Julien Buret created SPARK-11145: ------------------------------------ Summary: Cannot filter using a partition key and another column Key: SPARK-11145 URL: https://issues.apache.org/jira/browse/SPARK-11145 Project: Spark Issue Type: Bug Components: PySpark, SQL Affects Versions: 1.5.1 Reporter: Julien Buret
A Dataframe, loaded from partitionned parquet files, cannot be filtered by a predicate comparing a partition key and another column. In this case all records are returned Example {code} from pyspark.sql import SQLContext sqlContext = SQLContext(sc) d = [ {'name': 'a', 'YEAR': 2015, 'year_2': 2014, 'statut': 'a'}, {'name': 'b', 'YEAR': 2014, 'year_2': 2014, 'statut': 'a'}, {'name': 'c', 'YEAR': 2013, 'year_2': 2011, 'statut': 'a'}, {'name': 'd', 'YEAR': 2014, 'year_2': 2013, 'statut': 'a'}, {'name': 'e', 'YEAR': 2016, 'year_2': 2017, 'statut': 'p'} ] rdd = sc.parallelize(d) df = sqlContext.createDataFrame(rdd) df.write.partitionBy('YEAR').mode('overwrite').parquet('data') df2 = sqlContext.read.parquet('data') df2.filter(df2.YEAR == df2.year_2).show() {code} return {code} +----+------+------+----+ |name|statut|year_2|YEAR| +----+------+------+----+ | d| a| 2013|2014| | b| a| 2014|2014| | c| a| 2011|2013| | e| p| 2017|2016| | a| a| 2014|2015| +----+------+------+----+ {code} -- This message was sent by Atlassian JIRA (v6.3.4#6332) --------------------------------------------------------------------- To unsubscribe, e-mail: issues-unsubscr...@spark.apache.org For additional commands, e-mail: issues-h...@spark.apache.org