summaryrefslogtreecommitdiff
path: root/happybase/table.py
Commit message (Collapse)AuthorAgeFilesLines
* Update versionadded in docstrings (#190)Sean Morgan2018-04-031-1/+1
|
* Rename to reverse, minor code fixesSean Morgan2017-01-301-10/+13
|
* Adding reversed scanner, and new compat mode 0.98Sean Morgan2017-01-301-3/+21
|
* fixed typo in docstring of method scan()OldPanda2016-09-141-1/+1
|
* Avoid thriftpy import hookWouter Bolsterlee2016-08-011-1/+1
| | | | | | ...since it's a global thing. Make importing more explicit by specifying a path (using pkg_resources) to the .thrift file and loading it explicitly.
* Use byte strings and print() function in example codeWouter Bolsterlee2016-07-281-8/+9
|
* Rewrite str_increment() as bytes_increment()Wouter Bolsterlee2016-07-271-2/+2
| | | | ...and make it work on both Python 3 and 2.
* Column families are byte stringsWouter Bolsterlee2016-07-271-2/+2
| | | | ...hence require splitting using a byte string.
* Clean up helpers to transform rows and cell valuesWouter Bolsterlee2016-07-271-11/+16
|
* Use list comprehension instead of calling map()Wouter Bolsterlee2016-07-271-5/+3
| | | | ...which behaves differently on Python 3.
* Use six.iteritems()Wouter Bolsterlee2016-07-271-1/+3
|
* Port to thriftpyWouter Bolsterlee2016-03-271-1/+2
|
* Avoid useless round-trip when using scan(limit=...)Wouter Bolsterlee2014-11-291-2/+2
| | | | See #73.
* Simplify scanner logic: always use scannerGetListWouter Bolsterlee2014-11-291-5/+2
|
* Correctly handle scanner that are smaller than expectedWouter Bolsterlee2014-11-241-5/+4
| | | | Fixes #72.
* Fix inversed wording in error message about 'sorted_columns'Wouter Bolsterlee2014-05-181-1/+1
|
* Add missing 'versionadded' for scan_batching arg in docsWouter Bolsterlee2014-02-251-0/+3
|
* Add argument sanity check for Table.scan(scan_batching=...)Wouter Bolsterlee2014-02-251-0/+3
|
* No longer confuse batching/caching; add Table.scan(scan_batching=...)Wouter Bolsterlee2014-02-251-11/+34
| | | | | For details, see the comments added in this commit, and issues #54 and issue #56.
* Revert "Allow batch_size=None in Table.scan() to avoid filter incompatibilities"Wouter Bolsterlee2014-02-251-20/+11
| | | | | This reverts commit 8481d317804e060b12839b571ef22306074fba9c, since it is not a correct fix. See issue #56.
* Use 'argument' instead of 'parameter' in docstringsWouter Bolsterlee2014-01-251-6/+6
|
* Allow batch_size=None in Table.scan() to avoid filter incompatibilitiesWouter Bolsterlee2014-01-251-11/+20
| | | | | | | | Allow None as a valid value for the batch_size argument to Table.scan(), since HBase does not support specifying a batch size when some scanner filters are used. Fixes issue #54.
* Add missing arg docstrings for Table.scanWouter Bolsterlee2013-11-171-0/+2
|
* Minor cleanupsWouter Bolsterlee2013-11-171-4/+3
|
* Mention correct 'versionadded' for sorted_columnsWouter Bolsterlee2013-11-171-1/+1
|
* Add support for retrieving sorted columns in HBase >= 0.96Wouter Bolsterlee2013-11-081-6/+40
| | | | | This is possible with the HBase 0.96 Thrift API. This feature uses a new 'sorted_columns' argument to Table.scan(). Fixes issue #39.
* Compatibility fixes for HBase 0.96Wouter Bolsterlee2013-11-081-2/+3
| | | | | | | | HBase 0.96.0 is more strict wrt column family names without a column qualifier. In previous versions a trailing ':' character after a column family name (e.g. 'cf1:') did not make any difference compared to the version without it (e.g. 'cf1'), but from 0.96.0 this is no longer the case...
* Pass batch_size to underlying Thrift Scan instanceWouter Bolsterlee2013-11-031-0/+1
| | | | Fixes issue #38.
* Expose Mutation.writeToWAL in .put(), .delete(), and batch()Wouter Bolsterlee2013-11-031-6/+27
| | | | | | | | Various data manipulation methods on Table and Batch instances now support a 'wal' argument to influence whether the mutation is written to the Write-Ahead Log. This is only supported in recent HBase version. Fixes issue #36.
* Always delegate deletion to the Batch classWouter Bolsterlee2013-11-031-9/+2
| | | | | | This eliminates the use of the .deleteAllRow() and .deleteAllRowTs() Thrift API, both of which do not support flags to influence the WAL behaviour.
* Adapt to API changes in regenerated Thrift APIWouter Bolsterlee2013-11-031-13/+13
| | | | | Simply pass an empty dictionary to each method that takes an 'attributes' arg in the regenerated Thrift API.
* Fix typo in docstringWouter Bolsterlee2013-06-031-1/+1
|
* Also allow 'long' type for 'timestamp' argsWouter Bolsterlee2013-05-221-3/+4
| | | | Fixes issue #23.
* A few minor PEP8 cleanupsWouter Bolsterlee2013-05-031-5/+10
|
* Always access Thrift client instance through connectionWouter Bolsterlee2013-05-031-25/+31
|
* Split code into separate modulesWouter Bolsterlee2013-05-021-0/+482
Split the monolithic .api module into separate modules implementing the connection, table, and batch parts. The public API is still the same, since all functionality is available from the main 'happybase' namespace, so this is just an internal cleanup.