Bad SYSTEM for DTD DocType and Entity breaks the XML validation.

[Oobiewan]

When I start VS Code and an xml document is open, it is validated as expected, and problems are thrown where needed. However, if I close the file with problems and then open it again, no validation happens anymore, and no problems are shown. The same way, if I fix problems in a file with invalid structures and the validation does not find any problems anymore, I cannot trigger the validation anymore, even if I write content that is not only invalid according to the defined DTD but even invalid xml is not shown as a problem. Basically validation never works except when opening vs code.
I am trying to validate xml files against the DocBook 4.4 DTD.

[angelozerr]

Could you share your xml and dtd please.

I cannot reproduce it.

[Oobiewan]

Thanks for the quick reply. I tested some things and the problem seems to be related to some entity references.
An XML file where the validation fails has the following declarations:

<!DOCTYPE chapter PUBLIC "-//OASIS//DTD DocBook XML V4.4//EN" "http://www.docbook.org/xml/4.4/docbookx.dtd" [
    <!ENTITY % xinclude SYSTEM "http://www.docbook.org/xml/4.4/xinclude.mod">
    %xinclude;
    <!ENTITY % document SYSTEM "document.ent">
    %document;
]>

If I delete the %document; parameter entity reference, the validation works again. The contents of document.ent are:

<!ENTITY % project SYSTEM "project.ent"> %project;
<!-- Adjust the document-specific information! -->
<!ENTITY document-state            "&lt;Draft&gt;" >
<!ENTITY document-revision         "&lt;1.0.3&gt;" >

If I keep the %document; reference in the xml file but remove %project; in document.ent, the validation works again. The contents of project.ent are:

<!ENTITY % company SYSTEM "company.ent"> %company;

<!-- Adjust the project-specific information! -->
<!ENTITY project-name              "best_project">
<!ENTITY project-number            "112344">

Unlike in the previous file, removing the %company; reference does not help, the validation doesn't work. It seems like the validation has problem substituting entities on more than two levels. In Oxygen XML Author (22.1), all of the entity files and their content are resolved without problems, I can use and validate entities from all three levels/files that we have in this project.

[Oobiewan]

I have also tried referencing all entity files directly from the XML file instead, but the result is the same, that is, as long as I refer to %document; only, validation works if document.ent has not further entity references. Referencing %project; and/or %company; in any combination breaks the validation.

[Oobiewan]

Suddenly after restarting VS code (for about the 6th time since I first encountered this error), everything worked perfectly, even though all contents, declarations and references were the same as originally. Another restart of VS code, and everything is back to the previously described problem.

[Oobiewan]

Not sure this is useful but the referenced xinclude entity is mapped to a local file with a catalog. The local file has the following contents:

<!ELEMENT xi:include (xi:fallback?) >
<!ATTLIST xi:include
xmlns:xi            CDATA       #FIXED       "http://www.w3.org/2001/XInclude"
href                CDATA       #REQUIRED
parse               (xml|text)  "xml"
xpointer            CDATA       #IMPLIED
encoding            CDATA       #IMPLIED
accept              CDATA       #IMPLIED
accept-charset      CDATA       #IMPLIED
accept-language     CDATA       #IMPLIED >
<!ELEMENT xi:fallback ANY >
<!ATTLIST xi:fallback
xmlns:xi            CDATA       #FIXED "http://www.w3.org/2001/XInclude" >
<!ENTITY % local.preface.class    "| xi:include" >
<!ENTITY % local.part.class       "| xi:include" >
<!ENTITY % local.chapter.class    "| xi:include" >
<!ENTITY % local.divcomponent.mix "| xi:include" >
<!ENTITY % local.para.char.mix    "| xi:include" >
<!ENTITY % local.info.class       "| xi:include" >

[angelozerr]

Many thanks @Oobiewan for your detailed information.

I can reproduce the problem. I need some investigation to understand the problem.

[angelozerr]

@Oobiewan it seems http://www.docbook.org/xml/4.4/xinclude.mod is not available?

[Oobiewan]

@angelozerr correct, http://www.docbook.org/xml/4.4/xinclude.mod is not available at this specific address, but it is mapped with a catalog file to a local file in the vs code property xml.catalogs.

The contents of this catalog file are:

<?xml version="1.0"?>
<catalog xmlns="urn:oasis:names:tc:entity:xmlns:xml:catalog">
    <system systemId="http://www.docbook.org/xml/4.4/xinclude.mod"
        uri="C:\5.2.10\custom\dtd\4.4\xinclude.mod"/>
</catalog>

In a comment above I have copied the complete contents of this local .mod file. You can create it for yourself and add its location to the catalog if you want to reproduce this part of the implementation I have. But I think you can also refer to the .mod file directly from the XML, or simply remove the xinclude declaration and reference completely from the doctype definition as there is no xinclude element in my document anyway.

[Oobiewan]

One more test I just did confirms that it might be something with the multi-level entity files:
If I merge the contents of project.ent and company.ent into document.ent (and get rid of the declarations and references for those two files), the validation works fine. So it looks like the plugin cannot resolve entity references inside the referenced entity files.

[angelozerr]

Have you set https://github.com/redhat-developer/vscode-xml/blob/master/docs/Validation.md#resolve-external-entities to true?

[Oobiewan]

Yes, I'm pretty sure that was one of the first things I did.

[Oobiewan]

@angelozerr What I either haven't tried before or tried it when there were still other problems with the setup is triggering the re-validation manually with the "Revalidate current XML file" command. Apparently if I trigger this, the validation works properly even with the original setup of the entities. Strangely, if I remove the %document; entity reference, the document is validated in real-time continuously. With the %document; reference in place, I need to trigger it manually. So after all, the validation works, but the automatic validation doesn't. It would be great if I didn't have to trigger it manually each time, but at least it does work in general then.
Allow me to note that aside from this inconvenience, this plugin looks really good with very nice features, thanks a lot to you and all other contributors. And of course, thanks a lot for your support.