Escaped chars not translated to ELM correctly? about clinical_quality_language

[brynrhodes]

So, this was quite the rabbit-hole :) Yes, the CQL-to-ELM translator was not correctly dealing with escape sequences in strings and identifiers, so I added logic to escape/unescape CQL literals and identifiers, and a unit test to verify it. Along the way, I found an issue with the CQL grammar itself, the way strings and identifiers are defined slightly too greedily, which results in an edge case with escape characters, specifically the backslash escape (\\) isn't handled properly. Because that's a specification change, I can't fix that until the grammar changes in the spec, so I've submitted an STU comment to address that, and until then, I've commented out the affected tests in the unit test.

However, having fixed all that and verified that escape sequences are being handled correctly for all possible escape sequences in both strings and identifiers, the output in the XML is still Hello \'World\' instead of the simpler Hello 'World'. In the JSON, the value is the simpler, but apparently, the escape logic for XML escapes single-quotes.

So I'm committing the fix I have so far, but leaving this issue open until the full fix can be applied based on the spec change.

[cmoesel]

That's a reasonable approach. Thanks for looking into this. If it works in JSON, I'm happy for now!

[brynrhodes]

Submitted CQL STU comment #1317 to address the STRING and QUOTEDIDENTIFIER lexical rules. (A similar fix was applied to the FHIRPath specification).

[brynrhodes]

Released in v1.1.1-SNAPSHOT