Parse using regex

current linewise approach is simple-but-slow and a structure more like the following could be good:

* extract all using regex match with look-ahead/behind (or, validate open/close and then extract)
* validate the individual snippets, again possibly with regex or linewise

not sure how significant an issue performance is yet / ever

Start by locating the current linewise parsing entry point in the Python tool and measure its performance on representative source files. Compare a regex-based extraction and validation approach against the existing behavior, including open/close handling; done means the approach is justified by benchmarks and preserves snippet validation.

Written by the indexing model from the issue text.