total_num_pages = len(pages)
bar_widgets = [
progressbar.Bar(),
progressbar.Counter(format='%(value)i/%(max_value)i')
]
bar = progressbar.ProgressBar(max_value=total_num_pages, widgets=bar_widgets, redirect_stdout=True)
bar.start()
with open(input_filepath, "rb") as fp:
rsrcmgr = PDFResourceManager(caching=True)
device = TextBoxStripper(rsrcmgr, outfp)
interpreter = PDFPageInterpreter(rsrcmgr, device)
for (page_num, page) in PDFPage.get_pages2(fp,
pages,
password="",
caching=True,
check_extractable=True,
fallback=False):
try:
#print("===== Page {}".format(page_num))
# Text box processing:
device.text_boxes = []
device.tables = []
interpreter.process_page(page)
device.drop_empty_textboxes()
device.merge_textboxes()
# Table processing
device.build_tables()
## For now, we don't care about the title of the page.
## With table contents we have all the information
#for text_box in device.text_boxes:
