def run(directory, output):
"""
Compiles a directory of HTML into a single pdf.
:param directory: Directory to search.
:param output: filename to write to.
"""
files = [f for f in os.listdir(directory) if f[-5:] == '.html']
files.sort(key=lambda x: int(x[:4]))
with ChargingBar('Converting', max=len(files)) as progress:
for file in files:
html = weasyprint.HTML(filename=f'{directory}/{file}')
html.write_pdf(target=f"{directory}/{file[:-5]}.pdf")
progress.next()
merger = PdfFileMerger()
with ChargingBar('Merging', max=len(files) + 1) as progress:
for file in files:
reader = open(f"{directory}/{file[:-5]}.pdf")
merger.append(reader,
bookmark=f"{file[5:-5]}",
import_bookmarks=False)
progress.next()
merger.write(output)
progress.next()
for file in files:
os.remove(f"{directory}/{file[:-5]}.html")
os.remove(f"{directory}/{file[:-5]}.pdf")
def render(self, form_data):
merger = PdfFileMerger()
o = PdfTicketOutput(self.event)
qs = OrderPosition.objects.filter(
order__event=self.event,
order__status=Order.STATUS_PAID).select_related(
'order', 'item', 'variation')
for op in qs:
if op.addon_to_id and not self.event.settings.ticket_download_addons:
continue
if not op.item.admission and not self.event.settings.ticket_download_nonadm:
continue
with language(op.order.locale):
buffer = BytesIO()
p = o._create_canvas(buffer)
layout = o.layout_map.get(op.item_id, o.default_layout)
o._draw_page(layout, p, op, op.order)
p.save()
outbuffer = o._render_with_background(layout, buffer)
merger.append(ContentFile(outbuffer.read()))
outbuffer = BytesIO()
merger.write(outbuffer)
merger.close()
outbuffer.seek(0)
return '{}_tickets.pdf'.format(
self.event.slug), 'application/pdf', outbuffer.read()
def _create_pdf_from_rtf_files(self):
pdfs = []
self.progress.emit(0)
for count, file in enumerate(self.files):
changed_file = change_filetype(file, "pdf", self.engine)
pdfs.append(changed_file)
self.progress.emit(count + 1)
merger = PdfFileMerger()
pages = []
chapters = []
for file in pdfs:
read_pdf = PdfFileReader(file)
txt = read_pdf.getPage(0)
page_content = txt.extractText()
try:
chapter = helper_functions.get_chapter_from_pdf_txt(
page_content)
chapters.append(chapter)
except:
chapter = os.path.basename(file)
chapter = chapter.split(".")[0]
chapter = chapter.replace("_", " ")
chapters.append(chapter)
pages.append(read_pdf.getNumPages())
merger.append(fileobj=file)
self.pages = pages
self.chapters = chapters
if not self.create_toc:
merger.write(self.master_file_name)
else:
merger.write("tmp.pdf")
merger.close()
self.trash += pdfs
def PDFsMerger(pathx: str, pathy: str, dest_folder: str):
merger = PdfFileMerger()
merger.append(pathx)
merger.append(pathy)
name_pathx = pathx.split("\\")[-1].split(".")[0]
name_pathy = pathy.split("\\")[-1].split(".")[0]
merger.write("{}\\{}_{}_merged.pdf".format(dest_folder, name_pathx, name_pathy))
merger.close()
def render(self, form_data):
merger = PdfFileMerger()
qs = OrderPosition.objects.filter(
order__event__in=self.events
).prefetch_related(
'answers', 'answers__question'
).select_related('order', 'item', 'variation', 'addon_to')
if form_data.get('include_pending'):
qs = qs.filter(order__status__in=[Order.STATUS_PAID, Order.STATUS_PENDING])
else:
qs = qs.filter(order__status__in=[Order.STATUS_PAID])
if form_data.get('order_by') == 'name':
qs = qs.order_by('attendee_name_cached', 'order__code')
elif form_data.get('order_by') == 'code':
qs = qs.order_by('order__code')
elif form_data.get('order_by', '').startswith('name:'):
part = form_data['order_by'][5:]
qs = qs.annotate(
resolved_name=Coalesce('attendee_name_parts', 'addon_to__attendee_name_parts', 'order__invoice_address__name_parts')
).annotate(
resolved_name_part=JSONExtract('resolved_name', part)
).order_by(
'resolved_name_part'
)
o = PdfTicketOutput(Event.objects.none())
for op in qs:
if not op.generate_ticket:
continue
if op.order.event != o.event:
o = PdfTicketOutput(op.event)
with language(op.order.locale, o.event.settings.region):
layout = o.layout_map.get(
(op.item_id, op.order.sales_channel),
o.layout_map.get(
(op.item_id, 'web'),
o.default_layout
)
)
outbuffer = o._draw_page(layout, op, op.order)
merger.append(ContentFile(outbuffer.read()))
outbuffer = BytesIO()
