Issue
I'm using AWS Athena to query raw data from S3. Since Athena writes the query output into S3 output bucket I used to do:
df = pd.read_csv(OutputLocation)
But this seems like an expensive way. Recently I noticed the get_query_results
method of boto3
which returns a complex dictionary of the results.
client = boto3.client('athena')
response = client.get_query_results(
QueryExecutionId=res['QueryExecutionId']
)
I'm facing two main issues:
- How can I format the results of
get_query_results
intopandas
data frame? get_query_results
only returns 1000 rows. How can I use it to get two million rows?
Solution
get_query_results only returns 1000 rows. How can I use it to get two million rows into a Pandas dataframe?
If you try to add:
client.get_query_results(QueryExecutionId=res['QueryExecutionId'], MaxResults=2000)
You will obtain the next error:
An error occurred (InvalidRequestException) when calling the GetQueryResults operation: MaxResults is more than maximum allowed length 1000.
You can obtain millions of rows if you obtain the file directly from your bucket s3 (in the next example into a Pandas Dataframe):
def obtain_data_from_s3(self):
self.resource = boto3.resource('s3',
region_name = self.region_name,
aws_access_key_id = self.aws_access_key_id,
aws_secret_access_key= self.aws_secret_access_key)
response = self.resource \
.Bucket(self.bucket) \
.Object(key= self.folder + self.filename + '.csv') \
.get()
return pd.read_csv(io.BytesIO(response['Body'].read()), encoding='utf8')
The self.filename can be:
self.filename = response['QueryExecutionId'] + ".csv"
Because Athena names the files as the QueryExecutionId. I will write you all my code that takes a query and return a dataframe with all the rows and columns.
import time
import boto3
import pandas as pd
import io
class QueryAthena:
def __init__(self, query, database):
self.database = database
self.folder = 'my_folder/'
self.bucket = 'my_bucket'
self.s3_input = 's3://' + self.bucket + '/my_folder_input'
self.s3_output = 's3://' + self.bucket + '/' + self.folder
self.region_name = 'us-east-1'
self.aws_access_key_id = "my_aws_access_key_id"
self.aws_secret_access_key = "my_aws_secret_access_key"
self.query = query
def load_conf(self, q):
try:
self.client = boto3.client('athena',
region_name = self.region_name,
aws_access_key_id = self.aws_access_key_id,
aws_secret_access_key= self.aws_secret_access_key)
response = self.client.start_query_execution(
QueryString = q,
QueryExecutionContext={
'Database': self.database
},
ResultConfiguration={
'OutputLocation': self.s3_output,
}
)
self.filename = response['QueryExecutionId']
print('Execution ID: ' + response['QueryExecutionId'])
return response
except Exception as e:
print(e)
def run_query(self):
queries = [self.query]
for q in queries:
res = self.load_conf(q)
try:
query_status = None
while query_status == 'QUEUED' or query_status == 'RUNNING' or query_status is None:
query_status = self.client.get_query_execution(QueryExecutionId=res["QueryExecutionId"])['QueryExecution']['Status']['State']
print(query_status)
if query_status == 'FAILED' or query_status == 'CANCELLED':
raise Exception('Athena query with the string "{}" failed or was cancelled'.format(self.query))
time.sleep(10)
print('Query "{}" finished.'.format(self.query))
df = self.obtain_data()
return df
except Exception as e:
print(e)
def obtain_data(self):
try:
self.resource = boto3.resource('s3',
region_name = self.region_name,
aws_access_key_id = self.aws_access_key_id,
aws_secret_access_key= self.aws_secret_access_key)
response = self.resource \
.Bucket(self.bucket) \
.Object(key= self.folder + self.filename + '.csv') \
.get()
return pd.read_csv(io.BytesIO(response['Body'].read()), encoding='utf8')
except Exception as e:
print(e)
if __name__ == "__main__":
query = "SELECT * FROM bucket.folder"
qa = QueryAthena(query=query, database='myAthenaDb')
dataframe = qa.run_query()
Answered By - Eric Bellet
0 comments:
Post a Comment
Note: Only a member of this blog may post a comment.