Don't patch non-HTML data from a media playback iframe

waybackproxy.py

@@ -229,7 +229,8 @@ class Handler(socketserver.BaseRequestHandler):
 		else:
 			mode = 0
 
-		# Wayback will add its HTML to anything it thinks is HTML
+		# Check content type to determine if this is HTML we need to patch.
+		# Wayback will add its HTML to anything it thinks is HTML.
 		guessed_content_type = conn.info().get('X-Archive-Guessed-Content-Type')
 		if not guessed_content_type:
 			guessed_content_type = content_type
@@ -244,7 +245,7 @@ class Handler(socketserver.BaseRequestHandler):
 				_print('[r] [QI]', archived_url)
 				return self.send_redirect_page(http_version, archived_url, 301)
 
-			# check if the date is within tolerance
+			# Check if the date is within tolerance.
 			if DATE_TOLERANCE is not None:
 				match = re.search('''//web\\.archive\\.org/web/([0-9]+)''', conn.geturl())
 				if match:
@@ -254,10 +255,10 @@ class Handler(socketserver.BaseRequestHandler):
 						conn.close()
 						return self.send_error_page(http_version, 412, 'Snapshot ' + requested_date + ' not available')
 
-			# consume all data
+			# Consume all data.
 			data = conn.read()
 
-			# patch the page
+			# Patch the page.
 			if mode == 0: # wayback
 				if b'<title>Wayback Machine</title>' in data:
 					if b'<p>This URL has been excluded from the Wayback Machine.</p>' in data: # exclusion error (robots.txt?)
@@ -284,15 +285,29 @@ class Handler(socketserver.BaseRequestHandler):
 							else:
 								return self.send_error_page(http_version, e.code, e.reason)
 
+						# Identify content type so we don't modify non-HTML content.
 						content_type = conn.info().get('Content-Type')
-						if not CONTENT_TYPE_ENCODING and content_type.find(';') > -1:
-							content_type = content_type[:content_type.find(';')]
-						data = conn.read()
+						if not CONTENT_TYPE_ENCODING:
+							idx = content_type.find(';')
+							if idx > -1:
+								content_type = content_type[:idx]
+						if 'text/html' in content_type:
+							# Consume all data and proceed with patching the page.
+							data = conn.read()
+						else:
+							# Pass non-HTML data through.
+							self.send_response_headers(conn, http_version, content_type, request_url)
+							while True:
+								data = conn.read(1024)
+								if not data: break
+								self.request.sendall(data)
+							self.request.close()
+							return
 
 				if b'<title></title>' in data and b'<span class="label style-scope media-button"><!---->Wayback Machine<!----></span>' in data:
 					match = re.search(b'''<p class="impatient"><a href="(?:(?:https?:)?//web\\.archive\\.org)?/web/([^/]+)/([^"]+)">Impatient\\?</a></p>''', data)
 					if match:
-						# wayback redirect page, follow it
+						# This is a Wayback redirect page, follow the redirect.
 						match2 = re.search(b'<p class="code shift red">Got an HTTP ([0-9]+)', data)
 						try:
 							redirect_code = int(match2.group(1))
@@ -303,18 +318,18 @@ class Handler(socketserver.BaseRequestHandler):
 						_print('[r]', archived_url)
 						return self.send_redirect_page(http_version, archived_url, redirect_code)
 
-				# pre-toolbar scripts and CSS
+				# Remove pre-toolbar scripts and CSS.
 				data = re.sub(b'''<script src="//archive\\.org/.*<!-- End Wayback Rewrite JS Include -->\\r?\\n''', b'', data, flags=re.S)
-				# toolbar
+				# Remove toolbar.
 				data = re.sub(b'''<!-- BEGIN WAYBACK TOOLBAR INSERT -->.*<!-- END WAYBACK TOOLBAR INSERT -->''', b'', data, flags=re.S)
-				# comments on footer
+				# Remove comments on footer.
 				data = re.sub(b'''<!--\\r?\\n     FILE ARCHIVED .*$''', b'', data, flags=re.S)
-				# fix base tag
+				# Fix base tag.
 				data = re.sub(b'''(<base (?:[^>]*)href=(?:["\'])?)(?:(?:https?:)?//web.archive.org)?/web/(?:[^/]+)/''', b'\\1', data, flags=re.I + re.S)
 
-				# remove extraneous :80 from links
+				# Remove extraneous :80 from links.
 				data = re.sub(b'((?:(?:https?:)?//web.archive.org)?/web/)([^/]+)/([^:]+)://([^:]+):80/', b'\\1\\2/\\3://\\4/', data)
-				# fix links
+				# Fix links.
 				if QUICK_IMAGES:
 					# QUICK_IMAGES works by intercepting asset URLs (those
 					# with a date code ending in im_, js_...) and letting the
@@ -338,23 +353,24 @@ class Handler(socketserver.BaseRequestHandler):
 						return orig_url
 					data = re.sub(b'''(?:(?:https?:)?//web.archive.org)?/web/([^/]+)/([^"\\'#<>]+)''', add_to_date_cache, data)
 			elif mode == 1: # oocities
-				# viewport/cache-control/max-width code (header)
+				# Remove viewport/cache-control/max-width code from the header.
 				data = re.sub(b'''^.*?\n\n''', b'', data, flags=re.S)
-				# archive notice and tracking code (footer)
+				# Remove archive notice and tracking code from the footer.
 				data = re.sub(b'''<style> \n.zoomout { -webkit-transition: .*$''', b'', data, flags=re.S)
-				# clearly labeled snippets from Geocities
+				# Remove clearly labeled snippets from Geocities.
 				data = re.sub(b'''^.*<\\!-- text above generated by server\\. PLEASE REMOVE -->''', b'', data, flags=re.S)
 				data = re.sub(b'''<\\!-- following code added by server\\. PLEASE REMOVE -->.*<\!-- preceding code added by server\. PLEASE REMOVE -->''', b'', data, flags=re.S)
 				data = re.sub(b'''<\\!-- text below generated by server\\. PLEASE REMOVE -->.*$''', b'', data, flags=re.S)
 
-				# fix links
+				# Fix links.
 				data = re.sub(b'''//([^.]*)\\.oocities\\.com/''', b'//\\1.geocities.com/', data, flags=re.S)
 
+			# Send patched page.
 			self.send_response_headers(conn, http_version, content_type, request_url)
 			self.request.sendall(data)
-		else: # other data
+		else:
+			# Pass non-HTML data through.
 			self.send_response_headers(conn, http_version, content_type, request_url)
-
 			while True:
 				data = conn.read(1024)
 				if not data: break