n3 2.11.2 → 3.0.0-alpha.2

package/lib/N3Lexer.js

@@ -54,8 +54,7 @@ const localNameEscapeReplacements = {
 };
 const illegalIriChars = /[\x00-\x20<>\\"\{\}\|\^\`]/;
 // Characters that cannot occur in a prefixed name, not even escaped
-// (global, so that testAt searches the rest of the input from a position)
-const nonPrefixedNameChar = /[\s<>"{}|^`]/g;
+const nonPrefixedNameChar = /[\s<>"{}|^`]/;
 
 // A valid code point is a Unicode scalar value: at most U+10FFFF and not a surrogate
 function isValidCodePoint(charCode) {
@@ -71,48 +70,30 @@ const lineModeRegExps = {
   _whitespace: true
 };
 const invalidRegExp = /$0^/;
-const nonWhitespace = /\S*/y;
-
-// Matches a sticky regular expression at the given position of the input
-function execAt(regExp, input, pos) {
-  regExp.lastIndex = pos;
-  return regExp.exec(input);
-}
-function testAt(regExp, input, pos) {
-  regExp.lastIndex = pos;
-  return regExp.test(input);
-}
-// Matches the rest of the input followed by a space, as at the end of the input,
-// a token that can contain (but not end with) a dot needs a non-dot character after it
-function execAtEnd(regExp, input, pos) {
-  regExp.lastIndex = 0;
-  return regExp.exec(`${input.slice(pos)} `);
-}
 
 // ## Constructor
 class N3Lexer {
   constructor(options) {
     // ## Regular expressions
-    // It's slightly faster to have these as properties than as in-scope variables.
-    // They are sticky, so they only match at the `lastIndex` set by `execAt`.
-    this._iri = /<((?:[^ <>{}\\]|\\[uU])+)>[ \t]*/y; // IRI with escape sequences; needs sanity check after unescaping
-    this._unescapedIri = /<([^\x00-\x20<>\\"\{\}\|\^\`]*)>[ \t]*/y; // IRI without escape sequences; no unescaping
-    this._simpleQuotedString = /"([^"\\\r\n]*)"(?=[^"])/y; // string without escape sequences
-    this._simpleApostropheString = /'([^'\\\r\n]*)'(?=[^'])/y;
-    this._langcode = /@([a-z]+(?:-[a-z0-9]+)*)(?=[^a-z0-9])/iy;
-    this._prefix = /((?:[A-Za-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])*)?:(?=[#\s<])/y;
-    this._prefixed = /((?:[A-Za-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])*)?:((?:(?:[0-9:A-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-fA-F]{2}|\\[!#-\/;=?\-@_~])(?:(?:[\.\-0-9:A-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]|%[0-9a-fA-F]{2}|\\[!#-\/;=?\-@_~])*(?:[\-0-9:A-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]|%[0-9a-fA-F]{2}|\\[!#-\/;=?\-@_~]))?)?)(?:[ \t]+|(?=\.?[,;!\^\s#()\[\]\{\}"'<>]))/y;
