diff --git a/pywb/apps/frontendapp.py b/pywb/apps/frontendapp.py
index 2de16fa6..80469e5c 100644
--- a/pywb/apps/frontendapp.py
+++ b/pywb/apps/frontendapp.py
@@ -24,7 +24,7 @@ from pywb.warcserver.warcserver import WarcServer
from pywb.rewrite.templateview import BaseInsertView
from pywb.apps.static_handler import StaticHandler
-from pywb.apps.rewriterapp import RewriterApp
+from pywb.apps.rewriterapp import RewriterApp, UpstreamException
from pywb.apps.wbrequestresponse import WbResponse
import os
@@ -441,6 +441,13 @@ class FrontEndApp(object):
coll in self.warcserver.list_dynamic_routes())
def raise_not_found(self, environ, err_type, url):
+ """Utility function for raising a werkzeug.exceptions.NotFound execption with the supplied WSGI environment
+ and message.
+
+ :param dict environ: The WSGI environment dictionary for the request
+ :param str err_type: The identifier for type of error that occured
+ :param str url: The url of the archived page that was requested
+ """
raise AppPageNotFound(err_type, url)
def _check_refer_redirect(self, environ):
diff --git a/pywb/rewrite/html_rewriter.py b/pywb/rewrite/html_rewriter.py
index 92c4e8e0..bf28bf25 100644
--- a/pywb/rewrite/html_rewriter.py
+++ b/pywb/rewrite/html_rewriter.py
@@ -56,7 +56,7 @@ class HTMLRewriterMixin(StreamingRewriter):
'archive': 'oe_'},
'area': {'href': defmod},
'audio': {'src': 'oe_'},
- 'base': {'href': 'ba_'},
+ 'base': {'href': defmod},
'blockquote': {'cite': defmod},
'body': {'background': 'im_'},
'button': {'formaction': defmod},
diff --git a/pywb/rewrite/regex_rewriters.py b/pywb/rewrite/regex_rewriters.py
index 2bcf42d9..9403bb1d 100644
--- a/pywb/rewrite/regex_rewriters.py
+++ b/pywb/rewrite/regex_rewriters.py
@@ -25,8 +25,8 @@ class RxRules(object):
return lambda _, _2: string
@staticmethod
- def archival_rewrite():
- return lambda string, rewriter: rewriter.rewrite(string)
+ def archival_rewrite(mod=None):
+ return lambda string, rewriter: rewriter.rewrite(string, mod)
@staticmethod
def add_prefix(prefix):
@@ -327,13 +327,12 @@ class JSReplaceFuzzy(object):
class CSSRules(RxRules):
CSS_URL_REGEX = "url\\s*\\(\\s*(?:[\\\\\"']|(?:&.{1,4};))*\\s*([^)'\"]+)\\s*(?:[\\\\\"']|(?:&.{1,4};))*\\s*\\)"
- CSS_IMPORT_NO_URL_REGEX = ("@import\\s+(?!url)\\(?\\s*['\"]?" +
- "(?!url[\\s\\(])([\w.:/\\\\-]+)")
+ CSS_IMPORT_REGEX = ("@import\\s+(?:url\\s*)?\\(?\\s*['\"]?([\w.:/\\\\-]+)")
def __init__(self):
rules = [
- (self.CSS_URL_REGEX, self.archival_rewrite(), 1),
- (self.CSS_IMPORT_NO_URL_REGEX, self.archival_rewrite(), 1),
+ (self.CSS_URL_REGEX, self.archival_rewrite('oe_'), 1),
+ (self.CSS_IMPORT_REGEX, self.archival_rewrite('cs_'), 1),
]
super(CSSRules, self).__init__(rules)
diff --git a/pywb/rewrite/test/test_html_rewriter.py b/pywb/rewrite/test/test_html_rewriter.py
index 26f12248..cd7ab24b 100644
--- a/pywb/rewrite/test/test_html_rewriter.py
+++ b/pywb/rewrite/test/test_html_rewriter.py
@@ -25,23 +25,23 @@ r"""
# Base Tests -- w/ rewrite (default)
>>> parse('')
-
+
# Full Path
>>> parse('', urlrewriter=full_path_urlrewriter)
-
+
# Full Path Scheme Rel Base
>>> parse('
', urlrewriter=full_path_urlrewriter)
-
+
# Rel Base
>>> parse('', urlrewriter=full_path_urlrewriter)
-
+
# Rel Base + example
>>> parse('', urlrewriter=full_path_urlrewriter)
-
+
# Rel Base
>>> parse('
', urlrewriter=full_path_urlrewriter)
@@ -53,7 +53,7 @@ r"""
# ensure trailing slash added
>>> parse('')
-
+
# Base Tests -- no rewrite
>>> parse('', urlrewriter=no_base_canon_rewriter)
@@ -244,29 +244,29 @@ r"""
>>> parse('')
-
+
>>> parse('')
-
+
>>> parse('')
-
+
>>> parse('')
-
+
>>> parse('')
>>> parse("")
-
+
#>>> parse('')
# Style
>>> parse('')
-
+
# Unterminated style tag, handle and auto-terminate
>>> parse('