Merge multiple spark rows to one - sql

I have a dataframe which looks like one given below. All the values for a corresponding id is the same except for the mappingcol field.
+--------------------+----------------+--------------------+-------+
|misc |fruit |mappingcol |id |
+--------------------+----------------+--------------------+-------+
|ddd |apple |Map("name"->"Sameer"| 1 |
|ref |banana |Map("name"->"Riyazi"| 2 |
|ref |banana |Map("lname"->"Nikki"| 2 |
|ddd |apple |Map("lname"->"tenka"| 1 |
+--------------------+----------------+--------------------+-------+
I want to merge the rows with same row in such a way that I get exactly one row for one id and the value of mappingcol needs to be merged. The output should look like :
+--------------------+----------------+--------------------+-------+
|misc |fruit |mappingcol |id |
+--------------------+----------------+--------------------+-------+
|ddd |apple |Map("name"->"Sameer"| 1 |
|ref |banana |Map("name"->"Riyazi"| 2 |
+--------------------+----------------+--------------------+-------+
the value for mappingcol for id = 1 would be :
Map(
"name" -> "Sameer",
"lname" -> "tenka"
)
I know that maps can be merged using ++ operator, so thats not what im worried about. I just cant understand how to merge the rows, because if I use a groupBy, I have nothing to aggregate the rows on.

You can use by groupBy and then managing a little the map
df.groupBy("id", "fruit", "misc").agg(collect_list("mappingcol"))
.as[(Int, String, String, Seq[Map[String, String]])]
.map { case (id, fruit, misc, list) => (id, fruit, misc, list.reduce(_ ++ _)) }
.toDF("id", "fruit", "misc", "mappingColumn")
With the first line, tou group by your desired columns and aggregate the map pairs in the same element (an array)
With the second line (as), you convert your structure to a Dataset of a Tuple4 with the last element being a sequence of maps
With the third line (map), you merge all the elements to a single map
With the last line (toDF) to give the columns the original names
OUTPUT
+---+------+----+--------------------------------+
|id |fruit |misc|mappingColumn |
+---+------+----+--------------------------------+
|1 |apple |ddd |[name -> Sameer, lname -> tenka]|
|2 |banana|ref |[name -> Riyazi, lname -> Nikki]|
+---+------+----+--------------------------------+

You can definitely do the above with a Window function!
This is in PySpark not Scala but there's almost no difference when only using native Spark functions.
The below code only works on a map column that 1 one key, value pair per row, as it how your example data is, but it can be made to work with map columns with multiple entries.
from pyspark.sql import Window
map_col = 'mappingColumn'
group_cols = ['id', 'fruit', 'misc']
# or, a lazier way if you have a lot of columns to group on
cols = df.columns # save as list
group_cols_2 = cols.remove('mappingCol') # remove what you're not grouping by
w = Window.partitionBy(group_cols)
# unpack map value and key into a pair struct column
df1 = df.withColumn(map_col , F.struct(F.map_keys(map_col)[0], F.map_values(map_col)[0]))
# Collect all key values into an array of structs, here each row
# contains the map entries for all rows in the group/window
df1 = df1.withColumn(map_col , F.collect_list(map_col).over(w))
# drop duplicate values, as you only want one row per group
df1 = df1.dropDuplicates(group_cols)
# return the values for map type
df1 = df1.withColumn(map_col , F.map_from_entries(map_col))
You can save the output of each step to a new column to see how each step works, as I have done below.
from pyspark.sql import Window
map_col = 'mappingColumn'
group_cols = list('id', 'fruit', 'misc')
w = Window.partitionBy(group_cols)
df1 = df.withColumn('test', F.struct(F.map_keys(map_col)[0], F.map_values(map_col)[0]))
df1 = df1.withColumn('test1', F.collect_list('test').over(w))
df1 = df1.withColumn('test2', F.map_from_entries('test1'))
df1.show(truncate=False)
df1.printSchema()
df1 = df1.dropDuplicates(group_cols)

