Introduction to the Collectors part¶

Collectors are n6‘s entry points for any external data to flow in. Data harvested with a collector are then sent to its corresponding parser. So the pipeline at this stage looks like this:

(External Data) → [Collector] → [Parser] → ...

Note

For a broader overview, you may want to take a look at the n6 architecture and data flow diagram.

To describe collectors’ job more technically: a collector obtains data from its respective external source (e.g., by downloading files from a certain security-focused web site), and sends them further down the n6 data processing pipeline (by pushing those data – in a form as similar to the original as possible – into the appropriate RabbitMQ exchange).

RabbitMQ-based pipeline¶

Most of the n6 pipeline components – in particular, all collectors – send their output data to a RabbitMQ message broker instance (using the AMQP 0.9.1 protocol), so that the pipeline components that are supposed to consume those data (in particular, the parser corresponding to the given collector) will be able to take the data from the respective RabbitMQ queue.

Running a collector¶

Each collector is run as a separate program (OS process). Depending on the type of the external source a particular collector is related to, it may be spawned by a cron-like scheduler (the most typical way), by a procmail-like agent (this is typical for e-mail-based sources), or in some other way.

Collector base classes¶

Each collector needs to have a network connection to the RabbitMQ broker; however, typically, there is no need for you, as the programmer who implements a collector, to deal with that stuff directly. There are several base classes your collector class can derive from which take care of various repeatable tasks, especially of the stuff related to initialization of, communication by, and shutting down the RabbitMQ connection – so that you can focus solely on your collector’s logic.

Those base classes can be found in the N6Core subdirectory of the n6 source code repository, in the n6.collectors.generic module. (The module’s file path in the repository is N6Core/n6/collectors/generic.py.)

Stateful collectors¶

Some collectors need to maintain a state, stored between consecutive collector runs (for example, to remember the id and creation time of the last downloaded record, so that, on the next run, the collector will download and send, as its output, only newer records).

Each class of such a stateful collector needs to refer to its own state that will not be shared with other collectors. The n6.collectors.generic.CollectorWithStateMixin class helps to take care of that.