Change Log

0.1.23

Fixed deprecate 'in' operator to be compatible with pandas 1.2.0 onwards (PR #58)
Add argument to append() to control duplicates (PR #57)

0.1.22

Uses PYSTORE_PATH environment variable, if it exists, as the path when not calling store.set_path() (defaults to ~/pystore)

0.1.21

Updated PyPi install script (lib is the same as 0.1.20)

0.1.20

Fix: Resetting config._CLIENT to None

0.1.19

Fixed: Exposed set/get_partition_size and set/get_clients

0.1.18

Added support for dask.distributed via pystore.set_client(...)
Added store.item(...) for accessing single collection item directly (pull request #44)
Added store.set_partition_size(...) and store.get_partition_size(). Default is ~99MB.

0.1.17

Updated PyPi install script (lib is the same as 0.1.16)

0.1.16

Fixed npartition=None issues on .append()

0.1.15

Fixed append issues
Raising an error when trying to read invalid item
Fixed path issued (removed unnecessary os.path.join calls)

0.1.14

Auto-detection and handling of nano-second based data

0.1.13

collection.reload_items defaults to False
Default npartitions and chunksize are better optimized (~99MB/partition)
collection.apply() repartitions the dataframe based on new data size (~99MB/partition)
Option to specify the default engine for the store by specifying engine="fastparquet" or engine="pyarrow" (dafaults to fastparquet)
Solving fastparquet/numba issues when using Dask >= 2.2.0 by importing numba in __init__.py

0.1.12

Added reload_items (default True) to collection.write and collection.delete to explicitly re-read the collection's items' directory

0.1.11

Reversed list_snapshots() behaviour
Added collection.threaded_write(...) method
collection.items being updated using items.add() and an async/threaded directory read

0.1.10

Switched from dtype_str to str(dtype) (Pandas 0.25+ compatibility)
Implemented collection.items and collection.snapshots as @property to reduce initialization overhead
collection.items and collection.snapshots are now of type set()
Option to specify both npartitions and chunksize in collection.append()

0.1.9

Fixed issues #13 and #15

0.1.8

Added pystore.read-csv() to quickly load csv as dask dataframe, ready for storage

0.1.7

Using os.path.expanduser("~") to determine user's home directory
collection.write(...) accepts Dask dataframes

0.1.6

Misc improvements

0.1.5

Added support for Python 2.7

0.1.4

Added support for Python 3.7

0.1.3

Fixed support for nanosecond-level data

0.1.2

epochdate defaults to True when storing ns data
Switched to dtype_str instead of str(dtype)

0.1.1

Infer datetime format when converting to Pandas

0.1.0

Increased version to fix setup
Bugfixes

0.0.12

Switched path parsing to pathlib.Path to help with cross-platform compatibility
Minor code refactoring

0.0.11

Adding an index name when one is not available

0.0.10

Added pystore.delete_store(NAME), pystore.delete_stores(), and pystore.get_path()
Added Jupyter notebook example to Github repo
Minor code refactoring

0.0.9

Allowing _ and . in snapshot name

0.0.8

Changed license to Apache License, Version 2.0
Moduled seperated into files
Code refactoring

0.0.7

Added support for snapshots
collection.list_items() supports querying based on metadata
Some code refactoring

Exposing more methods
Path setting moved to pystore.set_path()
Store.collection() auto-creates collection
Updated readme to reflect changes
Minor code refactoring

0.0.5

Not converting datetimte to epoch by defaults (use epochdate=True to enable)
Using "snappy" compression by default
Metadata's "_updated" is now a YYYY-MM-DD HH:MM:SS.MS string

0.0.4

Can pass columns and filters to Item object
Faster append
Store.path is now public

0.0.3

Updated license version

0.0.2

Switched readme/changelog files from .md to .rst.

0.0.1

Initial release