Related

Extract key value from dataframe in PySpark

I have the below dataframe which I have read from a JSON file.
1
2
3
4
{"todo":["wakeup", "shower"]}
{"todo":["brush", "eat"]}
{"todo":["read", "write"]}
{"todo":["sleep", "snooze"]}
I need my output to be as below Key and Value. How do I do this? Do I need to create a schema?
ID
todo
1
wakeup, shower
2
brush, eat
3
read, write
4
sleep, snooze
The key-value which you refer to is a struct. "keys" are struct field names, while "values" are field values.
What you want to do is called unpivoting. One of the ways to do it in PySpark is using stack. The following is a dynamic approach, where you don't need to provide existent column names.
Input dataframe:
df = spark.createDataFrame(
[((['wakeup', 'shower'],),(['brush', 'eat'],),(['read', 'write'],),(['sleep', 'snooze'],))],
'`1` struct<todo:array<string>>, `2` struct<todo:array<string>>, `3` struct<todo:array<string>>, `4` struct<todo:array<string>>')
Script:
to_melt = [f"\'{c}\', `{c}`.todo" for c in df.columns]
df = df.selectExpr(f"stack({len(to_melt)}, {','.join(to_melt)}) (ID, todo)")
df.show()
# +---+----------------+
# | ID| todo|
# +---+----------------+
# | 1|[wakeup, shower]|
# | 2| [brush, eat]|
# | 3| [read, write]|
# | 4| [sleep, snooze]|
# +---+----------------+
Use from_json to convert string to array. Explode to cascade each unique element to row.
data
df = spark.createDataFrame(
[(('{"todo":"[wakeup, shower]"}'),('{"todo":"[brush, eat]"}'),('{"todo":"[read, write]"}'),('{"todo":"[sleep, snooze]"}'))],
('value1','values2','value3','value4'))
code
new = (df.withColumn('todo', explode(flatten(array(*[map_values(from_json(x, "MAP<STRING,STRING>")) for x in df.columns])))) #From string to array to indivicual row
.withColumn('todo', translate('todo',"[]",'')#Remove corner brackets
) ).show(truncate=False)
outcome
+---------------------------+-----------------------+------------------------+--------------------------+--------------+
|value1 |values2 |value3 |value4 |todo |
+---------------------------+-----------------------+------------------------+--------------------------+--------------+
|{"todo":"[wakeup, shower]"}|{"todo":"[brush, eat]"}|{"todo":"[read, write]"}|{"todo":"[sleep, snooze]"}|wakeup, shower|
|{"todo":"[wakeup, shower]"}|{"todo":"[brush, eat]"}|{"todo":"[read, write]"}|{"todo":"[sleep, snooze]"}|brush, eat |
|{"todo":"[wakeup, shower]"}|{"todo":"[brush, eat]"}|{"todo":"[read, write]"}|{"todo":"[sleep, snooze]"}|read, write |
|{"todo":"[wakeup, shower]"}|{"todo":"[brush, eat]"}|{"todo":"[read, write]"}|{"todo":"[sleep, snooze]"}|sleep, snooze |
+---------------------------+-----------------------+------------------------+--------------------------+--------------+

How to add multiple column dynamically based on filter condition

I am trying to create multiple columns dynamically based on filter condition after comparing two data frame with below code
source_df
+---+-----+-----+----+
|key|val11|val12|date|
+---+-----+-----+-----+
|abc| 1.1| john|2-3-21
|def| 3.0| dani|2-2-21
+---+-----+-----+------
dest_df
+---+-----+-----+------+
|key|val11|val12|date |
+---+-----+-----+------
|abc| 2.1| jack|2-3-21|
|def| 3.0| dani|2-2-21|
-----------------------
columns= source_df.columns[1:]
joined_df=source_df\
.join(dest_df, 'key', 'full')
for column in columns:
column_name="difference_in_"+str(column)
report = joined_df\
.filter((source_df[column] != dest_df[column]))\
.withColumn(column_name, F.concat(F.lit('[src:'), source_df[column], F.lit(',dst:'),dest_df[column],F.lit(']')))
The output I expect is
#Expected
+---+-----------------+------------------+
|key| difference_in_val11| difference_in_val12 |
+---+-----------------+------------------+
|abc|[src:1.1,dst:2.1]|[src:john,dst:jack]|
+---+-----------------+-------------------+
I get only one column result
#Actual
+---+-----------------+-
|key| difference_in_val12 |
+---+-----------------+-|
|abc|[src:john,dst:jack]|
+---+-----------------+-
How to generate multiple columns based on filter condition dynamically?
Dataframes are immutable objects. Having said that, you need to create another dataframe using the one that got generated in the 1st iteration. Something like below -
from pyspark.sql import functions as F
columns= source_df.columns[1:]
joined_df=source_df\
.join(dest_df, 'key', 'full')
for column in columns:
if column != columns[-1]:
column_name="difference_in_"+str(column)
report = joined_df\
.filter((source_df[column] != dest_df[column]))\
.withColumn(column_name, F.concat(F.lit('[src:'), source_df[column], F.lit(',dst:'),dest_df[column],F.lit(']')))
else:
column_name="difference_in_"+str(column)
report1 = report.filter((source_df[column] != dest_df[column]))\
.withColumn(column_name, F.concat(F.lit('[src:'), source_df[column], F.lit(',dst:'),dest_df[column],F.lit(']')))
report1.show()
#report.show()
Output -
+---+-----+-----+-----+-----+-------------------+-------------------+
|key|val11|val12|val11|val12|difference_in_val11|difference_in_val12|
+---+-----+-----+-----+-----+-------------------+-------------------+
|abc| 1.1| john| 2.1| jack| [src:1.1,dst:2.1]|[src:john,dst:jack]|
+---+-----+-----+-----+-----+-------------------+-------------------+
You could also do this with a union of both dataframes and then collect list only if collect_set size is greater than 1 , this can avoid joining the dataframes:
from pyspark.sql import functions as F
cols = source_df.drop("key").columns
output = (source_df.withColumn("ref",F.lit("src:"))
.unionByName(dest_df.withColumn("ref",F.lit("dst:"))).groupBy("key")
.agg(*[F.when(F.size(F.collect_set(i))>1,F.collect_list(F.concat("ref",i))).alias(i)
for i in cols]).dropna(subset = cols, how='all')
)
output.show()
+---+------------------+--------------------+
|key| val11| val12|
+---+------------------+--------------------+
|abc|[src:1.1, dst:2.1]|[src:john, dst:jack]|
+---+------------------+--------------------+

