Big refactoring and two new options.

giuliov commented 2 years ago

Don't be scared by the size of the PR: I did quite a lot of refactoring/reorganizing. I am sorry not being able to submit a smaller PR but I need to break down the WikiPDFExporter into manageable and unit-testable chunks.

It adds two new options:

--include-unlisted-pages use a different algorithm to find markdown pages; it should better match current Wiki behaviour.
--exclude-paths accepts a list of regular expressions and skip matching pages.

For both options, you'll find unit and integration tests. Part of the refactoring was to organize the test files, so it should be easier to add more integration tests.

I have a few more ideas like optimizing memory allocations for big Wikis, but it will be a different PR.

Hope you'll like my changes and approve them.

giuliov commented 2 years ago

The latest commits added to the PR bring in:

an major architectural change: Chrome loads the temporary HTML from file system instead of keeping it in memory and send it via SetContentAsync API. Why? To solve the 100MB limit. I tested with my main Wiki which generates a 160MB HTML file (670MB of PDF) in less than 5 mins on my machine.
--chrome-timeout option
log outputs relative time to help measuring the steps

MaxMelcher commented 2 years ago

wow - what big one... thanks a lot @giuliov

MaxMelcher / AzureDevOps.WikiPDFExport

Big refactoring and two new options. #105