Chrome introduced headless mode in v59 opening the possibility of using Chrome as a fast and elegant way of generating PDF data or PNG screenshots programatically via HTML. django-hardcopy is an alternative to other projects which leverage wkhtmltopdf, a great tool but one that lacks the portability, ease of installation, the performance, and reliability of Chrome.
Install the library:
pip install django-hardcopyInstall Chrome or a derivative:
apt-get install chromium-browserSet your Chrome path (optional):
# settings.py
CHROME_PATH = '/path/to/chrome-or-chromium'This can be useful if you want to use chrome-canary or chromium-browser (available by default in Ubuntu). Django-hardcopy will attempt to smartly default the appropriate chrome path for your os. If you're on Mac OSX, just upgrade to the latest Chrome and you're good to go!
Set the rendering window size (optional, default: 1280,720):
# settings.py
CHROME_WINDOW_SIZE = '800,600'The easiest way to use django-hardcopy is to use its CBV mixin:
from django.views.generic import TemplateView
from hardcopy.views import PDFViewMixin, PNGViewMixin
class MyPDFView(PDFViewMixin, TemplateView):
template_name = "pdf_me.html"
class MyPNGView(PNGViewMixin, TemplateView):
template_name = "png_me.html"
height = '1080'
width = '1920'
It works with any Django Class Based View, and implements PDF or PNG rendering on the GET HTTP method. Further, if the ?html querystring variable is provided the mixin will render the view normally for designing and debugging of the raw HTML. The CBV mixin supports several options for extension and customization covered in the FAQ section.
There are two methods which implement a lower level API which can be used directly for PDFs:
Arguments:
input_file: An open for reading "file-like" object to read HTML from for renderingoutput_file: An open for writing "file-like" object to write the PDF to after rendering**extra_args: See belowThis function will read the contents of input_file (an HTML bytestring), render it with Chrome and store the binary PDF data in output_file. Any kwargs are translated as commandline arguments to chrome when starting the headless browser for rendering, i.e.:
from hardcopy import file_to_pdf
extra_args = {
'virtual-time-budget': 6000
}
file_to_pdf(open('myfile.html'), open('myfile.pdf'), **extra_args)
# translates to --virtual-time-budget=6000 when starting chrome
extra_args = {
'disable-gpu': None
}
file_to_pdf(open('myfile.html'), open('myfile.pdf'), **extra_args)
# translates to --disable-gpu when starting chrome (currently on by default and required by Chrome)
Arguments:
html_data: A bytestring of HTML data. Note: We use bytestrings because the most common execution path is to generate a PDF from a rendered Django template responseoutput_file: An open for writing "file-like" object to write the PDF to after rendering**extra_args: See belowThis render the contents of html_data with Chrome and store the binary PDF data in output_file. Any kwargs are translated as commandline arguments to chrome when starting the headless browser for rendering, i.e.:
from hardcopy import bytestring_to_pdf
extra_args = {
'virtual-time-budget': 6000
}
bytestring_to_pdf(b"<html><h1>Hello Chrome!</h1></html>", open('myfile.pdf'), **extra_args)
# translates to --virtual-time-budget=6000 when starting chrome
extra_args = {
'disable-gpu': None
}
bytestring_to_pdf(b"<html><h1>Hello Chrome!</h1></html>", open('myfile.pdf'), **extra_args)
# translates to --disable-gpu when starting chrome (currently on by default and required by Chrome)Similar functions are available for PNG generation:
Arguments:
input_file: An open for reading "file-like" object to read HTML from for renderingoutput_file: An open for writing "file-like" object to write the PNG to after renderingwidth: width of the viewport in pixelsheight: height of the viewport in pixels**extra_args: See aboveArguments:
html_data: A bytestring of HTML data. Note: We use bytestrings because the most common execution path is to generate a PNG from a rendered Django template responseoutput_file: An open for writing "file-like" object to write the PNG to after renderingwidth: width of the viewport in pixelsheight: height of the viewport in pixels**extra_args: See belowHow do I configure a view to download the PDF/PNG file?
Set the download_attachment property to True:
class MyView(PDFViewMixin, TemplateView):
download_attachment = TrueHow do I override the chrome window size defined in a view?
Set the chrome_window_size property to a string of your choice:
class MyView(PDFViewMixin, TemplateView):
chrome_window_size = '1920,1600'How do I customize the file name of the generated PDF/PNG?
Override the get_filename method of your view:
class MyView(PDFViewMixin, TemplateView):
def get_filename(self):
return "my_file_{}.pdf".format(now().strftime('Y-m-d'))How do I add context data with django-hardcopy ?
There's no magic here, simply override the TemplateView.get_context_data method,
like you would do in a normal view:
class MyView(PDFViewMixin, TemplateView):
def get_context_data(self, **kwargs):
context = super(MyView, self).get_context_data(**kwargs)
context['example_data'] = self.request.GET.get('example')
return contextI want to process the rendered HTML content before it is converted to PDF or PNG, how to do this ?
Just override the mixin method process_html_content in your view:
class MyView(PDFViewMixin, TemplateView):
def process_html_content(self, content):
return make_absolute_paths(content)Under the hood, django-hardcopy writes HTML content to a temporary file which is loaded in chrome via the file:// protocol. This does no magic on handling static files, and since Django is not necessarily serving static files at the URL rendered in the template (say if STATIC_URL is set to a relative path like most development envirnoments). In production environments where static assets are served via a media server or s3 bucket at an absoulte URL, this is probably fine.
In local development however, Chrome will recieve a connection refused error on attempts to load static files in templates included like <link href="https://github.com/loftylabs/django-hardcopy/blob/master/{% static 'style.css' %}" rel="stylesheet">. The best workaround for this is to include static assets inline in PDF/PNG templates.
A nice feature for the roadmap of django-hardcopy would be dynamic parsing of templates to convert linked static assets to inline assets automatically. (PRs welcome :))