Join two dataframes based on common value in column (which is array)

I have one dataframe - df_similar_strings, which looks like this:
|---------------------|
| string_values |
|---------------------|
| ['catish', 'cat'] |
|---------------------|
| ['doggo', 'dogy'] |
|---------------------|
and the other one - df_source:
|-----------------------------|------------------|
| values | key_value |
|-----------------------------|------------------|
| ['catish', 'cat', 'cat-'] | cat |
|-----------------------------|------------------|
| ['doggo', 'dogy', 'dog'] | dog |
|-----------------------------|------------------|
I would like to join those data frames based on the column string_values and values so that there is at least one value matching.
I have no idea how to do this since the columns are nested as arrays.
Hey you just need to type cast your list to tuple. And then try merging. Since list is unhashable hence merge operation can't be applied. Try this
df_source.values = df_source["values"].apply(lambda x: tuple(x))
Similarly with the other df and try merging using pd merge.
You can solve it by first doing a cartesian-product between your two dataframes and then dropping from that dataframe all rows which doesn't have any shared value.
For simplicity, I assume the columns on both datasets have the same name ("values"). Also, I assume the lists doesn't have repeated values (all values appear once).
from collections import Counter
def find_duplicates(arr):
return [item for item,count in Counter(arr).items() if count==2]
df1['key']=1
df2['key']=1
cartes_prod_df = df1.merge(df2,on=['key'],how='outer').drop(columns=['key'])
duplicate_values = (cartes_prod_df.values_x + cartes_prod_df.values_y).apply(find_duplicates)
merged_df = cartes_prod_df[duplicate_values.apply(lambda x: len(x)>0)]
I've used a little trick in order to do the cartesian product (Adding the key column), and then the duplicate_values found from the joint array (using the + operator) are the values which appeared twice in the joint array.
UPDATE
In order to supply a full example, here's an example of df1 and df2:
d1 = {'values': [['A','B'],['B','C'],['D']],'otherkey':[1,2,3]}
d2 = {'values': [['A'],['B'],['A','C'],['D']],'otherkey':[4,5,3,6]}
df1 = pd.DataFrame(d1)
df2 = pd.DataFrame(d2)
Now, merged_df would give the output:

Adding column from dataframe(df1) to another dataframe (df2)

