Search code examples
apache-sparkpysparkapache-spark-sqlspark-streamingavro

Spark-Avro Error in PYCHARM [TypeError: 'RecordSchema' object is not iterable]

I am trying to run a simple spark program to read an avro file in PYCHARM environment. I keep getting this error which i am not able to resolve. I appreciate your help.

from environment_variables import *
import avro.schema
from pyspark.sql import SparkSession

Schema = avro.schema.parse(open(SCHEMA_PATH, "rb").read())
print(Schema)
spark = SparkSession.builder.appName("indu").getOrCreate()
df = spark.read.format("avro").load(list(Schema))
print(df)

The schema printed looks like this

{"type": "record", "name": "DefaultEventRecord", "namespace": "io.divolte.record", "fields": [{"type": "boolean", "name": "detectedDuplicate"}, {"type": "boolean", "name": "detectedCorruption"}, {"type": "boolean", "name": "firstInSession"}, {"type": "long", "name": "clientTimestamp"}, {"type": "long", "name": "timestamp"}, {"type": "string", "name": "remoteHost"}, {"type": ["null", "string"], "name": "referer", "default": null}, {"type": ["null", "string"], "name": "location", "default": null}, {"type": ["null", "int"], "name": "devicePixelRatio", "default": null}, {"type": ["null", "int"], "name": "viewportPixelWidth", "default": null}, {"type": ["null", "int"], "name": "viewportPixelHeight", "default": null}, {"type": ["null", "int"], "name": "screenPixelWidth", "default": null}, {"type": ["null", "int"], "name": "screenPixelHeight", "default": null}, {"type": ["null", "string"], "name": "partyId", "default": null}, {"type": ["null", "string"], "name": "sessionId", "default": null}, {"type": ["null", "string"], "name": "pageViewId", "default": null}, {"type": ["null", "string"], "name": "eventId", "default": null}, {"type": "string", "name": "eventType", "default": "unknown"}, {"type": ["null", "string"], "name": "userAgentString", "default": null}, {"type": ["null", "string"], "name": "userAgentName", "default": null}, {"type": ["null", "string"], "name": "userAgentFamily", "default": null}, {"type": ["null", "string"], "name": "userAgentVendor", "default": null}, {"type": ["null", "string"], "name": "userAgentType", "default": null}, {"type": ["null", "string"], "name": "userAgentVersion", "default": null}, {"type": ["null", "string"], "name": "userAgentDeviceCategory", "default": null}, {"type": ["null", "string"], "name": "userAgentOsFamily", "default": null}, {"type": ["null", "string"], "name": "userAgentOsVersion", "default": null}, {"type": ["null", "string"], "name": "userAgentOsVendor", "default": null}, {"type": ["null", "int"], "name": "cityIdField", "default": null}, {"type": ["null", "string"], "name": "cityNameField", "default": null}, {"type": ["null", "string"], "name": "countryCodeField", "default": null}, {"type": ["null", "int"], "name": "countryIdField", "default": null}, {"type": ["null", "string"], "name": "countryNameField", "default": null}]}

The error got is,

21/03/02 16:06:50 WARN NativeCodeLoader: Unable to load native-hadoop library for your platform... using builtin-java classes where applicable
Using Spark's default log4j profile: org/apache/spark/log4j-defaults.properties
Setting default log level to "WARN".
To adjust logging level use sc.setLogLevel(newLevel). For SparkR, use setLogLevel(newLevel).
Traceback (most recent call last):
  File "X:\Git_repo\Project_Red\spark_streaming\spark_scripting.py", line 15, in <module>
    df = spark.read.format("avro").load(list(jsonFormatSchema))
TypeError: 'RecordSchema' object is not iterable

I appreciate your help.

Solution

  • There must be 3 corrections in your code:

    1. You don't have to separately load a schema file because any Avro data file already contains it in its header.
    2. The load() method in your spark.read.format("avro").load(list(Schema)) expects a path to your Avro file, not a schema.
    3. print(df) won't give any meaningful output. Just use df.show() if you want to glance at the data in your Avro file.

    Having said that, you may have already got an idea of what must be changed in your code:

    from pyspark.sql import SparkSession
spark = SparkSession.builder.appName("indu").getOrCreate()
df = spark.read.format("avro").load(DATA_PATH)
df.printSchema()
df.show(truncate=False)