mirror of
https://github.com/kanzure/pdfparanoia.git
synced 2024-06-24 12:42:55 +02:00
56cc7719da
Also cleaned up some flakes noticed by pyflakes, and make the scrub() be @classmethod instead of @staticmethod so I could use the class for the verbose output. caveats: * there are no unit tests of this patch * now your logs of your stderr have potentially sensitive information in them * the implementation of arg parsing is very low-tech; (a *good* way to do arg parsing is the "argparse" module)
58 lines
1.7 KiB
Python
58 lines
1.7 KiB
Python
# -*- coding: utf-8 -*-
|
|
|
|
import sys
|
|
|
|
from copy import copy
|
|
|
|
from ..parser import parse_content
|
|
from ..eraser import remove_object_by_id
|
|
from ..plugin import Plugin
|
|
|
|
class AmericanInstituteOfPhysics(Plugin):
|
|
"""
|
|
American Institute of Physics
|
|
~~~~~~~~~~~~~~~
|
|
|
|
These watermarks are pretty basic, but sometimes they don't have indexes
|
|
attached for whatever reason.
|
|
"""
|
|
|
|
@classmethod
|
|
def scrub(cls, content, verbose=False):
|
|
evil_ids = []
|
|
|
|
# parse the pdf into a pdfminer document
|
|
pdf = parse_content(content)
|
|
|
|
# get a list of all object ids
|
|
xrefs = pdf._parser.read_xref()
|
|
xref = xrefs[0]
|
|
objids = xref.get_objids()
|
|
|
|
# check each object in the pdf
|
|
for objid in objids:
|
|
# get an object by id
|
|
obj = pdf.getobj(objid)
|
|
|
|
if hasattr(obj, "attrs"):
|
|
# watermarks tend to be in FlateDecode elements
|
|
if obj.attrs.has_key("Filter") and str(obj.attrs["Filter"]) == "/FlateDecode":
|
|
length = obj.attrs["Length"]
|
|
|
|
# the watermark is never very long
|
|
if length < 1000:
|
|
#rawdata = copy(obj.rawdata)
|
|
data = copy(obj.get_data())
|
|
|
|
if "Redistribution subject to AIP license or copyright" in data:
|
|
if verbose:
|
|
sys.stderr.write("%s: Found object with %r; omitting..." % (cls.__name__, data,))
|
|
|
|
evil_ids.append(objid)
|
|
|
|
for objid in evil_ids:
|
|
content = remove_object_by_id(content, objid)
|
|
|
|
return content
|
|
|