Quality facet - "quantity-contains-e" has many false positives

[CharlesNepote]

What

See: https://world.openfoodfacts.org/quality/quantity-contains-e

Example: "1 litre" in https://world.openfoodfacts.org/product/3464660002434/jus-d-aloe-vera-pur-aloe

The regexp recognize too much things:

openfoodfacts-server/lib/ProductOpener/SiteQuality_off.pm

Line 747 in aa25a7e

and ($product_ref->{quantity} =~ /(?:.*e$)|(?:[0-9]+\s*[kmc]?[gl]?\s*e)/i)

I replaced:

• /(?:.*e$)|(?:[0-9]+\s*[kmc]?[gl]?\s*e)/i by
• /(?:^e\s)|(?:\se[^a-z])|(?:\se$)/i

And it deleted more than 1500 false positives (see file attached).

4.txt

Part of

• 🧽 Data quality (tracker) #5538

[stephanegigandet]

I agree that the regexp is broken, but the proposed change also delete true positives.

What we really want to identify is cases like this:

500 ge

[VaiTon]

In the first place, why do we want to identify quantities containing "e"? @stephanegigandet

[benbenben2]

See: (https://world.openfoodfacts.org/quality/quantity-contains-e) -> https://world.openfoodfacts.org/data-quality-info/en:quantity-contains-e

For the aforementioned example, see also from the API: https://world.openfoodfacts.org/api/v2/search?code=3464660002434

The aforementioned line of code can now be found in /lib/ProductOpener/SiteQuality_off.pm

Finally, to find more examples:
https://mirabelle.openfoodfacts.org/products?sql=select+quantity+from+%5Ball%5D%0D%0AWHERE+quantity+%21%3D+%22%22+AND+quantity+LIKE+%22%25e%25%22%0D%0ALIMIT+100

[stephanegigandet]

In the first place, why do we want to identify quantities containing "e"? @stephanegigandet

That's a good question. :) (sorry @VaiTon I missed it 4 years ago).

The symbol is present on most products, I didn't see the point in including it in the quantity field.

[benbenben2]

Should we close this issue @CharlesNepote?

[alexgarel]

I think we can close.