Implement support for specifying string interopolation style variable substitution in the import ...

Implement support for specifying string interopolation style variable substitution in the import and source statements.

e.g:
sivamach[runtime]>more /workspace/asiva/expr/test1.dart
#library('test1.dart');
#import('${GOOGLE}$SRC/test2.dart');
#import('$DART/test3.dart');
test1() {
  print("test1");
}
main() {
  test1();
  test2();
  test3();
}
sivamach[runtime]>more /workspace/asiva/expr/src/test2.dart
#library('test2.dart');
#source('$GOOGLE$SRC/test2a.dart');
test2() {
  print("test2");
  test2a();
}
sivamach[runtime]>more /workspace/asiva/expr/src/test2a.dart
test2a() {
  print("test2a");
}
sivamach[runtime]>more /workspace/asiva/expr/src/test3.dart
#library('test3.dart');
#source('test3a.dart');
test3() {
  print("test3");
  test3a();
}
sivamach[runtime]>more /workspace/asiva/expr/src/test3a.dart
test3a() {
  print("test3a");
}

sivamach[runtime]>out/Debug_ia32/dart --import_map=GOOGLE,/workspace/asiva/expr --import_map=SRC,src --import_map=DART,/workspace/asiva/expr/src /workspace/asiva/expr/test1.dart
test1
test2
test2a
test3
test3a
Committed: https://code.google.com/p/dart/source/detail?r=4199

[hausner, 2012-02-08 18:28:46]

LGTM with comments.

http://codereview.chromium.org/9359009/diff/2024/vm/object.cc
File vm/object.cc (right):

http://codereview.chromium.org/9359009/diff/2024/vm/object.cc#newcode4569
vm/object.cc:4569: result.raw_ptr()->import_map_ = Array::Empty();
I don't see that this field is written to the snapshot. Is that intentional? If
so you still have to initialize it when the snapshot code reads the library in
Library::ReadFrom(), right?

http://codereview.chromium.org/9359009/diff/2024/vm/parser.cc
File vm/parser.cc (right):

http://codereview.chromium.org/9359009/diff/2024/vm/parser.cc#newcode7411
vm/parser.cc:7411: if (CurrentToken() == Token::kIDENT) {
I would check for IsIdentifier(). I guarantee you someone will insist that they
can't live without using pseudo keywords as import variable names.

http://codereview.chromium.org/9359009/diff/2024/vm/parser.cc#newcode7415
vm/parser.cc:7415: ExpectToken(Token::kINTERPOL_END);
I would add a more specific error message here, like:

if (CurrentToken() != Token::kINTERPOL_END) {
  ErrorMsg("'}' expected");
}

Your version would print: '' expected

http://codereview.chromium.org/9359009/diff/2024/vm/parser.cc#newcode7417
vm/parser.cc:7417: ErrorMsg("'%s' expected", Token::Str(Token::kIDENT));
How about just "idenitfier expected" or "import variable name expected"?

http://codereview.chromium.org/9359009/diff/2024/vm/parser.cc#newcode7424
vm/parser.cc:7424: ErrorMsg("'%s' expected", Token::Str(Token::kINTERPOL_VAR));
This else path is impossible due to the guard in the while statement.

[siva, 2012-02-08 23:23:47]

Thanks, addressed review comments.

I also uploaded the test case I had above into the tests/vm directory.

http://codereview.chromium.org/9359009/diff/2024/vm/object.cc
File vm/object.cc (right):

http://codereview.chromium.org/9359009/diff/2024/vm/object.cc#newcode4569
vm/object.cc:4569: result.raw_ptr()->import_map_ = Array::Empty();
On 2012/02/08 18:28:47, hausner wrote:
> I don't see that this field is written to the snapshot. Is that intentional?
If
> so you still have to initialize it when the snapshot code reads the library in
> Library::ReadFrom(), right? 


The field is written out in a snapshot (all object fields between from_ and to_
in the declaration are written out).

http://codereview.chromium.org/9359009/diff/2024/vm/parser.cc
File vm/parser.cc (right):

http://codereview.chromium.org/9359009/diff/2024/vm/parser.cc#newcode7411
vm/parser.cc:7411: if (CurrentToken() == Token::kIDENT) {
On 2012/02/08 18:28:47, hausner wrote:
> I would check for IsIdentifier(). I guarantee you someone will insist that
they
> can't live without using pseudo keywords as import variable names.

Nice :-)

I have changed it to IsIdentifier();

http://codereview.chromium.org/9359009/diff/2024/vm/parser.cc#newcode7415
vm/parser.cc:7415: ExpectToken(Token::kINTERPOL_END);
On 2012/02/08 18:28:47, hausner wrote:
> I would add a more specific error message here, like:
> 
> if (CurrentToken() != Token::kINTERPOL_END) {
>   ErrorMsg("'}' expected");
> }
> 
> Your version would print: '' expected

Done.

http://codereview.chromium.org/9359009/diff/2024/vm/parser.cc#newcode7417
vm/parser.cc:7417: ErrorMsg("'%s' expected", Token::Str(Token::kIDENT));
On 2012/02/08 18:28:47, hausner wrote:
> How about just "idenitfier expected" or "import variable name expected"?

Done.

http://codereview.chromium.org/9359009/diff/2024/vm/parser.cc#newcode7424
vm/parser.cc:7424: ErrorMsg("'%s' expected", Token::Str(Token::kINTERPOL_VAR));
On 2012/02/08 18:28:47, hausner wrote:
> This else path is impossible due to the guard in the while statement.

Good point, removed.