anteater/src/patch_scan.py

   1 #!/usr/bin/env python
   2 # -*- coding: utf-8 -*-
   3 ##############################################################################
   4 # Copyright (c) 2017 Luke Hinds <lhinds@redhat.com>, Red Hat
   5 #
   6 # All rights reserved. This program and the accompanying materials
   7 # are made available under the terms of the Apache License, Version 2.0
   8 # which accompanies this distribution, and is available at
   9 # http://www.apache.org/licenses/LICENSE-2.0
  10 ##############################################################################
  11
  12 """
  13     Accepts the --patchset argument and iterates through each line of the
  14     patchset file to perform various checks such as if the file is a binary, or
  15     contains a blacklisted string. If any violations are found, the script
  16     exits with code 1 and logs the violation(s) found.
  17 """
  18
  19 from __future__ import division, print_function, absolute_import
  20 from binaryornot.check import is_binary
  21 import logging
  22 import hashlib
  23 import six.moves.configparser
  24 import sys
  25 import re
  26
  27 from . import get_lists
  28
  29 logger = logging.getLogger(__name__)
  30 config = six.moves.configparser.RawConfigParser()
  31 config.read('anteater.conf')
  32 reports_dir = config.get('config', 'reports_dir')
  33 failure = False
  34 hasher = hashlib.sha256()
  35
  36
  37 def prepare_patchset(project, patchset):
  38     """ Create black/white lists and default / project waivers
  39         and iterates over patchset file """
  40
  41     # Get Various Lists / Project Waivers
  42     lists = get_lists.GetLists()
  43     # Get binary white list
  44     binary_list = lists.binary_list(project)
  45
  46     # Get file name black list and project waivers
  47     file_audit_list, file_audit_project_list = lists.file_audit_list(project)
  48
  49     # Get file content black list and project waivers
  50     master_list, project_list_re = lists.file_content_list(project)
  51
  52     # Get Licence Lists
  53     licence_ext = lists.licence_extensions()
  54     licence_ignore = lists.licence_ignore()
  55
  56     # Open patch set to get file list
  57     try:
  58         fo = open(patchset, 'r')
  59         lines = fo.readlines()
  60     except IOError:
  61         logger.error('%s does not exist', patchset)
  62         sys.exit(1)
  63
  64     for line in lines:
  65         patch_file = line.strip('\n')
  66         # Perform binary and file / content checks
  67         scan_patch(project, patch_file, binary_list,
  68                    file_audit_list, file_audit_project_list,
  69                    master_list, project_list_re, licence_ext,
  70                    licence_ignore)
  71
  72     # Process each file in patch set using waivers generated above
  73     # Process final result
  74     process_failure()
  75
  76
  77 def scan_patch(project, patch_file, binary_list, file_audit_list,
  78                file_audit_project_list, master_list,
  79                project_list_re, licence_ext, licence_ignore):
  80     """ Scan actions for each commited file in patch set """
  81     global failure
  82     if is_binary(patch_file):
  83         hashlist = get_lists.GetLists()
  84         binary_hash = hashlist.binary_hash(project, patch_file)
  85         if not binary_list.search(patch_file):
  86             with open(patch_file, 'rb') as afile:
  87                 buf = afile.read()
  88                 hasher.update(buf)
  89             if hasher.hexdigest() in binary_hash:
  90                 logger.info('Found matching file hash for file: %s',
  91                             patch_file)
  92             else:
  93                 logger.error('Non Whitelisted Binary file: %s',
  94                              patch_file)
  95                 logger.error('Submit patch with the following hash: %s',
  96                              hasher.hexdigest())
  97             failure = True
  98             with open(reports_dir + "binaries-" + project + ".log", "a") \
  99                     as gate_report:
 100                 gate_report.write('Non Whitelisted Binary file: {0}\n'.
 101                                   format(patch_file))
 102     else:
 103         # Check file names / extensions
 104         if file_audit_list.search(patch_file) and not \
 105                     file_audit_project_list.search(patch_file):
 106             match = file_audit_list.search(patch_file)
 107             logger.error('Blacklisted file: %s', patch_file)
 108             logger.error('Matched String: %s', match.group())
 109             failure = True
 110             with open(reports_dir + "file-names_" + project + ".log", "a") \
 111                     as gate_report:
 112                 gate_report.write('Blacklisted file: {0}\n'.
 113                                   format(patch_file))
 114                 gate_report.write('Matched String: {0}'.
 115                                   format(match.group()))
 116
 117         # Open file to check for blacklisted content
 118         try:
 119             fo = open(patch_file, 'r')
 120             lines = fo.readlines()
 121         except IOError:
 122             logger.error('%s does not exist', patch_file)
 123             sys.exit(1)
 124
 125         for line in lines:
 126             for key, value in master_list.iteritems():
 127                 regex = value['regex']
 128                 desc = value['desc']
 129                 if re.search(regex, line) and not re.search(project_list_re, line):
 130                     logger.error('File contains violation: %s', patch_file)
 131                     logger.error('Flagged Content: %s', line.rstrip())
 132                     logger.error('Matched Regular Exp: %s', regex)
 133                     logger.error('Rationale: %s', desc.rstrip())
 134                     failure = True
 135                     with open(reports_dir + "contents_" + project + ".log",
 136                               "a") as gate_report:
 137                         gate_report.write('File contains violation: {0}\n'.
 138                                           format(patch_file))
 139                         gate_report.write('Flagged Content: {0}'.
 140                                           format(line))
 141                         gate_report.write('Matched Regular Exp: {0}'.
 142                                           format(regex))
 143                         gate_report.write('Rationale: {0}'.
 144                                           format(desc.rstrip()))
 145         # Run license check
 146         licence_check(project, licence_ext, licence_ignore, patch_file)
 147
 148
 149 def licence_check(project, licence_ext,
 150                   licence_ignore, patch_file):
 151     """ Performs licence checks """
 152     global failure
 153     if patch_file.endswith(tuple(licence_ext)) \
 154             and patch_file not in licence_ignore:
 155         fo = open(patch_file, 'r')
 156         content = fo.read()
 157         # Note: Hardcoded use of 'copyright' & 'spdx' is the result
 158         # of a decision made at 2017 plugfest to limit searches to
 159         # just these two strings.
 160         patterns = ['copyright', 'spdx',
 161                     'http://creativecommons.org/licenses/by/4.0']
 162         if any(i in content.lower() for i in patterns):
 163             logger.info('Contains needed Licence string: %s', patch_file)
 164         else:
 165             logger.error('Licence header missing in file: %s', patch_file)
 166             failure = True
 167             with open(reports_dir + "licence-" + project + ".log", "a") \
 168                     as gate_report:
 169                 gate_report.write('Licence header missing in file: {0}\n'.
 170                                   format(patch_file))
 171
 172
 173 def process_failure():
 174     """ If any scan operations register a failure, sys.exit(1) is called
 175         to allow jjb to register a failure"""
 176     if failure:
 177         logger.error('Please visit: https://wiki.opnfv.org/x/5oey')
 178         sys.exit(1)