Fix incremental parsing of import combinators.

Fix incremental parsing of import combinators.

This bug was discovered by the incremental parse stress tester.

R=scheglov@google.com

Committed: https://code.google.com/p/dart/source/detail?r=41993

[Paul Berry, 2014-11-26 04:37:01]

paulberry@google.com changed reviewers:
+ scheglov@google.com

[scheglov, 2014-11-26 04:45:44]

LGTM

[Paul Berry, 2014-11-26 14:05:54]

Committed patchset #1 (id:1) manually as 41993 (presubmit successful).

[Brian Wilkerson, 2014-11-26 14:44:38]

brianwilkerson@google.com changed reviewers:
+ brianwilkerson@google.com

[Brian Wilkerson, 2014-11-26 14:44:38]

https://codereview.chromium.org/759183002/diff/1/pkg/analyzer/lib/src/generat...
File pkg/analyzer/lib/src/generated/parser.dart (right):

https://codereview.chromium.org/759183002/diff/1/pkg/analyzer/lib/src/generat...
pkg/analyzer/lib/src/generated/parser.dart:1252: return
_parser.parseCombinator();
Does this handle the case where one combinator is split into two? For example:

import 'foo.dart' show A;

import 'foo.dart' show B, hide A;

It looks like it would only parse one of the two combinators.

[Paul Berry, 2014-11-26 16:08:38]

https://codereview.chromium.org/759183002/diff/1/pkg/analyzer/lib/src/generat...
File pkg/analyzer/lib/src/generated/parser.dart (right):

https://codereview.chromium.org/759183002/diff/1/pkg/analyzer/lib/src/generat...
pkg/analyzer/lib/src/generated/parser.dart:1252: return
_parser.parseCombinator();
On 2014/11/26 14:44:38, Brian Wilkerson wrote:
> Does this handle the case where one combinator is split into two? For example:
> 
> import 'foo.dart' show A;
> 
> import 'foo.dart' show B, hide A;
> 
> It looks like it would only parse one of the two combinators.

Yes, it works correctly.  Note: I'm assuming you mean "import 'foo.dart' show B
hide A;".  ',' is only allowed between identifiers within a combinator, not
between combinators.

You are correct that the call to _parser.parseCombinator() only parses one of
the combinators, but that's not a problem, because when the result is returned
to IncrementalParser.reparse(), the first thing it does is to check that the new
node can replace the old node without gaining or losing tokens.  In this case,
it can't, because if "show A" is replaced by "show B", the tokens "hide A" will
be lost.  So it sets advanceToParent = true and tries again at the next level up
the parse tree.

Having said that, your example seems valuable to explicitly test.  I'll add that
test in a follow up CL.