Currently parsing the query is done in multiple steps: Firstly, the query is split at the slashes. Following that, the parts are split at the colons. Eventually, any resulting tests are split again: first, at the slashes, then at the colons ...
There is a problem with this: the first stage correctly skips escaped delimiters, and translates them into delimiters. This is done under the assumption that we are only doing one parse step and want to have the unescaped string. However, when parsing tests, the missing escape character leaves a valid delimiter, thereby creating bogus output.
Probably, the whole query should be parsed in a single step and transformed into a Query object with all the right properties.
Currently parsing the query is done in multiple steps: Firstly, the query is split at the slashes. Following that, the parts are split at the colons. Eventually, any resulting tests are split again: first, at the slashes, then at the colons ...
There is a problem with this: the first stage correctly skips escaped delimiters, and translates them into delimiters. This is done under the assumption that we are only doing one parse step and want to have the unescaped string. However, when parsing tests, the missing escape character leaves a valid delimiter, thereby creating bogus output.
Probably, the whole query should be parsed in a single step and transformed into a Query object with all the right properties.