[GTK] Implement printing using the Print portal

[GeorgesStavracas]

fd47d13

[GTK] Implement printing using the Print portal
https://bugs.webkit.org/show_bug.cgi?id=192748

Reviewed by Michael Catanzaro.

When running inside a sandboxed environment - which is fairly common
nowadays - printing is severely limited due to using the legacy GTK
print API (GtkPrintUnixDialog & family).

Due to WebKit's multi-process architecture, using the modern alternative,
GtkPrintOperation, is not really possible, because GtkPrintOperation has
the enconded assumption that the rendering of the document pages happens
in the same process as of printing.

Implement manual support for the Print portal. The portal has two well
defined steps:

 1. PreparePrint() is where WebKit sends the settings and page setup
    that are stored in WebKitPrintOperation to the portal. The portal
    presents a dialog to let users choose their print preferences. This
    step returns WebKit a token, which is the "authentication" mechanism
    for the next step.

 2. Print() takes a token, and a file descriptor pointing to the file
    that will be send to the printer. The file descriptor can and often
    does point to an in-memory buffer! In WebKit's case, it's a tmpfile.

This effectively means that webkit_print_operation_print(), which assumes
that step 1 can be skipped - cannot work under a sandbox. Deprecate this
function, and adjust the tests to ignore this deprecation, as the tests
should be adapted separately.

* Source/WebKit/UIProcess/API/gtk/WebKitPrintOperation.cpp:
(webkitPrintOperationFinished):
(webkitPrintOperationFailed):
(webkitPrintOperationPrintPagesForFrame):
(findFilePrinter):
(webkitPrintOperationSendPagesToPrintPortal):
(webkitPrintOperationPreparePrint):
(webkitPrintOperationRunPortalDialog):
(webkitPrintOperationRunDialogForFrame):
* Source/WebKit/UIProcess/API/gtk/WebKitPrintOperation.h.in:
* Tools/TestWebKitAPI/Tests/WebKitGtk/TestPrinting.cpp:
(testPrintOperationPrint):
(testPrintOperationErrors):

Canonical link: https://commits.webkit.org/277671@main

10a3cdb

Misc	iOS, tvOS & watchOS	macOS	Linux	Windows
✅ 🧪 style	✅ 🛠 ios	✅ 🛠 mac	✅ 🛠 wpe	✅ 🛠 wincairo
✅ 🧪 bindings	✅ 🛠 ios-sim	✅ 🛠 mac-AS-debug	✅ 🧪 wpe-wk2
✅ 🧪 webkitperl	✅ 🧪 ios-wk2	✅ 🧪 api-mac	✅ 🧪 api-wpe
	✅ 🧪 ios-wk2-wpt	✅ 🧪 mac-wk1	✅ 🛠 wpe-skia
	✅ 🧪 api-ios	✅ 🧪 mac-wk2	✅ 🛠 gtk
	✅ 🛠 tv	✅ 🧪 mac-AS-debug-wk2	✅ 🧪 gtk-wk2
	✅ 🛠 tv-sim		✅ 🧪 api-gtk
	✅ 🛠 watch
✅ 🛠 🧪 unsafe-merge	✅ 🛠 watch-sim

[webkit-early-warning-system]

EWS run on previous version of this PR (hash 8348b27)

Details

Misc	iOS, tvOS & watchOS	macOS	Linux	Windows
✅ 🧪 style	✅ 🛠 ios	✅ 🛠 mac	✅ 🛠 wpe	✅ 🛠 wincairo
	✅ 🛠 ios-sim	✅ 🛠 mac-AS-debug	✅ 🧪 wpe-wk2
✅ 🧪 webkitperl	❌ 🧪 ios-wk2	✅ 🧪 api-mac	✅ 🧪 api-wpe
	🧪 ios-wk2-wpt		✅ 🛠 wpe-skia
	⏳ 🧪 api-ios	✅ 🧪 mac-wk2	✅ 🛠 gtk
	🛠 tv	✅ 🧪 mac-AS-debug-wk2	✅ 🧪 gtk-wk2
	✅ 🛠 tv-sim		❌ 🧪 api-gtk
	🛠 watch
	✅ 🛠 watch-sim

[GeorgesStavracas]

I'm making this a draft because there is one unresolved problem with the Print portal that I think needs to be solved first: the Print portal offers various output formats (PDF, PS, and SVG) and Skia cannot handle that.

I gave this a stab at https://gitlab.gnome.org/GNOME/xdg-desktop-portal-gnome/-/merge_requests/148 but I'm not sure if this approach is even acceptable - it's likely just a workaround to the problem. Maybe we'll need a new parameter to the Print portal itself.

But please share your thoughts about the code, I'll apply review comments as usual.

[GeorgesStavracas]

The warning this patch introduces is making the test suite unhappy...

