n3 2.11.3 → 2.11.4

package/lib/N3Lexer.js

@@ -89,6 +89,11 @@ function execAtEnd(regExp, input, pos) {
   return regExp.exec(`${input.slice(pos)} `);
 }
 
+// Whitespace or the start of a comment
+function isSeparatorCode(code) {
+  return code === SPACE || code === TAB || code === LF || code === CR || code === HASH;
+}
+
 // ## Constructor
 class N3Lexer {
   constructor(options) {
@@ -194,6 +199,8 @@ class N3Lexer {
         }
       }
       if (pos >= input.length) {
+        // A datatype marker needs a type
+        if (inputFinished && this._previousMarker === '^^') return reportSyntaxError(this, input, pos);
         this._linePosition = currentLineLength;
         if (inputFinished) {
           emitToken('eof', '', '', this._line, currentLineLength, 0);
@@ -215,6 +222,8 @@ class N3Lexer {
         inconclusive = false;
       switch (firstChar) {
         case '^':
+          // A datatype marker separated from its type cannot be followed by another marker
+          if (this._previousMarker === '^^') return reportSyntaxError(this, input, pos);
           // We need at least 3 tokens lookahead to distinguish ^^<IRI> and ^^pre:fixed
           if (input.length - pos < 3) break;
           // Try to match a type
@@ -223,6 +232,8 @@ class N3Lexer {
             // Move to type IRI or prefixed name
             pos += 2;
             if (input[pos] !== '<') {
+              // Whitespace and comments may separate the marker from the type
+              if (isSeparatorCode(input.charCodeAt(pos))) continue; // eslint-disable-line no-continue
               inconclusive = true;
               break;
             }

package/lib/N3Writer.js

@@ -19,9 +19,15 @@ const {
   xsd
 } = _IRIs.default;
 
-// Characters in literals that require escaping
+// Characters that require escaping in Turtle, TriG, and N3,
+// where characters outside the Basic Multilingual Plane are escaped as well
 const escape = /["\\\t\n\r\b\f\u0000-\u0019\ud800-\udbff]/,
   escapeAll = /["\\\t\n\r\b\f\u0000-\u0019]|[\ud800-\udbff][\udc00-\udfff]/g,
+  // Characters that require escaping in canonical N-Triples and N-Quads:
+  // U+0000–U+001F, `"`, `\`, U+007F, U+FFFE, and U+FFFF.
+  // A negated class is as fast to test as a shorter class listing characters.
+  canonicalEscape = /[^ !#-\[\]-~\u0080-\uFFFD]/,
+  canonicalEscapeAll = /[^ !#-\[\]-~\u0080-\uFFFD]/g,
   escapedCharacters = {
     '\\': '\\\\',
     '"': '\\"',
@@ -75,6 +81,7 @@ class N3Writer {
     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;
       this._prefixIRIs = Object.create(null);
       // Escaped prefix IRIs and names for the prefix matcher, computed once per prefix
@@ -87,6 +94,9 @@ class N3Writer {
     } else {
       this._lineMode = true;
       this._writeQuad = this._writeQuadLine;
+      // N-Triples and N-Quads are written in their canonical form
+      this._escape = canonicalEscape, this._escapeAll = canonicalEscapeAll;
+      this._characterReplacer = canonicalCharacterReplacer;
     }
   }
 
@@ -176,7 +186,7 @@ class N3Writer {
       iri = this._baseIri.toRelative(iri);
     }
     // Escape special characters
-    if (escape.test(iri)) iri = iri.replace(escapeAll, characterReplacer);
+    if (this._escape.test(iri)) iri = iri.replace(this._escapeAll, this._characterReplacer);
     // Try to represent the IRI as prefixed name, unless no prefixes were added
     const prefixMatch = this._hasPrefixes ? this._prefixRegex.exec(iri) : null;
     return !prefixMatch ? `<${iri}>` : !prefixMatch[1] ? iri : this._prefixIRIs[prefixMatch[1]] + prefixMatch[2];
@@ -186,7 +196,7 @@ class N3Writer {
   _encodeLiteral(literal) {
     // Escape special characters
     let value = literal.value;
-    if (escape.test(value)) value = value.replace(escapeAll, characterReplacer);
+    if (this._escape.test(value)) value = value.replace(this._escapeAll, this._characterReplacer);
 
