Commits

Author Commit Message Labels Comments Date
Vinay Sajip
Fixed int/long bug: Now all tests pass on 2.7 and 3.2, except for multipool which is skipped on 3.2 for now.
Vinay Sajip
Interim update. All tests pass on 2.7, on 3.2 1 skip and 3 failures.
Matt Chaput
Fickle change to using hashlib.md5 instead of cdb_hash.
Matt Chaput
Merging changes from mainline.
Branches
betterq
Matt Chaput
Fixed miscalculation of avg_field_length() in a submatcher. Fixed nonapplication of filter/mask in Results._load_docs(). Bumped version to 1.8.4. Fixes issue #140.
Matt Chaput
Added lower-level method to take query string and return parse tree.
Branches
dawg
Matt Chaput
Work on TermInfo and better quality optimizations.
Branches
betterq
Matt Chaput
Cleanups and docs.
Branches
dawg
Matt Chaput
Starting "better quality" (betterq) branch.
Branches
betterq
Matt Chaput
Bumped version number to 1.8.3. Added release notes.
Matt Chaput
Make sure a file has a fileno() method before trying to call it.
Matt Chaput
Instead of ignoring all exceptions, just ignore OSError.
Matt Chaput
Revert changes to read lock blocking.
Matt Chaput
Ignore the exception if we can't acquire the "read lock".
Matt Chaput
Switched hash file to new more flexible format, switched back to cdb_hash, added backwards compatibility. Fixes issue #131.
Matt Chaput
Fixed bugs in dawg: lastword should be None instead of "" so it compares properly with lists, and reduce() was losing final flags.
Branches
dawg
Matt Chaput
Rewrote test_extend_empty() without len(). It still doesn't show the problem in issue #140.
Matt Chaput
More work on spelling.
Branches
dawg
Matt Chaput
Continuing work on DAWG.
Branches
dawg
Matt Chaput
Merging changes from default branch.
Branches
dawg
Matt Chaput
Added the ability to pass multiple values in a list or tuple to a keyword in add_document().
Matt Chaput
Merging changes from default branch.
Branches
dawg
Matt Chaput
Made acquiring the reader lock non-blocking to allow readers to open in parallel, added docs on threading, locking, and versioning. While this does create the possibility (though slimmer than before) of a writer deleting a segment just as a reader is trying to open it, the performance difference, at least on Windows, is enormous. Fixes issue #132.
Matt Chaput
Added filter argument to and fixed performance problem in Searcher.more_like(). Split filter argument to Collector.search() into two arguments, allow and restrict, and check them separately. This is more efficient for large collections than e.g. making a filter with all but one document turned on. Added mask argument to Searcher.search() which corresponds to restrict on Collector.search(). Fixes issues #138, #139.
Matt Chaput
Added copying to the test for extending an empty result set. See issue #140.
Matt Chaput
Added test for extending an empty result set. See issue #140.
Matt Chaput
Was adding IDs to Collector.docset in two different places, and one of them wasn't adding offsets. Changed Collector.pull_matches() to take an offset argument, yield offset IDs, and add the offset IDs to the docset. This should fix issue #106.
Matt Chaput
Merging changes from default branch.
Branches
dawg
Matt Chaput
Merge with thomaswaldmann's documentation fork. Thank Thomas! See issue #137.
Matt Chaput
Removed unnecessary calls to _test_field().
  1. Prev
  2. Next