-    this._variable = /\?(?:(?:[A-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])(?:[\-0-9:A-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])*)(?=[.,;!\^\s#()\[\]\{\}"'<>])/y;
-    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._n3Verb = /(?:has|is|of)(?=[\s#()\[\]\{\}"'<>?_+\-0-9])/y;
-    this._n3Id = /id(?=[\s#<])/y;
-    this._shortPredicates = /a(?=[\s#()\[\]\{\}"'<>])/y;
-    this._commentLine = /[ \t]*#([^\n\r]*)(?:\r\n|\n|\r)([ \t]*)/y;
-    this._whitespace = /[ \t]+/y;
+    // It's slightly faster to have these as properties than as in-scope variables
+    this._iri = /^<((?:[^ <>{}\\]|\\[uU])+)>[ \t]*/; // IRI with escape sequences; needs sanity check after unescaping
+    this._unescapedIri = /^<([^\x00-\x20<>\\"\{\}\|\^\`]*)>[ \t]*/; // IRI without escape sequences; no unescaping
+    this._simpleQuotedString = /^"([^"\\\r\n]*)"(?=[^"])/; // string without escape sequences
+    this._simpleApostropheString = /^'([^'\\\r\n]*)'(?=[^'])/;
+    this._langcode = /^@([a-z]+(?:-[a-z0-9]+)*)(?=[^a-z0-9])/i;
+    this._prefix = /^((?:[A-Za-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])*)?:(?=[#\s<])/;
+    this._prefixed = /^((?:[A-Za-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])*)?:((?:(?:[0-9:A-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-fA-F]{2}|\\[!#-\/;=?\-@_~])(?:(?:[\.\-0-9:A-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]|%[0-9a-fA-F]{2}|\\[!#-\/;=?\-@_~])*(?:[\-0-9:A-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]|%[0-9a-fA-F]{2}|\\[!#-\/;=?\-@_~]))?)?)(?:[ \t]+|(?=\.?[,;!\^\s#()\[\]\{\}"'<>]))/;
+    this._variable = /^\?(?:(?:[A-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])(?:[\-0-9:A-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])*)(?=[.,;!\^\s#()\[\]\{\}"'<>])/;
+    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#()\[\]\{\}"'<>]))/;
+    this._number = /^[\-+]?(?:(\d+\.\d*|\.?\d+)[eE][\-+]?\d+|(?=\.?\d)\d*(?:(\.)\d+)?)(?=\.?[,;:!\^\s#()\[\]\{\}"'<>])/;
+    this._boolean = /^(?:true|false)(?=[.,;!\^\s#()\[\]\{\}"'<>])/;
+    this._atKeyword = /^@[a-z]+(?=[\s#<:])/i;
+    this._keyword = /^(?:PREFIX|BASE|VERSION|GRAPH)(?=[\s#<])/i;
+    this._n3Verb = /^(?:has|is|of)(?=[\s#()\[\]\{\}"'<>?_+\-0-9])/;
+    this._n3Id = /^id(?=[\s#<])/;
+    this._shortPredicates = /^a(?=[\s#()\[\]\{\}"'<>])/;
+    this._commentLine = /^[ \t]*#([^\n\r]*)(?:\r\n|\n|\r)([ \t]*)/;
+    this._whitespace = /^[ \t]+/;
     options = options || {};
 
