Create utils/normalize-securecookie.py (EFForg#13864)

Chan Chak Shing · Hainish · commit 45dd7e40d311 · 2017-12-15T12:44:28.000-08:00
diff --git a/test/travis.sh b/test/travis.sh
@@ -42,8 +42,7 @@ if [ "$TEST" == "rules" ]; then
   docker run --rm -ti -v $(pwd):/opt httpse bash -c "utils/remove-obsolete-references.sh"
   docker run --rm -ti -v $(pwd):/opt httpse bash -c "utils/validate.sh"
   docker run --rm -ti -v $(pwd):/opt httpse bash -c "test/rules.sh"
-  docker run --rm -ti -v $(pwd):/opt node bash -c "cd /opt && node utils/normalize-securecookie.js"
-  [ `git diff --name-only $RULESETFOLDER | wc -l` -eq 0 ]
+  docker run --rm -ti -v $(pwd):/opt httpse python utils/normalize-securecookie.py
 fi
 
 if [ "$TEST" == "fetch" ]; then
diff --git a/utils/normalize-securecookie.js b/utils/normalize-securecookie.js
diff --git a/utils/normalize-securecookie.py b/utils/normalize-securecookie.py
@@ -0,0 +1,58 @@
+#!/usr/bin/env python2.7
+
+# This python utility check for wildcard securecookies which
+# can be normalized, warn and exit with non-zero when such 
+# rulesets exist. 
+
+# This is create in attempt to fix the issues on
+# https://github.com/EFForg/https-everywhere/pull/13840
+# https://github.com/EFForg/https-everywhere/pull/12493
+
+import argparse
+import glob
+import os
+import unicodedata
+import xml.etree.ElementTree
+
+def normalize_fn(fn):
+    """
+    OSX and Linux filesystems encode composite characters differently in 
+    filenames. We should normalize to NFC: https://unicode.org/reports/tr15/
+    """
+    fn = unicodedata.normalize("NFC", unicode(fn, "utf-8")).encode()
+    return fn
+
+def should_normalize_securecookie(host, name):
+    wildcards = [ ".", ".*" ]
+    return True if host in wildcards or name in wildcards else False
+
+# commandline arguments parsing (nobody use it, though)
+parser = argparse.ArgumentParser(description="Normalize wildcard securecookies")
+parser.add_argument("--source_dir", default="src/chrome/content/rules")
+
+args = parser.parse_args()
+
+# Exit code
+exit_with_non_zero = False
+
+# XML ruleset files
+filenames = map(normalize_fn, glob.glob(os.path.join(args.source_dir, "*.xml")))
+
+for filename in filenames:
+    tree = xml.etree.ElementTree.parse(filename)
+    root = tree.getroot()
+
+    for branch in root:
+        if branch.tag != "securecookie":
+            continue
+
+        host = branch.attrib["host"]
+        name = branch.attrib["name"]
+
+        if should_normalize_securecookie(host, name):
+            print ("ERROR %s: contains wildcard securecookies "\
+                    "which can be normalized." % filename)
+            exit_with_non_zero = True
+            break
+
+exit(0) if not exit_with_non_zero else exit(1)
