def split_into_small_pdfs(pdf_path,
output_directory=None,
start_page=1,
end_page=None,
small_pdf_pages=25):
pdf_name_stem = Path(pdf_path).stem
if output_directory == None:
output_directory = _get_ocr_dir(pdf_path)
# noinspection PyArgumentList
with Pdf.open(pdf_path) as pdf:
if end_page == None:
end_page = len(pdf.pages)
pages = range(start_page, end_page + 1)
page_sets = list_helper.divide_chunks(list_in=pages, n=small_pdf_pages)
for page_set in page_sets:
pages = [pdf.pages[i - 1] for i in page_set]
dest_pdf_path = os.path.join(
output_directory, "%s_%04d-%04d.pdf" %
(pdf_name_stem, page_set[0], page_set[-1]))
if not os.path.exists(dest_pdf_path):
# noinspection PyArgumentList
dest_pdf = Pdf.new()
dest_pdf.pages.extend(pages)
os.makedirs(os.path.dirname(dest_pdf_path), exist_ok=True)
dest_pdf.save(filename=dest_pdf_path)
else:
logging.warning("%s exists", dest_pdf_path)
def break_to_small_pdf_paths_original(pdf_path,
output_directory=None,
start_page=1,
end_page=None,
small_pdf_pages=25):
# logging.info("Splitting %s into segments of %d", pdf_path, 25)
pdf_name_stem = Path(pdf_path).stem
if output_directory == None:
output_directory = os.path.join(
os.path.dirname(pdf_path),
Path(pdf_path).stem + "_small_originals")
# noinspection PyArgumentList
with Pdf.open(pdf_path) as pdf:
if end_page == None:
end_page = len(pdf.pages)
pages = range(start_page, end_page + 1)
page_sets = list_helper.divide_chunks(list_in=pages, n=small_pdf_pages)
dest_pdfs = []
for page_set in page_sets:
pages = [pdf.pages[i - 1] for i in page_set]
dest_pdf_path = os.path.join(
output_directory,
"%s_%04d-%04d.pdf" %
(pdf_name_stem, page_set[0], page_set[-1]),
)
if not os.path.exists(dest_pdf_path):
# noinspection PyArgumentList
dest_pdf = Pdf.new()
dest_pdf.pages.extend(pages)
os.makedirs(os.path.dirname(dest_pdf_path), exist_ok=True)
dest_pdf.save(filename_or_stream=dest_pdf_path)
else:
logging.warning("%s exists", dest_pdf_path)
dest_pdfs.append(dest_pdf_path)
return dest_pdfs