Pysolr

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

- Update version to 3.0.2. [Chris Adams]

- Fix rich content extraction method & tests. [Chris Adams]

* Update test setup instructions with content extraction handler
dependencies
* Enable file upload support to _send_request
* Added simple extract test

- Fix field boosting, simplify _build_doc. [Chris Adams]

* Ensure that numbers are converted to strings to avoid
lxml choking when asked to serialize a number (in 2013!).
* Refactor logic to have a single code-path for both single and
multi-value fields
* Refactor use **kwargs style so there's a single Element() create
call

- Force_unicode support for non-string types. [Chris Adams]

Now force_unicode(1.0) will return u"1.0" for consistency and to avoid confusion
with the Django function of the same name

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

- Bumped to v3.0.1! [Daniel Lindsley]

- Updated README to include testing info & made sure the README gets
included n the package. [Daniel Lindsley]

- Updated ignores. [Daniel Lindsley]

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

- Bumped to v3.0.0, adding Python3 support! [Daniel Lindsley]

Dependencies have changed & been slimmed down.

- Bumped to v2.1.0! [Daniel Lindsley]

- Catch socket errors for httplib fallback path. [Chris Adams]

- Catch IOError in _send_request. [Chris Adams]

httplib2 can raise a bare socket.error in _send_request, which handles only
AttributeError. This change catches all IOError subclasses, tells logging to
include exception information and moves logging code outside of the try/except
block to avoid any possibility of an exception in a log handler being caught by
mistake.

- Fall back to HTML title when scraping error messages. [Chris Adams]

Solr 3.6 + Jetty is not reliably detected by the existing approach but it does
return a reasonably useful message in the title which is a lot more informative
than "None"

- Provide full headers & response to logging handlers. [Chris Adams]

This allows handlers such as Raven / Sentry to do something smart
with the full HTTP headers and/or response body. Among other things
this should provide more insight in situations when pysolr currently
logs "Response: None"

- Full exception logging for basic connection failures. [Chris Adams]

- Logging: use obvious exc_info= syntax. [Chris Adams]

As per the documentation, logging exc_info just needs to evaluate to
True. This change makes it obvious that the passed in value is not
actually used in any other way

- Added gthb to AUTHORS. [Daniel Lindsley]

- PEP-8 nitpicks. [Chris Adams]

- Don't bork on response with no "response" attr. [Gunnlaugur Þór Briem]

(happens e.g. in grouped queries)

- Support 'grouped' in Solr results. [Gunnlaugur Þór Briem]

- Added ``extra_requires`` to cover the ``BeautifulSoup`` dependency.
Thanks to kylemacfarlane for the report! [Daniel Lindsley]

- Added pabluk to AUTHORS. [Daniel Lindsley]

- Updated README file with optional requirement. [Pablo SEMINARIO]

- Added kwargs to extract() method. [Pablo SEMINARIO]

- Avoid forcing string interpolation when logging. [Chris Adams]

This allows aggregators like Sentry and other consumers to see the raw,
unformatted string and variables so they can e.g. group all instances of the
same message even if the specific request values differ.

- Added HTTPS support for httplib. [Richard Mitchell]

- Added a long description for PyPI. [Daniel Lindsley]

- Added support for Solr rich-content extraction. [Chris Adams]

This exposes Solr's http://wiki.apache.org/solr/ExtractingRequestHandler which
allows you to index text content from structured file formats like PDF,
Microsoft Office, etc.

- Bumped for the next round of beta. [Daniel Lindsley]

- Added cordmata to AUTHORS. [Daniel Lindsley]

- Updated suggest_terms so that it correctly handles response from Solr
3.x releases. [Matt Cordial]

- Edited README via GitHub. [Daniel Lindsley]

- Bumped to v2.0.15! [Daniel Lindsley]

- Fixed a bug where ``server_string`` could come back as ``None``.
Thanks to croddy for the report! [Daniel Lindsley]

- Added dourvais & soypunk to AUTHORS. [Daniel Lindsley]

- Unescape html entities in error messages. [David Cramer]

- Added support for getting at the Solr querying debug data when using
search(). [Shawn Medero]

Passing ``debug=True`` as kwarg, the ``search()`` method will activate this property in the JSON results.

