Previously, we would conduct regular expression searches by
enumerating all values of a given term, manually regex-matching each
one against our search regular expression, remember all the term
values that matched our regular expression, then do a big Xapian
OR-query that matched any of those term values. In constructing this
OR-query, however, we would split each term value on space and add a
separate Xapian phrase search term for each resulting word. This
approach worked fine most of the time, beacuse when we index a term,
we index both each word in a term and the whole term by itself.
This word splitting produced false negatives in some matches, however,
because Xapian and the Mu-level word splitting code do word splitting
slightly differently and apply different transformations to the text
while splitting. (For example, Xapian transforms fancy Unicode
apostrophes to ASCII apostrophes.)
This patch avoids the problem by not word splitting when constructing
the big Xapian OR-query for finding the results of regular
expression matching.
Update all cc code using .clang-format; please do so as well for future PRs
etc.; emacs has a handy 'clang-format' mode to make this automatic.
For comparing old changes with git blame, we can disregard this one using
--ignore-rev
(see https://www.moxio.com/blog/43/ignoring-bulk-change-commits-with-git-blame )
- Move the lib/query/ stuff up a level into lib/
- Associate directly with the Query object
- Rework the Query object to be C++ rather than mixed with C
- Update all dependencies, tests
