XML entities declared in a DTD are marked undeclared after XML file change

[ph777]

I have a DTD file base.dtd that declares an XML entity external:

<?xml version="1.0" encoding="utf-8"?>
<!ELEMENT root-element (#PCDATA)>
<!ENTITY external "EXTERNALLY DECLARED ENTITY">

I have an XML file file.xml that references the DTD file and declares an XML entity local. To make the case clear it also uses an undeclared entity undeclared.

<?xml version="1.0" encoding="utf-8"?>
<!DOCTYPE root-element SYSTEM "./base.dtd" [
<!ENTITY local "LOCALLY DECLARED ENTITY">
]>
<root-element>
&local;
&external;
&undeclared;
</root-element>

Now, if I open file.xml in VSCode, the DTD is recognized and the &undeclared; is correctly marked with an EntityNotDeclared error.

However, once I modify file.xml, e.g. by putting a new line or space anywhere (any safe place that does not break well-formdness of XML, e.g. after <root-element> opening tag), the &external; entity gets marked with an EntityNotDeclared error, which is NOT correct.

Is it possible to fix it? This error is really annoying in large XML files with a lot of externally declared XML entities. Once you modify your XML it gets red all over and you easily miss your real errors in XML. Or is this a lemminx problem?

Thanks

[ph777]

I need to either reload VSCode or save file.xml, close it and reopen again to have externally declared XML entities evaluated correctly.

[angelozerr]

DTD entity validation is indeed not very stable, we need to improve it. I need to investigate the problem.

[ph777]

Thanks, that would be great!

[angelozerr]

Thanks @ph777 for your detailed explanation.

I can reproduce it and the bug is not trivial. It comes from the custom LemMinx cache of DTD which is used to avoid parsing referenced DTD grammar each time you type something in the XML editor.

The first time DTD grammar is not in the cache, the validation error for entities is correct. The second time when DTD grammar is on the cache, the Xerces validator doesn't reparse the DTD (because it is on the cache), that's nice but as it doesn't parse the DTD, it doesn't collect the declared entities and you have crazy errors about entities.

But I have some ideas to fix this problem. I will ping you as soon as I will find a fix.

[ph777]

Great news! It sounds promising. Dne čt 26. bře 2020 10:43 uživatel Angelo <notifications@github.com> napsal:

…

Thanks @ph777 <https://github.com/ph777> for your detailed explanation. I can reproduce it and the bug is not trivial. It comes from the custom LemMinx cache of DTD which is used to avoid parsing referenced DTD grammar each time you type something in the XML editor. But I have some ideas to fix this problem. I will ping you as soon as I will find a fix. — You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub <#234 (comment)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/AE3ZQXBDH6YTUEZUSUDGKXDRJMPTJANCNFSM4LTLMFXQ> .

[angelozerr]

@ph777 I create a PR to fix this issue eclipse/lemminx#636, here a demo:

[ph777]

Great news! Thanks!

When can we expect the 0.12.0 version to be released?

[angelozerr]

When can we expect the 0.12.0 version to be released?

Perhaps in 3 weeks. But you can install the current version vsix at
https://download.jboss.org/jbosstools/vscode-xml/staging/?C=M;O=D

It should be good if you could do that to give us feedback before the release. Thanks.