rewrite: better fix for multiple ../ in urls, additional tests

pywb/rewrite/test/test_regex_rewriters.py

@@ -89,6 +89,9 @@ r"""
 >>> _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:#000 url('/static/styles/../../images/layout/logo.png')")
+"background:#000 url('/web/20131010em_/http://example.com/images/layout/logo.png')"
+
 >>> _test_css("background: url('')")
 "background: url('')"
 

pywb/rewrite/test/test_url_rewriter.py

@@ -1,4 +1,21 @@
 """
+# urljoin tests
+
+>>> UrlRewriter.urljoin('http://example.com/test/', '../file.html')
+'http://example.com/file.html'
+
+>>> UrlRewriter.urljoin('http://example.com/test/', '../path/../../../file.html')
+'http://example.com/file.html'
+
+>>> UrlRewriter.urljoin('http://example.com/test/', '/../file.html')
+'http://example.com/file.html'
+
+>>> UrlRewriter.urljoin('http://example.com/', '/abc/../../file.html')
+'http://example.com/file.html'
+
+>>> UrlRewriter.urljoin('http://example.com/path/more/', 'abc/../../file.html')
+'http://example.com/path/file.html'
+
 # UrlRewriter tests
 >>> do_rewrite('other.html', '20131010/http://example.com/path/page.html', 'https://web.archive.org/web/')
 'https://web.archive.org/web/20131010/http://example.com/path/other.html'

pywb/rewrite/url_rewriter.py

@@ -102,14 +102,18 @@ class UrlRewriter(object):
         scheme, netloc, path, query, frag = parts
 
         path_parts = path.split('/')
-        i = len(path_parts) - 1
+        i = 0
-        while i >= 0:
+        n = len(path_parts) - 1
+        while i < n:
             if path_parts[i] == '..':
                 del path_parts[i]
+                n -= 1
                 if i > 0:
                     del path_parts[i - 1]
+                    n -= 1
                     i -= 1
-            i -= 1
+            else:
+                i += 1
 
         if path_parts == ['']:
             path = '/'