     // Whether the log:isImpliedBy predicate is supported
@@ -140,71 +121,77 @@ class N3Lexer {
 
   // ### `_tokenizeToEnd` tokenizes as for as possible, emitting tokens through the callback
   _tokenizeToEnd(callback, inputFinished) {
-    // Continue parsing as far as possible; the loop will return eventually.
-    // Rather than slicing off every token, track the position of the remaining input;
-    // the regular expressions are sticky, so they match at that position.
-    const input = this._input;
-    let pos = 0;
+    // Continue parsing as far as possible; the loop will return eventually
+    let input = this._input;
     let currentLineLength = this._linePosition + input.length;
     while (true) {
       // Consume one separator line at a time, including its following indentation.
       while (true) {
-        let charCode = input.charCodeAt(pos),
+        let charCode = input.charCodeAt(0),
           separatorLength = 0;
         if (charCode === SPACE || charCode === TAB) {
-          const next = input.charCodeAt(pos + 1);
-          separatorLength = next === SPACE || next === TAB ? execAt(this._whitespace, input, pos)[0].length : 1;
-          charCode = input.charCodeAt(pos + separatorLength);
+          const next = input.charCodeAt(1);
+          separatorLength = next === SPACE || next === TAB ? this._whitespace.exec(input)[0].length : 1;
+          charCode = input.charCodeAt(separatorLength);
         }
         if (charCode === HASH) {
-          const comment = execAt(this._commentLine, input, pos);
+          const comment = this._commentLine.exec(input);
           if (comment) {
             const commentLength = comment[0].length;
             // Keep a trailing CR buffered in case the next chunk starts with LF.
-            if (!inputFinished && pos + commentLength === input.length && input.charCodeAt(input.length - 1) === CR) return this._suspend(input, pos, currentLineLength);
-            if (this.comments) emitComment(comment[1], this._line, currentLineLength - (input.length - pos) + separatorLength);
-            pos += commentLength;
-            currentLineLength = input.length - pos + comment[2].length;
+            if (!inputFinished && commentLength === input.length && input.charCodeAt(commentLength - 1) === CR) {
+              this._linePosition = currentLineLength - input.length;
+              return this._input = input;
+            }
+            if (this.comments) emitComment(comment[1], this._line, separatorLength);
+            input = input.slice(commentLength);
+            currentLineLength = input.length + comment[2].length;
             this._line++;
           } else {
             // A comment without a line ending stays buffered until EOF.
-            pos += separatorLength;
-            if (!inputFinished) return this._suspend(input, pos, currentLineLength);
-            if (this.comments) emitComment(input.slice(pos + 1), this._line, currentLineLength - (input.length - pos));
-            pos = input.length;
+            input = input.slice(separatorLength);
+            if (!inputFinished) {
+              this._linePosition = currentLineLength - input.length;
+              return this._input = input;
+            }
+            if (this.comments) emitComment(input.slice(1), this._line, 0);
+            input = '';
             break;
           }
         } else if (charCode === LF || charCode === CR) {
           // A CR at the end of a chunk may still be followed by LF.
-          if (!inputFinished && charCode === CR && pos + separatorLength + 1 === input.length) return this._suspend(input, pos, currentLineLength);
-          separatorLength += charCode === CR && input.charCodeAt(pos + separatorLength + 1) === LF ? 2 : 1;
+          if (!inputFinished && charCode === CR && separatorLength + 1 === input.length) {
+            this._linePosition = currentLineLength - input.length;
+            return this._input = input;
+          }
+          separatorLength += charCode === CR && input.charCodeAt(separatorLength + 1) === LF ? 2 : 1;
           // Indentation is consumed with the newline, but belongs to the next line's columns.
           let indentationLength = 0;
-          const next = input.charCodeAt(pos + separatorLength);
+          const next = input.charCodeAt(separatorLength);
           if (next === SPACE || next === TAB) {
-            const following = input.charCodeAt(pos + separatorLength + 1);
-            indentationLength = following === SPACE || following === TAB ? execAt(this._whitespace, input, pos + separatorLength)[0].length : 1;
+            const following = input.charCodeAt(separatorLength + 1);
+            indentationLength = following === SPACE || following === TAB ? this._whitespace.exec(input.slice(separatorLength))[0].length : 1;
           }
-          pos += separatorLength + indentationLength;
-          currentLineLength = input.length - pos + indentationLength;
+          input = input.slice(separatorLength + indentationLength);
+          currentLineLength = input.length + indentationLength;
           this._line++;
         } else {
-          pos += separatorLength;
+          if (separatorLength !== 0) input = input.slice(separatorLength);
           break;
         }
       }
-      if (pos >= input.length) {
-        this._linePosition = currentLineLength;
+      if (input.length === 0) {
         if (inputFinished) {
-          emitToken('eof', '', '', this._line, currentLineLength, 0);
-          return this._input = null;
+          input = null;
+          emitToken('eof', '', '', this._line, 0);
         }
-        return this._input = '';
+        this._linePosition = currentLineLength;
+        return this._input = input;
       }
 
