From 7032160cf9327565948315de15b65433458c6c68 Mon Sep 17 00:00:00 2001
From: Ilya Kreymer <ikreymer@gmail.com>
Date: Mon, 14 Jul 2014 19:13:19 -0700
Subject: [PATCH] rewrite: fix rel url resolution to better handle parent rel
 path. Explicitly resolve path when possible, remove only if at root level

---
 pywb/rewrite/test/test_regex_rewriters.py |  3 +++
 pywb/rewrite/url_rewriter.py              | 32 ++++++++++++++++++++++-
 2 files changed, 34 insertions(+), 1 deletion(-)

diff --git a/pywb/rewrite/test/test_regex_rewriters.py b/pywb/rewrite/test/test_regex_rewriters.py
index cbd2cb21..4391edee 100644
--- a/pywb/rewrite/test/test_regex_rewriters.py
+++ b/pywb/rewrite/test/test_regex_rewriters.py
@@ -86,6 +86,9 @@ r"""
 >>> _test_css("background: url(file.jpeg)")
 'background: url(/web/20131010em_/http://example.com/file.jpeg)'
 
+>>> _test_css("background:#abc url('/static/styles/../images/layout/logo.png')")
+"background:#abc url('/web/20131010em_/http://example.com/static/images/layout/logo.png')"
+
 >>> _test_css("background: url('')")
 "background: url('')"
 
diff --git a/pywb/rewrite/url_rewriter.py b/pywb/rewrite/url_rewriter.py
index 79136ff5..236aba96 100644
--- a/pywb/rewrite/url_rewriter.py
+++ b/pywb/rewrite/url_rewriter.py
@@ -57,7 +57,7 @@ class UrlRewriter(object):
         else:
             # optimize: join if not absolute url, otherwise just use that
             if not is_abs:
-                new_url = urlparse.urljoin(wburl.url, url).replace('../', '')
+                new_url = self.urljoin(wburl.url, url)
             else:
                 new_url = url
 
@@ -92,6 +92,36 @@ class UrlRewriter(object):
     def __repr__(self):
         return "UrlRewriter('{0}', '{1}')".format(self.wburl, self.prefix)
 
+    @staticmethod
+    def urljoin(orig_url, url):
+        new_url = urlparse.urljoin(orig_url, url)
+        if '../' not in new_url:
+            return new_url
+
+        parts = urlparse.urlsplit(new_url)
+        scheme, netloc, path, query, frag = parts
+
+        path_parts = path.split('/')
+        i = len(path_parts) - 1
+        while i >= 0:
+            if path_parts[i] == '..':
+                del path_parts[i]
+                if i > 0:
+                    del path_parts[i - 1]
+                    i -= 1
+            i -= 1
+
+        if path_parts == ['']:
+            path = '/'
+        else:
+            path = '/'.join(path_parts)
+
+        parts = (scheme, netloc, path, query, frag)
+
+
+        new_url = urlparse.urlunsplit(parts)
+        return new_url
+
 
 #=================================================================
 class HttpsUrlRewriter(object):