Rossmann Scraper for Product, Price and Grundpreis Data

Fields a Rossmann scrape returns

Every row starts with the identifier. A product URL on rossmann.de is a long descriptive slug followed by /p/ and a thirteen-digit code, and that code is a GTIN. Archived URLs show babydream articles under codes beginning 4305615 and ISANA articles under codes beginning 4068134, which is how own-brand items cluster on the company's own GS1 prefix. The code in the path is the stable join key. The slug in front of it is not, because it carries the category name and moves when merchandising moves.

A standard Rossmann data extract carries:

• Preis (selling price), parsed out of German number format, where the comma is the decimal separator and the currency symbol trails the figure.
• Grundpreis (base price per unit), kept as three columns - numeric value, reference amount and unit - so per 100 ml, per litre, per 100 g, per kilogram and per-piece bases stay comparable instead of collapsing into one string.
• Inhalt (pack size or content), normalised to a number and a unit, since the Grundpreis can only be checked against it.
• Marke (brand), with a flag for whether that brand is a Rossmann own brand.
• Verfügbarkeit (availability) as the shop states it, with the raw label kept beside the boolean.
• GTIN taken from the /p/ path, plus any internal article number the page prints.
• Category path as walked, together with the /c/ category code, for example olcat1_1 for Make-up and olcat1_4 for Lebensmittel.
• Bewertungen (customer reviews): rating value and review count where an article carries them.
• Inhaltsstoffe (ingredients) captured verbatim as text where the page prints them, with nothing added and nothing interpreted.
• Any struck-through reference price shown next to the selling price, kept in its own column rather than folded into Preis.
• Image URLs, canonical URL and the timestamp of the observation.

Collection covers public product and pricing data only. No customer data is touched.

What Rossmann scraping has to solve: own brands

Own brands are the second reason this is not a routine retail scrape. The company site at unternehmen.rossmann.de, stated a portfolio of 27 Marken (brands) and over 6.000 Produkte (products), and named ISANA, Alterra, domol, babydream, facelle, sunozon, enerBiO and altapharma among them. Those articles are sold nowhere else, and their GTINs are issued under Rossmann's own prefix, so no shared product code exists to join on. A matching pipeline keyed on EAN will line up the Garnier and Pantene rows and drop every own-brand row on the floor - which is precisely the part of the catalogue a competitor most wants to read.

Own-brand rows have to be matched on attributes instead: Marke, the normalised Inhalt, the product form, the category path, and the Grundpreis unit basis. That yields like against like, a 300 ml own-brand item set against a 300 ml branded one at the same base unit, rather than a code join that fails silently. We build the normalisation into the extract, so the columns arrive ready to match instead of arriving as free text. The shop also publishes a brand index at /de/Marken/c/brands, which gives the crawl a clean entry point for separating own brands from third-party brands.

Rossmann is a drugstore and not a pharmacy, and one of the listed own brands is named altapharma. We record the fields a page prints and nothing else. We do not describe, classify or infer what any article is for.