       // Look for specific token types based on the first character
       const line = this._line,
-        firstChar = input[pos];
+        firstChar = input[0];
       let type = '',
         value = '',
         prefix = '',
@@ -216,13 +203,13 @@ class N3Lexer {
       switch (firstChar) {
         case '^':
           // We need at least 3 tokens lookahead to distinguish ^^<IRI> and ^^pre:fixed
-          if (input.length - pos < 3) break;
+          if (input.length < 3) break;
           // Try to match a type
-          else if (input[pos + 1] === '^') {
+          else if (input[1] === '^') {
             this._previousMarker = '^^';
             // Move to type IRI or prefixed name
-            pos += 2;
-            if (input[pos] !== '<') {
+            input = input.slice(2);
+            if (input[0] !== '<') {
               inconclusive = true;
               break;
             }
@@ -238,53 +225,53 @@ class N3Lexer {
         // Fall through in case the type is an IRI
         case '<':
           // Try to find a full IRI without escape sequences
-          if (match = execAt(this._unescapedIri, input, pos)) {
+          if (match = this._unescapedIri.exec(input)) {
             type = 'IRI', value = match[1];
             lexicalLength = match[1].length + 2;
           }
           // Try to find a full IRI with escape sequences
-          else if (match = execAt(this._iri, input, pos)) {
+          else if (match = this._iri.exec(input)) {
             value = this._unescape(match[1], stringEscapeReplacements);
-            if (value === null || illegalIriChars.test(value)) return reportSyntaxError(this, input, pos);
+            if (value === null || illegalIriChars.test(value)) return reportSyntaxError(this);
             type = 'IRI';
             lexicalLength = match[1].length + 2;
           }
           // Try to find a triple term
-          else if (input.length - pos > 2 && input[pos + 1] === '<' && input[pos + 2] === '(') type = '<<(', matchLength = 3;
+          else if (input.length > 2 && input[1] === '<' && input[2] === '(') type = '<<(', matchLength = 3;
           // Try to find a reified triple
-          else if (!this._lineMode && input.length - pos > (inputFinished ? 1 : 2) && input[pos + 1] === '<') type = '<<', matchLength = 2;
+          else if (!this._lineMode && input.length > (inputFinished ? 1 : 2) && input[1] === '<') type = '<<', matchLength = 2;
           // Try to find a backwards implication arrow
-          else if (this._n3Mode && input.length - pos > 1 && input[pos + 1] === '=') {
+          else if (this._n3Mode && input.length > 1 && input[1] === '=') {
             matchLength = 2;
             if (this._isImpliedBy) type = 'abbreviation', value = '<';else type = 'inverse', value = '>';
           }
           // Try to find an inverted predicate marker
-          else if (this._n3Mode && input.length - pos > 1 && input[pos + 1] === '-') type = 'inversePredicate', matchLength = 2;
+          else if (this._n3Mode && input.length > 1 && input[1] === '-') type = 'inversePredicate', matchLength = 2;
           break;
         case '>':
           // Try to find a reified triple
-          if (input.length - pos > 1 && input[pos + 1] === '>') type = '>>', matchLength = 2;
+          if (input.length > 1 && input[1] === '>') type = '>>', matchLength = 2;
           break;
         case '_':
           // Try to find a blank node. Since it can contain (but not end with) a dot,
           // we always need a non-dot character before deciding it is a blank node.
           // Therefore, try inserting a space if we're at the end of the input.
-          if ((match = execAt(this._blank, input, pos)) || inputFinished && (match = execAtEnd(this._blank, input, pos))) {
+          if ((match = this._blank.exec(input)) || inputFinished && (match = this._blank.exec(`${input} `))) {
             type = 'blank', prefix = '_', value = match[1];
             lexicalLength = match[1].length + 2;
           }
           break;
         case '"':
           // Try to find a literal without escape sequences
-          if (match = execAt(this._simpleQuotedString, input, pos)) value = match[1];
+          if (match = this._simpleQuotedString.exec(input)) value = match[1];
           // Try to find a literal wrapped in three pairs of quotes
           else {
             ({
               value,
               matchLength,
               finalLineLength
-            } = this._parseLiteral(input, pos));
-            if (value === null) return reportSyntaxError(this, input, pos);
+            } = this._parseLiteral(input));
+            if (value === null) return reportSyntaxError(this);
           }
           if (match !== null || matchLength !== 0) {
             type = 'literal';
@@ -294,15 +281,15 @@ class N3Lexer {
         case "'":
           if (!this._lineMode) {
             // Try to find a literal without escape sequences
-            if (match = execAt(this._simpleApostropheString, input, pos)) value = match[1];
