BLN ETL

• Overview
• Install
• Usage
  • API
  • Git Repository
  • Archive
• Contributors

Overview

Utilities to assist with data gathering for and publishing to Big Local News. It supports common workflows used by the BLN core team for its own data gathering operations, but may also be useful to others working with our platform.

In particular, this package provides utility code to:

• Simplify data acquisition from GitHub repos
• Create a Zip archive
• Upload files to a BLN platform project

Install

• Install git CLI tools
• Install the bln_etl package from GitHub:

  pip install git+https://github.com/biglocalnews/bln-etl#egg=bln-etl

• Sign in to Big Local News and create an API key in Developer -> Manage Keys
• Set the BLN_API_KEY=<YOUR_KEY> environment variable obtained in prior step

Usage

Api

Below examples assume you've configured the BLN_API_KEY environment variable. Alternatively, you can pass an api_token keyword argument to either the Client or Project classes on instantiation, as well as class methods on Project.

Get projects.

from bln_etl.api import Client

# Set up the client
client = Client() # Or Client(api_token=<YOUR_TOKEN>)

# Get your own projects
client.user_projects

# Get our Open Projects
client.open_projects

Get a particular project.

from bln_etl.api import Project

# Projects can be looked up using an "id".
# (accessible, for example, from Client.user_projects)
Project.get('Uadfas19etc.etc.')

# Or pass an API token if BLN_API_KEY env var is not set
Project.get('Uadfas19etc.etc.', api_token=<YOUR_TOKEN>)

Create a project.

# Every project requires a name
project_name = 'Awesome Project'

# Optional  fields:
options = {
    'description': 'A truly awesome project.',
    'contact': '[email protected]',
    'is_open': True, # projects are private by default
}
Project.create(name, **options)

# Or pass an API token if BLN_API_KEY env var is not set
Project.create(name, api_token=<YOUR_TOKEN>)

Below Project.get examples assume BLN_API_KEY env var is set

Upload files to a project (silently overwrites pre-existing files of the same name).

project = Project.get(<uuid>)
to_upload = ['/tmp/test.csv']
project.upload_files(to_upload)

List project files.

project = Project.get(<uuid>)
for f in project.files:
    print(f)

Delete files.

project = Project.get(<uuid>)
for f in project.files:
    f.delete()

Git Repository

The Repository class is a light wrapper around basic Git command-line utilities. It is a context manager that:

• automatically creates the project folder if it doesn't exist
• switches the current working directory to repository folder before executing git commands
• restores the working directory to its original state upon exit

As such, you should always instantiate Repository using a with statement.

from bln_etl import Repository

with Repository('/path/to/data-project-repo') as repo:

  # Check if directory is initialized as a git repo
  repo.initialized

  # Initialize directory as a git repo...
  repo.init()

  # ...or clone a repo to the data-project-repo folder
  repo.clone('[email protected]:biglocalnews/bln-etl.git')

  # Stage all file changes
  repo.add()

  # Commit staged changes (a commit message is required)
  message = "Added some code"
  repo.commit(message)

  # Push changes (default: main branch of remote "origin")
  repo.push()

  # Customize the push
  repo.push(remote="upstream", branch="master")

  # Pull changes (current branch only)
  repo.pull()

Repository also provides a static method that doesn't require use of a with statement:

# Clone a repo to specified local directory
url = '[email protected]:biglocalnews/bln-etl.git'
target_dir = '/tmp/etl'
Repository.clone_to_dir(url, target_dir)

Archive

A wrapper class to help with creation of a Zipfile.

from bln_etl import Archive

archive = Archive('/tmp/data.zip')

# Add a single file
archive.add('/tmp/data.csv')

# Add a file but store using a different name
archive.add('/tmp/data.csv', rename='foo.csv')

# Add all files in directory tree
archive.add_dir('/tmp/folder-with-data')

# Add CSVs in directory tree using glob pattern
archive.add_dir('/tmp/folder-with-data', pattern='**/*.csv')

# Include hidden files in directory tree
archive.add_dir('/tmp/folder-with-data', skip_hidden=False)

# Extract files in zip to same dir as zip
archive.extractall()

# Extract files in zip to some other directory
archive.extractall(path="/tmp/some/other/dir")


# List files in archive
archive.list()

See the Archive class for additional usage details.

Write mode configuration

The add and add_dir methods operate in append mode by default, meaning they will add files to a new or pre-existing Zipfile.

These methods allow you to configure the write mode for Zipfiles. This can be useful in data scrapers that run on a schedule, allowing you to use write mode on the first call (thereby deleting a pre-existing Zipfile of the same name), and then use the default append mode on subsequent calls.

from bln_etl import Archive

archive = Archive('/tmp/data.zip')

# Add a single file using write mode
# to overwrite any previous archive at the same path
archive.add('/tmp/data.csv', mode='w')

# ...or add an entire dir in (over)write mode
archive.add_dir('/tmp/dir-with-lots-of-data', mode='w')

# On subsequent calls, default to append mode 
# to update the new zipfile
archive.add('/tmp/data2.csv')
archive.add_dir('/tmp/some-other-data-dir')

Compression

The Archive class attempts to use Python ZipFile's ZIP_DEFLATED compression type if the zlib library is installed. If it is installed, it uses zlib's default compression level.

If zlib or its system-level dependencies are not installed, Archive falls back to ZipFile's default ZIP_STORED compression type (i.e. uncompressed).

Contributors

Tests use pytest and vcrpy to record live web interactions. A small number of tests that always hit live web services are marked as webtest and require an additional command-line flag to force their execution.

All tests require a valid BLN_API_KEY environment variable to be set. See Install instructions for more details.

# Run tests
pytest

# Force running of all tests, including live "webtests"
pytest --webtest

To update semantic versioning:

setup.cfg specifies files to change on version update, and is configured to automatically commit and tag file changes

bumpversion [major|minor|patch]

# e.g., switch 0.1.0 -> 0.1.1
bumpversion patch