2.2 Acquisition via Extract
Since data formats are in a constant state of flux, it’s helpful to understand how to add and modify data formats. These projects will all build on Project 1.1 by adding features to the base application. The following projects are designed around alternative sources for data:
Project 1.2: ”Acquire Web Data from an API”. This project will acquire data from web services using JSON format.
Project 1.3: ”Acquire Web Data from HTML”. This project will acquire data from a web page by scraping the HTML.
Two separate projects are part of gathering data from a SQL database:
Project 1.4: ”Build a Local Database”. This is a necessary sidebar project to build a local SQL database. This is necessary because SQL databases accessible by the public are a rarity. It’s more secure to build our own demonstration database.
Project 1.5: ”Acquire Data from a Local Database”. Once a database is available, we can acquire data from a SQL extract.
These projects will focus on data represented as text. For CSV files, the data is text; an application must convert it to a more useful Python type. HTML pages, also, are pure text. Sometimes, additional attributes are provided to suggest the text should be treated as a number. A SQL database is often populated with non-text data. To be consistent, the SQL data should be serialized as text. The acquisition applications all share a common approach of working with text.
These applications will also minimize the transformations applied to the source data. To process the data consistently, it’s helpful to make a shift to a common format. As we’ll see in Chapter 3, Project 1.1: Data Acquisition Base Application the NDJSON format provides a useful structure that can often be mapped back to source files.
After acquiring new data, it’s prudent to do a manual inspection. This is often done a few times at the start of application development. After that, inspection is only done to diagnose problems with the source data. The next few chapters will cover projects to inspect data.