+            if (match = this._simpleApostropheString.exec(input)) value = match[1];
             // Try to find a literal wrapped in three pairs of quotes
             else {
               ({
                 value,
                 matchLength,
                 finalLineLength
-              } = this._parseLiteral(input, pos));
-              if (value === null) return reportSyntaxError(this, input, pos);
+              } = this._parseLiteral(input));
+              if (value === null) return reportSyntaxError(this);
             }
             if (match !== null || matchLength !== 0) {
               type = 'literal';
@@ -312,7 +299,7 @@ class N3Lexer {
           break;
         case '?':
           // Try to find a variable
-          if (this._n3Mode && (match = execAt(this._variable, input, pos))) type = 'var', value = match[0];
+          if (this._n3Mode && (match = this._variable.exec(input))) type = 'var', value = match[0];
           break;
         case '@':
           // Try to find a language code. A language code can contain dash-separated
@@ -320,16 +307,15 @@ class N3Lexer {
           // input is not finished, another subtag may still arrive in a later chunk and
           // the match would be premature; wait for more input in that case.
           // A double dash starts a direction code, which cannot extend the language code.
-          if (this._previousMarker === 'literal' && (match = execAt(this._langcode, input, pos)) && match[1] !== 'version') {
-            const end = pos + match[0].length;
-            if (!inputFinished && input[end] === '-' && input[end + 1] !== '-') match = null;else type = 'langcode', value = match[1];
+          if (this._previousMarker === 'literal' && (match = this._langcode.exec(input)) && match[1] !== 'version') {
+            if (!inputFinished && input[match[0].length] === '-' && input[match[0].length + 1] !== '-') match = null;else type = 'langcode', value = match[1];
           }
           // Try to find a keyword
-          else if (match = execAt(this._atKeyword, input, pos)) type = match[0];
+          else if (match = this._atKeyword.exec(input)) type = match[0];
           break;
         case '.':
           // Try to find a dot as punctuation
-          if (input.length - pos === 1 ? inputFinished : input[pos + 1] < '0' || input[pos + 1] > '9') {
+          if (input.length === 1 ? inputFinished : input[1] < '0' || input[1] > '9') {
             type = '.';
             matchLength = 1;
             break;
@@ -348,10 +334,10 @@ class N3Lexer {
         case '9':
         case '+':
         case '-':
-          if (input[pos + 1] === '-') {
+          if (input[1] === '-') {
             // Try to find a direction code
             if (this._previousMarker === 'langcode') {
-              if (input.startsWith('--ltr', pos)) type = 'dircode', value = 'ltr', matchLength = 5;else if (input.startsWith('--rtl', pos)) type = 'dircode', value = 'rtl', matchLength = 5;
+              if (input.startsWith('--ltr')) type = 'dircode', value = 'ltr', matchLength = 5;else if (input.startsWith('--rtl')) type = 'dircode', value = 'rtl', matchLength = 5;
             }
             break;
           }
@@ -359,7 +345,7 @@ class N3Lexer {
           // Try to find a number. Since it can contain (but not end with) a dot,
           // we always need a non-dot character before deciding it is a number.
           // Therefore, try inserting a space if we're at the end of the input.
-          if (match = execAt(this._number, input, pos) || inputFinished && (match = execAtEnd(this._number, input, pos))) {
+          if (match = this._number.exec(input) || inputFinished && (match = this._number.exec(`${input} `))) {
             type = 'literal', value = match[0];
             prefix = typeof match[1] === 'string' ? xsd.double : typeof match[2] === 'string' ? xsd.decimal : xsd.integer;
           }
@@ -373,42 +359,42 @@ class N3Lexer {
         case 'V':
         case 'v':
           // Try to find a SPARQL-style keyword
-          if (match = execAt(this._keyword, input, pos)) type = match[0].toUpperCase();else inconclusive = true;
+          if (match = this._keyword.exec(input)) type = match[0].toUpperCase();else inconclusive = true;
           break;
         case 'f':
         case 't':
           // Try to match a boolean
-          if (testAt(this._boolean, input, pos)) type = 'literal', value = firstChar === 't' ? 'true' : 'false', prefix = xsd.boolean, matchLength = value.length;else inconclusive = true;
+          if (this._boolean.test(input)) type = 'literal', value = firstChar === 't' ? 'true' : 'false', prefix = xsd.boolean, matchLength = value.length;else inconclusive = true;
           break;
         case 'a':
           // Try to find an abbreviated predicate
-          if (testAt(this._shortPredicates, input, pos)) type = 'abbreviation', value = 'a', matchLength = 1;else inconclusive = true;
+          if (this._shortPredicates.test(input)) type = 'abbreviation', value = 'a', matchLength = 1;else inconclusive = true;
