Utility belt to handle data on AWS.
Project description
AWS Data Wrangler
Utility belt to handle data on AWS.
Contents: Use Cases | Installation | Examples | Diving Deep
Use Cases
Pandas
- Pandas -> Parquet (S3) (Parallel :rocket:)
- Pandas -> CSV (S3) (Parallel :rocket:)
- Pandas -> Glue Catalog
- Pandas -> Athena (Parallel :rocket:)
- Pandas -> Redshift (Parallel :rocket:)
- CSV (S3) -> Pandas (One shot or Batching)
- Athena -> Pandas (One shot or Batching)
- CloudWatch Logs Insights -> Pandas (NEW :star:)
- Encrypt Pandas Dataframes on S3 with KMS keys (NEW :star:)
PySpark
- PySpark -> Redshift (Parallel :rocket:) (NEW :star:)
General
- List S3 objects (Parallel :rocket:)
- Delete S3 objects (Parallel :rocket:)
- Delete listed S3 objects (Parallel :rocket:)
- Delete NOT listed S3 objects (Parallel :rocket:)
- Copy listed S3 objects (Parallel :rocket:)
- Get the size of S3 objects (Parallel :rocket:)
- Get CloudWatch Logs Insights query results (NEW :star:)
Installation
pip install awswrangler
Runs only with Python 3.6 and beyond.
Runs anywhere (AWS Lambda, AWS Glue, EMR, EC2, on-premises, local, etc).
P.S. Lambda Layer bundle and Glue egg are available to download. It's just upload to your account and run! :rocket:
Examples
Pandas
Writing Pandas Dataframe to S3 + Glue Catalog
session = awswrangler.Session()
session.pandas.to_parquet(
dataframe=dataframe,
database="database",
path="s3://...",
partition_cols=["col_name"],
)
If a Glue Database name is passed, all the metadata will be created in the Glue Catalog. If not, only the s3 data write will be done.
Writing Pandas Dataframe to S3 as Parquet encrypting with a KMS key
extra_args = {
"ServerSideEncryption": "aws:kms",
"SSEKMSKeyId": "YOUR_KMY_KEY_ARN"
}
session = awswrangler.Session(s3_additional_kwargs=extra_args)
session.pandas.to_parquet(
path="s3://..."
)
Reading from AWS Athena to Pandas
session = awswrangler.Session()
dataframe = session.pandas.read_sql_athena(
sql="select * from table",
database="database"
)
Reading from AWS Athena to Pandas in chunks (For memory restrictions)
session = awswrangler.Session()
dataframe_iter = session.pandas.read_sql_athena(
sql="select * from table",
database="database",
max_result_size=512_000_000 # 512 MB
)
for dataframe in dataframe_iter:
print(dataframe) # Do whatever you want
Reading from S3 (CSV) to Pandas
session = awswrangler.Session()
dataframe = session.pandas.read_csv(path="s3://...")
Reading from S3 (CSV) to Pandas in chunks (For memory restrictions)
session = awswrangler.Session()
dataframe_iter = session.pandas.read_csv(
path="s3://...",
max_result_size=512_000_000 # 512 MB
)
for dataframe in dataframe_iter:
print(dataframe) # Do whatever you want
Reading from CloudWatch Logs Insights to Pandas
session = awswrangler.Session()
dataframe = session.pandas.read_log_query(
log_group_names=[LOG_GROUP_NAME],
query="fields @timestamp, @message | sort @timestamp desc | limit 5",
)
Typical Pandas ETL
import pandas
import awswrangler
df = pandas.read_... # Read from anywhere
# Typical Pandas, Numpy or Pyarrow transformation HERE!
session = awswrangler.Session()
session.pandas.to_parquet( # Storing the data and metadata to Data Lake
dataframe=dataframe,
database="database",
path="s3://...",
partition_cols=["col_name"],
)
PySpark
Loading PySpark Dataframe to Redshift
session = awswrangler.Session(spark_session=spark)
session.spark.to_redshift(
dataframe=df,
path="s3://...",
connection=conn,
schema="public",
table="table",
iam_role="IAM_ROLE_ARN",
mode="append",
)
General
Deleting a bunch of S3 objects (parallel :rocket:)
session = awswrangler.Session()
session.s3.delete_objects(path="s3://...")
Get CloudWatch Logs Insights query results
session = awswrangler.Session()
results = session.cloudwatchlogs.query(
log_group_names=[LOG_GROUP_NAME],
query="fields @timestamp, @message | sort @timestamp desc | limit 5",
)
Diving Deep
Pandas to Redshift Flow
Spark to Redshift Flow
Project details
Release history Release notifications | RSS feed
Download files
Download the file for your platform. If you're not sure which to choose, learn more about installing packages.
Source Distribution
awswrangler-0.0.1.tar.gz
(27.4 kB
view hashes)
Built Distribution
Close
Hashes for awswrangler-0.0.1-py36,py37-none-any.whl
Algorithm | Hash digest | |
---|---|---|
SHA256 | 926187002bfd7a1bbc20d8b684b8465d35e6fd937c418cf67d7bfe0c243fdff5 |
|
MD5 | 984d0dcb675649d270d8cea9f036d798 |
|
BLAKE2b-256 | a25fb736f766463f870aae6c046a7ab48291f3476f5093d35d21c4416519184b |