If you pull this request, then the Stemmer will work with Seq[String], rather than merely String. This allows easy (easier?) use with Tokenizer and other transformers that work with arrays. I also updated to Spark 2.2 though it seemed to compile fine with the original Spark version as well.
This PR fixes an issue seen when using with Array(StringType): java.lang.IllegalArgumentException: requirement failed: Input type must be string type but got ArrayType(StringType,true). See e.g. the discussion at https://spark-packages.org/package/master/spark-stemming
If you pull this request, then the Stemmer will work with Seq[String], rather than merely String. This allows easy (easier?) use with Tokenizer and other transformers that work with arrays. I also updated to Spark 2.2 though it seemed to compile fine with the original Spark version as well.
This PR fixes an issue seen when using with Array(StringType): java.lang.IllegalArgumentException: requirement failed: Input type must be string type but got ArrayType(StringType,true). See e.g. the discussion at https://spark-packages.org/package/master/spark-stemming