Regression: String functions

[Snugug]

node-sass 3.0.0-beta which gives me access to Libsass 3.2.0-beta.2 and I've discovered a regression: In Libsass 3.1, This str-replace function works without a hitch. I believe it may be related to #987, but this particular function worked in 3.1 and no longer works in 3.2:

@function str-replace($string, $search, $replace: '') {
  $index: str-index($string, $search);

  @if $index {
    @return str-slice($string, 1, $index - 1) + $replace + str-replace(str-slice($string, $index + str-length($search)), $search, $replace);
  }

  @return $string;
}


$string: 'Foo Bar Baz Qux';

.foo {
  content: str-replace($string, ' ', '-');
}

Original Output

.foo {
  content: "Foo-Bar-Baz-Qux"; }

**Regressed Output in Libsass 3.2

.foo {
  content: "Foo"-Bar-Baz-Qux; }

How should I go about writing a test to catch this? It's not Sass function, but it's a regression that should be tested for.

[mgreter]

I already have this on my radar and a proper fix will need some more re-factoring, as it will cause another spec-test to fail otherwise. Altough I have a hotfix which I probably will include in 3.2.

in functions.cpp line 945

- return new (ctx.mem) String_Constant(pstate, newstr);
+ if (newstr == "\"\"") new (ctx.mem) String_Constant(pstate, newstr);
+ return new (ctx.mem) String_Quoted(pstate, newstr);

[Snugug]

@mgreter Great, thanks. That's exactly where it's choking, on \". I've got a URL Encode function I'm using and the error I'm getting happens when I'm trying to encode that bit.

[mgreter]

If you can, please test with PR #989 applied and report if it fixes your issues! Thanks!

[Snugug]

I'm not entirely sure if I'm able to do that on pure Libsass :\ I compile through node-sass and not entirely sure what the process is for testing Libsass updates through that.

[drewwells]

@mgreter pr #989 fixes this issue 👍

/* line 19, stdin */
.foo {
  content: "Foo-Bar-Baz-Qux"; }
Rebuilt: test.scss
2015/03/27 13:16:39 main.go:94: Compilation took: 8.654ms