n3 2.12.0 → 2.12.2

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;
@@ -236,7 +236,8 @@ 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 +311,8 @@ class N3Lexer {
             ({
               value,
               matchLength,
-              finalLineLength
+              finalLineLength,
+              tripleQuoted
             } = this._parseLiteral(input, pos));
             if (value === null) return reportSyntaxError(this, input, pos);
           }
@@ -328,7 +330,8 @@ class N3Lexer {
               ({
                 value,
                 matchLength,
-                finalLineLength
+                finalLineLength,
+                tripleQuoted
               } = this._parseLiteral(input, pos));
               if (value === null) return reportSyntaxError(this, input, pos);
             }
@@ -521,7 +524,21 @@ 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 +688,8 @@ class N3Lexer {
           return {
             value: this._unescape(raw, stringEscapeReplacements),
             matchLength,
-            finalLineLength
+            finalLineLength,
+            tripleQuoted: openingLength === 3
           };
         }
         closingPos++;
@@ -681,14 +699,34 @@ class N3Lexer {
     return {
       value: '',
       matchLength: 0,
-      finalLineLength: 0
+      finalLineLength: 0,
+      tripleQuoted: false
     };
   }
 
+  // ### `_tryTokenizeToEnd` tokenizes as far as possible, reporting failures through the callback
+  _tryTokenizeToEnd(callback, inputFinished) {
+    // Keep track of errors thrown by the callback, which must reach the caller unchanged
+    let callbackError;
+    try {
+      this._tokenizeToEnd((error, token) => {
+        try {
+          return callback(error, token);
+        } catch (thrown) {
+          throw callbackError = thrown;
+        }
+      }, inputFinished);
+    } catch (error) {
+      // Matching an extremely long token can exhaust the regular expression stack
+      if (error === callbackError || !(error instanceof RangeError)) throw error;
+      callback(this._syntaxError(null, `Token too long on line ${this._line}.`));
+    }
+  }
+
   // ### `_syntaxError` creates a syntax error for the given issue
-  _syntaxError(issue) {
+  _syntaxError(issue, message = `Unexpected "${issue}" on line ${this._line}.`) {
     this._input = null;
-    const err = new Error(`Unexpected "${issue}" on line ${this._line}.`);
+    const err = new Error(message);
     err.context = {
       token: undefined,
       line: this._line,
@@ -729,13 +767,13 @@ class N3Lexer {
       this._input = this._readStartingBom(input);
       // If a callback was passed, asynchronously call it
       if (typeof callback === 'function') queueMicrotask(() => {
-        if (this._tokenization === tokenization) this._tokenizeToEnd(callback, true);
+        if (this._tokenization === tokenization) this._tryTokenizeToEnd(callback, true);
       });
       // If no callback was passed, tokenize synchronously and return
       else {
         const tokens = [];
         let error;
-        this._tokenizeToEnd((e, t) => e ? error = e : tokens.push(t), true);
+        this._tryTokenizeToEnd((e, t) => e ? error = e : tokens.push(t), true);
         if (error) throw error;
         return tokens;
       }
@@ -762,7 +800,7 @@ class N3Lexer {
           // Tokenize as far as possible. When a previous attempt left a long unfinished token,
           // wait until the buffered input has doubled, so the token is not rescanned for every chunk.
           if (this._input.length >= retryLength) {
-            this._tokenizeToEnd(callback, false);
+            this._tryTokenizeToEnd(callback, false);
             retryLength = this._input !== null && this._input.length > MIN_RESCAN_LENGTH ? 2 * this._input.length : 0;
           }
         }
@@ -773,7 +811,7 @@ class N3Lexer {
           // Decode any incomplete character left at the end
           const rest = decoder ? decoder.decode() : '';
           if (rest) this._input = typeof this._input === 'string' ? this._input + rest : rest;
-          if (typeof this._input === 'string') this._tokenizeToEnd(callback, true);
+          if (typeof this._input === 'string') this._tryTokenizeToEnd(callback, true);
         }
       });
       input.on('error', error => {

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/package.json

@@ -1,6 +1,6 @@
 {
   "name": "n3",
-  "version": "2.12.0",
+  "version": "2.12.2",
   "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;
@@ -208,7 +208,7 @@ 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 +293,7 @@ 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 +310,7 @@ 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 +550,15 @@ 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;
@@ -705,19 +710,44 @@ 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
+  _tryTokenizeToEnd(callback, inputFinished) {
+    // Keep track of errors thrown by the callback, which must reach the caller unchanged
+    let callbackError;
+    try {
+      this._tokenizeToEnd((error, token) => {
+        try {
+          return callback(error, token);
+        }
+        catch (thrown) {
+          throw (callbackError = thrown);
+        }
+      }, inputFinished);
+    }
+    catch (error) {
+      // Matching an extremely long token can exhaust the regular expression stack
+      if (error === callbackError || !(error instanceof RangeError))
+        throw error;
+      callback(this._syntaxError(null, `Token too long on line ${this._line}.`));
+    }
   }
 
   // ### `_syntaxError` creates a syntax error for the given issue
-  _syntaxError(issue) {
+  _syntaxError(issue, message = `Unexpected "${issue}" on line ${this._line}.`) {
     this._input = null;
-    const err = new Error(`Unexpected "${issue}" on line ${this._line}.`);
+    const err = new Error(message);
     err.context = {
       token: undefined,
       line: this._line,
@@ -760,13 +790,13 @@ export default class N3Lexer {
       if (typeof callback === 'function')
         queueMicrotask(() => {
           if (this._tokenization === tokenization)
-            this._tokenizeToEnd(callback, true);
+            this._tryTokenizeToEnd(callback, true);
         });
       // If no callback was passed, tokenize synchronously and return
       else {
         const tokens = [];
         let error;
-        this._tokenizeToEnd((e, t) => e ? (error = e) : tokens.push(t), true);
+        this._tryTokenizeToEnd((e, t) => e ? (error = e) : tokens.push(t), true);
         if (error) throw error;
         return tokens;
       }
@@ -793,7 +823,7 @@ export default class N3Lexer {
           // Tokenize as far as possible. When a previous attempt left a long unfinished token,
           // wait until the buffered input has doubled, so the token is not rescanned for every chunk.
           if (this._input.length >= retryLength) {
-            this._tokenizeToEnd(callback, false);
+            this._tryTokenizeToEnd(callback, false);
             retryLength = this._input !== null && this._input.length > MIN_RESCAN_LENGTH ?
               2 * this._input.length : 0;
           }
@@ -807,7 +837,7 @@ export default class N3Lexer {
           if (rest)
             this._input = typeof this._input === 'string' ? this._input + rest : rest;
           if (typeof this._input === 'string')
-            this._tokenizeToEnd(callback, true);
+            this._tryTokenizeToEnd(callback, true);
         }
       });
       input.on('error', error => {

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))