From c28e38718c1fa8424113ec1469ce8317231e52c2 Mon Sep 17 00:00:00 2001
From: John Berlin <n0tan3rd@gmail.com>
Date: Wed, 10 Oct 2018 18:24:34 -0400
Subject: [PATCH] Updated html_rewriter.py to correctly handle self-closing
 <script> elements: (#392)

- adding the 'xlink:href' attribute to script element attributes to rewrite
Updated html_rewriter.py to better handle self closing tags:
- added boolean set_parsing_context arg to _rewrite_tag_attrs to indicate if the parsing context is to be set
- the call to _rewrite_tag_attrs from handle_startendtag now sets set_parsing_context to false
Added a test to test_html_rewriter.py for rewriting SVGScriptElements
---
 pywb/rewrite/html_rewriter.py           | 24 ++++++++++++++++++++----
 pywb/rewrite/test/test_html_rewriter.py |  8 ++++++++
 2 files changed, 28 insertions(+), 4 deletions(-)

diff --git a/pywb/rewrite/html_rewriter.py b/pywb/rewrite/html_rewriter.py
index f8848c2d..41be3364 100644
--- a/pywb/rewrite/html_rewriter.py
+++ b/pywb/rewrite/html_rewriter.py
@@ -63,7 +63,7 @@ class HTMLRewriterMixin(StreamingRewriter):
             'param':   {'value': 'oe_'},
             'q':       {'cite': defmod},
             'ref':     {'href': 'oe_'},
-            'script':  {'src': 'js_'},
+            'script':  {'src': 'js_', 'xlink:href': 'js_'},  # covers both HTML and SVG script tags
             'source':  {'src': 'oe_'},
             'video':   {'src': 'oe_',
                         'poster': 'im_'},
@@ -310,7 +310,22 @@ class HTMLRewriterMixin(StreamingRewriter):
 
         return None
 
-    def _rewrite_tag_attrs(self, tag, tag_attrs):
+    def _rewrite_tag_attrs(self, tag, tag_attrs, set_parsing_context=True):
+        """Rewrite a tags attributes.
+
+        If set_parsing_context is false then the parsing context will not set.
+        If the head insert has not been added to the HTML being rewritten, there
+        is no parsing context and the tag is not in BEFORE_HEAD_TAGS then the
+        head_insert will be "inserted" and set to None
+
+        :param str tag: The name of the tag to be rewritten
+        :param list[tuple[str, str]] tag_attrs: A list of tuples representing
+        the tags attributes
+        :param bool set_parsing_context: Boolean indicating if the parsing
+        context should be set
+        :return: True
+        :rtype: bool
+        """
         # special case: head insertion, before-head tags
         if (self.head_insert and
               not self._wb_parse_context
@@ -318,7 +333,8 @@ class HTMLRewriterMixin(StreamingRewriter):
             self.out.write(self.head_insert)
             self.head_insert = None
 
-        self._set_parse_context(tag, tag_attrs)
+        if set_parsing_context:
+            self._set_parse_context(tag, tag_attrs)
 
         # attr rewriting
         handler = self.rewrite_tags.get(tag)
@@ -604,7 +620,7 @@ class HTMLRewriter(HTMLRewriterMixin, HTMLParser):
             self.out.write('>')
 
     def handle_startendtag(self, tag, attrs):
-        self._rewrite_tag_attrs(tag, attrs)
+        self._rewrite_tag_attrs(tag, attrs, False)
 
         if tag != 'head' or not self._rewrite_head(True):
             self.out.write('/>')
diff --git a/pywb/rewrite/test/test_html_rewriter.py b/pywb/rewrite/test/test_html_rewriter.py
index d85b7b78..371867a8 100644
--- a/pywb/rewrite/test/test_html_rewriter.py
+++ b/pywb/rewrite/test/test_html_rewriter.py
@@ -223,6 +223,14 @@ r"""
 >>> parse('<script>window.location = "http://example.com/a/b/c.html"</sc>')
 <script>window.WB_wombat_location = "/web/20131226101010/http://example.com/a/b/c.html"</sc></script>
 
+# SVG Script tag
+>>> parse('<script xlink:href="/js/scripts.js"/>')
+<script xlink:href="/web/20131226101010js_/http://example.com/js/scripts.js"/>
+
+# SVG Script tag with other elements
+>>> parse('<svg><defs><script xlink:href="/js/scripts.js"/><defs/><title>I\'m a title tag in svg!</title></svg>')
+<svg><defs><script xlink:href="/web/20131226101010js_/http://example.com/js/scripts.js"/><defs/><title>I'm a title tag in svg!</title></svg>
+
 >>> parse('<script>/*<![CDATA[*/window.location = "http://example.com/a/b/c.html;/*]]>*/"</script>')
 <script>/*<![CDATA[*/window.WB_wombat_location = "/web/20131226101010/http://example.com/a/b/c.html;/*]]>*/"</script>