mode, enable N3 functionality by default
     else {

package/lib/N3Parser.js

@@ -11,6 +11,10 @@ function _interopRequireDefault(e) { return e && e.__esModule ? e : { default: e
 // **N3Parser** parses N3 documents.
 
 let blankNodePrefix = 0;
+// Detects `.` and `..` path segments in an IRI
+const dotSegments = /(^|\/)\.\.?($|[/#?])/;
+// Detects `.` and `..` segments in the path of an IRI, ignoring its query and fragment
+const pathDotSegments = /^[^?#]*(?:^|\/)\.\.?(?:$|[/#?])/;
 
 // ## Constructor
 class N3Parser {
@@ -91,6 +95,7 @@ class N3Parser {
     if (!baseIRI) {
       this._base = '';
       this._basePath = '';
+      this._basePathHasDotSegments = false;
     } else {
       // Remove fragment if present
       const fragmentPos = baseIRI.indexOf('#');
@@ -107,6 +112,8 @@ class N3Parser {
       // If the base has an authority but an empty path,
       // relative IRIs merge under the path '/' (RFC 3986 §5.2.3)
       if (baseIRI[2] !== undefined && (base.length === this._baseRoot.length || base[this._baseRoot.length] === '?')) this._basePath = `${this._baseRoot}/`;
+      // Check once whether resolving against the base path needs to remove dot segments
+      this._basePathHasDotSegments = dotSegments.test(this._basePath);
     }
   }
 
@@ -153,7 +160,7 @@ class N3Parser {
     // Prefix and base declarations are scoped to their formula,
     // so record prefix changes to undo them when the formula ends
     if (type === 'formula') {
-      context.base = [this._base, this._basePath, this._baseRoot, this._baseScheme];
+      context.base = [this._base, this._basePath, this._baseRoot, this._baseScheme, this._basePathHasDotSegments];
       this._prefixChanges = [];
     }
     this._contextStack.push(context);
@@ -193,7 +200,7 @@ class N3Parser {
     if (this._n3Mode) {
       this._inversePredicate = context.inverse;
       this._expectOf = context.expectOf;
-      if (type === 'formula') [this._base, this._basePath, this._baseRoot, this._baseScheme] = context.base;
+      if (type === 'formula') [this._base, this._basePath, this._baseRoot, this._baseScheme, this._basePathHasDotSegments] = context.base;
       if (this._prefixChanges !== context.prefixChanges) {
         undoChanges(this._prefixes, this._prefixChanges);
         this._prefixChanges = context.prefixChanges;
@@ -1455,14 +1462,17 @@ class N3Parser {
       // Resolve all other IRIs at the base IRI's path
       default:
         // Relative IRIs cannot contain a colon in the first path segment
-        return /^[^/:]*:/.test(iri) ? null : this._removeDotSegments(this._basePath + iri);
+        if (/^[^/:]*:/.test(iri)) return null;
+        // Only scan the joined IRI for dot segments if either part can contain them,
+        // as the base path can be long and the joined IRI would need to be copied
+        return this._basePathHasDotSegments || pathDotSegments.test(iri) ? this._removeDotSegments(this._basePath + iri) : this._basePath + iri;
     }
   }
 
   // ### `_removeDotSegments` resolves './' and '../' path segments in an IRI as per RFC3986
   _removeDotSegments(iri) {
     // Don't modify the IRI if it does not contain any dot segments
-    if (!/(^|\/)\.\.?($|[/#?])/.test(iri)) return iri;
+    if (!dotSegments.test(iri)) return iri;
 
     // Start with an imaginary slash before the IRI in order to resolve trailing './' and '../'
     const length = iri.length;

package/lib/N3Writer.js

@@ -41,6 +41,9 @@ const escape = /["\\\t\n\r\b\f\u0000-\u0019\ud800-\udbff]/,
     '\f': '\\f'
   };
 
+// Characters that a version label cannot contain
+const invalidVersionLabel = /["\\\u0000-\u001f\u007f]|[\ud800-\udbff](?![\udc00-\udfff])|(?:^|[^\ud800-\udbff])[\udc00-\udfff]/;
+
 // A local name (PN_LOCAL) as written in a prefixed name, without backslash escapes,
 // so that it denotes the same characters as the IRI it is taken from.
 // Characters outside the Basic Multilingual Plane are escaped before matching, so they never occur.
@@ -88,12 +91,17 @@ class N3Writer {
       this._endStream = options.end === undefined ? true : !!options.end;
     }
 
