HTML/DOM Collection

[benjamingr]

Talking to @gioragutt regarding HTML collection and writing things down:

• add a per-step hook (like screenshots/console-logs) that calls Page.captureSnapshot as mhtml and save it (don't use fast-mhtml yet). Make sure to add it to the config file and make sure it's off by default.
• Open a PR.
• On top of it - add fast-mhtml and add a step (in the viewer) that parses the mhtml using it to display the output of Page.captureSnapshot.

Takeaways:

• Probably better to use Page.captureSnapshot and not DOMSnapshot given the standard format and the lack of info on parsing the output of DOMSnapshot.
• Probably better to save the MHTML as a separate file like screenshots rather than inline it in the results file.

[gioragutt]

Impl related question - CDPSession, can I acquire it early (as early as attach) and keep reference to it in the hook? Or should I get it every time I invoke the hook (probably not)?

In the util.ts code you showed me, I didn't see you closing the session, and I don't get the feeling from CDPSession['attach'] that you have to call to free up resources or whatever.

[benjamingr]

You can probably access the existing CDP session from the page - though caching (a WeakMap of Page is fine)

[Bnaya]

This is merged, but we leave that issue open until we have docs in place