- Changed: Upgrade Parso to 2.0.11.
- Changed: Support Spark 3.x / Scala 2.12.x. (#59 Thanks @thesuperzapper)
- Added: Support for externally compressed sas7bdat files. (#50 Thanks @thadeusb)
- Changed: Upgrade Parso to 2.0.10.
- Fixed: Incorrect reading of internally compressed SAS files. (See: #38 Thanks @thesuperzapper)
- Added: Extraction of SAS column labels as comments for Parquet/Hive.
- Added: Ability to force SAS column names to lowercase.
- Added: Schema inference for Long, Decimal, Int, Float, Short. (Based on SAS column formatters)
- Added: Ability to specify the maximum/minimum size of input splits.
- Added: Timeout for schema inference (handles corrupt files)
- Improved: Non-valid SAS files now throw errors on schema inference.
- Improved: User provided schema is now honoured, if conversion to the provided type is possible.
- Removed: Ability to provide jobConf in the implicit reader. (Instead use
SparkContext.hadoopConfiguration
)
- Changed: Upgrade Parso to 2.0.8. (Thanks @mulya, @Tagar, @printsev)
- Changed: Package is now licensed under Apache License 2.0.
- Changed: Package now compiles against Spark 2.0.1. (Thanks @chappers)
- Fixed: Remove dependency on GenericMutableRow so package works for 1.4 and above.
- Added: Add date datatype support and Date will now return as
java.sql.Date
. - Changed: Package now compiles against 1.5.0.
- Added: Add proper support for load via sqlContext.
- Fixed: Force javac version to 1.7 for better runtime compatibility.
- Fixed: Override parso sas7bdat parser to fill data page fully during read. This fixes exceptions thrown when reading data from AWS S3n.
- Changed: Migrate Hadoop 2.x (mapreduce) implementation to Hadoop 1.x (mapred) for better compatibility. As a result, the library is now runnable with all Hadoop versions.