newspapers | R package to import articles from newspaper databases

Your code means that the number of items in a row must be exactly equal to header.length. But in your CSV file, each row has a different length from the others.

You must check the number of items in each row separately from header length.

Your issue stems from the fact that List's selection mode uses the id: property to track your selection. Since you are declaring your List as List(..., id: \.name, ...), your selection var needs to be of type String. If you change it to List(..., id: \.self, ...), it will work, but using self in a list like that brings it own problems. In keeping with best practice, and forgetting the selection for a moment, you should be using an Identifiable struct. List should then identify the elements by the id parameter on the struct. (I used a UUID)

Working up to the selection, that means you need to define it as @State private var selection = Set(). That leaves dealing with your favorites computed variable. Instead of returning an array of your selection, you simply filter the newspapers array for those element contained in selection. In the end, that leaves you with this:

For this particular web page the pages are served from a predictable url:

• https://www.sbt.ti.ch/aqp_pdf/gdp/2005/12/gdp_2005-12-01/gdp_2005-12-01_001.pdf
• https://www.sbt.ti.ch/aqp_pdf/gdp/2005/12/gdp_2005-12-01/gdp_2005-12-01_002.pdf
• etc

This is so regular I wouldn't even bother extracting it from the page for this problem: I'd just generate the url myself, do a requests.get() for each of them, and splice them together with PyPdf2.

The more general question is: how did I know that url? Have a look at your browser's devtools:

There are basically two solutions to this kind of problem:

• extract the required parameters from the page (look at how the page builds up the urls it needs), or
• run a real browser with something like selenium, and automate it.

Sometimes you get lucky and there's a real api designed to help you do this. It's quite common when looking at public archive data like this (in France, the apis of the BNF are excellent, but I don't know what, if anything, would be the Italian equivalent).

I am not sure if the two queries are supposed to be same, but they are not.

Anyway for the second query I think this should be better

None of your data is numeric, all are factors. So you plot against a discrete factor value y and not as desired to numeric continuous scale.

So, your schema generator has a limitation that it ignores xsi:type attributes. That's not a great surprise, it's hard to know what it would do with them.

You obviously know something that the tool doesn't ("LibraryAssets are Books, Newspapers and Patents, and LibraryAsset is the abstract class they derive from") and it's normal practice that when you generate a schema, what you get is only a first cut, and you're expected to edit it to take account of your knowledge of the semantics of the application domain.

I am looking at your example output (thanks for including it). The first row is showing 82% in pre-assessment and 96% in the post-assessment. 82 - 96 = -15%, so the difference should indeed be negative.

You can, however, flip the estimate by multiplying it by -1. Example below!

In case anybody stumbles upon this question because they have the same problem, here's the answer: