def extractAWAMIndicators(pdf, password=''):
""" Check whether the given PDF document is accessible """
# Takes an optional password which can be used to
# unlock the document for encrypted documents.
try:
pdfobj = PdfFileReader(pdf, password)
print 'Producer=>',pdfobj.producer
print 'Creator=>',pdfobj.creator
except DecryptionFailedException:
# We are unable to decrypt document.
# We have got no parsed pdfobj, and cannot do much more,
# unfortunately...
# Tell that the document was not accessible due to encryption, at least
print "Decryption failed"
return {'EIAO.A.10.8.1.4.PDF.1.1':{(0, 1): 0}}
except NotImplementedError:
# pyPdf only supports algorithm version 1 and 2.
# Version 3 and 4 are not yet supported.
print "Unsupported decryption algorithm."
return {'EIAO.A.10.8.1.4.PDF.1.1':{(0, 1): 0}}
try:
# Fix indirect object references
pdfobj.fixIndirectObjectXref()
structTreeRoot=pdfobj.root['/StructTreeRoot'].getObject()
roleMap=structTreeRoot['/RoleMap'].getObject()
except (KeyError, ValueError), e:
roleMap=None