- Fixed bug, qtime wasn't set when it was 0. [Daniel Dourvaris]

- Added query time to results as attribute. [Daniel Dourvaris]

- Bumped revision for dev on the next release. [Daniel Lindsley]

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

- V2.0.14. [Daniel Lindsley]

- Always send commit if its not-null. [David Cramer]

- Add support for waitFlush and waitSearcher on update queries. Added
support for expungeDeletes on commit(). Added support for maxSegments
on optimize() [David Cramer]

- Ensure port is coerced to an integer as (at least some version of)
socket does not handle unicode ports nicely. [David Cramer]

- Add support for commitWithin on Solr.add. [David Cramer]

- Better compatibility with the latest revisions of lxml. Thanks to
ghostmob for pointing this out! [Daniel Lindsley]

- Fixed occasionally trying to call ``lower`` on ``None``. Thanks to
girasquid for the report & original patch! [Daniel Lindsley]

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

- Cleaned up how parameters are checked. Thanks to zyegfryed for the
patch. v2.0.13. [Daniel Lindsley]

- Fixed a bug in the weighting when given a string field that's
weighted. Thanks to akaihola for the report. [Daniel Lindsley]

- Fixed the case where the data being converted would be clean unicode.
Thanks to acdha for submitting another version of this patch. [Daniel
Lindsley]

- Fixed the long URL support to correctly deal with sequences. [Daniel
Lindsley]

- Fixed a bug where additional parameters could cause the URL to be
longer than 1024 even if the query is not. Thanks to zyegfryed for the
report & patch! [Daniel Lindsley]

- Boost values are now coerced into a string. Thanks to notanumber for
the patch! [Daniel Lindsley]

- All params are now safely encoded. Thanks to acdha for the patch!
[Daniel Lindsley]

- Added term suggestion. Requires Solr 1.4+. Thanks to acdha for the
patch! [Daniel Lindsley]

- If invalid characters are found, replace them. Thanks to stugots for
the report and fix. [Daniel Lindsley]

- Slicing ``None`` doesn't work. Make it a string... [Daniel Lindsley]

- Added basic logging support. Thanks to sjaday for the suggestion.
[Daniel Lindsley]

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

- Releasing version v2.0.12. [Daniel Lindsley]

- Added a more helpful message for the ever classic "'NoneType' object
has no attribute 'makefile'" error when providing an incorrect URL.
[Daniel Lindsley]

- Added better error support when using Tomcat. Thanks to bochecha for
the original patch. [Daniel Lindsley]

- Fixed a long-standing TODO, allowing commits to happen without a
second request. Thanks to lyblandin for finally chiding me into fixing
it. [Daniel Lindsley]

- Fixed a bug when sending long queries. Thanks to akaihola & gthb for
the report and patch. [Daniel Lindsley]

- Corrected a bug where Unicode character might not transmit correctly.
Thanks to anti-social for the initial patch. [Daniel Lindsley]

- Added field-based boost support. Thanks to notanumber for the patch.
[David Sauve]

- Better error messages are now provided when things go south. Thanks to
bochecha for the patch. [Daniel Lindsley]

- Added support for working with Solr cores. Thanks to james.colin.brady
for the original patch. [Daniel Lindsley]

- Fixed a bug where empty strings/``None`` would be erroneously sent.
Thanks to Chak for the patch. [Daniel Lindsley]

- Added support for the Stats component. Thanks to thomas.j.lee for the
original patch. [Daniel Lindsley]

- Fixed datetime/date handling to use ``isoformat`` instead of manually
constructing the string. Thanks to joegermuska for the suggestion.
[Daniel Lindsley]

- Added document boost support. Thanks to Tomasz.Wegrzanowski for the
patch. [Daniel Lindsley]

- Fixed pysolr to add documents explicitly using UTF-8. Thanks to jarek
& dekstop for the patch. [Daniel Lindsley]