Possible to parse XBRL from string?

[ahoward-ch]

I have a database containing many iXBRL instances in string format. With the current code, is it possible to parse those strings directly into the reader instead of having to save them to an HTML file on disk and then have the reader load them back in?

[manusimidt]

Hmm, that's probably a good suggestion however not that easy to implement.
The main issue is that an XBRL submission does not necessarily consist of only one document.
It can contain relative imports to additional files (taxonomy-schemas, linkbases e.t.c).
Therefore this can not naively be implemented.

Here is an actual example:

Let's say you store the latest iXBRL submission from Apple Inc as a string into a database and want to parse it via py-xbrl.
The submission contains the following XML code:

<link:schemaref xlink:type="simple" xlink:href="./aapl-20220924.xsd"></link:schemaref>

But how should py-xbrl know where to download the file 'aapl-20220924.xsd' ?
Currently, it does this by looking into the same directory as the original file was in. Therefore the file path currently is essential for parsing with py-xbrl.

But yes, with UK submission it would theoretically be perfectly possible to parse them from string because they just consist of one file.

[ahoward-ch]

That makes sense, and I can see it certainly poses issues for the general case.

I have a couple modifications to the taxonomy (I found the UK pre-2010 taxonomy is identical to the 2010 one, but has a different name) which I will submit as a pull request, but then I will create a custom fork for what I need my specific version to do. I can probably unpick the XBRL reader and implement my own.

If you have any suggestions on how best to do this (try to get it to read a string directly instead of loading from file, or try to convert to an XML object in mem and then try to get it to read) I would appreciate the advice!