     // Write a language-tagged literal
     const language = literal.language;
@@ -256,7 +266,9 @@ class N3Writer {
     object,
     graph
   }) {
-    return `<<(${this._encodeSubject(subject)} ${this._encodePredicate(predicate)} ${this._encodeObject(object)}${(0, _N3Util.isDefaultGraph)(graph) ? '' : ` ${this._encodeIriOrBlank(graph)}`})>>`;
+    // Canonical N-Triples and N-Quads put spaces inside the delimiters
+    const space = this._lineMode ? ' ' : '';
+    return `<<(${space}${this._encodeSubject(subject)} ${this._encodePredicate(predicate)} ${this._encodeObject(object)}${(0, _N3Util.isDefaultGraph)(graph) ? '' : ` ${this._encodeIriOrBlank(graph)}`}${space})>>`;
   }
 
   // ### `_blockedWrite` replaces `_write` after the writer has been closed
@@ -412,4 +424,15 @@ function characterReplacer(character) {
     }
   }
   return result;
+}
+
+// Replaces a character by its canonical N-Triples escape:
+// a short escape where one exists, and an upper-case `\uXXXX` escape otherwise
+function canonicalCharacterReplacer(character) {
+  let result = escapedCharacters[character];
+  if (result === undefined) {
+    result = character.charCodeAt(0).toString(16).toUpperCase();
+    result = '\\u0000'.substr(0, 6 - result.length) + result;
+  }
+  return result;
 }

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "n3",
-  "version": "2.11.3",
+  "version": "2.11.4",
   "description": "Lightning fast, asynchronous, streaming Turtle / N3 / RDF library.",
   "author": "Ruben Verborgh <ruben.verborgh@gmail.com>",
   "keywords": [

package/src/N3Lexer.js

@@ -54,6 +54,11 @@ function execAtEnd(regExp, input, pos) {
   return regExp.exec(`${input.slice(pos)} `);
 }
 
+// Whitespace or the start of a comment
+function isSeparatorCode(code) {
+  return code === SPACE || code === TAB || code === LF || code === CR || code === HASH;
+}
+
 // ## Constructor
 export default class N3Lexer {
   constructor(options) {
@@ -170,6 +175,9 @@ export default class N3Lexer {
         }
       }
       if (pos >= input.length) {
+        // A datatype marker needs a type
+        if (inputFinished && this._previousMarker === '^^')
+          return reportSyntaxError(this, input, pos);
         this._linePosition = currentLineLength;
         if (inputFinished) {
           emitToken('eof', '', '', this._line, currentLineLength, 0);
@@ -185,6 +193,9 @@ export default class N3Lexer {
           finalLineLength = 0, inconclusive = false;
       switch (firstChar) {
       case '^':
+        // A datatype marker separated from its type cannot be followed by another marker
+        if (this._previousMarker === '^^')
+          return reportSyntaxError(this, input, pos);
         // We need at least 3 tokens lookahead to distinguish ^^<IRI> and ^^pre:fixed
         if (input.length - pos < 3)
           break;
@@ -194,6 +205,9 @@ export default class N3Lexer {
           // Move to type IRI or prefixed name
           pos += 2;
           if (input[pos] !== '<') {
+            // Whitespace and comments may separate the marker from the type
+            if (isSeparatorCode(input.charCodeAt(pos)))
+              continue; // eslint-disable-line no-continue
             inconclusive = true;
             break;
           }

package/src/N3Writer.js

@@ -9,9 +9,15 @@ const DEFAULTGRAPH = N3DataFactory.defaultGraph();
 
 const { rdf, xsd } = namespaces;
 
