Add message for generic parsing exceptions

[jelovirt]

Description

When the input XML is in some way invalid but it can still be parsed, for example

<topic id="%50">

the XML parser will throw an exception with message

Attribute value "%50" of type ID must be an NCName when namespaces are enabled.

We currently capture this and log out as an error message.

In order to give as many messages an ID as possible, we add a message ID for generic parsing errors where the <reason> of the message is provided by the XML parser and the <response> is left empty. The logged message will be

Error: file:/Volumes/tmp/src/topic.dita:3:17: [DOTJ088E] Attribute value "%50" of type ID must be an NCName when namespaces are enabled.

or for example with incorrect attribute name that goes against the DTD

Error: file:/Volumes/tmp/src/topic.dita:3:36: [DOTJ088E] Attribute "xml:langx" must be declared for element type "topic".

Motivation and Context

Add error ID for parsing exceptions to improve debugging.

How Has This Been Tested?

Manual testing.

Type of Changes

• New feature (non-breaking change which adds functionality)

Documentation and Compatibility

This is a generic error message, so we don't know in advance what the error message will be. Users may override the message configuration to be e.g.

<reason>Error in parsing file: %1</reason>

if they want to add more context to the message.

[chrispy-snps]

@jelovirt - this is a nice enhancement, especially now being able to override this class of message as desired.

[infotexture]

If we know this is an XML parsing error, should we say so?

Suggested change

	<reason>%1</reason>
	<response></response>
	<reason>XML parsing error: </reason>
	<response>%1</response>

[jelovirt]

The %1 in this case is the error message thrown by the XML parser. It describes what the error condition is, like "Attribute value "%50" of type ID must be an NCName when namespaces are enabled.".

This means it's not a response that describes what the user should do next. Because this is a generic parsing, we don't know what the next step should be. It could be "fix your specialization" or "fix your content".

This error message has been thrown before, just without any message ID. This allows us to make it easier for users to locate a description of this generic error in the message reference. There we can describe that there was something wrong with your DITA file, check that; or, if you're using a custom specialization, there may be some issue there.

[infotexture]

OK, so we leave the placeholder in the <reason> element, but shall we add the intro for context:

Suggested change

	<reason>%1</reason>
	<response></response>
	<reason>XML parsing error: %1</reason>
	<response></response>

[jelovirt]

I would prefer to not use a prefix in the reason, but it may be useful for users 🤷🏽. I'll need to get more examples of those messages to help in deciding whether it's helpful of not.

[jelovirt]

Xerces produces error messages listed at https://github.com/apache/xerces2-j/blob/trunk/src/org/apache/xerces/impl/msg/XMLMessages.properties#L230. The messages are in sentence format.

[jelovirt]

Added XML parsing error: prefix to the message.