Python Hive query framework
Project description
# Apiarist
A python 2.5+ package for defining Hive queries which can be run on AWS EMR.
It is, in its current form, only addressing a very narrow use-case.
Reading large CSV files into a Hive database, running a Hive query, and outputting the results to a CSV file.
Future versions will endeavour to extend the input/output formats and be runnable locally.
It is modeled on [mrjob](https://github.com/Yelp/mrjob) and attempts to present a similar API and use similar common variables to cooperate with `boto`.
## A simple Hive job
You will need to provide four methods:
- `table()` the name of the table that your query will select from.
- `input_columns()` the columns in the source data file.
- `output_columns()` the columns that your query will output.
- `query` the HiveQL query.
This code lives in `/examples`.
```python
from apiarist.job import HiveJob
class EmailRecipientsSummary(HiveJob):
def table(self):
return 'emails_sent'
def input_columns(self):
return [
('day', 'STRING'),
('weekday', 'INT'),
('sent', 'BIGINT')
]
def output_columns(self):
return [
('year', 'INT'),
('weekday', 'INT'),
('sent', 'BIGINT')
]
def query(self):
return "SELECT YEAR(day), weekday, SUM(sent) FROM emails_sent GROUP BY YEAR(day), weekday;"
if __name__ == "__main__":
EmailRecipientsSummary().run()
```
### Try it out
Locally (must have a Hive server available):
python email_recipients_summary.py -r local /path/to/your/local/file.csv
EMR:
python email_recipients_summary.py -r emr s3://path/to/your/S3/files/
## Command-line options
Arguments can be passed to jobs on the command line, or programmatically with an array of options. Argument handling uses the [optparse](https://docs.python.org/2/library/optparse.html) module.
Various options can be passed to control the running of the job. In particular the AWS/EMR options.
- `-r` the run mode. Either `local` or `emr` (default is `local`)
- `--output-dir` where the results of the job will go.
- `--s3-scratch-uri` the bucket in which all the temporary files can go.
- `--ec2-instance-type` the base instance type. Default is `m3.xlarge`
- `--ec2-master-instance-type` if you want the master type to be different.
- `--num-ec2-instances` number of instances (including the master). Default is `2`.
- `--ami-version` the ami version. Default is `latest`.
- `--hive-version`. Default is `latest`.
### Passing options to your jobs
Jobs can be configured to accept arguments.
To do this, add the following method to your job class to configutr the options:
def configure_options(self):
super(EmailRecipientsSummary, self).configure_options()
self.add_passthrough_option('--year', dest='year')
And then use the option by providing it in the command line arguments, like this:
python email_recipients_summary.py -r local /path/to/your/local/file.csv --year 2014
Then incorporating it into your HiveQL query like this:
def query(self):
q = "SELECT YEAR(day), weekday, SUM(sent) "
q += "FROM emails_sent "
q += "WHERE YEAR(day) = {0} ".format(self.options.year)
q += "GROUP BY YEAR(day), weekday;"
return q
## License
Apiarist source code is released under Apache 2 License. Check LICENSE file for more information.
A python 2.5+ package for defining Hive queries which can be run on AWS EMR.
It is, in its current form, only addressing a very narrow use-case.
Reading large CSV files into a Hive database, running a Hive query, and outputting the results to a CSV file.
Future versions will endeavour to extend the input/output formats and be runnable locally.
It is modeled on [mrjob](https://github.com/Yelp/mrjob) and attempts to present a similar API and use similar common variables to cooperate with `boto`.
## A simple Hive job
You will need to provide four methods:
- `table()` the name of the table that your query will select from.
- `input_columns()` the columns in the source data file.
- `output_columns()` the columns that your query will output.
- `query` the HiveQL query.
This code lives in `/examples`.
```python
from apiarist.job import HiveJob
class EmailRecipientsSummary(HiveJob):
def table(self):
return 'emails_sent'
def input_columns(self):
return [
('day', 'STRING'),
('weekday', 'INT'),
('sent', 'BIGINT')
]
def output_columns(self):
return [
('year', 'INT'),
('weekday', 'INT'),
('sent', 'BIGINT')
]
def query(self):
return "SELECT YEAR(day), weekday, SUM(sent) FROM emails_sent GROUP BY YEAR(day), weekday;"
if __name__ == "__main__":
EmailRecipientsSummary().run()
```
### Try it out
Locally (must have a Hive server available):
python email_recipients_summary.py -r local /path/to/your/local/file.csv
EMR:
python email_recipients_summary.py -r emr s3://path/to/your/S3/files/
## Command-line options
Arguments can be passed to jobs on the command line, or programmatically with an array of options. Argument handling uses the [optparse](https://docs.python.org/2/library/optparse.html) module.
Various options can be passed to control the running of the job. In particular the AWS/EMR options.
- `-r` the run mode. Either `local` or `emr` (default is `local`)
- `--output-dir` where the results of the job will go.
- `--s3-scratch-uri` the bucket in which all the temporary files can go.
- `--ec2-instance-type` the base instance type. Default is `m3.xlarge`
- `--ec2-master-instance-type` if you want the master type to be different.
- `--num-ec2-instances` number of instances (including the master). Default is `2`.
- `--ami-version` the ami version. Default is `latest`.
- `--hive-version`. Default is `latest`.
### Passing options to your jobs
Jobs can be configured to accept arguments.
To do this, add the following method to your job class to configutr the options:
def configure_options(self):
super(EmailRecipientsSummary, self).configure_options()
self.add_passthrough_option('--year', dest='year')
And then use the option by providing it in the command line arguments, like this:
python email_recipients_summary.py -r local /path/to/your/local/file.csv --year 2014
Then incorporating it into your HiveQL query like this:
def query(self):
q = "SELECT YEAR(day), weekday, SUM(sent) "
q += "FROM emails_sent "
q += "WHERE YEAR(day) = {0} ".format(self.options.year)
q += "GROUP BY YEAR(day), weekday;"
return q
## License
Apiarist source code is released under Apache 2 License. Check LICENSE file for more information.
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
apiarist-0.0.11.tar.gz
(38.6 kB
view hashes)