[Poppler-bugs] [Bug 79936] Add support for printing to a Windows printer from pdftocairo

[bugzilla-daemon at freedesktop.org]

https://bugs.freedesktop.org/show_bug.cgi?id=79936

--- Comment #5 from rodrigo <rodrigorivascosta at gmail.com> ---
(In reply to comment #2)

> If an existing option does not support all the functionality we can
> make the -printopt version override the old one (maybe with a warning
> about duplicated options).

Done!

> I would like the win32 code to be in a separate file...

Done!

> The #ifdefs should test for CAIRO_HAS_WIN32_SURFACE instead of _WIN32...

Done!

> I'm not sure how easy it is to make the build system compile
> pdftocairowin32.cc only when CAIRO_HAS_WIN32_SURFACE is
> defined...

I simply added `#include "pdftocairowin32.cc"` to the main source. Some people
will frown upon it, but I think it fits nicely the
one-source-file-but-not-quite character of pdftocairo. Moreover, this way we
avoid the need to extern-declare all the globals and functions.

> I recommend using git diff...

Done!

> It would be nicer if parseSource used a table...

Done!

> The paper size in devmode should be set to the values in paperWidth
> and paperHeight...

Done, I think.

> The paper size is also used to scale and translate... the pdf page
> depending on the -expand/-noshrink/nocenter options. For this to work
> correctly the paper width and height returned by getOutputSize()
> should match the device context. ie the values obtain from
> GetDeviceCaps should be used.

Sorry, I get lost with all these options and transformations. I tried to do
what you suggested but I couldn't get it right. Sometimes the text is fine but
the images go crazy, sometimes is the other way around. The only way that I got
it to work for every PDF I tested is the one in the original patch: modifying
the cairo matrix.

I suspect that cairo may be playing with the HDC transformation too, but I
didn't check it...

> Note that the windows printing device context positions the origin at
> the top left of the printable area. This works well for the default
> pdftocairo settings where large pages are shrunk to fit the paper
> size. If -noshrink is used it would be better to use the full width
> and height of the paper. ie printing an A4 pdf to A4 paper with
> -noshrink should not scale the document even if it means any content
> outside the printable area will be lost.
> 
> I would recommend doing the scaling from physical size to points on
> the hdc. This avoid the need make win32 specific changes to the cairo
> matrix transformation code. I suggest something like this after creating the
> hdc:
> 
> [code snip] 
>
> and use the outputPaperWidth/Height in getOutputSize().

I tried that, too, but it didn't work. One problem is that `getOutputSize()` is
called *before* CreateDC(), but all these computations require the HDC. So you
have a kind of chicken-egg situation...

> EndDoc and DeleteDC should be called after cairo_surface_finish and
> cairo_surface_destroy.

I tried moving it, but I fails badly. The problem is that after
`cairo_surface_finish()` calling `cairo_win32_surface_get_dc()` will return
NULL. I don't know if the HDC is really destroyed, or if it would leak. With a
display HDC it would be bad enough, but with a printer HDC it is terrible, as
the printer job may never finish.


> The pdftocairo.1 man page needs to be updated...

Done! I'm notably bad at writing man pages, so feel free to improve.

> In DOCINFO put the actual pdf filename if available. If reading from
> standard input I suggest using the document name "pdftocairo <stdin>".

Done!

> Win32 calls that have both ansi and unicode versions should have an
> "A" or "W" to specify which one we are using. eg DeviceCapabilitiesA.

I don't see the point, but it is easy, so, done!
I also added a check to the return value of `StartDocA()` because it will
return an error with the XPS driver if the user presses "Cancel" in the "Save
File" dialog.

> The "Error: one of the output format ..." error message should only
> include "-print" if this option has been compiled in.

That would be easy enough, but I feel it a bit unfair. -ps, -eps, -pdf and -svg
options can also be disabled at compile time but they are unconditionally
listed. (Just half-kidding).

--

That all said, there is still room for improvement. For example, I found that
there is that function ResetDC(), that can be used just before `StartPage()` to
change the DEVMODE for each page. That is way beyond my original intents, but
it is nice to know.

-- 
You are receiving this mail because:
You are the assignee for the bug.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.freedesktop.org/archives/poppler-bugs/attachments/20140730/159d097e/attachment.html>