           break;
         case 'h':
         case 'o':
           // Try to find an N3 verb keyword
-          if (this._n3Mode && (match = this._matchN3Verb(input, pos, inputFinished))) type = match[0];else inconclusive = true;
+          if (this._n3Mode && (match = this._matchN3Verb(input, inputFinished))) type = match[0];else inconclusive = true;
           break;
         case 'i':
           // Try to find an IRI property list identifier or N3 verb keyword
-          if (this._n3Mode && testAt(this._n3Id, input, pos)) type = 'id', matchLength = 2;else if (this._n3Mode && (match = this._matchN3Verb(input, pos, inputFinished))) type = match[0];else inconclusive = true;
+          if (this._n3Mode && this._n3Id.test(input)) type = 'id', matchLength = 2;else if (this._n3Mode && (match = this._matchN3Verb(input, inputFinished))) type = match[0];else inconclusive = true;
           break;
         case '=':
           // Try to find an implication arrow or equals sign
-          if (this._n3Mode && input.length - pos > 1) {
+          if (this._n3Mode && input.length > 1) {
             type = 'abbreviation';
-            if (input[pos + 1] !== '>') matchLength = 1, value = '=';else matchLength = 2, value = '>';
+            if (input[1] !== '>') matchLength = 1, value = '=';else matchLength = 2, value = '>';
           }
           break;
         case '!':
           if (!this._n3Mode) break;
         case ')':
-          if (!inputFinished && (input.length - pos === 1 || input.length - pos === 2 && input[pos + 1] === '>')) {
+          if (!inputFinished && (input.length === 1 || input.length === 2 && input[1] === '>')) {
             // Don't consume yet, as it *could* become a triple term end.
             break;
           }
           // Try to find a triple term
-          if (input.length - pos > 2 && input[pos + 1] === '>' && input[pos + 2] === '>') {
+          if (input.length > 2 && input[1] === '>' && input[2] === '>') {
             type = ')>>', matchLength = 3;
             break;
           }
@@ -426,15 +412,15 @@ class N3Lexer {
           break;
         case '{':
           // We need at least 2 tokens lookahead to distinguish "{|" and "{ "
-          if (!this._lineMode && input.length - pos >= 2) {
+          if (!this._lineMode && input.length >= 2) {
             // Try to find a quoted triple annotation start
-            if (input[pos + 1] === '|') type = '{|', matchLength = 2;else type = firstChar, matchLength = 1;
+            if (input[1] === '|') type = '{|', matchLength = 2;else type = firstChar, matchLength = 1;
           }
           break;
         case '|':
           // We need 2 tokens lookahead to parse "|}"
           // Try to find a quoted triple annotation end
-          if (input.length - pos >= 2 && input[pos + 1] === '}') type = '|}', matchLength = 2;
+          if (input.length >= 2 && input[1] === '}') type = '|}', matchLength = 2;
           break;
         default:
           inconclusive = true;
@@ -443,11 +429,11 @@ class N3Lexer {
       // Some first characters do not allow an immediate decision, so inspect more
       if (inconclusive) {
         // Try to find a prefix
-        if ((this._previousMarker === '@prefix' || this._previousMarker === 'PREFIX') && (match = execAt(this._prefix, input, pos))) type = 'prefix', value = match[1] || '';
+        if ((this._previousMarker === '@prefix' || this._previousMarker === 'PREFIX') && (match = this._prefix.exec(input))) type = 'prefix', value = match[1] || '';
         // Try to find a prefixed name. Since it can contain (but not end with) a dot,
         // we always need a non-dot character before deciding it is a prefixed name.
         // Therefore, try inserting a space if we're at the end of the input.
-        else if ((match = execAt(this._prefixed, input, pos)) || inputFinished && (match = execAtEnd(this._prefixed, input, pos))) {
+        else if ((match = this._prefixed.exec(input)) || inputFinished && (match = this._prefixed.exec(`${input} `))) {
           type = 'prefixed', prefix = match[1] || '';
           value = this._unescape(match[2], localNameEscapeReplacements);
           lexicalLength = prefix.length + match[2].length + 1;
@@ -473,14 +459,16 @@ class N3Lexer {
         // We could be in streaming mode, and then we just wait for more input to arrive.
         // Otherwise, a syntax error has occurred in the input.
         // One exception: error on an unaccounted linebreak (= not inside a triple-quoted literal).
-        if (inputFinished || !input.startsWith("'''", pos) && !input.startsWith('"""', pos) && /\n|\r/.test(input.slice(pos))) return reportSyntaxError(this, input, pos);else return this._suspend(input, pos, currentLineLength);
+        if (inputFinished || !/^'''|^"""/.test(input) && /\n|\r/.test(input)) return reportSyntaxError(this);else {
+          this._linePosition = currentLineLength - input.length;
+          return this._input = input;
+        }
       }
 
