For years (between January 2016 and October 2021), TECTRA Ltd utilized a spreadsheet template for issuing invoices to customers, without a codification of the invoiced product items, or any data linkage to a database regarding the information of the invoiced parties.

With the advent of mandatory electronic invoicing, and building upon the “heavy lifting” done by the ProductItems API software, I developed a parsing system that accessed the hundreds of sheets (each representing an invoice) across tens of spreadsheet files, and parsed the information spanning multiple rows (instead of text with line breaks within a single cell).

Due to occassional manual modifications and inconsistent data entry over a number of years, the parsing rules were written with sanity checks that reported potentially missing or misplaced information for manual correction. In the end, around 20 files had to be manually edited so that the entire dataset could be generated without interruption.

The resulting dataset was serialized as a Python pickle object, later accessed by the early versions of the company’s bespoke Timologia API , and ultimately converted into an ODS spreadsheet that came to represents the commercial (i.e., non-accounting) single source of truth of invoices since January 2016.

This software was therefore written as a “one-off” operation that delivered the basis upon which the management of TECTRA Ltd, in 2022, i.e. after 30 years as a going concern, was finally able to perform analyses of commercial transactions using pivot charts, and thus gain an understanding of the evolution of pricing and revenues per product brand, category, family, manufacturing origin, and other attributes.