Problem with base-uri in assertion

[sgtraptor]

Hi,
first of all thanks for 3.0.0 release! Right now we have a problem with validation against assertion that check filename.

$xml_file_path := upper-case(string(base-uri())),

Validate function acts as if there was no filename. Is there any way to set this up so base-uri returns filename properly?

[brunato]

Hi,

It could be an issue of the XPath processor (elementpath). The function base-uri() returns the base-uri property of the XPath node (if the context item is an XPath node).
Usually this property is related with the presence of a xml:base attribute in the XML data.
I need a chunk of the XML data for reproducing the issue.

thank you

[sgtraptor]

Hi @brunato,
in archive there is:

• test.xml - has schemalocation set to xmlschema-test.xsd
• xmlschema-test.xsd - there is one test that checks if xml's filename is "test"
• script-test.py - thats a script that we run the validation with

base-uri-test.zip

[brunato]

A resolution for this will require an enhancement of the XPath processor (maybe an additional argument for DocumentNode class), and then a change in xmlschema.resources.XMLResource.
It could be in a next minor version.

As a workaround until that, you could add a specific xml:base attribute in XML root element, e.g.:

import xmlschema
my_schema = xmlschema.XMLSchema11('xmlschema-test.xsd',use_xpath3=True)
resource = xmlschema.XMLResource('test.xml')
resource.root.attrib['{http://www.w3.org/XML/1998/namespace}base'] = resource.uri
my_schema.validate(resource)

[sgtraptor]

Thanks for quick reply, but we seem to have a problem with parsing our xsd first (we thought this happens because of base-uri but if we only leave XMLSchema11 we get an error as well), so this happens even before xml is validated. This is XSD is valid because we can validate it with Altova.

`Traceback (most recent call last):
File "/Users/def/Library/Python/3.9/lib/python/site-packages/xmlschema/validators/assertions.py", line 115, in build
self.parse_error(err)
File "/Users/def/Library/Python/3.9/lib/python/site-packages/xmlschema/validators/xsdbase.py", line 194, in parse_error
raise error
xmlschema.validators.exceptions.XMLSchemaParseError: [err:FORG0006] the second argument must be xs:numeric:

Schema component:

<xs:assert xmlns:xs="http://www.w3.org/2001/XMLSchema" test="let $xml_file_path := string(base-uri()), $last_slash_pos := index-of(string-to-codepoints(upper-case(string(base-uri()))), string-to-codepoints('/'))[last()], $xml_file_name := substring($xml_file_path, $last_slash_pos + 1), $schema_file_location := @xsi:noNamespaceSchemaLocation, $schema_last_slash_pos := index-of(string-to-codepoints($schema_file_location), string-to-codepoints('/'))[last()], $schema_name := substring($schema_file_location, $schema_last_slash_pos + 1), $schema_type := substring-before(substring-after(substring-after($schema_name, '.'), '.'), '.'), $schema_type_upper := upper-case($schema_type), $regex_pattern := concat('^', $schema_type_upper, '_.*') return matches($xml_file_name, $regex_pattern)">
xs:annotation
<xs:documentation source="Definition" xml:lang="PL">Typ raportu z nazwy pliku nie zgadza się z typem raportu wybranej XML schema. Typ raportu powinien być określony w nazwie pliku XML wielkimi literami.</xs:documentation>
<xs:documentation source="Name" xml:lang="PL">SPRAWOZDANIE_00001</xs:documentation>
</xs:annotation>
</xs:assert>`

XSD that is causing these problems:
knf.fi.DniWycenyFIZ.zip

[brunato]

The problem maybe the same and happens during the XPath static analysis. I've no workaround for this, you have to wait for a fix in elementpath package.

[brunato]

Hi,
the latest version (v3.0.2) should process the assertions of knf.fi.DniWycenyFIZ schema. The schema analysis in elementpath has been changed so many errors are not considered anymore during this phase (see the release notes), so the assertions parsing do not fails anymore.

Probably there could be other thing to adjust on this, also for assertion dynamic evaluation on XML data, so I wait for your feedback on XML validation cases.

[brunato]

Seem to be fine now, if not re-open this issue.