api_stats tool
==============
api_stats is a Python tool aimed at recording statistics from an API endpoint. For example, the GitHub API gives out a single number of downloads for each release; by running this tool, you can easily graph rolling averages and historical figures.
Quickstart
----------
api_stats loads a "configuration file" that tells it what to fetch and which fields to record. This file is a Python script that will be loaded by api_stats, with a single global ``stats`` that acts as the interface with the program. This interface currently offers two functions: ``record(key, value)``, and ``get_json(url)``.
Here is an example configuration file, that records the number of downloads of each file from GitHub releases:
.. code-block:: python
releases = stats.get_json(
"https://api.github.com/repos/remram44/api_stats/releases")
for release in releases:
relname = release.get('tag_name') or release['name']
for asset in release['assets']:
stats.record('%s/%s' % (relname, asset['name']),
asset['download_count'])
Then you can plot the data, for example using matplotlib:
.. code-block:: bash
python -m api_stats.plot -m '^(.+)$' 'api_stats:\g<1>' data.jsonl
A more complex example
----------------------
You might want to use functions to factor the retrieval of data points from a specific source:
.. code-block:: python
def github(repo):
releases = stats.get_json(
"https://api.github.com/repos/%s/releases" % repo)
for release in releases:
if release['draft']:
continue
relname = release.get('tag_name') or release['name']
for asset in release['assets']:
stats.record('github/%s/%s/%s' % (repo, relname, asset['name']),
asset['download_count'])
github('VisTrails/VisTrails')
github('remram44/api_stats')
def pypi(pkg):
releases = stats.get_json(
"https://pypi.python.org/pypi/%s/json" % pkg)['releases']
for relname, release in releases.items():
for asset in release:
stats.record('pypi/%s/%s/%s' % (pkg, relname, asset['filename']),
asset['downloads'])
pypi('VisTrails')
pypi('api_stats')
This will record data under ``pypi/<pkg_name>/<version>/<filename>`` and ``github/<owner>/<repo>/<version>/<filename>``. You can then plot the data, one figure per project, aggregated by version but adding PyPI and GitHub together, using:
.. code-block:: bash
python -m api_stats.plot \
-m '^pypi/([^/]+)/([^/]+)/.+$' '\g<1>:\g<2>' \
-m '^github/[^/]+/([^/]+)/v([^/]+)/.+$' '\g<1>:\g<2>' \
releases.jsonl
.. image:: https://cloud.githubusercontent.com/assets/426784/10232482/0550d7e0-6857-11e5-835f-86631dd89b4d.png
:alt: Example Plot with matplotlib
