forked from s434650/CatOrNot
127 lines
3.5 KiB
Python
127 lines
3.5 KiB
Python
|
# -*- coding: utf-8 -*-
|
||
|
#
|
||
|
# Copyright 2011 Sybren A. Stüvel <sybren@stuvel.eu>
|
||
|
#
|
||
|
# Licensed under the Apache License, Version 2.0 (the "License");
|
||
|
# you may not use this file except in compliance with the License.
|
||
|
# You may obtain a copy of the License at
|
||
|
#
|
||
|
# https://www.apache.org/licenses/LICENSE-2.0
|
||
|
#
|
||
|
# Unless required by applicable law or agreed to in writing, software
|
||
|
# distributed under the License is distributed on an "AS IS" BASIS,
|
||
|
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
|
||
|
# See the License for the specific language governing permissions and
|
||
|
# limitations under the License.
|
||
|
|
||
|
"""Functions that load and write PEM-encoded files."""
|
||
|
|
||
|
import base64
|
||
|
|
||
|
from rsa._compat import is_bytes, range
|
||
|
|
||
|
|
||
|
def _markers(pem_marker):
|
||
|
"""
|
||
|
Returns the start and end PEM markers, as bytes.
|
||
|
"""
|
||
|
|
||
|
if not is_bytes(pem_marker):
|
||
|
pem_marker = pem_marker.encode('ascii')
|
||
|
|
||
|
return (b'-----BEGIN ' + pem_marker + b'-----',
|
||
|
b'-----END ' + pem_marker + b'-----')
|
||
|
|
||
|
|
||
|
def load_pem(contents, pem_marker):
|
||
|
"""Loads a PEM file.
|
||
|
|
||
|
:param contents: the contents of the file to interpret
|
||
|
:param pem_marker: the marker of the PEM content, such as 'RSA PRIVATE KEY'
|
||
|
when your file has '-----BEGIN RSA PRIVATE KEY-----' and
|
||
|
'-----END RSA PRIVATE KEY-----' markers.
|
||
|
|
||
|
:return: the base64-decoded content between the start and end markers.
|
||
|
|
||
|
@raise ValueError: when the content is invalid, for example when the start
|
||
|
marker cannot be found.
|
||
|
|
||
|
"""
|
||
|
|
||
|
# We want bytes, not text. If it's text, it can be converted to ASCII bytes.
|
||
|
if not is_bytes(contents):
|
||
|
contents = contents.encode('ascii')
|
||
|
|
||
|
(pem_start, pem_end) = _markers(pem_marker)
|
||
|
|
||
|
pem_lines = []
|
||
|
in_pem_part = False
|
||
|
|
||
|
for line in contents.splitlines():
|
||
|
line = line.strip()
|
||
|
|
||
|
# Skip empty lines
|
||
|
if not line:
|
||
|
continue
|
||
|
|
||
|
# Handle start marker
|
||
|
if line == pem_start:
|
||
|
if in_pem_part:
|
||
|
raise ValueError('Seen start marker "%s" twice' % pem_start)
|
||
|
|
||
|
in_pem_part = True
|
||
|
continue
|
||
|
|
||
|
# Skip stuff before first marker
|
||
|
if not in_pem_part:
|
||
|
continue
|
||
|
|
||
|
# Handle end marker
|
||
|
if in_pem_part and line == pem_end:
|
||
|
in_pem_part = False
|
||
|
break
|
||
|
|
||
|
# Load fields
|
||
|
if b':' in line:
|
||
|
continue
|
||
|
|
||
|
pem_lines.append(line)
|
||
|
|
||
|
# Do some sanity checks
|
||
|
if not pem_lines:
|
||
|
raise ValueError('No PEM start marker "%s" found' % pem_start)
|
||
|
|
||
|
if in_pem_part:
|
||
|
raise ValueError('No PEM end marker "%s" found' % pem_end)
|
||
|
|
||
|
# Base64-decode the contents
|
||
|
pem = b''.join(pem_lines)
|
||
|
return base64.standard_b64decode(pem)
|
||
|
|
||
|
|
||
|
def save_pem(contents, pem_marker):
|
||
|
"""Saves a PEM file.
|
||
|
|
||
|
:param contents: the contents to encode in PEM format
|
||
|
:param pem_marker: the marker of the PEM content, such as 'RSA PRIVATE KEY'
|
||
|
when your file has '-----BEGIN RSA PRIVATE KEY-----' and
|
||
|
'-----END RSA PRIVATE KEY-----' markers.
|
||
|
|
||
|
:return: the base64-encoded content between the start and end markers, as bytes.
|
||
|
|
||
|
"""
|
||
|
|
||
|
(pem_start, pem_end) = _markers(pem_marker)
|
||
|
|
||
|
b64 = base64.standard_b64encode(contents).replace(b'\n', b'')
|
||
|
pem_lines = [pem_start]
|
||
|
|
||
|
for block_start in range(0, len(b64), 64):
|
||
|
block = b64[block_start:block_start + 64]
|
||
|
pem_lines.append(block)
|
||
|
|
||
|
pem_lines.append(pem_end)
|
||
|
pem_lines.append(b'')
|
||
|
|
||
|
return b'\n'.join(pem_lines)
|