rewrite: js obj proxy rewrite improvements:

- add general ' = this' rewriting to check for proxy obj
- add tests for js obj proxy regex rewriting (without first or last wrapper)

pywb/rewrite/regex_rewriters.py

@@ -21,6 +21,10 @@ class RegexRewriter(StreamingRewriter):
     def format(template):
         return lambda string: template.format(string)
 
+    @staticmethod
+    def fixed(string):
+        return lambda _: string
+
     @staticmethod
     def remove_https(string):
         return string.replace("https", "http")
@@ -33,9 +37,6 @@ class RegexRewriter(StreamingRewriter):
     def archival_rewrite(rewriter):
         return lambda string: rewriter.rewrite(string)
 
-    # @staticmethod
-    # def replacer(other):
-    #    return lambda m, string: other
 
     HTTPX_MATCH_STR = r'https?:\\?/\\?/[A-Za-z0-9:_@.-]+'
 
@@ -178,12 +179,17 @@ if (!self.__WB_pmw) {{ self.__WB_pmw = function(obj) {{ return obj; }} }}\n\
                   'frames',
                   'opener']
 
+    THIS_RW = '(this && this._WB_wombat_obj_proxy || this)'
+
     def __init__(self, rewriter, rules=[]):
+        func_rw = 'Function("return {0}")'.format(self.THIS_RW)
+
         rules = rules + [
-           (r'Function\(["\']return this["\']\)', RegexRewriter.format('Function("return this._WB_wombat_obj_proxy || this")'), 0),
-           (r'(?<![$.])\bthis\b(?=(?:\.(?:{0})))'.format('|'.join(self.local_objs)),
-            RegexRewriter.format('(this && this._WB_wombat_obj_proxy || this)'), 0),
-           (r'(?<=\.)postMessage\b\(', RegexRewriter.add_prefix('__WB_pmw(self).'), 0),
+           (r'Function\(["\']return this["\']\)', self.fixed(func_rw), 0),
+           (r'(?<![$.])\bthis\b(?=(?:\.(?:{0})\b))'.format('|'.join(self.local_objs)),
+            self.fixed(self.THIS_RW), 0),
+           (r'(?<=\.)postMessage\b\(', self.add_prefix('__WB_pmw(self).'), 0),
+           (r'(?<=[=])\s*this\b\s*(?![.$])', self.fixed(self.THIS_RW), 0),
         ]
 
         super(JSWombatProxyRewriterMixin, self).__init__(rewriter, rules)

pywb/rewrite/test/test_regex_rewriters.py

@@ -126,6 +126,32 @@ r"""
 >>> _test_js('"a=b&http:\/\/example.com/;c=d"')
 '"a=b&/web/20131010/http:\\/\\/example.com/;c=d"'
 
+#=================================================================
+# JS Obj Proxy Rewriter
+#=================================================================
+
+>>> _test_js_obj_proxy('var foo = this;   location = bar')
+'var foo =(this && this._WB_wombat_obj_proxy || this);   location = bar'
+
+>>> _test_js_obj_proxy('var foo = this.location')
+'var foo = (this && this._WB_wombat_obj_proxy || this).location'
+
+>>> _test_js_obj_proxy('var foo = this.location2')
+'var foo = this.location2'
+
+>>> _test_js_obj_proxy('func(Function("return this"));')
+'func(Function("return (this && this._WB_wombat_obj_proxy || this)"));'
+
+>>> _test_js_obj_proxy('this.document.location = foo')
+'(this && this._WB_wombat_obj_proxy || this).document.location = foo'
+
+# not rewritten
+>>> _test_js_obj_proxy('var window = this$')
+'var window = this$'
+
+>>> _test_js_obj_proxy('var window = $this')
+'var window = $this'
+
 #=================================================================
 # XML Rewriting
 #=================================================================
@@ -206,6 +232,7 @@ r"""
 #=================================================================
 from pywb.rewrite.url_rewriter import UrlRewriter
 from pywb.rewrite.regex_rewriters import RegexRewriter, JSRewriter, CSSRewriter, XMLRewriter
+from pywb.rewrite.regex_rewriters import JSWombatProxyRewriter
 
 
 urlrewriter = UrlRewriter('20131010/http://example.com/', '/web/', 'https://localhost/web/')
@@ -214,6 +241,12 @@ urlrewriter = UrlRewriter('20131010/http://example.com/', '/web/', 'https://loca
 def _test_js(string, extra = []):
     return JSRewriter(urlrewriter, extra).rewrite(string)
 
+def _test_js_obj_proxy(string):
+    rw = JSWombatProxyRewriter(urlrewriter)
+    rw.first_buff = ''
+    rw.close_string = ''
+    return rw.rewrite(string)
+
 def _test_xml(string):
     return XMLRewriter(urlrewriter).rewrite(string)