add support for PDF export

[godiard]

Paola Bruccoleri from Uruguay has reported a solution using a web site:

...usé: http://www.web2pdfconvert.com/

entonces.. en el archivo webtoolbar.py agregué:

el botón:
       self._savepdf = ToolButton('filesave')
       self._savepdf.set_tooltip(_('Save PDF'))
       self._savepdf.connect('clicked', self._savepdf_clicked)
       self.insert(self._savepdf, -1)
       self._savepdf.show()

la función:
   def _savepdf_clicked(self, button):
       uri = self._browser.web_navigation.currentURI
       dir = 'http://www.web2pdfconvert.com/?u='+uri
       dir = 'http://html-pdf-converter.com/en/convert?u='+uri
       self._browser.load_uri(dir)
       self._browser.grab_focus()

[greenfeld]

I would not recommend using a solution which depends on a web site; we will never know if we can access it or not, if the site may go down, start sticking advertisements in files, etc.

If we integrate such functionality using an Internet system, we also probably should get permission from the site's owner.

Firefox in Linux can save natively to a PDF file. Can we access that from libxulrunner?

[godiard]

I agree. I think the right solution is use gecko to print to a pdf file. I have added the code from the teacher only for reference,

[godiard]

This information can be useful

 http://www.geckofx.org/viewtopic.php?id=980

[sascha_silbe]

For reading web pages offline there are other, potentially better solutions (e.g. Lucians  Webified work, wwwoffle). Nevertheless I took a stab at implementing PDF support.

This is the code I added to browser.Browser:

    def export_pdf(self):
        cls = components.classes['@mozilla.org/gfx/printsettings-service;1']
        setting_service = cls.getService(interfaces.nsIPrintSettingsService)
        req = self.get_dom_window().QueryInterface(interfaces.nsIInterfaceRequestor)
        print_iface = req.getInterface(interfaces.nsIWebBrowserPrint)

        temp_dir = os.path.join(activity.get_activity_root(), 'instance')
        pdf_file = tempfile.NamedTemporaryFile(suffix='.pdf', dir=temp_dir,
                                               delete=False)

        settings = setting_service.newPrintSettings
        settings.printSilent = True
        settings.printToFile = True
        settings.toFileName = pdf_file.name
        settings.outputFormat = interfaces.nsIPrintSettings.kOutputFormatPDF

#        print_iface.print(printSettings, None)
        getattr(print_iface, 'print')(printSettings, None)

It fails with the following error:

Traceback (most recent call last):
  File "/home/sascha.silbe/sugar-jhbuild/install/share/sugar/activities/Browse.activity/webtoolbar.py", line 494, in _export_pdf_cb
    browser.export_pdf()
  File "/home/sascha.silbe/sugar-jhbuild/install/share/sugar/activities/Browse.activity/browser.py", line 320, in export_pdf
    logging.debug('print: %r', getattr(print_iface, 'print'))
  File "/usr/lib/pymodules/python2.6/xpcom/client/__init__.py", line 374, in __getattr__
    return getattr(interface, attr)
  File "/usr/lib/pymodules/python2.6/xpcom/client/__init__.py", line 466, in __getattr__
    unbound_method = BuildMethod(method_info, self._iid_)
  File "/usr/lib/pymodules/python2.6/xpcom/client/__init__.py", line 125, in BuildMethod
    codeObject = compile(method_code, "<XPCOMObject method '%s'>" % (name,), "exec")
  File "<XPCOMObject method 'print'>", line 2
    def print(self, Param1, Param2):
            ^
SyntaxError: invalid syntax

The problem is that print is a reserved keyword in Python 2. XPCOM would need to rename print to something else, e.g. print_. I'm not sure how feasible it is to get that fix rolled out to the distributions that still ship a python-xpcom we can use in the first place (Ubuntu doesn't ship it at all and recent versions of Fedora are broken, too).

[lucian]

The way I see it, converting to PDF isn't ideal: 1) it's a lossy conversion and 2) it'll be opened by a different activity.

As Sacha says, in Webified there's a proof of concept for saving complete web pages (just like firefox), putting them in a zip and later opening them with Browse:
 http://git.sugarlabs.org/browse/webified/commit/25450e1f18701402eb3bdba3a7c3297f61ac752a
 http://git.sugarlabs.org/browse/webified/commit/7005fb13a31f26704ec561d8e71bb4f82274fb3c

I don't have time for it now, but it shouldn't be hard to port it to Browse mainline. The main problem with the patch(es) as it is now is that it doesn't clean up after it correctly.

[garycmartin]

Replying to lucian:

The way I see it, converting to PDF isn't ideal: 1) it's a lossy conversion and 2) it'll be opened by a different activity.

