Hi. This is just me fooling around trying to come up a better way to scrape election results. The tricky logic has been refined in other Texas Tribune projects, but they were deeply tied to other logic.
The idea is to split the process up into multiple logical steps that other people might find useful:
- Ingest results: Typically either with curl or cat or anything that pipes output to stdout.
- Serialize the output html as JSON: Does not attempt to extract information. Just separates data from the html. This is the hard part that scrapers have trouble with.
- Interpret the serialized output: Turns the raw serialized data into something you might expect to see from a nice API.
In a Extract, transform, load (ETL) process, this just covers the extractions, with support for minor transforming.