D17162: Harmonize handling of underscore in query parser

[Stefan Brüns]

bruns created this revision.
bruns added reviewers: Baloo, Frameworks, ngraham, poboiko.
Herald added projects: Frameworks, Baloo.
Herald added a subscriber: kde-frameworks-devel.
bruns requested review of this revision.

REVISION SUMMARY
  The engine differentiates between independent terms, and adjacent terms
  forming a phrase. The latter is e.g. used for mail addresses and file
  names, "ice at fire.com" or "foo_bar.png". In the database these are stored
  without any punctuation characters, but with their relative positions.
  
  While most punctuation characters (e.g. ".", ",", "/", "-", ...) are
  treated the same, underscore was handled differently. A query for
  "foo-bar.png" searches for the three adjacent terms, while "foo_bar png"
  searches for the three terms in any order.
  
  Treat "_" like any other punctuation character. For this to work, the
  delimiter has to be checked from the original input string, not the
  text mangled for QTextBoundaryFinder.

TEST PLAN
  ctest

REPOSITORY
  R293 Baloo

BRANCH
  submit

REVISION DETAIL
  https://phabricator.kde.org/D17162

AFFECTED FILES
  autotests/unit/engine/queryparsertest.cpp
  src/engine/queryparser.cpp

To: bruns, #baloo, #frameworks, ngraham, poboiko
Cc: kde-frameworks-devel, ashaposhnikov, michaelh, astippich, spoorun, ngraham, bruns, abrahams
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mail.kde.org/pipermail/kde-frameworks-devel/attachments/20181125/c561789a/attachment-0001.html>