Pysolr

-------------------

New
~~~

- Expose the full Solr response in `Results` [Chris Adams]

This makes life easier for anyone using custom extensions by
removing the need to create a `Results` subclass just to get
access to an extra dictionary key.

- More flexible control of request handlers. [nuarhu]

This allows configuring the default search handler and overriding it for every query method

Thanks to nuarhu for the patch

- Start maintaining a changelog from gitchangelog. [Chris Adams]

- Overwrite flag for Solr.add (closes 182) [Chris Adams]

Thanks to robinsonkwame for the patch

- SolrCloud support (see 138) [Chris Adams]

This optionally adds support for SolrCloud using the Kazoo client
library.

Thanks to upayavira

Other
~~~~~

- V3.5.0. [Chris Adams]

- Merge pull request 192 from dhruvpathak/optimize_commit_flag. [Chris
Adams]

chg: `optimize()` also accepts `commit` flag

- Included commit flag in optimize() to let optimize call run with or
without commit. [dhruv.pathak]

- Merge pull request 188 from TigorC/master. [Chris Adams]

Removed py26 from tox.ini

- Removed py26 from tox.ini. [Igor Tokarev]

- Tests: avoid timeout-based CI failures. [Chris Adams]

These caused sporadic CI build failures and weren’t
otherwise testing actual functionality since we don’t have a
test which does something like SIGSTOP the test Solr server
long enough to confirm a timeout.

We’ll confirm that the timeout is passed through but
otherwise use the defaults.

- Update Travis CI badge in the README. [Chris Adams]

- Merge pull request 184 from atuljangra/master. [Chris Adams]

Correct documentation for `_update`

Thanks to atuljangra for the patch!

- Merge branch 'master' of https://github.com/atuljangra/pysolr.
[atuljangra]

- Misleading comments. [atuljangra]

- Travis: use build matrix for regular and SolrCloud tests. [Chris
Adams]

- Test_cloud: remove dead code. [Chris Adams]

The first instance of test_custom_results_class was broken because it
used the wrong port but this wasn’t failing because the same method name
was redefined further down in the file and that used the updated port
config.

- PEP-8. [Chris Adams]

- ZooKeeper: log unexpected format changes to watched aliases. [Chris
Adams]

- ZooKeeper: restore JSON blob decoding. [Chris Adams]

- PEP-8. [Chris Adams]

- PEP-8 unused imports. [Chris Adams]

- PEP-8. [Chris Adams]

- PEP-8. [Chris Adams]

- PEP-8. [Chris Adams]

- Setup.cfg: add pep8 and isort config. [Chris Adams]

- Tear down requests.Session instance at close. [Chris Adams]

This avoids log-spew on modern unittest implementations
which report unclosed file handles at the end of a run.

- Remove Python 2.6 from Travis test matrix. [Chris Adams]

- Add __future__ absolute_import. [Chris Adams]

This is currently moot but avoids any chance of regression
between Python 2.x and 3.x.

- PEP-8. [Chris Adams]

- Drop support for Python 2.6. [Chris Adams]

We have some old import dances and other overhead for Python
2.6 support, which the CPython developers dropped support
for in 2013:

http://www.curiousefficiency.org/posts/2015/04/stop-supporting-python26.html

- Allow queries to be directed to different search handlers. [Chris
Adams]

The `search` method now allows you override the default `select` handler
when your Solr instance has multiple search handlers.

Thanks to k-patel for the patch.

Closes 174
Closes 175

-------------------

- Update version numbers for v3.4.0. [Chris Adams]

- Logging: better message for HTTP status != 200. [Chris Adams]

We already extract error message from Solr responses and that is
great. Unfortunately it can contain the data that may change with
every request (like document id).

This creates an issue when user uses Sentry or other solution
that captures logging or exceptions. Previous implementation
causes many duplicated events in Sentry if message extracted
using `self._extract_error(resp)` contained such variable data.

This change uses 'non-mutable' message that is complemented
with extracted data that using string formatting option supplied
by Python logging. Thanks to this, Sentry and other solutions
can perform better grouping of logging messages (by status code).

This is approach that is already used in handling other errors.

- Fix response error handling on Python 3 (closes 162) [Chris Adams]

Previously the error handling did not work correctly on Python 3 because
a byte-string response wasn't decoded before processing.

Thanks to Emmanuel Leblond (touilleMan) for the patch.

- Merge pull request 167 from swistakm/master. [Chris Adams]

Refactor common response processing to Results class

- Move response manipulation responsibility to Results class and allow
custom results classes. [mjaworski]

- Add Python 3.5 to automated test matrix. [Chris Adams]

-------------------

- V3.3.3. [Chris Adams]

- Fix response error handling on Python 3 (closes 162) [Chris Adams]

Previously the error handling did not work correctly on Python 3 because
a byte-string response wasn't decoded before processing.

Thanks to Emmanuel Leblond (touilleMan) for the patch.

