Improve error message for unrecognized escape sequences.

dist/ohm-grammar.js

@@ -54,7 +54,9 @@ module.exports = ohm.makeRecipe(function() {
     .define('escapeChar_tab', [], this.prim('\\t'))
     .define('escapeChar_unicodeEscape', [], this.seq(this.prim('\\u'), this.app('hexDigit'), this.app('hexDigit'), this.app('hexDigit'), this.app('hexDigit')))
     .define('escapeChar_hexEscape', [], this.seq(this.prim('\\x'), this.app('hexDigit'), this.app('hexDigit')))
-    .define('escapeChar', [], this.alt(this.app('escapeChar_backslash'), this.app('escapeChar_doubleQuote'), this.app('escapeChar_singleQuote'), this.app('escapeChar_backspace'), this.app('escapeChar_lineFeed'), this.app('escapeChar_carriageReturn'), this.app('escapeChar_tab'), this.app('escapeChar_unicodeEscape'), this.app('escapeChar_hexEscape')))
+    .define('escapeChar_unrecognized', [], this.seq(this.prim('\\'), this.not(this.app('controlChar')), this.not(this.prim(' ')), this.app('_')))
+    .define('escapeChar', [], this.alt(this.app('escapeChar_backslash'), this.app('escapeChar_doubleQuote'), this.app('escapeChar_singleQuote'), this.app('escapeChar_backspace'), this.app('escapeChar_lineFeed'), this.app('escapeChar_carriageReturn'), this.app('escapeChar_tab'), this.app('escapeChar_unicodeEscape'), this.app('escapeChar_hexEscape'), this.app('escapeChar_unrecognized')))
+    .define('controlChar', [], this.prim(/[\000-\037]/))
     .define('regExp', [], this.seq(this.prim('/'), this.app('reCharClass'), this.prim('/')), 'a regular expression')
     .define('reCharClass_unicode', [], this.seq(this.prim('\\p{'), this.plus(this.prim(/[A-Za-z]/)), this.prim('}')))
     .define('reCharClass_ordinary', [], this.seq(this.prim('['), this.star(this.alt(this.prim('\\]'), this.seq(this.not(this.prim(']')), this.app('_')))), this.prim(']')))

src/Interval.js

@@ -49,7 +49,8 @@ Interval.prototype = {
   },
 
   getLineAndColumnMessage: function() {
-    return util.getLineAndColumnMessage(this.inputStream.source, this.startIdx);
+    var range = [this.startIdx, this.endIdx];
+    return util.getLineAndColumnMessage(this.inputStream.source, this.startIdx, range);
   },
 
   // Returns a new Interval which contains the same contents as this one,

src/errors.js

@@ -56,6 +56,18 @@ var GrammarSyntaxError = makeCustomError(
     }
 );
 
+var UnrecognizedEscapeSequence = makeCustomError(
+    'ohm.error.UnrecognizedEscapeSequence',
+    function(interval) {
+      this.interval = interval;
+      var seq = interval.contents;
+      var c = seq[1];
+      this.message = interval.getLineAndColumnMessage() +
+                     'Unrecognized escape sequence "' + seq + '". Did you mean ' +
+                     JSON.stringify(seq) + ' or ' + JSON.stringify(c) + '?';
+    }
+);
+
 // Undeclared grammar
 
 var UndeclaredGrammar = makeCustomError(
@@ -261,6 +273,7 @@ module.exports = {
   MultipleErrors: MultipleErrors,
   UndeclaredGrammar: UndeclaredGrammar,
   UndeclaredRule: UndeclaredRule,
+  UnrecognizedEscapeSequence: UnrecognizedEscapeSequence,
   WrongNumberOfParameters: WrongNumberOfParameters,
 
   throwErrors: function(errors) {

src/main.js

@@ -243,10 +243,13 @@ function buildGrammar(match, namespace, optOhmGrammarForTesting) {
     },
 
     strChar: function(_) {
-      return this.interval.contents;
+      return _.visit();
     },
 
-    escapeChar: function(_) {
+    escapeChar: function(child) {
+      if (child.ctorName === 'escapeChar_unrecognized') {
+        throw new errors.UnrecognizedEscapeSequence(child.interval);
+      }
       return this.interval.contents;
     },
 

src/ohm-grammar.ohm

@@ -101,6 +101,10 @@ Ohm {
     | "\\t"                                      -- tab
     | "\\u" hexDigit hexDigit hexDigit hexDigit  -- unicodeEscape
     | "\\x" hexDigit hexDigit                    -- hexEscape
+    | "\\" ~controlChar ~" " _                   -- unrecognized
+
+  controlChar
+    = /[\000-\037]/
 
   regExp  (a regular expression)
     = "/" reCharClass "/"

test/ohm-spec.js

@@ -390,41 +390,30 @@ test('char', function(t) {
 test('string', function(t) {
   var m = ohm.grammar('M { foo = "foo\\b\\n\\r\\t\\\\\\\"\\u01bcff\\x8f" }');
 