       // Emit the parsed token
       // Consumption includes separator whitespace; lexicalLength excludes it
-      // and any synthetic EOF space. Consumption is clamped to the input below.
+      // and any synthetic EOF space. slice below clamps consumption to the input.
       const length = matchLength || match[0].length;
-      const start = currentLineLength - (input.length - pos);
       let token;
       if (finalLineLength) {
         token = {
@@ -488,22 +476,23 @@ class N3Lexer {
           value,
           prefix,
           line,
-          start,
+          start: currentLineLength - input.length,
           end: finalLineLength,
           endLine: this._line
         };
         callback(null, token);
-      } else token = emitToken(type, value, prefix, line, start, lexicalLength || length);
+      } else token = emitToken(type, value, prefix, line, lexicalLength || length);
       this.previousToken = token;
       this._previousMarker = type;
 
       // Advance to next part to tokenize
-      pos = Math.min(pos + length, input.length);
-      if (finalLineLength) currentLineLength = input.length - pos + finalLineLength;
+      input = input.slice(length);
+      if (finalLineLength) currentLineLength = input.length + finalLineLength;
     }
 
     // Emits a comment at its exact position within matched whitespace.
-    function emitComment(value, line, start) {
+    function emitComment(value, line, offset) {
+      const start = currentLineLength - input.length + offset;
       callback(null, {
         type: 'comment',
         value,
@@ -514,53 +503,49 @@ class N3Lexer {
       });
     }
     // Emits the token through the callback
-    function emitToken(type, value, prefix, line, start, length) {
+    function emitToken(type, value, prefix, line, length) {
+      const start = input ? currentLineLength - input.length : currentLineLength;
+      const end = start + length;
       const token = {
         type,
         value,
         prefix,
         line,
         start,
-        end: start + length
+        end
       };
       callback(null, token);
       return token;
     }
     // Signals the syntax error through the callback
-    function reportSyntaxError(self, input, pos) {
-      callback(self._syntaxError(execAt(nonWhitespace, input, pos)[0]));
+    function reportSyntaxError(self) {
+      callback(self._syntaxError(/^\S*/.exec(input)[0]));
     }
   }
 
-  // ### `_suspend` keeps the unconsumed input until more input arrives
-  _suspend(input, pos, currentLineLength) {
-    this._linePosition = currentLineLength - (input.length - pos);
-    return this._input = input.slice(pos);
-  }
-
   // ### `_matchN3Verb` matches an N3 verb unless the input is a longer prefixed name
-  _matchN3Verb(input, pos, inputFinished) {
-    const verb = execAt(this._n3Verb, input, pos);
+  _matchN3Verb(input, inputFinished) {
+    const verb = this._n3Verb.exec(input);
     if (!verb) return null;
 
     // Most verb boundaries cannot be part of a prefix, so keep the common path fast.
-    const next = input[pos + verb[0].length];
+    const next = input[verb[0].length];
     if (next !== '-' && next !== '_' && (next < '0' || next > '9')) return verb;
 
     // A prefix can start with a verb and continue with characters that are also
     // valid verb boundaries. Prefer the longer prefixed name when it is complete.
-    if (execAt(this._prefixed, input, pos)) return null;
+    if (this._prefixed.exec(input)) return null;
     // Appending to the input only matters when a prefixed name could run up to
     // its end, which a character that cannot occur in prefixed names rules out.
     // This avoids copying the rest of the document for every such verb.
-    if (testAt(nonPrefixedNameChar, input, pos)) return verb;
-    if (execAtEnd(this._prefixed, input, pos)) return null;
+    if (nonPrefixedNameChar.test(input)) return verb;
+    if (this._prefixed.exec(`${input} `)) return null;
 
