0

I get lot of pdfs in my system. I need to check if all of these files are:-

  • digitally signed
  • their integrity maintained(by comparing the hash of file content with the message digest embedded in /Contents.signature dictionary.

I am using python to do this. Till now I have been able to get the /Content from signature dictionary using PyPDF2. The content is pkcs7--der encoded. Is there a way I can extract the signed message digest?

Similar operation was done in C as this answer

Community
  • 1
  • 1
Barun Sharma
  • 1,452
  • 2
  • 15
  • 20

2 Answers2

1

DER is binary format, its structure is called ASN.1. PEM format is Base64 encoded DER.

This online PEM decoder is very useful: http://lapo.it/asn1js/ After you identify the message signature in it, you can write code to extract it by any ASN.1 library.

zakjan
  • 2,391
  • 1
  • 19
  • 29
  • Wel thanks for the direction. I have the decoded DER from the link you provided. 1) How do identify the signature? 2) Would the ASN.1 library let me parse this file and get what I desire? Have not worked much on all of it. I am still checking if all that helps. Will update once I am done. Thanks again. – Barun Sharma Feb 06 '15 at 06:47
  • I'd look into the corresponding [RFC2315](https://tools.ietf.org/html/rfc2315) for correct signature structure. The ASN.1 library will expect only the DER/PEM encoded data, you need to extract it out of your PDF. – zakjan Feb 06 '15 at 07:29
1

The SignedData subtype of a CADES, CMS or PKCS#7 1.5 signature has a collection of SignerInfo blocks defined here contains basically:

  • SignerIdentifier: key into certificates collection
  • DigestAlgorithmIdentifier: which algorithm was used to calculate message digest
  • SignedAttributes (optional): the sealed data:
  • SignatureAlgorithmIdentifier: which algorithm was used to calculate the signature (over SignedAttributes)
  • SignatureValue: the signature value
  • UnsignedAttributes (optional)

The SignedAttributes may contain, depending on the type of signature:

  • ContentType: type of signed content
  • MessageDigest
  • SigningTime
  • CounterSignature

If we would simplify this using just the first signature found and using my fork of pyx509 this could be some type of code like this one (not tested):

from pyx509.models import PKCS7
pkcs7 = PKCS7.from_der(here_goes_your_pks7_signature_data_der_encoded)
signer_info = pkcs7.content.signerInfos[0]
auth_attrs = signer_info.auth_attributes
for attr in auth_attrs.attributes:
    if attr.type == '1.2.840.113549.1.9.4':  # Message Digest OID
        message_digest = attr.value
        print "Digest: %s#%s" % (signer_info.oid2name(signer_info.digest_algorithm), messageDigest)
Community
  • 1
  • 1
erny
  • 1,296
  • 1
  • 13
  • 28
  • *If we would simplify this using just the first signature found* - integrated PDF signatures allow only a single `SignerInfo` in each `SignedData`; so, no simplification warning required in this respect. – mkl Sep 22 '15 at 12:10