+    // A version label is written as-is, so it cannot contain characters that need escaping,
+    // nor unpaired surrogates (which cannot be encoded)
+    if (options.version && invalidVersionLabel.test(options.version)) throw new Error(`Invalid version label: ${JSON.stringify(options.version)}`);
+
     // Initialize writer, depending on the format
     this._subject = null;
     if (!/triple|quad/i.test(options.format)) {
       this._lineMode = false;
       this._escape = escape, this._escapeAll = escapeAll, this._characterReplacer = characterReplacer;
       this._graph = DEFAULTGRAPH;
+      if (options.version) this._write(`@version "${options.version}".\n`);
       this._prefixIRIs = Object.create(null);
       // Escaped prefix IRIs and names for the prefix matcher, computed once per prefix
       this._prefixPatterns = Object.create(null);
@@ -108,6 +116,7 @@ class N3Writer {
       // N-Triples and N-Quads are written in their canonical form
       this._escape = canonicalEscape, this._escapeAll = canonicalEscapeAll;
       this._characterReplacer = canonicalCharacterReplacer;
+      if (options.version) this._write(`VERSION "${options.version}"\n`);
     }
   }
 
@@ -123,6 +132,15 @@ class N3Writer {
     this._outputStream.write(string, 'utf8', callback);
   }
 
+  // ### `_endStatement` finishes a pending statement and closes an open graph block
+  _endStatement() {
+    if (this._subject !== null) {
+      this._write(this._inDefaultGraph ? '.\n' : '\n}\n');
+      this._subject = null;
+      this._graph = DEFAULTGRAPH;
+    }
+  }
+
   // ### `_writeQuad` writes the quad to the output stream
   _writeQuad(subject, predicate, object, graph, done) {
     try {
@@ -325,10 +343,7 @@ class N3Writer {
       if (typeof iri !== 'string') iri = iri.value;
       hasPrefixes = true;
       // Finish a possible pending quad
-      if (this._subject !== null) {
-        this._write(this._inDefaultGraph ? '.\n' : '\n}\n');
-        this._subject = null, this._graph = '';
-      }
+      this._endStatement();
       // Store and write the prefix
       this._prefixIRIs[iri] = prefix += ':';
       this._prefixPatterns[iri] = [(0, _Util.escapeRegex)(iri), (0, _Util.escapeRegex)(prefix)];
@@ -407,10 +422,7 @@ class N3Writer {
   // ### `end` signals the end of the output stream
   end(done) {
     // Finish a possible pending quad
-    if (this._subject !== null) {
-      this._write(this._inDefaultGraph ? '.\n' : '\n}\n');
-      this._subject = null;
-    }
+    this._endStatement();
     // Disallow further writing
     this._write = this._blockedWrite;
 

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "n3",
-  "version": "2.12.2",
+  "version": "2.13.1",
   "description": "Lightning fast, asynchronous, streaming Turtle / N3 / RDF library.",
   "author": "Ruben Verborgh <ruben.verborgh@gmail.com>",
   "keywords": [

package/src/N3Lexer.js

@@ -101,6 +101,8 @@ export default class N3Lexer {
         if (!(key in lineModeRegExps) && this[key] instanceof RegExp)
           this[key] = invalidRegExp;
       }
+      // The only keyword in N-Triples and N-Quads is VERSION, which is case-sensitive
+      this._keyword = /VERSION(?=[\s#<])/y;
     }
     // When not in line mode, enable N3 functionality by default
     else {

package/src/N3Parser.js

@@ -4,6 +4,10 @@ import N3DataFactory from './N3DataFactory';
 import namespaces from './IRIs';
 
 let blankNodePrefix = 0;
+// Detects `.` and `..` path segments in an IRI
+const dotSegments = /(^|\/)\.\.?($|[/#?])/;
+// Detects `.` and `..` segments in the path of an IRI, ignoring its query and fragment
+const pathDotSegments = /^[^?#]*(?:^|\/)\.\.?(?:$|[/#?])/;
 
 // ## Constructor
 export default class N3Parser {
@@ -83,6 +87,7 @@ export default class N3Parser {
     if (!baseIRI) {
       this._base = '';
       this._basePath = '';
+      this._basePathHasDotSegments = false;
     }
     else {
       // Remove fragment if present
@@ -103,6 +108,8 @@ export default class N3Parser {
       // relative IRIs merge under the path '/' (RFC 3986 §5.2.3)
       if (baseIRI[2] !== undefined && (base.length === this._baseRoot.length || base[this._baseRoot.length] === '?'))
         this._basePath = `${this._baseRoot}/`;
+      // Check once whether resolving against the base path needs to remove dot segments
+      this._basePathHasDotSegments = dotSegments.test(this._basePath);
     }
   }
 