As Sacha says, in Webified there's a proof of concept for saving complete web pages (just like firefox), putting them in a zip and later opening them with Browse:
 http://git.sugarlabs.org/browse/webified/commit/25450e1f18701402eb3bdba3a7c3297f61ac752a
 http://git.sugarlabs.org/browse/webified/commit/7005fb13a31f26704ec561d8e71bb4f82274fb3c

I don't have time for it now, but it shouldn't be hard to port it to Browse mainline. The main problem with the patch(es) as it is now is that it doesn't clean up after it correctly.

...support for zipped up web pages would make a _very_ useful way to quickly provide/distribute browser compatible content (and keep it in a format that could be edited and re-shared).

[godiard]

Replying to lucian:

The way I see it, converting to PDF isn't ideal: 1) it's a lossy conversion and 2) it'll be opened by a different activity.

As Sacha says, in Webified there's a proof of concept for saving complete web pages (just like firefox), putting them in a zip and later opening them with Browse:
 http://git.sugarlabs.org/browse/webified/commit/25450e1f18701402eb3bdba3a7c3297f61ac752a
 http://git.sugarlabs.org/browse/webified/commit/7005fb13a31f26704ec561d8e71bb4f82274fb3c

I don't have time for it now, but it shouldn't be hard to port it to Browse mainline. The main problem with the patch(es) as it is now is that it doesn't clean up after it correctly.

Yeah, probably you are right, and I think firefox already can open zipped media (jar support)

[lucian]

Replying to godiard:

Replying to lucian:

The way I see it, converting to PDF isn't ideal: 1) it's a lossy conversion and 2) it'll be opened by a different activity.

As Sacha says, in Webified there's a proof of concept for saving complete web pages (just like firefox), putting them in a zip and later opening them with Browse:
 http://git.sugarlabs.org/browse/webified/commit/25450e1f18701402eb3bdba3a7c3297f61ac752a
 http://git.sugarlabs.org/browse/webified/commit/7005fb13a31f26704ec561d8e71bb4f82274fb3c

I don't have time for it now, but it shouldn't be hard to port it to Browse mainline. The main problem with the patch(es) as it is now is that it doesn't clean up after it correctly.

Yeah, probably you are right, and I think firefox already can open zipped media (jar support)

At first I'd tried jar:, but it turns out it doesn't work as expected. Mozilla didn't intend it to be used for anything other than extensions and because of certain security features, they aren't usable for this. Unzipping files in Browse is a better choice, and it's not a big issue.

Replying to sascha_silbe:

The problem is that print is a reserved keyword in Python 2. XPCOM would need to rename print to something else, e.g. print_. I'm not sure how feasible it is to get that fix rolled out to the distributions that still ship a python-xpcom we can use in the first place (Ubuntu doesn't ship it at all and recent versions of Fedora are broken, too).

For actually printing PDFs, a workaround like setattr(self, 'print', print_) should work.

[sascha_silbe]

Replying to sascha_silbe:

The problem is that print is a reserved keyword in Python 2. XPCOM would need to rename print to something else, e.g. print_. I'm not sure how feasible it is to get that fix rolled out to the distributions that still ship a python-xpcom we can use in the first place (Ubuntu doesn't ship it at all and recent versions of Fedora are broken, too).

For actually printing PDFs, a workaround like setattr(self, 'print', print_) should work.

As expected, this still fails in getattr():

        setattr(print_iface, 'print_', getattr(print_iface, 'print'))

Traceback (most recent call last):
  File "/home/sascha.silbe/sugar-jhbuild/install/share/sugar/activities/Browse.activity/webtoolbar.py", line 506, in _export_pdf_cb
    browser.export_pdf()
  File "/home/sascha.silbe/sugar-jhbuild/install/share/sugar/activities/Browse.activity/browser.py", line 323, in export_pdf
    logging.debug('print: %r', getattr(print_iface, 'print'))
  File "/usr/lib/pymodules/python2.6/xpcom/client/__init__.py", line 374, in __getattr__
    return getattr(interface, attr)
  File "/usr/lib/pymodules/python2.6/xpcom/client/__init__.py", line 466, in __getattr__
    unbound_method = BuildMethod(method_info, self._iid_)
  File "/usr/lib/pymodules/python2.6/xpcom/client/__init__.py", line 125, in BuildMethod
    codeObject = compile(method_code, "<XPCOMObject method '%s'>" % (name,), "exec")
  File "<XPCOMObject method 'print'>", line 2
    def print(self, Param1, Param2):
            ^
SyntaxError: invalid syntax

The problem is that the code generated on-the-fly by python-xpcom uses the reserved keyword print.

[sascha_silbe]

I have reported the syntax error issue at Debian as  Debian#628484 and found a hack to work around it.

[manuq]

This is doable in current Browse that features WebKit.

[erikos]

Not a regression, moving out at this point of the cycle.

[manuq]

This can be a good reference:

 https://raw.github.com/potyl/Webkit/master/screenshot.pl

See the usage of Gtk.OffscreenWindow with the webkit view inside, and the cairo PdfSurface.

[manuq]

Does not fit current goals, moving to 1.0 .