jsoniq 2.0.0 alpha 1

• When returning a single-column DataFrame with atomic values, the name is now __value and not value to avoid collisions with user-defined columns.
• Improved schema inferrence: DataFrames can be returned in a wider range of cases.
• Improved error display in notebooks when errors happen upon collecting the results and not already upon calling jsoniq().

jsoniq 0.2.0 alpha 9

• Stability improvements.

jsoniq 0.2.0 alpha 8

• Variables can now be bound to JSON values, pandas DataFrames or pyspark DataFrames with extra parameters to the rumble.jsoniq() call. It is no longer necessary to explicitly call bind(). This is similar to how DataFrames can be attached to views with extra parameters to spark.sql().
• rumble.lastResult is now correctly assigned also when partial data is returned (only with the partial data).
• Fixed issue with empty array constructors.

jsoniq 0.2.0 alpha 7

• rumble.lastResult now returns a pyspark/pandas DataFrame or rdd or tuple and no longer the sequence object.
• Enhance schema detection. When the detected static type of the overall query is DataFrame-compatible, it is now automatically possible to obtain the output as a DataFrame without explicitly giving a schema.
• It is now possible to access a table previously registered as a view via a table() function call. This is an alternative to binding variables.
• Enhancements in the JSONiq Update Facility support to update delta files and Hive metastore tables.

jsoniq 0.2.0 alpha 6

• Fix a bug with the config() call of the builder.
• add withDelta() to configure Delta Lake tables and files, for use with the JSONiq Update Facility.

jsoniq 0.2.0 alpha 5

• If the initialization of the Spark session fails, we now check if SPARK_HOME is set and if it may be invalid or pointing to a different Spark version than 4.0, and output a more informative error message.

jsoniq 0.2.0 alpha 4

• Added parameters to the jsoniq magic to select the desired output to print: -j, -df, -pdf
• Added informative error message with a hint on how to fix when trying to get a DataFrame and there is no schema.
• Added parameter -t to the jsoniq magic to measure the response time
• The RumbleSession object now saves the latest result (sequence of items) in a field called lastResult. This is particularly useful in notebooks for post-processing a result in Python after obtained it through the jsoniq magic.
• Improved static type detection upon binding a pandas or pyspark DataFrame as an input variable to a JSONiq queries.
• Now also accepts pandas version 2.2.

0.2.0a2: Merge pull request #9 from RumbleDB/JSONiqMagic

• You can change the result size cap through to the now accessible Rumble configuration (for example rumble .getRumbleConf().setResultSizeCap(10)). This controls how many items can be retrieved at most with a json() call. You can increase it to whichever number you would like if you reach the cap.
• Add the JSONiq magic to execute JSONiq queries directly in a notebook cell, using the RumbleDB instance shipped with the library.
• RumbleSession.builder.getOrCreate() now correctly reuses an existing session instead of creating a new object. It preserves the configuration.

jsoniq 0.2.0 alpha 1

• Allow to bind JSONiq variables to pandas dataframes
• Allow to retrieve the output of a JSONiq query as a pandas dataframe (if the output is available as a dataframe, i.e., availableOutputs() returns a list that contains "DataFrame")
• Clean up the mapping to strictly map tuples to sequence of items, and lists ot array items. This will avoid confusion between arrays and sequences.
• As a consequence, json() now returns a tuple, not a list.
• Calling bind() with a single list will return an informative error. Use bind() with a tuple instead, or call bindOne() to interpret the list as a sequence of one array item.

jsoniq 0.1.0 alpha 12

• Allow to bind JSONiq variables to Python values (mapping Python lists to sequences of items). This makes it possible to manipulate - Python values directly with JSONiq and even without any knowledge of Spark at all.
• renamed bindDataFrameAsVariable() to bind(), which can be used both with DataFrames and Python lists.
• add bindOne() for binding a single value to a JSONiq variable.
• wrapping df() in a Pyspark DataFrame to make sure it can be used with pyspark DataFrame transformations.