在Spark Java API中联接行数据集

Question

I want two join two DataSets DS1 and DS2 to get DS3

DS1 :

+---------+--------------------+-----------+------------+
|Compte   |         Lib        |ReportDebit|ReportCredit|
+---------+--------------------+-----------+------------+
|   447105|Autres impôts, ta...|    77171.0|         0.0|
|   753000|Jetons de présenc...|     6839.0|         0.0|
|   511107|Valeurs à l’encai...|        0.0|     77171.0|
+---------+--------------------+-----------+------------+

DS2:

+---------+------------+
|Compte   |SoldeBalance|
+---------+------------+
| 447105  |      992.13|
| 111111  |     35065.0|

I want to get DS3 like this:

+---------+--------------------+-----------+------------+------------+
|Compte   |           CompteLib|ReportDebit|ReportCredit|SoldeBalance|
+---------+--------------------+-----------+------------+------------+
|   447105|Autres impôts, ta...|    77171.0|         0.0|      992.13|
|   753000|Jetons de présenc...|    6839.0 |         0.0|         0.0|
|   511107|Valeurs à l’encai...|        0.0|     77171.0|         0.0|
    111111|                    |        0.0|         0.0|     35065.0|
+---------+--------------------+-----------+------------+------------+

Can somebody guide me with a sample Spark Java expression. Thanks in advance.

Answer 1

You can achieve this by applying a full outer join and then replacing null values with the desired values.

import static org.apache.spark.sql.functions.*;

...

ds1.join(ds2, ds1.col("Compte").equalTo(ds2.col("Compte")), "full_outer")
                .select(ds1.col("Compte").alias("Compte1"),
                        ds2.col("Compte").alias("Compte2"),
                        ds1.col("Lib"),
                        ds1.col("ReportDebit"),
                        ds1.col("ReportCredit"),
                        ds2.col("SoldeBalance"))
                .withColumn("Compte", when(col("Compte1").isNull(), col("Compte2")).otherwise(col("Compte1")))
                .drop("Compte1", "Compte2")
                .na().fill(0.0, new String[] { "ReportDebit", "ReportCredit", "SoldeBalance" })
                .na().fill("", new String[] { "Lib" })
                .show();

Output:

+--------------------+-----------+------------+------------+------+
|                 Lib|ReportDebit|ReportCredit|SoldeBalance|Compte|
+--------------------+-----------+------------+------------+------+
|Valeurs à l’encai...|        0.0|     77171.0|         0.0|511107|
|Autres impôts, ta...|    77171.0|         0.0|      992.13|447105|
|                    |        0.0|         0.0|     35065.0|111111|
|Jetons de présenc...|     6839.0|         0.0|         0.0|753000|
+--------------------+-----------+------------+------------+------+