banner: add back inner frame update of banner on load, if html

rewrite: banner only mode encodes to utf-8, adjusts length

pywb/rewrite/rewrite_content.py

@@ -118,9 +118,22 @@ class RewriteContent:
 
             if head_insert_func:
                 head_insert_str = head_insert_func(rule, cdx)
+                head_insert_str = head_insert_str.encode('utf-8')
 
             if wb_url.is_banner_only:
                 gen = self._head_insert_only_gen(head_insert_str, stream)
+
+                content_len = headers.get_header('Content-Length')
+                try:
+                    content_len = int(content_len)
+                except Exception:
+                    content_len = None
+
+                if content_len and content_len >= 0:
+                    content_len = str(content_len + len(head_insert_str))
+                    status_headers.replace_header('Content-Length',
+                                                  content_len)
+
                 return (status_headers, gen, False)
 
             rewriter = rewriter_class(urlrewriter,
@@ -162,7 +175,8 @@ class RewriteContent:
         matcher = self.HEAD_REGEX.search(buff)
 
         if matcher:
-            yield buff[:matcher.end()] + insert_str
+            yield buff[:matcher.end()]
+            yield insert_str
             yield buff[matcher.end():]
         else:
             yield insert_str

pywb/static/wb.js

@@ -119,14 +119,36 @@ function remove_event(name, func, object) {
     }
 }
 
-if ((window.self == window.top) && (window.self.top == window.top) && wbinfo) {
-    if (wbinfo.canon_url && (window.location.href != wbinfo.canon_url)) {
+function notify_top() {
+    if (window.parent != window.top) {
+        return;
+    }
+
+    if (!window.WB_wombat_location) {
+        return;
+    }
+
+    if (typeof(window.WB_wombat_location.href) != "string") {
+        return;
+    }
+
+    window.parent.update_wb_url(window.WB_wombat_location.href,
+                                wbinfo.timestamp,
+                                wbinfo.is_live);
+
+    remove_event("readystatechange", notify_top, document);
+}
+
+if ((window.self == window.top) && wbinfo) {
+    if (wbinfo.canon_url && (window.location.href != wbinfo.canon_url) && wbinfo.mod != "bn_") {
         // Auto-redirect to top frame
         window.location.replace(wbinfo.canon_url);
     } else {
         // Init Banner (no frame or top frame)
         add_event("readystatechange", init_banner, document);
     }
+} else if (window.self != window.parent && window.parent.update_wb_url) {
+    add_event("readystatechange", notify_top, document);
 }
 
 

pywb/static/wb_frame.js

@@ -1,7 +1,7 @@
-var update_wb_url = push_state;
-
 var LIVE_COOKIE_REGEX = /pywb.timestamp=([\d]{1,14})/;
 
+var curr_state = {};
+
 
 function make_outer_url(url, ts)
 {
@@ -41,13 +41,13 @@ function push_state(url, timestamp, capture_str, is_live) {
     
     window.history.replaceState(state, "", state.outer_url);
 
-    update_status(state.capture_str, is_live);
+    set_state(state);
 }
 
 function pop_state(state) {
-    update_status(state.capture_str, state.is_live);
+    set_state(state);
     
-    window.frames[0].src = state.outer_url;
+    window.frames[0].src = state.inner_url;
 }
 
 function extract_ts(url)
@@ -75,27 +75,29 @@ function extract_replay_url(url) {
     return url.substring(inx + 1);
 }
 
-function update_status(str, is_live) {
+function set_state(state) {
     var capture_info = document.getElementById("_wb_capture_info");
     if (capture_info) {
-        capture_info.innerHTML = str;
+        capture_info.innerHTML = state.capture_str;
     }
 
     var label = document.getElementById("_wb_label");
     if (label) {
-        if (is_live) {
+        if (state.is_live) {
             label.innerHTML = _wb_js.labels.LIVE_MSG;
         } else {
             label.innerHTML = _wb_js.labels.REPLAY_MSG;
         }
     }
+
+    curr_state = state;
 }
 
 window.onpopstate = function(event) {
-    var curr_state = event.state;
+    var state = event.state;
     
-    if (curr_state) {
-        pop_state(curr_state);
+    if (state) {
+        pop_state(state);
     }
 }
 
@@ -132,7 +134,16 @@ function iframe_loaded(event) {
             ts = extract_ts(iframe.location.href);
         }
     }
+
+    update_wb_url(url, ts, is_live);
+}
+
+function update_wb_url(url, ts, is_live) {
+    if (curr_state.url == url && curr_state.timestamp == ts) {
+        return;
+    }
+
     capture_str = _wb_js.ts_to_date(ts, true);
 
-    update_wb_url(url, ts, capture_str, is_live);
+    push_state(url, ts, capture_str, is_live);
 }

pywb/static/wombat.js

@@ -699,7 +699,7 @@ WB_wombat_init = (function() {
         wb_replay_prefix = replay_prefix;
         
         if (wb_replay_prefix) {
-            wb_replay_date_prefix = replay_prefix + capture_date + "em_/";
+            wb_replay_date_prefix = replay_prefix + capture_date + "mp_/";
             
             if (capture_date.length > 0) {
                 wb_capture_date_part = "/" + capture_date + "/";

pywb/ui/head_insert.html

@@ -14,7 +14,7 @@
   wbinfo = {}
   wbinfo.timestamp = "{{ cdx.timestamp }}";
   wbinfo.prefix = "{{ wbrequest.wb_prefix }}";
-  wbinfo.is_frame_mp = {{"true" if wbrequest.wb_url.mod == 'mp_' else "false"}};
+  wbinfo.mod = "{{ wbrequest.wb_url.mod }}";
   wbinfo.canon_url = "{{ canon_url }}";
   wbinfo.is_live = {{ "true" if cdx.is_live else "false" }};
   wbinfo.is_proxy_mode = {{ "true" if wbrequest.options.is_proxy else "false" }};