-// Characters in literals that require escaping
+// Characters that require escaping in Turtle, TriG, and N3,
+// where characters outside the Basic Multilingual Plane are escaped as well
 const escape    = /["\\\t\n\r\b\f\u0000-\u0019\ud800-\udbff]/,
     escapeAll = /["\\\t\n\r\b\f\u0000-\u0019]|[\ud800-\udbff][\udc00-\udfff]/g,
+    // Characters that require escaping in canonical N-Triples and N-Quads:
+    // U+0000–U+001F, `"`, `\`, U+007F, U+FFFE, and U+FFFF.
+    // A negated class is as fast to test as a shorter class listing characters.
+    canonicalEscape    = /[^ !#-\[\]-~\u0080-\uFFFD]/,
+    canonicalEscapeAll = /[^ !#-\[\]-~\u0080-\uFFFD]/g,
     escapedCharacters = {
       '\\': '\\\\', '"': '\\"', '\t': '\\t',
       '\n': '\\n', '\r': '\\r', '\b': '\\b', '\f': '\\f',
@@ -57,6 +63,7 @@ export default class N3Writer {
     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;
       this._prefixIRIs = Object.create(null);
       // Escaped prefix IRIs and names for the prefix matcher, computed once per prefix
@@ -71,6 +78,9 @@ export default class N3Writer {
     else {
       this._lineMode = true;
       this._writeQuad = this._writeQuadLine;
+      // N-Triples and N-Quads are written in their canonical form
+      this._escape = canonicalEscape, this._escapeAll = canonicalEscapeAll;
+      this._characterReplacer = canonicalCharacterReplacer;
     }
   }
 
@@ -176,8 +186,8 @@ export default class N3Writer {
       iri = this._baseIri.toRelative(iri);
     }
     // Escape special characters
-    if (escape.test(iri))
-      iri = iri.replace(escapeAll, characterReplacer);
+    if (this._escape.test(iri))
+      iri = iri.replace(this._escapeAll, this._characterReplacer);
     // Try to represent the IRI as prefixed name, unless no prefixes were added
     const prefixMatch = this._hasPrefixes ? this._prefixRegex.exec(iri) : null;
     return !prefixMatch ? `<${iri}>` :
@@ -188,8 +198,8 @@ export default class N3Writer {
   _encodeLiteral(literal) {
     // Escape special characters
     let value = literal.value;
-    if (escape.test(value))
-      value = value.replace(escapeAll, characterReplacer);
+    if (this._escape.test(value))
+      value = value.replace(this._escapeAll, this._characterReplacer);
 
     // Write a language-tagged literal
     const language = literal.language;
@@ -260,11 +270,13 @@ export default class N3Writer {
 
   // ### `_encodeQuad` encodes an RDF-star quad
   _encodeQuad({ subject, predicate, object, graph }) {
-    return `<<(${
+    // Canonical N-Triples and N-Quads put spaces inside the delimiters
+    const space = this._lineMode ? ' ' : '';
+    return `<<(${space}${
       this._encodeSubject(subject)} ${
       this._encodePredicate(predicate)} ${
       this._encodeObject(object)}${
-      isDefaultGraph(graph) ? '' : ` ${this._encodeIriOrBlank(graph)}`})>>`;
+      isDefaultGraph(graph) ? '' : ` ${this._encodeIriOrBlank(graph)}`}${space})>>`;
   }
 
   // ### `_blockedWrite` replaces `_write` after the writer has been closed
@@ -428,3 +440,14 @@ function characterReplacer(character) {
   }
   return result;
 }
+
+// Replaces a character by its canonical N-Triples escape:
+// a short escape where one exists, and an upper-case `\uXXXX` escape otherwise
+function canonicalCharacterReplacer(character) {
+  let result = escapedCharacters[character];
+  if (result === undefined) {
+    result = character.charCodeAt(0).toString(16).toUpperCase();
+    result = '\\u0000'.substr(0, 6 - result.length) + result;
+  }
+  return result;
+}