- Tests: upgrade Solr to 4.10.4. [Chris Adams]

* Resync test Solar script with django-haystack
These are still not quite the same; at some point it would be nice to
look into a common tool which both projects could use
* Update Solr configuration script to set correct libpath for solr-cell
to avoid lazy-load failures during testing as was reported on e.g. 162

- Tests: update Solr download script for recent API change. [Chris
Adams]

- Merge pull request 142 from yspanchal/master. [Chris Adams]

Add support for cursormark pagination

- Added cursormark deep pagination support. [Yogesh Panchal]

-------------------

- Version 3.3.2. [Chris Adams]

- Python 2.6 backwards compatibility. [Chris Adams]

Python 2.6 shipped with ElementTree 1.2.x. Among other differences, it
lacks support for the attribute selectors used to process valid XML
error messages, which was added in ElementTree 1.3.

- Merge pull request 155 from domenkozar/solr4/exceptions. [Chris
Adams]

Support Solr 4 XML error format parsing

Thanks domenkozar for the patch

- Overhaul Travis config. [Chris Adams]

* Sidestep use of Tox in favor of Travis-managed Python versions
* Enable container-based builds
* Enable caching for Solr server downloads

- Use builtin unittest2 runner on Python 2.7 as well. [Chris Adams]

- Simple error extraction. [Chris Adams]

Previously pysolr depended on lxml and cssselect to extract
text from Tomcat’s error messages, which was unreliable.

This change uses regular expressions to deal with invalid
XML rather than lxml’s salvaging parser and avoids having
to maintain the code which attempted to find the main error
message in tag soup

Closes 149

- Update test Solr download script to work with default Python 3. [Chris
Adams]

-------------------

- Version 3.3.1. [Chris Adams]

- Prepare for 3.3.1 release. [Chris Adams]

- Convert all HTTP client errors to SolrError. [Chris Adams]

This commit ensures that an outside caller can handle all HTTP-related errors by catching SolrError without knowing whether the exception class is owned by requests, urllib3, or httplib.

- Merge pull request 146 from gryphius/fix_doc_typo. [Chris Adams]

Fix typo in ExtractingRequestHandler documentation

Thanks gryphius

- Doc fix: a very simply model -> a very simple model. [Oli]

- Merge pull request 139 from upayavira/feature/no-optimize. [Daniel
Lindsley]

Optimize is no longer recommended

- Optimize is no longer recommended. [Upayavira]

Since Solr 3.6, Solr has used the TieredMergePolicy which makes,
in most scenarios, optimization a harmful rather than beneficial
step.

-------------------

- Bumped to v3.3.0! [Daniel Lindsley]

- Added acdha to primaries for all his hard work. [Daniel Lindsley]

- Support Solr 4+ individual field updates (closes 129) [Chris Adams]

Now fields can be updated individually:

conn.add(docs, fieldUpdates={'myfield1_ss': 'add',
'myfield2_s': 'set',
'myfield3_i': 'inc'})

Thanks to Çağatay Çallı (faraday) for the patch.

- Merge pull request 137 from LuRsT/patch-1. [Chris Adams]

Fixed syntax error in README.rst example (thanks LuRsT)

- Fixed syntax error in README.rst example. [Gil Gonçalves]

- Add softCommit support (closes 98) [Chris Adams]

add() and commit() may now be called with softCommit=True

Thanks to sicarrots for the patch

- Merge pull request 123 from ulivedit/master. [Chris Adams]

Python 3 compatibility for error message extraction (thanks ulivedit)

- Fix python 3.4 error with forcing unicode strings. [Eric Hagman]

- Merge pull request 135 from Grokzen/master. [Chris Adams]

Use DEBUG_PYSOLR environmental variable to configure logging

This offers an alternative to editing pysolr.py or reconfiguring logging elsewhere

- Make it easier to debug pysolr via environment variable. [Johan
Andersson]

- Merge pull request 131 from andreif/highlighted-readme. [Chris Adams]

Highlight Python code in README.rst (thanks andreif)

- Highlight Python code in README.rst. [Andrei Fokau]

- Add support for error responses in JSON format (closes 113) [Chris
Adams]

Thanks to andreif for the patch and tests

- Merge pull request 125 from phill-tornroth/patch-1. [Chris Adams]

Fix get-solr-download-url.py for Python 2.6

- Fixes 'zero field length' error from `format()` [Phill Tornroth]

Unless I'm missing something... :)

- Travis: download Solr before starting tests. [Chris Adams]

This should avoid download errors being presented as test failures

- Tests: increase Solr startup timeout. [Chris Adams]

- Add test Solr tarball downloads to .gitignore. [Chris Adams]

- Tests: add Python 3.4 targets. [Chris Adams]

- Tests: use Solr 4.7.2 from nearest mirror (closes 115) [Chris Adams]

- Tests: add a script to retrieve the closest Apache mirror. [Chris
Adams]

See 115

- Merge pull request 111 from redjack/py26-tests. [Chris Adams]

Update 'run-tests.py' to invoke unittest2 correctly on Python 2.6

- Update 'run-tests.py' to invoke unittest2 correctly on Python 2.6.
[Andy Freeland]

- Expanded testing section of the README. [Chris Adams]

- Merge pull request 36 from glenbot/master. [Chris Adams]

Update to SolrCoreAdmin.create to use correct action

- Updated create command in SolrCoreAdmin to use correct action.
[glenbot]

- Fix type in SolrAdmin.create default parameter. [Chris Adams]

See 36

- Updated ignores. [Daniel Lindsley]