n3 2.12.1 → 2.13.0

package/lib/N3Lexer.js

@@ -116,8 +116,8 @@ class N3Lexer {
     this._blank = /_:((?:[0-9A-Z_a-z\xc0-\xd6\xd8-\xf6\xf8-\u02ff\u0370-\u037d\u037f-\u1fff\u200c\u200d\u2070-\u218f\u2c00-\u2fef\u3001-\ud7ff\uf900-\ufdcf\ufdf0-\ufffd]|[\ud800-\udb7f][\udc00-\udfff])(?:\.?[\-0-9A-Z_a-z\xb7\xc0-\xd6\xd8-\xf6\xf8-\u037d\u037f-\u1fff\u200c\u200d\u203f\u2040\u2070-\u218f\u2c00-\u2fef\u3001-\ud7ff\uf900-\ufdcf\ufdf0-\ufffd]|[\ud800-\udb7f][\udc00-\udfff])*)(?:[ \t]+|(?=\.?[,;:!\^\s#()\[\]\{\}"'<>]))/y;
     this._number = /[\-+]?(?:(\d+\.\d*|\.?\d+)[eE][\-+]?\d+|(?=\.?\d)\d*(?:(\.)\d+)?)(?=\.?[,;:!\^\s#()\[\]\{\}"'<>])/y;
     this._boolean = /(?:true|false)(?=[.,;!\^\s#()\[\]\{\}"'<>])/y;
-    this._atKeyword = /@[a-z]+(?=[\s#<:])/iy;
-    this._keyword = /(?:PREFIX|BASE|VERSION|GRAPH)(?=[\s#<])/iy;
+    this._atKeyword = /@[a-z]+(?=[\s#<:"'])/iy;
+    this._keyword = /(?:PREFIX|BASE|VERSION|GRAPH)(?=[\s#<"'])/iy;
     this._n3Verb = /(?:has|is|of)(?=[\s#()\[\]\{\}"'<>?_+\-0-9])/y;
     this._n3Id = /id(?=[\s#<])/y;
     this._shortPredicates = /a(?=[\s#()\[\]\{\}"'<>])/y;
@@ -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 {
@@ -236,7 +238,8 @@ class N3Lexer {
         matchLength = 0,
         lexicalLength = 0,
         finalLineLength = 0,
-        inconclusive = false;
+        inconclusive = false,
+        tripleQuoted = false;
       switch (firstChar) {
         case '^':
           // A datatype marker separated from its type cannot be followed by another marker
@@ -310,7 +313,8 @@ class N3Lexer {
             ({
               value,
               matchLength,
-              finalLineLength
+              finalLineLength,
+              tripleQuoted
             } = this._parseLiteral(input, pos));
             if (value === null) return reportSyntaxError(this, input, pos);
           }
@@ -328,7 +332,8 @@ class N3Lexer {
               ({
                 value,
                 matchLength,
-                finalLineLength
+                finalLineLength,
+                tripleQuoted
               } = this._parseLiteral(input, pos));
               if (value === null) return reportSyntaxError(this, input, pos);
             }
@@ -521,7 +526,21 @@ class N3Lexer {
           line,
           start,
           end: finalLineLength,
-          endLine: this._line
+          endLine: this._line,
+          tripleQuoted
+        };
+        callback(null, token);
+      }
+      // Triple-quoted strings are marked, since version declarations do not allow them
+      else if (tripleQuoted) {
+        token = {
+          type,
+          value,
+          prefix,
+          line,
+          start,
+          end: start + length,
+          tripleQuoted
         };
         callback(null, token);
       } else token = emitToken(type, value, prefix, line, start, lexicalLength || length);
@@ -671,7 +690,8 @@ class N3Lexer {
           return {
             value: this._unescape(raw, stringEscapeReplacements),
             matchLength,
-            finalLineLength
+            finalLineLength,
+            tripleQuoted: openingLength === 3
           };
         }
         closingPos++;
@@ -681,7 +701,8 @@ class N3Lexer {
     return {
       value: '',
       matchLength: 0,
-      finalLineLength: 0
+      finalLineLength: 0,
+      tripleQuoted: false
     };
   }
 

package/lib/N3Parser.js

@@ -1025,7 +1025,8 @@ class N3Parser {
   // ### `_readVersion` reads version string declaration
   _readVersion(token) {
     if (token.type !== 'literal') return this._error('Expected literal to follow version declaration', token);
-    if (token.end - token.start !== token.value.length + 2) return this._error('Version declarations must use single quotes', token);
+    // Only short strings are allowed, so no numbers or booleans (which have a datatype prefix)
+    if (token.prefix !== '' || token.tripleQuoted) return this._error('Version declarations must use single quotes', token);
     this._versionCallback(token.value);
     if (!this._isValidVersion(token.value)) return this._error(`Detected unsupported version: "${token.value}"`, token);
     return this._readDeclarationPunctuation;

