A lightweight XMP parser for extracting PDF metadata in Python

Metadata (title, author, etc.) can be embedded in PDF files in a number of different ways, and can be a bit of a pain to extract. Older PDFs use “Info” in the XRefs trailer, whereas newer ones use XMP metadata. Using the Python PDFMiner library, it’s possible to extract the “Info” as a python dictionary, but the XMP metadata is just extracted as raw XML.

I couldn’t find a nice lightweight XMP parser in Python, so I put together something that seemed to work on all the PDFs I threw at it.

You can install PDFMiner by downloading the source, then doing:

cd pdfminer
make cmap
python setup.py install

Once installed, use PDFMiner to open the PDF and get the XMP.

from pdfminer.pdfparser import PDFParser, PDFDocument
from pdfminer.pdftypes import resolve1
from xmp import xmp_to_dict

fp = open('mypdf.pdf', 'rb')
parser = PDFParser(fp)
doc = PDFDocument()

print doc.info        # The "Info" metadata

if 'Metadata' in doc.catalog:
    metadata = resolve1(doc.catalog['Metadata']).get_data()
    print metadata  # The raw XMP metadata
    print xmp_to_dict(metadata)

The xmp_to_dict function is defined follows:

#!/usr/bin/env python

    Parses XMP metadata from PDF files.

from collections import defaultdict
from xml.etree import ElementTree as ET

RDF_NS = '{http://www.w3.org/1999/02/22-rdf-syntax-ns#}'
XML_NS = '{http://www.w3.org/XML/1998/namespace}'
NS_MAP = {
    'http://www.w3.org/1999/02/22-rdf-syntax-ns#'    : 'rdf',
    'http://purl.org/dc/elements/1.1/'               : 'dc',
    'http://ns.adobe.com/xap/1.0/'                   : 'xap',
    'http://ns.adobe.com/pdf/1.3/'                   : 'pdf',
    'http://ns.adobe.com/xap/1.0/mm/'                : 'xapmm',
    'http://ns.adobe.com/pdfx/1.3/'                  : 'pdfx',
    'http://prismstandard.org/namespaces/basic/2.0/' : 'prism',
    'http://crossref.org/crossmark/1.0/'             : 'crossmark',
    'http://ns.adobe.com/xap/1.0/rights/'            : 'rights',
    'http://www.w3.org/XML/1998/namespace'           : 'xml'

class XmpParser(object):
    Parses an XMP string into a dictionary.


        parser = XmpParser(xmpstring)
        meta = parser.meta

    def __init__(self, xmp):
        self.tree = ET.XML(xmp)
        self.rdftree = self.tree.find(RDF_NS+'RDF')

    def meta(self):
        """ A dictionary of all the parsed metadata. """
        meta = defaultdict(dict)
        for desc in self.rdftree.findall(RDF_NS+'Description'):
            for el in desc.getchildren():
                ns, tag =  self._parse_tag(el)
                value = self._parse_value(el)
                meta[ns][tag] = value
        return dict(meta)

    def _parse_tag(self, el):
        """ Extract the namespace and tag from an element. """
        ns = None
        tag = el.tag
        if tag[0] == "{":
            ns, tag = tag[1:].split('}',1)
            if ns in NS_MAP:
                ns = NS_MAP[ns]
        return ns, tag

    def _parse_value(self, el):
        """ Extract the metadata value from an element. """
        if el.find(RDF_NS+'Bag') is not None:
            value = []
            for li in el.findall(RDF_NS+'Bag/'+RDF_NS+'li'):
        elif el.find(RDF_NS+'Seq') is not None:
            value = []
            for li in el.findall(RDF_NS+'Seq/'+RDF_NS+'li'):
        elif el.find(RDF_NS+'Alt') is not None:
            value = {}
            for li in el.findall(RDF_NS+'Alt/'+RDF_NS+'li'):
                value[li.get(XML_NS+'lang')] = li.text
            value = el.text
        return value

def xmp_to_dict(xmp):
    """ Shorthand function for parsing an XMP string into a python dictionary. """
    return XmpParser(xmp).meta

2 notes

  1. mcswain posted this