I need some help with this Apache Spark (pyspark) issue.
I've a dataFrame (df1) which has a single column & a single row, it contains max_timestamp
+------------------+
|max_timestamp |
+-------------------+
|2019-10-24 21:18:26|
+-------------------+
I've another DataFrame, which contains 2 Columns - EmpId & Timestamp
masterData = [(1, '1999-10-24 21:18:23',), (1, '2019-10-24 21:18:26',), (2, '2020-01-24 21:18:26',)]
df_masterdata = spark.createDataFrame(masterData, ['dsid', 'txnTime_str'])
df_masterdata = df_masterdata.withColumn('txnTime_ts', col('txnTime_str').cast(TimestampType())).drop('txnTime_str')
df_masterdata.show(5, False)
+----+-------------------+
|dsid|txnTime_ts |
+----+-------------------+
|1 |1999-10-24 21:18:23|
|1 |2019-10-24 21:18:26|
|2 |2020-01-24 21:18:26|
+----+-------------------+
Object is to filter the records in the 2nd Dataframe, based on condition txnTime_ts < max_timestamp
What i'm trying to do -> add the column 'max_timestamp' to the 2nd DataFrame, and filter records by comparing the 2 values.
df_masterdata1 = df_masterdata.withColumn('maxTime', maxTS2['TEMP_MAX'])
Pyspark does not let me add the column from maxTS2 to the dataFrame - df_masterdata
Error -
AnalysisException: 'Resolved attribute(s) TEMP_MAX#207255 missing from dsid#207263L,txnTime_ts#207267 in operator
!Project [dsid#207263L, txnTime_ts#207267, TEMP_MAX#207255 AS maxTime#207280].;;\n!Project [dsid#207263L,
txnTime_ts#207267, TEMP_MAX#207255 AS maxTime#207280]\n+- Project [dsid#207263L, txnTime_ts#207267]\n +- Project
[dsid#207263L, txnTime_str#207264, cast(txnTime_str#207264 as timestamp) AS txnTime_ts#207267]\n +- LogicalRDD
[dsid#207263L, txnTime_str#207264], false\n'
Any ideas on how to resolve this issue?
If you actually have a DF with a single row/column, the most efficient way to accomplish this would be to extract the value from the dataframe and then filter df_masterdata against it. If you nevertheless need to do this within the context of a dataframe, you should us join , e.g.:
df_masterdata1 = df_masterdata.join(df1, df_masterdata.txnTime_ts <= df1.max_timestamp)

Fetching distinct values on a column using Spark DataFrame

Using Spark 1.6.1 version I need to fetch distinct values on a column and then perform some specific transformation on top of it. The column contains more than 50 million records and can grow larger.
I understand that doing a distinct.collect() will bring the call back to the driver program. Currently I am performing this task as below, is there a better approach?
import sqlContext.implicits._
preProcessedData.persist(StorageLevel.MEMORY_AND_DISK_2)
preProcessedData.select(ApplicationId).distinct.collect().foreach(x => {
val applicationId = x.getAs[String](ApplicationId)
val selectedApplicationData = preProcessedData.filter($"$ApplicationId" === applicationId)
// DO SOME TASK PER applicationId
})
preProcessedData.unpersist()
Well to obtain all different values in a Dataframe you can use distinct. As you can see in the documentation that method returns another DataFrame. After that you can create a UDF in order to transform each record.
For example:
val df = sc.parallelize(Array((1, 2), (3, 4), (1, 6))).toDF("age", "salary")
// I obtain all different values. If you show you must see only {1, 3}
val distinctValuesDF = df.select(df("age")).distinct
// Define your udf. In this case I defined a simple function, but they can get complicated.
val myTransformationUDF = udf(value => value / 10)
// Run that transformation "over" your DataFrame
val afterTransformationDF = distinctValuesDF.select(myTransformationUDF(col("age")))
In Pyspark try this,
df.select('col_name').distinct().show()
This solution demonstrates how to transform data with Spark native functions which are better than UDFs. It also demonstrates how dropDuplicates which is more suitable than distinct for certain queries.
Suppose you have this DataFrame:
+-------+-------------+
|country| continent|
+-------+-------------+
| china| asia|
| brazil|south america|
| france| europe|
| china| asia|
+-------+-------------+
Here's how to take all the distinct countries and run a transformation:
df
.select("country")
.distinct
.withColumn("country", concat(col("country"), lit(" is fun!")))
.show()
+--------------+
| country|
+--------------+
|brazil is fun!|
|france is fun!|
| china is fun!|
+--------------+
You can use dropDuplicates instead of distinct if you don't want to lose the continent information:
df
.dropDuplicates("country")
.withColumn("description", concat(col("country"), lit(" is a country in "), col("continent")))
.show(false)
+-------+-------------+------------------------------------+
|country|continent |description |
+-------+-------------+------------------------------------+
|brazil |south america|brazil is a country in south america|
|france |europe |france is a country in europe |
|china |asia |china is a country in asia |
+-------+-------------+------------------------------------+
See here for more information about filtering DataFrames and here for more information on dropping duplicates.
Ultimately, you'll want to wrap your transformation logic in custom transformations that can be chained with the Dataset#transform method.
df = df.select("column1", "column2",....,..,"column N").distinct.[].collect()
in the empty list, you can insert values like [ to_JSON()] if you want the df in a JSON format.