def get_pdf_metadata(self, pdf_file_stream):
metadata = {
'author': 'UNKNOWN_AUTHOR',
'title': 'UNKNOWN_TITLE',
'year': 'UNKNOWN_YEAR'
}
pdf_parser = PDFParser(pdf_file_stream)
pdf_doc = PDFDocument(pdf_parser)
pdf_metadata = pdf_doc.info[0]
author = make_pdf_metadata_str(pdf_metadata['Author'] if 'Author' in
pdf_metadata else '')
if author and author != '':
metadata['author'] = author
title = make_pdf_metadata_str(pdf_metadata['Title'] if 'Title' in
pdf_metadata else '')
if title and title != '':
metadata['title'] = title
year = pdf_metadata_moddate_to_year(
make_pdf_metadata_str(pdf_metadata['ModDate'] if 'ModDate' in
pdf_metadata else ''))
if year and year != '':
metadata['year'] = year
return metadata
def get_pdf_metadata(pdf):
"""Get PDF metadata with PDF content
Args:
pdf: PDF content (in bytes)
Returns:
metadata: PDF metadata dictionary
"""
temp_pdf_file = tempfile.TemporaryFile()
temp_pdf_file.write(pdf)
metadata = {
'author': 'UNKNOWN_AUTHOR',
'title': 'UNKNOWN_TITLE',
'year': 'UNKNOWN_YEAR'
}
pdf_parser = PDFParser(temp_pdf_file)
try:
pdf_doc = PDFDocument(pdf_parser)
pdf_metadata = pdf_doc.info[0]
author = make_pdf_metadata_str(pdf_metadata.get('Author', ''))
if author and author != '':
metadata['author'] = author
title = make_pdf_metadata_str(pdf_metadata.get('Title', ''))
if title and title != '':
metadata['title'] = title
year = pdf_metadata_moddate_to_year(
make_pdf_metadata_str(pdf_metadata.get('ModDate', '')))
if year and year != '':
metadata['year'] = year
except Exception as e:
pass
temp_pdf_file.close()
return metadata
