From 1106f02e192976315abe7ac02223cba83e660773 Mon Sep 17 00:00:00 2001
From: Tatuya Kamada <tatuya@nexedi.com>
Date: Tue, 31 Jan 2012 15:32:48 +0900
Subject: [PATCH] Fix the problem that web checking result can be duplicated.
---
erp5/util/webchecker/__init__.py | 21 +++++++++++++++++++++
1 file changed, 21 insertions(+)
diff --git a/erp5/util/webchecker/__init__.py b/erp5/util/webchecker/__init__.py
index 335354f064..c9b8e0afc0 100644
--- a/erp5/util/webchecker/__init__.py
+++ b/erp5/util/webchecker/__init__.py
@@ -363,6 +363,25 @@ class HTTPCacheCheckerTestSuite(object):
(header, read_value, reference_value)
self.report_dict.setdefault(url, []).append(message)
+ def _isSameUrl(self, url):
+ """
+ Return whether the url is already checked or not.
+
+ Example case):
+ http://example.com/login_form
+ http://example.com/login_form/
+ """
+ if url in (None, ''):
+ return False
+ same_url = None
+ if url.endswith('/'):
+ same_url = url.rstrip('/')
+ else:
+ same_url = '%s/' % url
+ if same_url in self.report_dict:
+ return True
+ return False
+
def _parseWgetLogs(self, wget_log_file, discarded_url_list=_MARKER,
prohibited_file_name_list=None,
prohibited_folder_name_list=None):
@@ -386,6 +405,8 @@ class HTTPCacheCheckerTestSuite(object):
# URL already checked during first pass
logging.debug('%r Discarded' % url)
discarded = True
+ elif self._isSameUrl(url):
+ discarded = True
if discarded:
# keep reading wget process without doing anything
continue
--
2.30.9