-  test('direct match, no stream', function(t) {
-    it('recognition', function() {
-      t.ok(m.match('foo\b\n\r\t\\"'));
-      t.equal(m.match('foo1').failed(), true);
-      t.equal(m.match('bar').failed(), true);
-      t.equal(m.match(null).failed(), true);
-    });
-
-    it('semantic actions', function() {
-      var s = m.semantics().addAttribute('v', {});
-      var cst = m.match('foo\b\n\r\t\\"\u01bcff\x8f');
-      t.equal(s(cst).v, 'foo\b\n\r\t\\"\u01bcff\x8f');
-    });
-
-    it('unrecognized escape characters are parse errors', function() {
-      t.throws(function() { ohm.grammar('G { r = "\\w" }'); }, /Failed to parse grammar/);
-    });
+  it('recognition', function() {
+    t.ok(m.match('foo\b\n\r\t\\"'));
+    t.equal(m.match('foo1').failed(), true);
+    t.equal(m.match('bar').failed(), true);
+    t.equal(m.match(null).failed(), true);
+  });
 
-    t.end();
+  it('semantic actions', function() {
+    var s = m.semantics().addAttribute('v', {});
+    var cst = m.match('foo\b\n\r\t\\"\u01bcff\x8f');
+    t.equal(s(cst).v, 'foo\b\n\r\t\\"\u01bcff\x8f');
   });
 
-  test('match in string stream', function(t) {
-    it('recognition', function() {
-      t.ok(m.match('foo\b\n\r\t\\"\u01bcff\x8f'));
-      t.equal(m.match('foo1').failed(), true);
-      t.equal(m.match('bar').failed(), true);
-    });
+  // Printable chars after "\" throw UnrecognizedEscapeSequence.
+  t.throws(function() { ohm.grammar('G { r = "\\w" }'); }, /Unrecognized escape sequence/);
+  t.throws(function() { ohm.grammar('G { r = "\\!" }'); }, /Unrecognized escape sequence/);
+  t.throws(function() { ohm.grammar('G { r = "\\~" }'); }, /Unrecognized escape sequence/);
+
+  // Non-printable chars or " " after "\" are just a regular parse error.
+  t.throws(function() { ohm.grammar('G { r = "\\\0" }'); }, /Failed to parse grammar/);
+  t.throws(function() { ohm.grammar('G { r = "\\\n" }'); }, /Failed to parse grammar/);
+  t.throws(function() { ohm.grammar('G { r = "\\\t" }'); }, /Failed to parse grammar/);
+  t.throws(function() { ohm.grammar('G { r = "\\ " }'); }, /Failed to parse grammar/);
 
-    it('semantic actions', function() {
-      var s = m.semantics().addAttribute('v', {});
-      var cst = m.match('foo\b\n\r\t\\"\u01bcff\x8f');
-      t.equal(s(cst).v, 'foo\b\n\r\t\\"\u01bcff\x8f');
-    });
-    t.end();
-  });
   t.end();
 });
 

test/test-errors.js

@@ -89,7 +89,7 @@ test('many expressions with nullable operands', function(t) {
     t.equal(e.message, [
       'Line 1, col 14:',
       '> 1 | G { start = ("a"?)*}',
-      '                   ^',
+      '                   ^~~~',
       'Nullable expression "a"? is not allowed inside \'*\' (possible infinite loop)'].join('\n'));
   }
 
@@ -100,7 +100,7 @@ test('many expressions with nullable operands', function(t) {
     t.equal(e.message, [
       'Line 1, col 14:',
       '> 1 | G { start = ("a"?)+}',
-      '                   ^',
+      '                   ^~~~',
       'Nullable expression "a"? is not allowed inside \'+\' (possible infinite loop)'].join('\n'));
   }
 
@@ -142,3 +142,19 @@ test('errors from makeGrammar()', function(t) {
 
   t.end();
 });
+
+test('unrecognized escape sequences', function(t) {
+  function getExceptionMessage(ruleBody) {
+    try {
+      ohm.grammar('G { r = ' + ruleBody + ' }');
+      t.fail('Expected an exception to be thrown');
+    } catch (e) {
+      return e.message;
+    }
+  }
+
+  t.ok(getExceptionMessage('"\\!"').match(/Did you mean "\\\\!" or "!"?/), '"\\!"');
+  t.ok(getExceptionMessage('"\\w"').match(/Did you mean "\\\\w" or "w"?/), '"\\w"');
+
+  t.end();
+});