Enable Boolean queries for BASE

[pkraker]

This PR enables all Boolean queries described in the BASE Interface Guide (see section 4 on page 13) in verbatim mode. The changes have been extensively tested, but further tests would be very useful. Examples include:

• 2019-ncov or sars-cov-2
• (cats and dogs) or (sun and moon)
• research -science
• "knowledge domain visualization" or ("knowledge map" and (research or science))

[chreman]

Thank you! In general this leads to a much better retrieval for complex queries, and allows a better fine tuning of search results.
The query parsing can be tricky, and may require more extensive documentation so that users can get the best out of it.
Examples:

• 2019-ncov or sars-cov-2: fails - here the problem is that - is understood as an operator (textus:2019-textus:ncov or textus:sars-textus:cov-textus:2)
• "2019-ncov" or "sars-cov-2": success
• "2019-ncov" and "sars-cov-2": success
• "2019-ncov" + "sars-cov-2": fails - can be solved, see below
• "2019-ncov"+"sars-cov-2": success (identical map as with and but different ID)

Likewise, (cats+dogs) or (sun+moon) == (cats and dogs) or (sun and moon) but (cats + dogs) or (sun + moon) fails.
A quick test showed that wrapping the + shorthand with optional space characters (changing the last matching group to \\s?\\+\\s?) leads to the same verbatim query.

Additionally, it may make sense to add to the documentation that negative keywords should be added with consecutive -, e.g. science -research -knowledge -theory, as something like science -(research or knowledge or theory) is not correctly parsed
(-> textus:science -textus:(textus:research or textus:knowledge or textus:theory)

[pkraker]

Thanks for the thorough review! I updated the regex to cover the problems you described. In addition, I added a preprocessing step that removes unnecessary pluses and minuses. The regex now also deals correctly with multiple parentheses and inverted commas (single and double).

Please check again. There is now a test file in /test that provides a number of test cases.

[chreman]

Thank you very much, I think this is a great improvement. I took liberty to add functionality to the test script, it now makes a comparison between actual and expected results and notifies of errors.
Please check that the specifications of expected results in the queries list are correct.
I already made at one change here for the following test case:
"Processing fails for a+b . Result: textus:a+textus:b Expected: textus:a textus:b"

[pkraker]

Thanks, I introduced several changes to the preprocessing to accommodate the queries that did not produce the expected result; as a results, pluses are now always replaced by spaces. I also made sure that phrases are not altered in the preprocessing step to enable the user to force searching for the exact term. Please check again!

[chreman]

Thank you for the expansion, all tests are now successful. In my view, this is good to go.