v3.0.0 - Fails to extract facts for Older FIlings

[rj3185]

XBRL document
The issue is observed with multiple older filings, as an example please look at 0001104659-23-073356 Also, while iterating consider it as an inline XBRL .. hence reference .. filing/0001104659-23-073356\\vod-20230331x20f_htm.xml

Describe the bug
This process fails at the line highlighted ..

with the error
` root: ET.Element | None = parse_file(instance_path).getroot()
if root is None:
raise InstanceParseException("Could not parse instance file!")
# get the link to the taxonomy schema and parse it
schema_ref: ET.Element | None = root.find(LINK_NS + "schemaRef")
if schema_ref is None:

      raise InstanceParseException("Document does not import any taxonomy schema")

E xbrl.InstanceParseException: Document does not import any taxonomy schema`

@manusimidt - Please look into this issue :) Thanks so much

[manusimidt]

Hi,

unfortunately the XBRL file is just invalid, since it has no schemaRef.

According to XBRL Specification, every XBRL document must have a schemaRef defined, without it, there is no way of parsing the document.

So the parser is correctly throwing a xbrl.InstanceParseException.

Interestingly the iXBRL version works without issues. I would anyways always recommend iXBRL over XBRL. Since 2018 the SEC also requires all filers to use iXBRL, not XBRL.

[rj3185]

@manusimidt - Thanks for the information .. let me check and get back to you .. Can we keep this ticket open .. for at least one more week ? The ONLY reason .. is that we are presented the filing by PDS as a "dissem" so .. I would like to look at our code and figure.. what are we doing here ..

[bailey-coding]

I unfortunately don't have a solution, but am looking at a UK Companies House XBRL file that might be hitting the same issue. There is a schemaRef field in the document, so I'll share if I find a solution.

I've made a bit more progress, where parsing the file with the library doesn't work, but this does:

from lxml import etree
from xbrl.instance import LINK_NS

doc = etree.parse("./example_file.xml")
root = doc.getroot()

# This first throws, because of an unrelated 'TypeError: empty namespace prefix is not supported in XP'
root.xpath("link:schemaRef", namespaces=root.nsmap)

# So if we copy the nsmap, and drop the None key
nss = root.nsmap
nss.pop(None)

# This still returns None, since it's not a direct child
root.xpath("link:schemaRef", namespaces=root.nsmap)

# But if we look deeper, we can find it!
root.xpath("//link:schemaRef", namespaces=nss)

[manusimidt]

Hmm ok interesting. Maybe the schemaRef is nested deeper in those particular documents.
@bailey-coding could you share the file you are trying to parse? Then I can develop a solution.
Either attach it here or you can also send it to me at hello@schmidt-manuel.de :)

[bailey-coding]

Hopefully these details help a little!

So here's what my debug script prints using https://github.com/manusimidt/py-xbrl/blob/main/tests/data/example.xml from the tests:

# python ./example.py ./working_example.xml
Processing ./working_example.xml
ERROR:root:The taxonomy with namespace Could not find taxonomy schema at example.xsd could not be found. Please check if it is imported in the schema file
DEBUG:root:link:schemaRef -> /xbrl/schemaRef
DEBUG:root://link:schemaRef -> /xbrl/schemaRef

and for one of the files I'm trying to parse:

# python ./example.py ./09923116_aa_2025-03-28.xhtml
Processing ./09923116_aa_2025-03-28.xhtml
INFO:xbrl.taxonomy:Ignoring unsupported linkbase: http://www.xbrl.org/2003/role/referenceLinkbaseRef
...
DEBUG:root:link:schemaRef -> []
DEBUG:root://link:schemaRef -> /html/body/div/header/references/schemaRef

Example script is here - https://gist.github.com/bailey-coding/57734eadc0ee9d3405504b55c78ca5c7

Note: I managed to make some progress with the XML parsing in https://github.com/uktrade/stream-read-xbrl, but it's much more opinionated in some ways so also hitting some challenges with that one.