@@ -140,7 +147,8 @@ export default class N3Parser {
     // Prefix and base declarations are scoped to their formula,
     // so record prefix changes to undo them when the formula ends
     if (type === 'formula') {
-      context.base = [this._base, this._basePath, this._baseRoot, this._baseScheme];
+      context.base = [this._base, this._basePath, this._baseRoot, this._baseScheme,
+        this._basePathHasDotSegments];
       this._prefixChanges = [];
     }
     this._contextStack.push(context);
@@ -182,7 +190,8 @@ export default class N3Parser {
       this._inversePredicate = context.inverse;
       this._expectOf = context.expectOf;
       if (type === 'formula')
-        [this._base, this._basePath, this._baseRoot, this._baseScheme] = context.base;
+        [this._base, this._basePath, this._baseRoot, this._baseScheme,
+          this._basePathHasDotSegments] = context.base;
       if (this._prefixChanges !== context.prefixChanges) {
         undoChanges(this._prefixes, this._prefixChanges);
         this._prefixChanges = context.prefixChanges;
@@ -1576,14 +1585,19 @@ export default class N3Parser {
     // Resolve all other IRIs at the base IRI's path
     default:
       // Relative IRIs cannot contain a colon in the first path segment
-      return (/^[^/:]*:/.test(iri)) ? null : this._removeDotSegments(this._basePath + iri);
+      if (/^[^/:]*:/.test(iri))
+        return null;
+      // Only scan the joined IRI for dot segments if either part can contain them,
+      // as the base path can be long and the joined IRI would need to be copied
+      return this._basePathHasDotSegments || pathDotSegments.test(iri) ?
+        this._removeDotSegments(this._basePath + iri) : this._basePath + iri;
     }
   }
 
   // ### `_removeDotSegments` resolves './' and '../' path segments in an IRI as per RFC3986
   _removeDotSegments(iri) {
     // Don't modify the IRI if it does not contain any dot segments
-    if (!/(^|\/)\.\.?($|[/#?])/.test(iri))
+    if (!dotSegments.test(iri))
       return iri;
 
     // Start with an imaginary slash before the IRI in order to resolve trailing './' and '../'

package/src/N3Writer.js

@@ -24,6 +24,9 @@ const escape    = /["\\\t\n\r\b\f\u0000-\u0019\ud800-\udbff]/,
       '\n': '\\n', '\r': '\\r', '\b': '\\b', '\f': '\\f',
     };
 
+// Characters that a version label cannot contain
+const invalidVersionLabel = /["\\\u0000-\u001f\u007f]|[\ud800-\udbff](?![\udc00-\udfff])|(?:^|[^\ud800-\udbff])[\udc00-\udfff]/;
+
 // A local name (PN_LOCAL) as written in a prefixed name, without backslash escapes,
 // so that it denotes the same characters as the IRI it is taken from.
 // Characters outside the Basic Multilingual Plane are escaped before matching, so they never occur.
@@ -70,12 +73,19 @@ export default class N3Writer {
       this._endStream = options.end === undefined ? true : !!options.end;
     }
 
+    // A version label is written as-is, so it cannot contain characters that need escaping,
+    // nor unpaired surrogates (which cannot be encoded)
+    if (options.version && invalidVersionLabel.test(options.version))
+      throw new Error(`Invalid version label: ${JSON.stringify(options.version)}`);
+
     // Initialize writer, depending on the format
     this._subject = null;
     if (!(/triple|quad/i).test(options.format)) {
       this._lineMode = false;
       this._escape = escape, this._escapeAll = escapeAll, this._characterReplacer = characterReplacer;
       this._graph = DEFAULTGRAPH;
+      if (options.version)
+        this._write(`@version "${options.version}".\n`);
       this._prefixIRIs = Object.create(null);
       // Escaped prefix IRIs and names for the prefix matcher, computed once per prefix
       this._prefixPatterns = Object.create(null);
@@ -92,6 +102,8 @@ export default class N3Writer {
       // N-Triples and N-Quads are written in their canonical form
       this._escape = canonicalEscape, this._escapeAll = canonicalEscapeAll;
       this._characterReplacer = canonicalCharacterReplacer;
+      if (options.version)
+        this._write(`VERSION "${options.version}"\n`);
     }
   }
 
@@ -107,6 +119,15 @@ export default class N3Writer {
     this._outputStream.write(string, 'utf8', callback);
   }
 
+  // ### `_endStatement` finishes a pending statement and closes an open graph block
+  _endStatement() {
+    if (this._subject !== null) {
+      this._write(this._inDefaultGraph ? '.\n' : '\n}\n');
+      this._subject = null;
+      this._graph = DEFAULTGRAPH;
+    }
+  }
+
   // ### `_writeQuad` writes the quad to the output stream
   _writeQuad(subject, predicate, object, graph, done) {
     try {
@@ -340,10 +361,7 @@ export default class N3Writer {
         iri = iri.value;
       hasPrefixes = true;
       // Finish a possible pending quad
-      if (this._subject !== null) {
-        this._write(this._inDefaultGraph ? '.\n' : '\n}\n');
-        this._subject = null, this._graph = '';
-      }
+      this._endStatement();
       // Store and write the prefix
       this._prefixIRIs[iri] = (prefix += ':');
       this._prefixPatterns[iri] = [escapeRegex(iri), escapeRegex(prefix)];
@@ -426,10 +444,7 @@ export default class N3Writer {
   // ### `end` signals the end of the output stream
   end(done) {
     // Finish a possible pending quad
-    if (this._subject !== null) {
-      this._write(this._inDefaultGraph ? '.\n' : '\n}\n');
-      this._subject = null;
-    }
+    this._endStatement();
     // Disallow further writing
     this._write = this._blockedWrite;