streaming rewriter improvements:

- add optional 'first_buff' defaulting to ''
- rename close() -> final_read()
- add rewrite_complete() for single-pass complete rewrite (including first buff and final_read()
- rewrite_text_stream_to_gen() uses first_buff, uses member funcs directly
- remove unused close() from other rewriters, only needed for HTMLParser interface

pywb/rewrite/content_rewriter.py

@@ -201,49 +201,48 @@ class BufferedRewriter(object):
 
 # ============================================================================
 class StreamingRewriter(object):
-    def __init__(self, url_rewriter, align_to_line=True):
+    def __init__(self, url_rewriter, align_to_line=True, first_buff=''):
         self.url_rewriter = url_rewriter
         self.align_to_line = align_to_line
+        self.first_buff = first_buff
 
     def __call__(self, rwinfo):
-        gen = self.rewrite_text_stream_to_gen(rwinfo.content_stream,
+        return self.rewrite_text_stream_to_gen(rwinfo.content_stream)
-                                              rewrite_func=self.rewrite,
-                                              final_read_func=self.close,
-                                              align_to_line=self.align_to_line)
-
-        return gen
 
     def rewrite(self, string):
         return string
 
-    def close(self):
+    def rewrite_complete(self, string):
+        return self.first_buff + self.rewrite(string) + self.final_read()
+
+    def final_read(self):
         return ''
 
-    def rewrite_text_stream_to_gen(cls, stream,
+    def rewrite_text_stream_to_gen(self, stream):
-                                   rewrite_func,
-                                   final_read_func,
-                                   align_to_line):
         """
         Convert stream to generator using applying rewriting func
         to each portion of the stream.
         Align to line boundaries if needed.
         """
         try:
-            buff = ''
+            buff = self.first_buff
+
+            if buff:
+                yield buff.encode('iso-8859-1')
 
             while True:
                 buff = stream.read(BUFF_SIZE)
                 if not buff:
                     break
 
-                if align_to_line:
+                if self.align_to_line:
                     buff += stream.readline()
 
-                buff = rewrite_func(buff.decode('iso-8859-1'))
+                buff = self.rewrite(buff.decode('iso-8859-1'))
                 yield buff.encode('iso-8859-1')
 
             # For adding a tail/handling final buffer
-            buff = final_read_func()
+            buff = self.final_read()
             if buff:
                 yield buff.encode('iso-8859-1')
 

pywb/rewrite/html_rewriter.py

@@ -237,7 +237,7 @@ class HTMLRewriterMixin(StreamingRewriter):
 
     def _rewrite_css(self, css_content):
         if css_content:
-            return self.css_rewriter.rewrite(css_content)
+            return self.css_rewriter.rewrite_complete(css_content)
         else:
             return ''
 
@@ -245,7 +245,7 @@ class HTMLRewriterMixin(StreamingRewriter):
         if not script_content:
             return ''
 
-        content = self.js_rewriter.rewrite(script_content)
+        content = self.js_rewriter.rewrite_complete(script_content)
         if ensure_window:
             content = self.ADD_WINDOW.sub('window.\\1', content)
 
@@ -456,7 +456,7 @@ class HTMLRewriterMixin(StreamingRewriter):
 
         return result
 
-    def close(self):
+    def final_read(self):
         self.out = self.AccumBuff()
 
         self._internal_close()
@@ -468,6 +468,9 @@ class HTMLRewriterMixin(StreamingRewriter):
 
         return result
 
+    def close(self):
+        return self.final_read()
+
     def _internal_close(self):  # pragma: no cover
         raise NotImplementedError('Base method')
 
@@ -562,8 +565,7 @@ class HTMLRewriter(HTMLRewriterMixin, HTMLParser):
             comment_rewriter = HTMLRewriter(self.url_rewriter,
                                             defmod=self.defmod)
 
-            data = comment_rewriter.rewrite(data)
+            data = comment_rewriter.rewrite_complete(data)
-            data += comment_rewriter.close()
             self.out.write(data)
         else:
             self.parse_data(data)

pywb/rewrite/jsonp_rewriter.py

@@ -21,8 +21,3 @@ class JSONPRewriter(StreamingRewriter):
         string = m_callback.group(1) + string[m_json.end(1):]
         return string
 
-    def close(self):
-        return ''
-
-
-

pywb/rewrite/regex_rewriters.py

@@ -61,9 +61,6 @@ class RegexRewriter(StreamingRewriter):
     def rewrite(self, string):
         return self.regex.sub(lambda x: self.replace(x), string)
 
-    def close(self):
-        return ''
-
     def replace(self, m):
         i = 0
         for _, op, count in self.rules: