def __init__(self, filename, basename, tmp_dir, parent):
self.render_lock = threading.Lock()
self.filename = os.path.abspath(filename)
self.mtime = os.path.getmtime(filename)
if basename is None: # When importing files
self.basename = os.path.basename(filename)
else: # When copy-pasting
self.basename = basename
self.password = ""
filemime = mimetypes.guess_type(self.filename)[0]
if not filemime:
raise PDFDocError(_("Unknown file format"))
if filemime == "application/pdf":
if self.filename.startswith(tmp_dir) and basename is None:
# In the "Insert Blank Page" we don't need to copy self.filename
self.copyname = self.filename
self.basename = ""
else:
fd, self.copyname = tempfile.mkstemp(suffix=".pdf",
dir=tmp_dir)
os.close(fd)
shutil.copy(self.filename, self.copyname)
try:
self.__from_file(parent, self.basename)
except GLib.Error as e:
raise PDFDocError(e.message + ": " + filename)
elif filemime.split("/")[0] == "image":
if not img2pdf:
raise PDFDocError(
_("Image files are only supported with img2pdf"))
if mimetypes.guess_type(filename)[0] in img2pdf_supported_img:
fd, self.copyname = tempfile.mkstemp(suffix=".pdf",
dir=tmp_dir)
os.close(fd)
with open(self.copyname, "wb") as f:
img = img2pdf.Image.open(filename)
if img.mode != "RGBA" and "transparency" in img.info:
# TODO: Find a way to keep image in P or L format and remove transparency.
# This will work but converting from 1, L, P to RGB is not optimal.
img = img.convert("RGBA")
if img.mode == "RGBA":
bg = img2pdf.Image.new("RGB", img.size,
(255, 255, 255))
bg.paste(img, mask=img.split()[-1])
imgio = img2pdf.BytesIO()
bg.save(imgio, "PNG")
imgio.seek(0)
f.write(img2pdf.convert(imgio))
else:
f.write(img2pdf.convert(filename))
uri = pathlib.Path(self.copyname).as_uri()
self.document = Poppler.Document.new_from_file(uri, None)
else:
raise PDFDocError(
_("Image format is not supported by img2pdf"))
else:
raise PDFDocError(_("File is neither pdf nor image"))
async def download_images(session, directory, links):
files = []
for i, link in enumerate(links):
async with session.head(link, allow_redirects=True) as resp:
size = int(resp.headers.get('Content-Length', -1))
if size > 5e6: # 5 MB
continue
image_filename = f'{directory}/{i}.jpg'
async with session.get(link) as resp:
content = await resp.content.read()
imgio = img2pdf.BytesIO(content)
image = Image.open(imgio).convert('RGB')
# Unknown ExifOrientationError on PNG format
image.save(image_filename, format='JPEG')
files.append(image_filename)
return files