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": "2.12.1",
+  "version": "2.13.0",
   "description": "Lightning fast, asynchronous, streaming Turtle / N3 / RDF library.",
   "author": "Ruben Verborgh <ruben.verborgh@gmail.com>",
   "keywords": [

package/src/N3Lexer.js

@@ -81,8 +81,8 @@ export default class N3Lexer {
     this._blank = /_:((?:[0-9A-Z_a-z\xc0-\xd6\xd8-\xf6\xf8-\u02ff\u0370-\u037d\u037f-\u1fff\u200c\u200d\u2070-\u218f\u2c00-\u2fef\u3001-\ud7ff\uf900-\ufdcf\ufdf0-\ufffd]|[\ud800-\udb7f][\udc00-\udfff])(?:\.?[\-0-9A-Z_a-z\xb7\xc0-\xd6\xd8-\xf6\xf8-\u037d\u037f-\u1fff\u200c\u200d\u203f\u2040\u2070-\u218f\u2c00-\u2fef\u3001-\ud7ff\uf900-\ufdcf\ufdf0-\ufffd]|[\ud800-\udb7f][\udc00-\udfff])*)(?:[ \t]+|(?=\.?[,;:!\^\s#()\[\]\{\}"'<>]))/y;
     this._number = /[\-+]?(?:(\d+\.\d*|\.?\d+)[eE][\-+]?\d+|(?=\.?\d)\d*(?:(\.)\d+)?)(?=\.?[,;:!\^\s#()\[\]\{\}"'<>])/y;
     this._boolean = /(?:true|false)(?=[.,;!\^\s#()\[\]\{\}"'<>])/y;
-    this._atKeyword = /@[a-z]+(?=[\s#<:])/iy;
-    this._keyword = /(?:PREFIX|BASE|VERSION|GRAPH)(?=[\s#<])/iy;
+    this._atKeyword = /@[a-z]+(?=[\s#<:"'])/iy;
+    this._keyword = /(?:PREFIX|BASE|VERSION|GRAPH)(?=[\s#<"'])/iy;
     this._n3Verb = /(?:has|is|of)(?=[\s#()\[\]\{\}"'<>?_+\-0-9])/y;
     this._n3Id = /id(?=[\s#<])/y;
     this._shortPredicates = /a(?=[\s#()\[\]\{\}"'<>])/y;
@@ -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 {
@@ -208,7 +210,7 @@ export default class N3Lexer {
       const line = this._line, firstChar = input[pos];
       let type = '', value = '', prefix = '',
           match = null, matchLength = 0, lexicalLength = 0,
-          finalLineLength = 0, inconclusive = false;
+          finalLineLength = 0, inconclusive = false, tripleQuoted = false;
       switch (firstChar) {
       case '^':
         // A datatype marker separated from its type cannot be followed by another marker
@@ -293,7 +295,7 @@ export default class N3Lexer {
           value = match[1];
         // Try to find a literal wrapped in three pairs of quotes
         else {
-          ({ value, matchLength, finalLineLength } = this._parseLiteral(input, pos));
+          ({ value, matchLength, finalLineLength, tripleQuoted } = this._parseLiteral(input, pos));
           if (value === null)
             return reportSyntaxError(this, input, pos);
         }
@@ -310,7 +312,7 @@ export default class N3Lexer {
             value = match[1];
           // Try to find a literal wrapped in three pairs of quotes
           else {
-            ({ value, matchLength, finalLineLength } = this._parseLiteral(input, pos));
+            ({ value, matchLength, finalLineLength, tripleQuoted } = this._parseLiteral(input, pos));
             if (value === null)
               return reportSyntaxError(this, input, pos);
           }
@@ -550,10 +552,15 @@ export default class N3Lexer {
       if (finalLineLength) {
         token = {
           type, value, prefix, line, start,
-          end: finalLineLength, endLine: this._line,
+          end: finalLineLength, endLine: this._line, tripleQuoted,
         };
         callback(null, token);
       }
+      // Triple-quoted strings are marked, since version declarations do not allow them
+      else if (tripleQuoted) {
+        token = { type, value, prefix, line, start, end: start + length, tripleQuoted };
+        callback(null, token);
+      }
       else
         token = emitToken(type, value, prefix, line, start, lexicalLength || length);
       this.previousToken = token;
@@ -705,13 +712,16 @@ export default class N3Lexer {
             break;
           this._line += lineCount;
           const finalLineLength = lineCount === 0 ? 0 : lines[lines.length - 1].length + openingLength;
-          return { value: this._unescape(raw, stringEscapeReplacements), matchLength, finalLineLength };
+          return {
+            value: this._unescape(raw, stringEscapeReplacements), matchLength, finalLineLength,
+            tripleQuoted: openingLength === 3,
+          };
         }
         closingPos++;
       }
       this._literalClosingPos = input.length - pos - openingLength + 1;
     }
-    return { value: '', matchLength: 0, finalLineLength: 0 };
+    return { value: '', matchLength: 0, finalLineLength: 0, tripleQuoted: false };
   }
 
   // ### `_tryTokenizeToEnd` tokenizes as far as possible, reporting failures through the callback

package/src/N3Parser.js

@@ -1111,7 +1111,8 @@ export default class N3Parser {
   _readVersion(token) {
     if (token.type !== 'literal')
       return this._error('Expected literal to follow version declaration', token);
-    if ((token.end - token.start) !== token.value.length + 2)
+    // Only short strings are allowed, so no numbers or booleans (which have a datatype prefix)
+    if (token.prefix !== '' || token.tripleQuoted)
       return this._error('Version declarations must use single quotes', token);
     this._versionCallback(token.value);
     if (!this._isValidVersion(token.value))

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;