     // If a stream chunk ends partway through such a prefix, wait for the colon
     // instead of prematurely emitting the verb. Appending ": " lets the prefix
     // grammar determine whether all input seen so far can be a complete prefix.
     if (!inputFinished) {
-      const prefix = execAt(this._prefix, `${input.slice(pos)}: `, 0);
+      const prefix = this._prefix.exec(`${input}: `);
       if (prefix) return null;
     }
     return verb;
@@ -607,19 +592,19 @@ class N3Lexer {
     return result + item.slice(start);
   }
 
-  // ### `_parseLiteral` parses a literal at the given position into an unescaped value
-  _parseLiteral(input, pos) {
+  // ### `_parseLiteral` parses a literal into an unescaped value
+  _parseLiteral(input) {
     // Ensure we have enough lookahead to identify triple-quoted strings
-    if (input.length - pos >= 3) {
+    if (input.length >= 3) {
       // The caller has already identified a single or double quote.
-      const quote = input[pos];
-      const openingLength = input[pos + 1] === quote && input[pos + 2] === quote ? 3 : 1;
+      const quote = input[0];
+      const openingLength = input[1] === quote && input[2] === quote ? 3 : 1;
       let opening = quote;
       if (openingLength === 3) opening = quote === '"' ? '"""' : "'''";
 
       // Find the next candidate closing quotes
-      let closingPos = pos + Math.max(this._literalClosingPos, openingLength);
-      while ((closingPos = input.indexOf(opening, closingPos)) > pos) {
+      let closingPos = Math.max(this._literalClosingPos, openingLength);
+      while ((closingPos = input.indexOf(opening, closingPos)) > 0) {
         // Count backslashes right before the closing quotes
         let backslashCount = 0;
         while (input[closingPos - backslashCount - 1] === '\\') backslashCount++;
@@ -628,10 +613,10 @@ class N3Lexer {
         // means these are actual, non-escaped closing quotes
         if (backslashCount % 2 === 0) {
           // Extract and unescape the value
-          const raw = input.substring(pos + openingLength, closingPos),
+          const raw = input.substring(openingLength, closingPos),
             lines = raw.split(/\r\n|\r|\n/),
             lineCount = lines.length - 1;
-          const matchLength = closingPos - pos + openingLength;
+          const matchLength = closingPos + openingLength;
           // Only triple-quoted strings can be multi-line
           if (openingLength === 1 && lineCount !== 0 || openingLength === 3 && this._lineMode) break;
           this._line += lineCount;
@@ -644,7 +629,7 @@ class N3Lexer {
         }
         closingPos++;
       }
-      this._literalClosingPos = input.length - pos - openingLength + 1;
+      this._literalClosingPos = input.length - openingLength + 1;
     }
     return {
       value: '',

package/lib/N3Parser.js

@@ -44,10 +44,11 @@ class N3Parser {
     // Whether the log:isImpliedBy predicate is supported
     this._isImpliedBy = options.isImpliedBy;
     // Whether an undeclared empty prefix resolves against the document IRI
-    this._implicitEmptyPrefix = !!options.implicitEmptyPrefix;
+    // (enabled unless explicitly disabled)
+    this._implicitEmptyPrefix = options.implicitEmptyPrefix !== false;
     // Whether an empty formula is read as the boolean literal true,
-    // as in the N3 spec tests (opt-in until the next major version)
-    this._emptyFormulaAsTrue = !!options.emptyFormulaAsTrue;
+    // as in the N3 spec tests (enabled unless explicitly disabled)
+    this._emptyFormulaAsTrue = options.emptyFormulaAsTrue !== false;
     // Disable relative IRIs in N-Triples or N-Quads mode
     if (isLineMode) this._resolveRelativeIRI = iri => {
       return null;
@@ -872,7 +873,7 @@ class N3Parser {
     // Restore the parent context containing this formula
     this._restoreContext('formula', token);
 
-    // When the emptyFormulaAsTrue option is set, an empty formula
+    // Unless the emptyFormulaAsTrue option is false, an empty formula
     // is read as the boolean literal true, following the N3 spec tests
     // and the direction discussed in https://github.com/w3c-cg/N3/issues/185
     if (empty && this._emptyFormulaAsTrue) {