1.1.8 - Thursday 24th July 2014 -----------------------------
- More rigorous unittests for parsing queries containing UTF-8.
- Fix another bug when generating CQL from HTML forms.
- toCQL() consistently returns unicode
- More unittests for web.queryFactory
- Fix another bug when generating CQL from HTML forms introduced in 1.1.5.
- Unittests added for web.queryFactory
- Fix bug when generating CQL from HTML forms introduced in 1.1.5.
- Fix bug in phrase extraction when generating CQL from HTML forms.
- Update package, download and dependency links to avoid DNS problem causing builds to fail for Cheshire3 and packages dependent on it.
- Improved handling of deleted records:
- Relevance score calculations no longer raise TypeError.
- SRU supplies a surrogate diagnostic for deleted results.
- OAI-PMH:
- Checks RecordStore for deletion support, and reports this in the Identify response.
- Includes an appropriate status in the header for deleted records.
- Don't mask errors due to a Logger being unavailable.
- Warn instead of error when a subconfig file is missing.
- Init additional tables used for linking in PostgreSQL base Stores.
- Avoid non-namespaced Records from inheriting the OAI-PMH namespace be prefixing the OAI-PMH namespace in responses.
cheshire3-unregister
command to unregister existing databases.
- Improved out-of-the-box indexing capabilities.
- Support for a number of common file formats. This is achieved by preParsing to XML where possible, and wrapping all formats in METS.
- HTML
- plain-text
- OpenDocument Format (LibreOffice, OpenOffice 3+)
- Office Open XML (Microsoft Office 2007+ - docx, pptx, xlsx etc.)
- Easily load and index data from an iRODS data grid.
- Attempts to create title index entries by default.
- Faster retrieval1 by compressing stored records using the lz4 algorithm to reduce read time from disk.
- Support for a number of common file formats. This is achieved by preParsing to XML where possible, and wrapping all formats in METS.
- Store low and high values for each Record when
sortStore
setting is given for an Index. This provides more intuitive results when ordering ResultSets. NLTK Integration enabling configuration of indexes for automatically extracted named entities. This feature can be enabled by installing cheshire3 with 'nlp' or 'textmining' extras, e.g.:
pip install cheshire3[nlp] >= 1.1.0
- Improved speed, readability and security of
sql
sub-package through use ofpsycopg2
. - Better support for custom OAI-PMH servers (available as part of 'web' extras).
- Fixed major bug with indexing on 64-bit platforms.
- Many more minor bug fixes.
New regression unittests:
- Workflows
- ResultSets
- ResultSetStores
- Loggers
- Indexes
For fuller details see the GitHub Issue Tracker <https://github.com/cheshire3/cheshire3/issues?milestone=8&state=closed>
- Usability improvements in the
cheshire3
interactive console.
- Fixed assumed end datetime for ranges in
DateRangeTokenizer
- Fixed
UnicodeDecodeError
when logging errors inBdbIndexStore
- Improved "Configuring Indexes" tutorial:
- Fixed incorrect information regarding
ProximityIndex
es. - Completed truncated section on
sortStore
setting.
- Fixed incorrect information regarding
- Replaced documentation in docs/ folder with Sphinx-based documentation.
Fixed typo in
index.SimpleIndex.construct_resultSetItem
rsitype -> rsiType
- Fixed ResultSet ordering by XPath
- Fixed IndexError when Workflows log a zero-length message
- Eventually fixed build bugs when discovering version number in setup.py Read in version from VERSION.txt instead of trying to import from package
python setup.py test
now works with Python 2.6
- Attempts to fix build bugs when discovering version number in setup.py
Fixed typo in cheshire3.resultSet:
ValueErorr -> ValueError
- Fixed mutable type default data argument to SimpleResultSet constructor
- Updated installations instructions in README.
- Added CHANGES file.
- Fixed bug in serialization of ResultSet class for storage in cheshire3.sql.resultSetStore.
- Updated download URL in package info.
- cheshireVersion reinstated for backward compatibility.
- Fixed missing import of cheshire3.exceptions in cheshire3.sql.resultSetStore.
- Fixed incorrect version number in package info which could break dependency version resolution.
- Fixed missing import of CONFIG_NS in cheshire3.web.transformer.
- Allowed all configured paths to be specified relative to user's home directory (i.e. by use of ~/).
- Added an implementation agnostic XMLSyntaxError to cheshire3.exceptions.
- Fixed permission error bug in
cheshire3-init
andcheshire3-register
when Cheshire3 was installed as root. Solution creates a.cheshire3-server
directory in the users home directory in which to create server-level config plugins, log files and persistent data stores.
- Standardized installation process. Installable from PyPI.
- Unittest suite for the majority of processing objects.
- Command-line UI
cheshire3-init
cheshire3-load
cheshire3-load
cheshire3-search
cheshire3-serve
Faster retrieval assuming reasonable processing power (>=2.5GHz) and non solid-state storage.↩