inconsistent error/warning throwing when transforming invalid HTML text

[DevTy2977]

Hello All,

I'm currently maintaining a project using Dita 4.2.2. This project consists of using dita to transform user created files. Some times these user created files can contain plain text invalid html. Things such as <b> tags not having closing tags. Due to this, we can encounter Dita error:
Error: file:/C:/foobar..xml:1:349: The content of element type \"b\" must match \"(boolean|cite|k....

I'm attempting to handle, catch and propagate this error back to the user. The issue in my attempt of this is that running the same bad input can result in different errors or warnings. For instance, I chose to log the std out and the log would consists of huge list of warnings looking like:
[DOTX009W]: Cannot retrieve a title from 'concepts/... [DOTX023W]: Unable to retrieve navtitle from target: '... [DOTX027W]: Unable to retrieve linktext from target: '....
Since these are warning, the process continues and produces output with broken links and pages not transformed.

When I don't log the std out, I have to institute a process timeout or the DITA process just hangs forever. When this timeout occurs and I manually kill the process and log the stderror output, I receive:
[DOTJ013E] Failed to parse the referenced resource 'file:....

Any one experienced such behavior and know why logging the stdout vs not logging it can result in Error vs Warnings. And is there anyway to stop the process from continuing when encountering an warning? Seems 'processing-mode=strict' still doesn't catch warnings. I know the issue is unmatching HTML tags but don't understand the discrepancies in DITA behavior.