Source

whoosh / tests / test_writing.py

Author Commit Message Labels Comments Date
Matt Chaput
Added test for the behavior of writer(procs=0).
Matt Chaput
Fixed Field.spellable_words() to work with a list of pre-analyzed words. Fixes issue #372.
Matt Chaput
Reduced iterations in various tests to speed up test suite.
Matt Chaput
Added functionality to clear the existing index contents when writing. Fixes issue #279.
Matt Chaput
Converted tests to use py.test instead of nose.
Comments 1
Thomas Waldmann
white-space-only source cosmetics, details see below Changes were made by running a script that did the cleanups automatically: - no trailing blanks - exactly one linefeed at file end, see PEP8 - DOS line endings on .bat and .cmd files, unix line endings everywhere else
Matt Chaput
Fixed tests on Python 3.2. Added py2.5 and py3.2 to tox.ini.
Matt Chaput
Checking in changes to make Whoosh 3 codec pass tests. Added memory codec, replaced use of ramindex in BufferedWrtier.
Matt Chaput
Removed RamIndex and commented out BufferedWriter pending replacement.
Matt Chaput
Reorganized modules. Changed util module to a package. Moved several modules from support package to util package to return support to a place for 3rd party code. Moved and renamed several modules to the top level package.
Matt Chaput
Added support for writing multiple segments in MpWriter (like old MultiSegmentWriter). Cleaned up writer code. Replaced old multiproc module with contents of multiproc2.
Matt Chaput
Writer now assembles segment files into a single compound file by default. (Use myindex.writer(compound=False) to disable.) Removed uses of mmap other than with compound file. Fixed bugs in file cleanup, RamIndex.
Matt Chaput
Big refactoring to make filedb use a pluggable codec for writing and reading to disk. This is still unstable. Multiprocessing isn't done and it might not work with old indices.
Matt Chaput
- In process of replacing multiprocessing writers -- non-functional in this commit. - Min/max/total field lengths now stored in lengths file instead of segment. - Greatly simplified file indexing code. - Simplified testing objects. - Replaced filedb.pools with support.externalsort. - Decoupled segments from generations, in anticipation of future work. - Rewrote "separate spelling" code path. - Replaced LRU decorator with simpler "double barrel" implementation. - Added create_temp…
Thomas Waldmann
transformed all *.py files to lf lineends, remove trailing blanks, normalize EOF
Matt Chaput
SegmentWriter's implementation of Writer.add_field() wasn't passing keyword args up. Fixes issue #170.
Matt Chaput
Removed error and warning flags.
Matt Chaput
Merging bug fixes and Python 3 compatibility from mainline.
Branches
betterq
Vinay Sajip
Interim update. All tests pass on 2.7, on 3.2 1 skip and 3 failures.
Matt Chaput
Cleaned up Collector. Started to implement true matcher/block quality. Commented out PL2 and added DFree.
Branches
betterq
Matt Chaput
Work on TermInfo and better quality optimizations.
Branches
betterq
Matt Chaput
Converted tests to nose.
Matt Chaput
Minor change to speed up tests.
Matt Chaput
Fixed test to close its resources.
Matt Chaput
Fixed missing import for test_read_inline.
Matt Chaput
Matt Chaput
filewriting: Fixed reversed arguments in calls to getlen(). pools: Partially regressed to more accurate but still inexact guess at posting size for speed. structfile: Added guard for mmap error on 0 length files. test_writing: Added test for inlined posting lists.
Matt Chaput
Added test for multiple threads using BufferedWriter. Fixed bug in Reading.first_id().
Matt Chaput
Replaced writing.BatchWriter with BufferedWriter, which uses a two-tiered index schema to allow quasi-real-time availability of indexed documents. Reader.is_atomic is now a method instead of an attribute.
Matt Chaput
Reconfigured tests to use temp dir context managers so they won't leave dirs behind. SegmentReader.close() wasn't closing vector postings file.
  1. Prev
  2. 1
  3. 2
  4. Next