merger.write(outbuffer)
merger.close()
outbuffer.seek(0)
if self.is_multievent:
return '{}_tickets.pdf'.format(self.events.first().organizer.slug), 'application/pdf', outbuffer.read()
else:
return '{}_tickets.pdf'.format(self.event.slug), 'application/pdf', outbuffer.read()
def render(self, form_data):
merger = PdfFileMerger()
o = PdfTicketOutput(self.event)
qs = OrderPosition.objects.filter(
order__event=self.event
).prefetch_related(
'answers', 'answers__question'
).select_related('order', 'item', 'variation', 'addon_to')
if form_data.get('include_pending'):
qs = qs.filter(order__status__in=[Order.STATUS_PAID, Order.STATUS_PENDING])
else:
qs = qs.filter(order__status__in=[Order.STATUS_PAID])
if form_data.get('order_by') == 'name':
qs = qs.order_by('attendee_name_cached', 'order__code')
elif form_data.get('order_by') == 'code':
qs = qs.order_by('order__code')
elif form_data.get('order_by', '').startswith('name:'):
part = form_data['order_by'][5:]
qs = qs.annotate(
resolved_name=Coalesce('attendee_name_parts', 'addon_to__attendee_name_parts', 'order__invoice_address__name_parts')
).annotate(
resolved_name_part=JSONExtract('resolved_name', part)
).order_by(
'resolved_name_part'
)
for op in qs:
if op.addon_to_id and not self.event.settings.ticket_download_addons:
continue
if not op.item.admission and not self.event.settings.ticket_download_nonadm:
continue
with language(op.order.locale):
buffer = BytesIO()
p = o._create_canvas(buffer)
layout = o.layout_map.get(
(op.item_id, op.order.sales_channel),
o.layout_map.get(
(op.item_id, 'web'),
o.default_layout
)
)
o._draw_page(layout, p, op, op.order)
p.save()
outbuffer = o._render_with_background(layout, buffer)
merger.append(ContentFile(outbuffer.read()))
outbuffer = BytesIO()
merger.write(outbuffer)
merger.close()
outbuffer.seek(0)
return '{}_tickets.pdf'.format(self.event.slug), 'application/pdf', outbuffer.read()
def create_toc_pdf_and_append_it(self):
link_locations, page_locations = self._create_toc_pdf_for_rtf()
link_locations = [
change_coordinates(x, self.toc_orientation) for x in link_locations
] # Change the coordinate
merger = PdfFileMerger()
merger.append("toc.pdf")
merger.append("tmp.pdf")
merger.write("tmp2.pdf")
merger.close()
self._create_hyperlinks(link_locations, page_locations)
self.trash += ["tmp.pdf", "tmp2.pdf", "toc.pdf"]
def prepare(self, address_override=None):
"""Prepare the PDF to be sent by appending attachments"""
# generate the pdf and merge all pdf attachments
# keep track of any problematic attachments
self.generate()
merger = PdfFileMerger(strict=False)
merger.append(BytesIO(self.output(dest="S").encode("latin-1")))
total_pages = self.page
files = []
for file_ in self.comm.files.all():
if file_.get_extension() == "pdf":
try:
pages = PdfFileReader(file_.ffile).getNumPages()
if pages + total_pages > self.page_limit:
# too long, skip
files.append((file_, "skipped", pages))
else:
merger.append(file_.ffile)
files.append((file_, "attached", pages))
total_pages += pages
except (PdfReadError, ValueError):
files.append((file_, "error", 0))
else:
files.append((file_, "skipped", 0))
single_pdf = BytesIO()
try:
self._resize_pages(merger.pages)
merger.write(single_pdf)
except PdfReadError:
return (None, None, files, None)
# create the mail communication object
address = address_override if address_override else self.comm.foia.address
mail, _ = MailCommunication.objects.update_or_create(
communication=self.comm,
defaults={
"to_address": address,
"sent_datetime": timezone.now()
},
)
single_pdf.seek(0)
mail.pdf.save("{}.pdf".format(self.comm.pk),
ContentFile(single_pdf.read()))
# return to begining of merged pdf before returning
single_pdf.seek(0)
return (single_pdf, total_pages, files, mail)
def render(self, form_data):
merger = PdfFileMerger()
o = PdfTicketOutput(self.event)
qs = OrderPosition.objects.filter(
order__event=self.event).prefetch_related(
'answers',
'answers__question').select_related('order', 'item',
'variation', 'addon_to')
if form_data.get('include_pending'):
qs = qs.filter(
order__status__in=[Order.STATUS_PAID, Order.STATUS_PENDING])
else:
qs = qs.filter(order__status__in=[Order.STATUS_PAID])
if form_data.get('order_by') == 'name':
