avroconvert

Utility to convert avro files to csv, json and parquet formats


Keywords
avroconvert
License
MIT
Install
pip install avroconvert==0.1.0

Documentation

avroconvert

codecov docs docs

Utility to convert avro files to csv, json and parquet formats

ReadtheDocs Documentation

  • Installation

Using pypi

pip install avroconvert

Using git:

git clone https://github.com/shrinivdeshmukh/avroconvert
make install
  • Usage

Using CLI

CLI can be used to interact with the tool. As the first argument, the source has to be passed. The source can be gs (google cloud storage bucket), s3 (amazon s3 bucket) or fs (local filesystem)

To read from cloud bucket (google cloud or amazon s3):

google cloud storage example:

avroconvert gs -b <BUCKET_NAME> -f <FORMAT> -o <OUTPUT_FOLDER>

amazon s3 example:

avroconvert s3 -b <BUCKET_NAME> -f <FORMAT> -o <OUTPUT_FOLDER>

The tool reads all avro files from the bucket specified by the -b parameter, converts them to the format specified by the -f parameter, and writes the output format files to the output folder specified by the -o parameter with the above command.

The cli accepts a few additional parameters to authenticate the tool with cloud providers. These parameters are only required if you haven't already been authenticated.

For google cloud, we have --auth-file:

avroconvert gs -b <BUCKET_NAME> -f <FORMAT> -o <OUTPUT_FOLDER> --auth-file <SERVICE_ACCOUNT_FILE_PATH>.json (or .p12)

For amazon s3, we have --access-key, --secret-key, --session-token:

avroconvert s3 -b <BUCKET_NAME> -f <FORMAT> -o <OUTPUT_FOLDER> --access-key <AWS_ACCESS_KEY_ID> --secret-key <AWS_SECRET_ACCESS_KEY> --session-token <AWS_SESSION_TOKEN> 

To read from local filesystem

avroconvert fs  -i <INPUT_DATA_FOLDER> -o <OUTPUT_FOLDER> -f <OUTPUT_FORMAT>

The tool reads all avro files from the input folder specified by the -i parameter, converts them to the format specified by the -f parameter, and writes the output format files to the output folder specified by the -o parameter with the above command.

Output folder structure

The tool replicates the cloud bucket's or local filesystem's directory structure. For example, suppose the output format is parquet and cloud bucket (or local filesystem) has the following structure:

BUCKET
├── 2021-06-17
│   └── file1.avro
│   └── file2.avro
│ 
├── 2021-06-16
│   └── data
│       └── file3.avro
│       └── file4.avro

the output files will then be saved as:

OUTPUT_FOLDER
├── 2021-06-17
│   └── file1.parquet
│   └── file2.parquet
│ 
├── 2021-06-16
│   └── data
│       └── file3.parquet
│       └── file4.parquet

Filter files to read

A parameter called -p or —-prefix can be passed as well. All three data sources, gs, s3, and fs, share this parameter. Only files with names that begin with the specified prefix will be read; all other files will be filtered out.

google cloud example with -p:

avroconvert gs -b <BUCKET_NAME> -f <FORMAT> -o <OUTPUT_FOLDER> -p 2021-06-17/file

amazon s3 example with -p:

avroconvert s3 -b <BUCKET_NAME> -f <FORMAT> -o <OUTPUT_FOLDER> -p 2021-06-17/file

local filesystem example with -p:

avroconvert fs  -i <INPUT_DATA_FOLDER> -o <OUTPUT_FOLDER> -f <OUTPUT_FORMAT> -p 2021-06-17/file

Using the API in code

    from avroconvert import Execute

    # for amazon s3 storage bucket reader
    output = Execute(source='gs', bucket='<BUCKET_NAME>, dst_format='parquet', auth_file='<SERVICE_ACCOUNT.json>',
                     outfolder='OUTPUT_FOLDER', access_key='<AWS ACCESS KEY>', secret_key='<AWS SECRET KEY>', 
                     session_token='<AWS SESSION TOKEN>(if any)', bucket='<S3 BUCKET>', prefix='<FILE PREFIX>').run()

    # google storage bucket reader
    output = Execute(source='gs', bucket='<BUCKET_NAME>, dst_format='parquet', auth_file='<SERVICE_ACCOUNT.json>',
                     outfolder='OUTPUT_FOLDER').run()

    # Local file system reader
    output = Execute(source='fs', bucket='<LOCAL_FOLDER NAME> dst_format='parquet', outfolder='OUTPUT_FOLDER').run()

For more details on using the API, please visit readthedocs

  • Credits

This package was created with Cookiecutter and the audreyr/cookiecutter-pypackage project template.