def process_set_list(self, *args, text="", search=False):
from kivymd.uix.progressbar import MDProgressBar
pb = MDProgressBar(type="determinate",
running_duration=1,
catching_duration=1.5)
status_bar = get_app().modellist_controller.screen.status_bar
status_bar.clear_widgets()
status_bar.add_widget(pb)
pb.start()
Clock.schedule_once(
partial(self.set_list, self, *args, text=text, search=search))
pb.stop()
def pdfimages(pdfpath, cmds, instance, *args):
pb = MDProgressBar(color=get_app().theme_cls.primary_color,
type="indeterminate")
status_bar = get_app().image_selection_controller.status_bar
status_bar.clear_widgets()
status_bar.add_widget(pb)
pb.start()
pdfdir = Path(pdfpath.split('.')[0])
makedirs(pdfdir, exist_ok=True)
params = []
children = instance.parent.parent.parent.parent.content_cls.children
process = cmds["pdfimages"]
for idx, child in enumerate(reversed(children)):
if idx == 6:
for fileformat in child.children:
if fileformat.state == 'down':
params.extend([f"-{fileformat.text}"])
if idx == 2 and child.text != "":
params.extend(["-f", child.text])
if idx == 4 and child.text != "":
params.extend(["-l", child.text])
if idx == 9 and child.ids['_left_container'].children[0].active:
params.extend(["-p"])
if idx == 8:
for convprocess in child.children:
if convprocess.state == 'down':
if convprocess.text == "rendering":
process = cmds["pdftoppm"]
else:
process = cmds["pdfimages"]
fileformat.text = "j" if fileformat.text == "jpeg" else fileformat.text
fileformat.text = "jpeg" if fileformat.text == "jp2" else fileformat.text
p1 = Popen([process, *params, pdfpath, pdfdir.joinpath(pdfdir.name)])
p1.communicate()
get_app().image_selection_controller.file_chooser._update_files()
get_app().image_selection_controller.add_images([pdfdir])
pb.stop()
def pdfimages(pdfpath, cmds, instance, ocr, *args):
""" Converts the pdf to images"""
pb = MDProgressBar(color=get_app().theme_cls.primary_color,
type="indeterminate")
status_bar = get_app().image_selection_controller.status_bar
status_bar.clear_widgets()
status_bar.add_widget(pb)
pb.start()
if ocr:
tmpdir = tempfile.TemporaryDirectory()
pdfdir = Path(tmpdir.name)
else:
pdfdir = Path(pdfpath.split('.')[0])
makedirs(pdfdir, exist_ok=True)
params = []
children = instance.parent.parent.parent.parent.content_cls.children
process = cmds["pdfimages"]
for idx, child in enumerate(reversed(children)):
if idx == 6:
for fileformat in child.children:
if fileformat.state == 'down':
params.extend([f"-{fileformat.text}"])
if idx == 2 and child.text != "":
params.extend(["-f", child.text])
if idx == 4 and child.text != "":
params.extend(["-l", child.text])
if idx == 9 and child.ids['_left_container'].children[0].active:
params.extend(["-p"])
if idx == 8:
for convprocess in child.children:
if convprocess.state == 'down':
if convprocess.text == "rendering":
process = cmds["pdftoppm"]
else:
process = cmds["pdfimages"]
params = " ;".join(params).replace('-jpeg',
'-j').split(' ;')
p1 = Popen([
process, *params, pdfpath,
pdfdir.joinpath(Path(pdfpath.split('.')[0]).name)
])
p1.communicate()
get_app().image_selection_controller.file_chooser._update_files()
if not ocr:
get_app().image_selection_controller.add_images([pdfdir])
else:
images = list(pdfdir.glob('*.*'))
tc_screen = get_app().tesseract_controller
thread = tc_screen.recognize_thread(None,
file_list=images,
profile={
'outputformats': ['pdf'],
'groupfolder': '',
'subforlder': False,
'print_on_screen': False
})
thread.join()
p2 = Popen([
cmds["pdfunite"], *sorted(list(pdfdir.glob('*.pdf'))),
pdfpath[:-3] + "ocr.pdf"
])
p2.communicate()
get_app().image_selection_controller.file_chooser._update_files()
pb.stop()