How to Remove header and footer from Dataframe?

Question

I am reading a text (not CSV) file that has header, content and footer using

spark.read.format(\"text\").option(\"delimiter\",\"|\")...load(file)

Answer 1

Assuming your text file has JSON header and Footer, Spark SQL way,

Sample Data

{"":[{<field_name>:<field_value1>},{<field_name>:<field_value2>}]}

Here the header can be avoided by following 3 lines (Assumption No Tilda in data),

jsonToCsvDF=spark.read.format("com.databricks.spark.csv").option("delimiter", "~").load(<Blob Path1/ ADLS Path1>)

jsonToCsvDF.createOrReplaceTempView("json_to_csv")

spark.sql("SELECT SUBSTR(`_c0`,5,length(`_c0`)-5) FROM json_to_csv").coalesce(1).write.option("header",false).mode("overwrite").text(<Blob Path2/ ADLS Path2>)

Now the output will look like,

[{<field_name>:<field_value1>},{<field_name>:<field_value2>}]

Hope it helps.

Answer 2

In addition to above answer, below solution fits good for files with multiple header and footer lines :-

val data_delimiter = "|"
val skipHeaderLines = 5
val skipHeaderLines = 3

//-- Read file into Dataframe and convert to RDD
val dataframe = spark.read.option("wholeFile", true).option("delimiter",data_delimiter).csv(s"hdfs://$in_data_file")

val rdd = dataframe.rdd

//-- RDD without header and footer
val dfRdd = rdd.zipWithIndex().filter({case (line, index) => index != (cnt - skipFooterLines) && index > (skipHeaderLines - 1)}).map({case (line, index) => line})

//-- Dataframe without header and footer
val df = spark.createDataFrame(dfRdd, dataframe.schema)

Hope this is helpful.

Answer 3

Sample data:

col1|col2|col3
100|hello|asdf
300|hi|abc
200|bye|xyz
800|ciao|qwerty
This is the footer line

Processing logic:

#load text file
txt = sc.textFile("path_to_above_sample_data_text_file.txt")

#remove header
header = txt.first()
txt = txt.filter(lambda line: line != header)

#remove footer
txt = txt.map(lambda line: line.split("|"))\
    .filter(lambda line: len(line)>1)

#convert to dataframe
df=txt.toDF(header.split("|"))
df.show()

Output is:

+----+-----+------+
|col1| col2|  col3|
+----+-----+------+
| 100|hello|  asdf|
| 300|   hi|   abc|
| 200|  bye|   xyz|
| 800| ciao|qwerty|
+----+-----+------+

Hope this helps!

Answer 4

assuming the file is not so large we can use collect to get the dataframe as iterator and the access the last element as follows:

df = df.collect()[data.count()-1]

avoid using collect on large datasets.

or

we can use take to cut off the last row.

df = df.take(data.count()-1)