[webkit-early-warning-system]

EWS run on previous version of this PR (hash 003af88)

Details

Misc	iOS, tvOS & watchOS	macOS	Linux	Windows
✅ 🧪 style	✅ 🛠 ios	✅ 🛠 mac	✅ 🛠 wpe	✅ 🛠 wincairo
	✅ 🛠 ios-sim	✅ 🛠 mac-AS-debug	✅ 🧪 wpe-wk2
✅ 🧪 webkitperl	✅ 🧪 ios-wk2	✅ 🧪 api-mac	✅ 🧪 api-wpe
	✅ 🧪 ios-wk2-wpt		✅ 🛠 wpe-skia
	🧪 api-ios	✅ 🧪 mac-wk2	✅ 🛠 gtk
	✅ 🛠 tv	✅ 🧪 mac-AS-debug-wk2	✅ 🧪 gtk-wk2
	✅ 🛠 tv-sim		❌ 🧪 api-gtk
	✅ 🛠 watch
	✅ 🛠 watch-sim

[mcatanzaro]

Nice, this is way simpler than I was expecting it would be. Using a temporary file was smart.

I think you should deprecate webkit_print_operation_print(). There's no point in not doing so; applications shouldn't do things that don't work when sandboxed.

[GeorgesStavracas]

I don't really know why the style checker skipped so many NULLs and TRUEs and FALSEs

[mcatanzaro]

I don't really know why the style checker skipped so many NULLs and TRUEs and FALSEs

The style checker is not smart enough to check for gboolean vs. bool.

It should be able to check for nullptr, though, because using NULL is never allowed. I'm not sure why it doesn't catch that. Maybe it's just something nobody ever bothered to implement....

[webkit-early-warning-system]

EWS run on previous version of this PR (hash 0517e8f)

Details

Misc	iOS, tvOS & watchOS	macOS	Linux	Windows
✅ 🧪 style	✅ 🛠 ios	✅ 🛠 mac	✅ 🛠 wpe	✅ 🛠 wincairo
✅ 🧪 bindings	✅ 🛠 ios-sim	✅ 🛠 mac-AS-debug	🧪 wpe-wk2
✅ 🧪 webkitperl	✅ 🧪 ios-wk2	✅ 🧪 api-mac	✅ 🧪 api-wpe
	⏳ 🧪 ios-wk2-wpt	✅ 🧪 mac-wk1	✅ 🛠 wpe-skia
	⏳ 🧪 api-ios	✅ 🧪 mac-wk2	✅ 🛠 gtk
	✅ 🛠 tv	✅ 🧪 mac-AS-debug-wk2	🧪 gtk-wk2
	🛠 tv-sim		✅ 🧪 api-gtk
	🛠 watch
	✅ 🛠 watch-sim

[webkit-early-warning-system]

EWS run on previous version of this PR (hash 6a1de80)

Details

Misc	iOS, tvOS & watchOS	macOS	Linux	Windows
✅ 🧪 style	✅ 🛠 ios	✅ 🛠 mac	✅ 🛠 wpe	✅ 🛠 wincairo
✅ 🧪 bindings	✅ 🛠 ios-sim	✅ 🛠 mac-AS-debug	✅ 🧪 wpe-wk2
✅ 🧪 webkitperl	✅ 🧪 ios-wk2	✅ 🧪 api-mac	✅ 🧪 api-wpe
	✅ 🧪 ios-wk2-wpt	✅ 🧪 mac-wk1	✅ 🛠 wpe-skia
	✅ 🧪 api-ios	✅ 🧪 mac-wk2	✅ 🛠 gtk
	✅ 🛠 tv	✅ 🧪 mac-AS-debug-wk2	✅ 🧪 gtk-wk2
	✅ 🛠 tv-sim		✅ 🧪 api-gtk
	✅ 🛠 watch
	✅ 🛠 watch-sim

[carlosgcampos]

Wouldn't it be possible to make the portal print without a dialog? I don't understand why we have to deprecated the feature.

[GeorgesStavracas]

Wouldn't it be possible to make the portal print without a dialog? I don't understand why we have to deprecated the feature.

I don't know for sure, but I don't think this would be possible. The challenge here is that, in a tight sandbox, apps cannot "just" send a print job to the Print portal, because they cannot choose a printer - they don't have access to the hardware in a tight sandbox.

The only printer available will effectively be the file printer, but even in that case, the app won't have an output file path. That's selected by the user through the PreparePrint method.

So, all in all, I don't see how an tightly sandboxed app would be able to skip the PreparePrint method. Any alternative would require opening sandbox holes - be it the CUPS permission, or add more filesystem permissions for printing into a file.

[carlosgcampos]

What I meant is whether it would be possible to add api to the portal to allow printing without showing a dialog.

