pdf-meta.py - extract metadata from first PDF file given as argument

pdf-meta.py

#!/usr/bin/env python3
# extract metadata from first PDF file given as argument
# scruss - 2017
# -*- coding: utf-8 -*-

import sys
from PyPDF2 import PdfFileReader

inputPdf = PdfFileReader(open(sys.argv[1], "rb"))
docInfo = inputPdf.getDocumentInfo()
# PyPDF2 has very limited docinfo fields;
# not much more than
print('Title:    ', docInfo.title)
print('Subject:  ', docInfo.subject)
print('Author: ', docInfo.author)

# XMP stream, if present, can have much more.
# mostly under Dublin Core Metadata standards:
#  http://dublincore.org/
xmpInfo=inputPdf.getXmpMetadata()
if xmpInfo is not None:
    print("\n*** XMP data -\n")
    print("dc_contributor: ", xmpInfo.dc_contributor)
    print("dc_coverage: ", xmpInfo.dc_coverage)
    print("dc_creator: ", xmpInfo.dc_creator)
    print("dc_date: ", xmpInfo.dc_date)
    print("dc_description: ", xmpInfo.dc_description)
    print("dc_format: ", xmpInfo.dc_format)
    print("dc_identifier: ", xmpInfo.dc_identifier)
    print("dc_language: ", xmpInfo.dc_language)
    print("dc_publisher: ", xmpInfo.dc_publisher)
    print("dc_relation: ", xmpInfo.dc_relation)
    print("dc_rights: ", xmpInfo.dc_rights)
    print("dc_source: ", xmpInfo.dc_source)
    print("dc_subject: ", xmpInfo.dc_subject) # tags/keywords
    print("dc_title: ", xmpInfo.dc_title)
    print("dc_type: ", xmpInfo.dc_type)
    print("pdf_keywords: ", xmpInfo.pdf_keywords)
    print("pdf_pdfversion: ", xmpInfo.pdf_pdfversion)
    print("pdf_producer: ", xmpInfo.pdf_producer)
    print("xmp_createDate: ", xmpInfo.xmp_createDate)
    print("xmp_creatorTool: ", xmpInfo.xmp_creatorTool)
    print("xmp_metadataDate: ", xmpInfo.xmp_metadataDate)
    print("xmpmm_documentId: ", xmpInfo.xmpmm_documentId)
    print("xmpmm_instanceId: ", xmpInfo.xmpmm_instanceId)
    print("xmp_modifyDate: ", xmpInfo.xmp_modifyDate)

# for
#  https://www.epa.gov/sites/production/files/2016-09/documents/climate-change-hi.pdf
# this returns:
#  dc_contributor:  []
#  dc_coverage:  None
#  dc_creator:  ['US EPA', 'OAR', 'Climate Change Division']
#  dc_date:  []
#  dc_description:  {'x-default': 'This fact sheet provides a concise overview of the observed and projected effects and impacts of climate change on Hawaii.'}
#  dc_format:  application/pdf
#  dc_identifier:  None
#  dc_language:  []
#  dc_publisher:  []
#  dc_relation:  []
#  dc_rights:  {}
#  dc_source:  None
#  dc_subject:  ['EPA', 'climate change', 'state', 'impacts', 'fact sheet', 'summary']
#  dc_title:  {'x-default': 'What Climate Change Means for Hawaii'}
#  dc_type:  []
#  pdf_keywords:  EPA; climate change; state; impacts; fact sheet; summary
#  pdf_pdfversion:  None
#  pdf_producer:  Adobe PDF Library 15.0
#  xmp_createDate:  2016-04-08 14:41:23
#  xmp_creatorTool:  Adobe InDesign CC 2015 (Windows)
#  xmp_metadataDate:  2016-08-24 15:05:46
#  xmpmm_documentId:  xmp.id:43618963-6411-e44a-8538-2301fb5e6821
#  xmpmm_instanceId:  uuid:196fd5fa-8ecf-44b0-870b-086d08707697
#  xmp_modifyDate:  2016-08-24 15:05:46