Package Feeds

The binary produced by cmd/scheduled-feed/main.go can be used to monitor various package repositories for changes and publish data to external services for further processing.

Additionally, the repo contains a few subprojects to aid in the analysis of these open source packages, in particular to look for malicious software.

These are:

Feeds to watch package registries (PyPI, NPM, etc.) for changes to packages and to make that data available via a single standard interface.

Publisher provides the functionality to push package details from feeds towards external services such as GCP Pub/Sub. Package details are formatted inline with a versioned json-schema.

This repo used to contain several other projects, which have since been split out into github.com/ossf/package-analysis.

The goal is for all of these components to work together and provide extensible, community-run infrastructure to study behavior of open source packages and to look for malicious software. We also hope that the components can be used independently, to provide package feeds or runtime behavior data for anyone interested.

Configuration

A YAML configuration file can be provided with the following format:

feeds:
- type: pypi
- type: npm
- type: goproxy
- type: rubygems
- type: crates

publisher:
  type: 'gcp_pubsub'
  config:
    url: "gcppubsub://foobar.com"

http_port: 8080

poll_rate: 5m

timer: false

poll_rate string formatted for duration parser.This is used as an initial value to generate a cutoff point for feed events relative to the given time at execution, with subsequent events using the previous time at execution as the cutoff point. timer will configure interal polling of the feeds at the given poll_rate period, individual feeds configured with a poll_rate will poll on an interval regardless of these options. To specify this configuration file, define its path in your environment under the PACKAGE_FEEDS_CONFIG_PATH variable.

An event handler can be configured through the events field, this is documented in the events README.

FeedOptions

Feeds can be configured with additional options, not all feeds will support these features. Check feeds/README.md for more information on feed specific configurations.

Below is an example of such options with pypi being configured to poll a specific set of packages

feeds:
- type: pypi
  options:
    packages:
    - fooPackage
    - barPackage

Legacy Configuration

Legacy configuration methods are still supported. By default, without a configuration file all feeds will be enabled. The environment variable OSSMALWARE_TOPIC_URL can be used to select the GCP pubsub publisher and PORT will configure the port for the HTTP server. The default poll_rate is 5 minutes, it is assumed that an external service is dispatching requests to the configured HTTP server at this frequency.

Running Locally

To start an instance of package-feeds running on a local machine, run the following commands:

$ docker build . -t local-package-feeds        ## Build the container
$ mkdir /tmp/feedconfig/                       ## Create the feeds.yml config
$ cat << EOF >> /tmp/feedconfig/feeds.yml
publisher:
  type: stdout
poll_rate: 5m
http_port: 8080
EOF
$ docker run -v /tmp/feedconfig:/config/ \
  -e "PACKAGE_FEEDS_CONFIG_PATH=/config/feeds.yml" \
  -p 8080:8080 --rm -ti local-package-feeds    ## Start the container

Contributing

If you want to get involved or have ideas you'd like to chat about, we discuss this project in the OSSF Securing Critical Projects Working Group meetings.

See the Community Calendar for the schedule and meeting invitations.