def analyze_pdf(path):
try:
with open(path, 'rb') as f:
metadata = PdfFileReader(f).getDocumentInfo()
metadata = {re.sub('[^A-Za-z]+', '', k): v for k, v in metadata.items()}
author = metadata.get('Author')
title = metadata.get('Title')
date = metadata.get('CreationDate')
year = get_year_from_date_string(date)
return (author, title, year, path, 'pdf')
except:
return None
def _get_pdf_page_dimensions(pdf_file_path, page_no):
"""
Gets the height and width of the pdf at the given page no after the rotation is applied. The default height and
width are swapped when the pdf has a rotation of 90/270(vertical).
:param pdf_file_path: File path of the input pdf
:param page_no: Page no whose dimensions are returned
:return: A tuple of the form (width, height)
"""
with open(pdf_file_path, 'rb') as file:
pdf_file = PdfFileReader(file).getPage(page_no)
media_box = pdf_file.mediaBox
rotation = pdf_file.get('/Rotate')
if utils.is_horizontal_orientation(rotation):
w, h = media_box.getWidth(), media_box.getHeight()
else:
w, h = media_box.getHeight(), media_box.getWidth()
return w, h