qs = qs.order_by('attendee_name_cached', 'order__code')
elif form_data.get('order_by') == 'code':
qs = qs.order_by('order__code')
elif form_data.get('order_by', '').startswith('name:'):
part = form_data['order_by'][5:]
qs = qs.annotate(resolved_name=Coalesce(
'attendee_name_parts', 'addon_to__attendee_name_parts',
'order__invoice_address__name_parts')).annotate(
resolved_name_part=JSONExtract(
'resolved_name', part)).order_by('resolved_name_part')
for op in qs:
if op.addon_to_id and not self.event.settings.ticket_download_addons:
continue
if not op.item.admission and not self.event.settings.ticket_download_nonadm:
continue
with language(op.order.locale):
buffer = BytesIO()
p = o._create_canvas(buffer)
layout = o.layout_map.get(op.item_id, o.default_layout)
o._draw_page(layout, p, op, op.order)
p.save()
outbuffer = o._render_with_background(layout, buffer)
merger.append(ContentFile(outbuffer.read()))
outbuffer = BytesIO()
merger.write(outbuffer)
merger.close()
outbuffer.seek(0)
return '{}_tickets.pdf'.format(
self.event.slug), 'application/pdf', outbuffer.read()
def prepare(self):
"""Prepare the PDF to be sent by appending attachments"""
# generate the pdf and merge all pdf attachments
# keep track of any problematic attachments
self.generate()
merger = PdfFileMerger(strict=False)
merger.append(StringIO(self.output(dest='S')))
files = []
for file_ in self.comm.files.all():
if file_.get_extension() == 'pdf':
try:
pages = PdfFileReader(file_.ffile).getNumPages()
merger.append(file_.ffile)
files.append((file_, 'attached', pages))
except (PdfReadError, ValueError):
files.append((file_, 'error', 0))
else:
files.append((file_, 'skipped', 0))
single_pdf = StringIO()
try:
merger.write(single_pdf)
except PdfReadError:
return (None, None, files)
# create the mail communication object
mail, _ = MailCommunication.objects.update_or_create(
communication=self.comm,
defaults={
'to_address': self.comm.foia.address,
'sent_datetime': timezone.now(),
}
)
single_pdf.seek(0)
mail.pdf.save(
'{}.pdf'.format(self.comm.pk),
ContentFile(single_pdf.read()),
)
# return to begining of merged pdf before returning
single_pdf.seek(0)
return (single_pdf, self.page, files, mail)
def mergePdfs(self, paths, fileName =None):
""" Takes a list of paths to existing PDF files and merges them into a single pdf with
the given file name.
[fileName] :: String :: None
The name of the file to be written. If not specified, a file name will be created
using the name of this class. """
merger = PdfFileMerger()
for p in paths:
with open(p, 'rb') as f:
merger.append(PdfFileReader(f))
if not fileName:
fileName = '%s-Report.pdf' % self.__class__.__name__
if not StringUtils.toStr2(fileName).endswith('.pdf'):
fileName += '.pdf'
with open(self.getPath(fileName, isFile=True), 'wb') as f:
merger.write(f)
print "\t-f[int]\tFiles you want to merge. [int] is the page it shall be (for the right order)"
print "\t-o\tThe output file."
exit(0)
i = 0
documents = {}
outfile = ""
while i < len(argv):
if argv[i].startswith('-f'):
documents[argv[i].strip('-f')] = argv[i + 1]
if argv[i].startswith('-o'):
outfile = argv[i + 1]
i += 1
for pdf in documents:
if not isfile(documents[pdf]):
print "Cannot find the file '%s'" % documents[pdf]
exit(0)
if isfile(outfile):
print "The file '%s' already exists." % outfile
pdfmerger = PdfFileMerger(False)
for pdf in documents:
pdfmerger.merge(int(pdf) - 1, documents[pdf])
pdfmerger.write(outfile)
pdfmerger.close()
print "Done."
from PyPDF2.merger import PdfFileMerger
import os
pdfs = []
merger = PdfFileMerger()
for file in os.scandir(os.getcwd()):
if file.path.endswith(".pdf"): # check if path leads to a pdf
file_name = file.path.split(os.sep)[-1]
if 'result_' not in file_name: # Save all pdfs except 'result_' ones
pdfs.append(file_name)
print('list of pdfs to merge: ', pdfs)
for pdf in pdfs:
merger.append(pdf)
for n in range(1, 10):
output_pdf_name = "result_" + str(n) + ".pdf" # result_1, result_2 ...
path_check = "./" + output_pdf_name # './result_1.pdf' ...
if not os.path.exists(path_check): # Check if pdf exists
merger.write(output_pdf_name)
merger.close()
print("Merged files into ", output_pdf_name)
break
else:
print(output_pdf_name, "already exists")
print("\nDone.")