[GeorgesStavracas]

What I meant is whether it would be possible to add api to the portal to allow printing without showing a dialog.

The lack of a dialog is the problem. How would WebKit, or any sandboxed app, tell the portal which printer or output file it should print to?

[mcatanzaro]

I don't see the point of ever allowing an app to print without first showing the print dialog. Why would an app ever want to do that?

And Debian codesearch indicates the only user of this API is our own API test, so it won't be missed.

[webkit-early-warning-system]

EWS run on previous version of this PR (hash be461e0)

Details

Misc	iOS, tvOS & watchOS	macOS	Linux	Windows
✅ 🧪 style	✅ 🛠 ios	✅ 🛠 mac	✅ 🛠 wpe	✅ 🛠 wincairo
✅ 🧪 bindings	✅ 🛠 ios-sim	✅ 🛠 mac-AS-debug	⏳ 🧪 wpe-wk2
✅ 🧪 webkitperl	🧪 ios-wk2	⏳ 🧪 api-mac	✅ 🧪 api-wpe
	⏳ 🧪 ios-wk2-wpt	✅ 🧪 mac-wk1	✅ 🛠 wpe-skia
	⏳ 🧪 api-ios	✅ 🧪 mac-wk2	✅ 🛠 gtk
	🛠 tv	✅ 🧪 mac-AS-debug-wk2	⏳ 🧪 gtk-wk2
	✅ 🛠 tv-sim		✅ 🧪 api-gtk
	🛠 watch
	🛠 watch-sim

[GeorgesStavracas]

I took the liberty to take a few executive decisions about this PR:

• Removed the code that depended on newer portal features. I'll create a new issue, and submit a separate PR for that.
  • This code was the Skia-related stuff, so this PR should be good to backport again
• Reverted back to iterating the current main context. Pushing a new main context freezes the UI and triggers the unresponsive app dialog.
• Fixed all remaining FIXMEs in code

I think this is good to go now.

[webkit-early-warning-system]

EWS run on previous version of this PR (hash 7c14592)

Details

Misc	iOS, tvOS & watchOS	macOS	Linux	Windows
✅ 🧪 style	✅ 🛠 ios	✅ 🛠 mac	✅ 🛠 wpe	✅ 🛠 wincairo
✅ 🧪 bindings	✅ 🛠 ios-sim	✅ 🛠 mac-AS-debug	❌ 🧪 wpe-wk2
✅ 🧪 webkitperl	✅ 🧪 ios-wk2	✅ 🧪 api-mac	✅ 🧪 api-wpe
	✅ 🧪 ios-wk2-wpt	✅ 🧪 mac-wk1	✅ 🛠 wpe-skia
	✅ 🧪 api-ios	✅ 🧪 mac-wk2	✅ 🛠 gtk
	✅ 🛠 tv	✅ 🧪 mac-AS-debug-wk2	✅ 🧪 gtk-wk2
	✅ 🛠 tv-sim		✅ 🧪 api-gtk
	✅ 🛠 watch
	✅ 🛠 watch-sim

[mcatanzaro]

Please check with @carlosgcampos before landing, because he might want to review this first. But you pass my review.

[carlosgcampos]

Maybe it woulld be easier if this is a struct instead of a class, and you can create instances using { } without adding explicit constructors

[GeorgesStavracas]

Done, it this what you had in mind?

[webkit-early-warning-system]

EWS run on current version of this PR (hash 10a3cdb)

Details

Misc	iOS, tvOS & watchOS	macOS	Linux	Windows
✅ 🧪 style	✅ 🛠 ios	✅ 🛠 mac	✅ 🛠 wpe	✅ 🛠 wincairo
✅ 🧪 bindings	✅ 🛠 ios-sim	✅ 🛠 mac-AS-debug	✅ 🧪 wpe-wk2
✅ 🧪 webkitperl	✅ 🧪 ios-wk2	✅ 🧪 api-mac	✅ 🧪 api-wpe
	✅ 🧪 ios-wk2-wpt	✅ 🧪 mac-wk1	✅ 🛠 wpe-skia
	✅ 🧪 api-ios	✅ 🧪 mac-wk2	✅ 🛠 gtk
	✅ 🛠 tv	✅ 🧪 mac-AS-debug-wk2	✅ 🧪 gtk-wk2
	✅ 🛠 tv-sim		✅ 🧪 api-gtk
	✅ 🛠 watch
✅ 🛠 🧪 unsafe-merge	✅ 🛠 watch-sim

[webkit-commit-queue]

Committed 277671@main (fd47d13): https://commits.webkit.org/277671@main

Reviewed commits have been landed. Closing PR #26765 and removing active labels.

[aperezdc]

Backported as 2de9573 into the 2.44 release branch 👏🏼