Add frameId to executeScript/insertCSS

extensions/common/api/extension_types.json

Index: extensions/common/api/extension_types.json
diff --git a/extensions/common/api/extension_types.json b/extensions/common/api/extension_types.json
index 3efc2f23ae3e455cd3b5a724a85b7d0c794f95d9..88c04a96da84f95b3fb94cd2dd46841e3bdde92b 100644
--- a/extensions/common/api/extension_types.json
+++ b/extensions/common/api/extension_types.json
@@ -34,7 +34,17 @@
         "properties": {
           "code": {"type": "string", "optional": true, "description": "JavaScript or CSS code to inject.<br><br><b>Warning:</b><br>Be careful using the <code>code</code> parameter. Incorrect use of it may open your extension to <a href=\"https://en.wikipedia.org/wiki/Cross-site_scripting\">cross site scripting</a> attacks."},
           "file": {"type": "string", "optional": true, "description": "JavaScript or CSS file to inject."},
-          "allFrames": {"type": "boolean", "optional": true, "description": "If allFrames is <code>true</code>, implies that the JavaScript or CSS should be injected into all frames of current page. By default, it's <code>false</code> and is only injected into the top frame."},
+          "allFrames": {
+            "type": "boolean",
+            "optional": true,
+            "description": "If allFrames is <code>true</code>, implies that the JavaScript or CSS should be injected into all frames of current page. By default, it's <code>false</code> and is only injected into the top frame. If <code>true</code> and <code>frameId</code> is set, then the code is inserted in the selected frame and all of its child frames."
+          },
+          "frameId": {
+            "type": "integer",
+            "optional": true,
+            "minimum": 0,
+            "description": "The <a href='webNavigation#frame_ids'>frame</a> where the script or CSS should be injected."

[nasko, 2015/02/23 18:39:02]

A frame id on its own is not unique and cannot fully qualify a frame. It should
be a pair of process id and frame id.

[robwu, 2015/02/23 18:59:52]

Within a tab, frameIds are supposed to be unique.
The fact that this is not always the case (e.g. with OOPIF) is a bug:
https://crbug.com/432875.

[not at google - send to devlin, 2015/02/23 20:41:24]

Worst case we can have our own (process id, frame id) --> (extension frame id)
mapping for all classes, but it would be nice not to. Requiring both process and
frame IDs for all these APIs would be a PITA, like Rob says.

In other circumstances I'd include "frame IDs are unique within tabs" in the
description, but it sounds like that's not the case.

[robwu, 2015/02/23 21:54:06]

If I had to create a central place to do this mapping, what would be the best
location for that? It must certainly be available in the browser, but having the
logic in the renderer would also be nice.

[not at google - send to devlin, 2015/02/23 22:04:14]

I was thinking of making a BrowserContextKeyedService or whatever. Pretty
simple.

I can't think of a reason why the renderer would need to do much else than
believe what it was told from an API perspective (internally it could still use
the frame ID), but if you did, you'd need to IPC it I suppose.

[dcheng, 2015/02/24 21:54:28]

We already have an API (WebNavigation) that uses (processId, frameId). What's
the plan for that then? Or is web navigation just going to be the odd one out?

+          },
           "matchAboutBlank": {"type": "boolean", "optional": true, "description": "If matchAboutBlank is true, then the code is also injected in about:blank and about:srcdoc frames if your extension has access to its parent document. Code cannot be inserted in top-level about:-frames. By default it is <code>false</code>."},
           "runAt": {
             "type": "string",