Ignore file/URL paths in same check

trans/checks/same.py

@@ -215,6 +215,7 @@ DOMAIN_RE = re.compile(
     re.IGNORECASE
 )
 
+PATH_RE = re.compile(r'(/[a-zA-Z0-9=:?._-]+)+')
 
 class SameCheck(TargetCheck):
     '''
@@ -260,6 +261,9 @@ class SameCheck(TargetCheck):
         # Strip domain names/URLs
         stripped = DOMAIN_RE.sub('', stripped)
 
+        # Strip file/URL paths
+        stripped = PATH_RE.sub('', stripped)
+
         # Remove some html entities
         stripped = stripped.replace(
             ' ', ' '

trans/tests/same_checks.py

@@ -199,3 +199,21 @@ class SameCheckTest(CheckTestCase):
                 ''
             )
         )
+
+    def test_same_path(self):
+        self.do_test(
+            False,
+            (
+                '/cgi-bin/koha/catalogue/search.pl?q=',
+                '/cgi-bin/koha/catalogue/search.pl?q=',
+                ''
+            )
+        )
+        self.do_test(
+            True,
+            (
+                'File/directory',
+                'File/directory',
+                ''
+            )
+        )