Exceptions

[releng-anteater.git] / anteater / src / project_scan.py

diff --git a/anteater/src/project_scan.py b/anteater/src/project_scan.py
index 09a4791..f33ccca 100644 (file)
--- a/anteater/src/project_scan.py
+++ b/anteater/src/project_scan.py
@@ -16,19 +16,22 @@
 """
 
 from __future__ import division, print_function, absolute_import
-import ConfigParser
+import hashlib
+import six.moves.configparser
 import os
 import re
-import anteater.utils.anteater_logger as antlog
-import anteater.src.get_lists as get_lists
+import logging
 from binaryornot.check import is_binary
 
-logger = antlog.Logger(__name__).getLogger()
-config = ConfigParser.RawConfigParser()
+from . import get_lists
+
+logger = logging.getLogger(__name__)
+config = six.moves.configparser.RawConfigParser()
 config.read('anteater.conf')
 reports_dir = config.get('config', 'reports_dir')
-gate_checks = config.get('config', 'gate_checks')
+master_list = config.get('config', 'master_list')
 ignore_dirs = ['.git']
+hasher = hashlib.sha256()
 
 
 def prepare_project(project, project_dir):
@@ -38,7 +41,7 @@ def prepare_project(project, project_dir):
     lists = get_lists.GetLists()
 
     # Get binary white list
-    binary_list, binary_project_list = lists.binary_list(project)
+    binary_list = lists.binary_list(project)
 
     # Get file name black list and project waivers
     file_audit_list, file_audit_project_list = lists.file_audit_list(project)
@@ -51,8 +54,8 @@ def prepare_project(project, project_dir):
     licence_ignore = lists.licence_ignore()
 
     # Perform rudimentary scans
-    scan_file(project_dir, project, binary_list, binary_project_list,
-              file_audit_list, file_audit_project_list, file_content_list,
+    scan_file(project_dir, project, binary_list,file_audit_list,
+              file_audit_project_list, file_content_list,
               project_content_list)
 
     # Perform licence header checks
@@ -60,8 +63,8 @@ def prepare_project(project, project_dir):
     licence_root_check(project_dir, project)
 
 
-def scan_file(project_dir, project, binary_list, binary_project_list,
-              file_audit_list, file_audit_project_list, file_content_list,
+def scan_file(project_dir, project, binary_list, file_audit_list,
+              file_audit_project_list, file_content_list,
               project_content_list):
     """Searches for banned strings and files that are listed """
     for root, dirs, files in os.walk(project_dir):
@@ -73,10 +76,8 @@ def scan_file(project_dir, project, binary_list, binary_project_list,
             if file_audit_list.search(full_path) and not \
                     file_audit_project_list.search(full_path):
                 match = file_audit_list.search(full_path)
-                logger.error('Blacklisted filename: {0}'.
-                             format(full_path))
-                logger.error('Matched String: {0}'.
-                             format(match.group()))
+                logger.error('Blacklisted filename: %s', full_path)
+                logger.error('Matched String: %s', match.group())
                 with open(reports_dir + "file-names_" + project + ".log",
                           "a") as gate_report:
                             gate_report. \
@@ -94,12 +95,9 @@ def scan_file(project_dir, project, binary_list, binary_project_list,
                     if file_content_list.search(line) and not \
                             project_content_list.search(line):
                         match = file_content_list.search(line)
-                        logger.error('File contains violation: {0}'.
-                                     format(full_path))
-                        logger.error('Flagged Content: {0}'.
-                                     format(line.rstrip()))
-                        logger.error('Matched String: {0}'.
-                                     format(match.group()))
+                        logger.error('File contains violation: %s', full_path)
+                        logger.error('Flagged Content: %s', line.rstrip())
+                        logger.error('Matched String: %s', match.group())
                         with open(reports_dir + "contents-" + project + ".log",
                                   "a") \
                                 as gate_report:
@@ -114,24 +112,32 @@ def scan_file(project_dir, project, binary_list, binary_project_list,
                                               format(match.group()))
             else:
                 # Check if Binary is whitelisted
-                if not binary_list.search(full_path) \
-                        and not binary_project_list.search(full_path):
-                    logger.error('Non Whitelisted Binary: {0}'.
-                                 format(full_path))
-                    with open(reports_dir + "binaries-" + project + ".log",
-                              "a") \
-                            as gate_report:
-                        gate_report.write('Non Whitelisted Binary: {0}\n'.
-                                          format(full_path))
+                hashlist = get_lists.GetLists()
+                binary_hash = hashlist.binary_hash(project, full_path)
+                if not binary_list.search(full_path):
+                    with open(full_path, 'rb') as afile:
+                        buf = afile.read()
+                        hasher.update(buf)
+                    if hasher.hexdigest() in binary_hash:
+                        logger.info('Found matching file hash for file: %s',
+                                    full_path)
+                    else:
+                        logger.error('Non Whitelisted Binary file: %s',
+                                     full_path)
+                        logger.error('Please submit patch with this hash: %s',
+                                     hasher.hexdigest())
+                        with open(reports_dir + "binaries-" + project + ".log",
+                                  "a") \
+                                as gate_report:
+                            gate_report.write('Non Whitelisted Binary: {0}\n'.
+                                              format(full_path))
 
 
 def licence_root_check(project_dir, project):
     if os.path.isfile(project_dir + '/LICENSE'):
-        logger.info('LICENSE file present in: {0}'.
-                    format(project_dir))
+        logger.info('LICENSE file present in: %s', project_dir)
     else:
-        logger.error('LICENSE file missing in: {0}'.
-                     format(project_dir))
+        logger.error('LICENSE file missing in: %s', project_dir)
         with open(reports_dir + "licence-" + project + ".log",
                   "a") \
                 as gate_report:
@@ -154,14 +160,11 @@ def licence_check(licence_ext, licence_ignore, project, project_dir):
                     # of a decision made at 2017 plugfest to limit searches to
                     # just these two strings.
                     if re.search("copyright", content, re.IGNORECASE):
-                        logger.info('Licence string present: {0}'.
-                                    format(full_path))
+                        logger.info('Licence string present: %s', full_path)
                     elif re.search("spdx", content, re.IGNORECASE):
-                        logger.info('Licence string present: {0}'.
-                                    format(full_path))
+                        logger.info('Licence string present: %s', full_path)
                     else:
-                        logger.error('Licence header missing: {0}'.
-                                     format(full_path))
+                        logger.error('Licence header missing: %s', full_path)
                         with open(reports_dir + "licence-" + project + ".log",
                                   "a") \
                                 as gate_report: