activityLogPrivate API: schema & events

This implements half of the activityLogPrivate API. The API is only available to a single whitelisted extension. It includes the schema and events. It is missing the implementation of its function, which I'll add as a separate CL.

BUG=241672
Committed: https://src.chromium.org/viewvc/chrome?view=rev&revision=204796

[felt, 2013-05-26 23:10:27]

Hi Matt & Ben,

The implementation isn't finished yet, but the schema & docs are now available
if you want to take a look at them before I finish writing the implementation.

Adrienne

[Matt Perry, 2013-05-28 19:00:37]

Can I ask you to break this CL up into 3 parts? The bugfix, removing the old UI,
and adding the API. They seem like independent pieces, and it'd be easier to
review them separately.

[felt, 2013-05-28 20:16:05]

Sure. I will break it up into two: removing the old UI, and adding the API. The
old UI was the cause of the bug so the bugfix is inherently part of that CL.

On 2013/05/28 19:00:37, Matt Perry wrote:
> Can I ask you to break this CL up into 3 parts? The bugfix, removing the old
UI,
> and adding the API. They seem like independent pieces, and it'd be easier to
> review them separately.

[mvrable, 2013-05-31 18:55:23]

I've only looked at a few of the files so far, but have a few comments on in the
WebRequest and event router components.

I'll read through the rest and send any additional comments I have later.

--Michael Vrable

https://codereview.chromium.org/16061002/diff/19001/chrome/browser/extensions...
File chrome/browser/extensions/activity_log/activity_log.cc (right):

https://codereview.chromium.org/16061002/diff/19001/chrome/browser/extensions...
chrome/browser/extensions/activity_log/activity_log.cc:201: if (!IsLogEnabled()
|| extension_id == kActivityLogExtensionId) return;
Alternately, you could define an IsLogEnabledForExtension(extension_id) method
to encapsulate this check.

https://codereview.chromium.org/16061002/diff/19001/chrome/browser/extensions...
File chrome/browser/extensions/api/web_request/web_request_api.cc (right):

https://codereview.chromium.org/16061002/diff/19001/chrome/browser/extensions...
chrome/browser/extensions/api/web_request/web_request_api.cc:1594: if
(!extensions::ActivityLog::GetInstance(profile)->IsLogEnabled())
I was told while writing this code that it's not safe to dereference profile
while on the I/O thread, which is why the code doesn't call
ActivityLog::GetInstance(profile) until on the UI thread.  (See: Matt Perry's
comment in
https://codereview.chromium.org/11946028/diff/24010/chrome/browser/extensions...)

I think this was one of the reasons for having a global, rather than
per-profile, IsLogEnabled() method.

https://codereview.chromium.org/16061002/diff/19001/chrome/browser/extensions...
chrome/browser/extensions/api/web_request/web_request_api.cc:1638:
LogExtensionActivity(static_cast<Profile*>(profile),
When writing this part of the patch, I had been asked to make the call to
LogExtensionActivity conditional on ActivityLog::IsLogEnabled() so it wouldn't
slow down the WebRequest API in the non-activity log case (the call to
SummarizeResponseDelta could do a fair bit of work).

Just making you aware.  This may be less of an issue going forward if we expect
the activity log to be enabled by default (eventually).

https://codereview.chromium.org/16061002/diff/19001/chrome/browser/extensions...
File chrome/browser/extensions/event_router.cc (right):

https://codereview.chromium.org/16061002/diff/19001/chrome/browser/extensions...
chrome/browser/extensions/event_router.cc:127: Profile* profile =
reinterpret_cast<Profile*>(profile_id);
Same comment as before about accessing the Profile object from a non-UI thread.

[felt, 2013-05-31 20:48:36]

On 2013/05/31 18:55:23, mvrable wrote:
> I've only looked at a few of the files so far, but have a few comments on in
the
> WebRequest and event router components.
> 
> I'll read through the rest and send any additional comments I have later.
> 
> --Michael Vrable
> 
>
https://codereview.chromium.org/16061002/diff/19001/chrome/browser/extensions...
> File chrome/browser/extensions/activity_log/activity_log.cc (right):
> 
>
https://codereview.chromium.org/16061002/diff/19001/chrome/browser/extensions...
> chrome/browser/extensions/activity_log/activity_log.cc:201: if
(!IsLogEnabled()
> || extension_id == kActivityLogExtensionId) return;
> Alternately, you could define an IsLogEnabledForExtension(extension_id) method
> to encapsulate this check.
> 
>
https://codereview.chromium.org/16061002/diff/19001/chrome/browser/extensions...
> File chrome/browser/extensions/api/web_request/web_request_api.cc (right):
> 
>
https://codereview.chromium.org/16061002/diff/19001/chrome/browser/extensions...
> chrome/browser/extensions/api/web_request/web_request_api.cc:1594: if
> (!extensions::ActivityLog::GetInstance(profile)->IsLogEnabled())
> I was told while writing this code that it's not safe to dereference profile
> while on the I/O thread, which is why the code doesn't call
> ActivityLog::GetInstance(profile) until on the UI thread.  (See: Matt Perry's
> comment in
>
https://codereview.chromium.org/11946028/diff/24010/chrome/browser/extensions...)
> 
> I think this was one of the reasons for having a global, rather than
> per-profile, IsLogEnabled() method.
> 
>
https://codereview.chromium.org/16061002/diff/19001/chrome/browser/extensions...
> chrome/browser/extensions/api/web_request/web_request_api.cc:1638:
> LogExtensionActivity(static_cast<Profile*>(profile),
> When writing this part of the patch, I had been asked to make the call to
> LogExtensionActivity conditional on ActivityLog::IsLogEnabled() so it wouldn't
> slow down the WebRequest API in the non-activity log case (the call to
> SummarizeResponseDelta could do a fair bit of work).
> 
> Just making you aware.  This may be less of an issue going forward if we
expect
> the activity log to be enabled by default (eventually).
> 
>
https://codereview.chromium.org/16061002/diff/19001/chrome/browser/extensions...
> File chrome/browser/extensions/event_router.cc (right):
> 
>
https://codereview.chromium.org/16061002/diff/19001/chrome/browser/extensions...
> chrome/browser/extensions/event_router.cc:127: Profile* profile =
> reinterpret_cast<Profile*>(profile_id);
> Same comment as before about accessing the Profile object from a non-UI
thread.

Thanks Michael! That's all very helpful. Since it seems like modifying
IsLogEnabled will be non-trivial, I'm going to move that into its own separate
CL.

[mvrable, 2013-06-01 00:15:45]

On 2013/05/31 18:55:23, mvrable wrote:
> I'll read through the rest and send any additional comments I have later.

Looking over the rest of the files nothing else jumped out at me.

--Michael Vrable

[felt, 2013-06-05 00:38:39]

Hi Matt,

Can you please review this CL? It includes:

- the schema for the whole API and the overall API framework code
- the implementation of the events
- unit and browser tests for the events
- small modifications to the activity_log classes to support the API

I have left out two other things which I will put in two other CLs:

- the implementation for the GetActions function
- changing how IsEnabled works to account for the presence of the extension

Thanks,
Adrienne

[felt, 2013-06-05 00:41:24]

Hi Ilya,

Adding you as a TBR reviewer, but sending this in case you want to take a look.
This CL makes a tiny change to histograms.xml due to introduction of new API.

Adrienne

[Matt Perry, 2013-06-05 01:24:47]

https://codereview.chromium.org/16061002/diff/66001/chrome/browser/extensions...
File chrome/browser/extensions/activity_log/activity_actions.h (right):

https://codereview.chromium.org/16061002/diff/66001/chrome/browser/extensions...
chrome/browser/extensions/activity_log/activity_actions.h:32: virtual
std::map<std::string, std::string> GetFieldsAsMap() = 0;
Why not have this allocate and return an ExtensionActivity instead?

https://codereview.chromium.org/16061002/diff/66001/chrome/browser/extensions...
chrome/browser/extensions/activity_log/activity_actions.h:40: const { return
activity_type_; }
nit: wrap before the function name

https://codereview.chromium.org/16061002/diff/66001/chrome/browser/extensions...
File
chrome/browser/extensions/api/activity_log_private/activity_log_private_api.cc
(right):

https://codereview.chromium.org/16061002/diff/66001/chrome/browser/extensions...
chrome/browser/extensions/api/activity_log_private/activity_log_private_api.cc:54:
extension_id == kActivityLogTestExtensionId)
nit: just return the result of this if clause

https://codereview.chromium.org/16061002/diff/66001/chrome/browser/extensions...
chrome/browser/extensions/api/activity_log_private/activity_log_private_api.cc:92:
DispatchEventToExtension(extension_id, event.Pass());
indent to 4 when wrapping

https://codereview.chromium.org/16061002/diff/66001/chrome/browser/extensions...
chrome/browser/extensions/api/activity_log_private/activity_log_private_api.cc:92:
DispatchEventToExtension(extension_id, event.Pass());
Just use BroadcastEvent instead of DispatchEventToExtension. That way you don't
need to know the extension IDs. The EventRouter and other code should ensure
unprivileged extensions don't listen to these events.

https://codereview.chromium.org/16061002/diff/66001/chrome/browser/extensions...
File
chrome/browser/extensions/api/activity_log_private/activity_log_private_api.h
(right):

https://codereview.chromium.org/16061002/diff/66001/chrome/browser/extensions...
chrome/browser/extensions/api/activity_log_private/activity_log_private_api.h:56:
const char extension_id[]);
pass a const std::string& here.

https://codereview.chromium.org/16061002/diff/66001/chrome/common/extensions/...
File chrome/common/extensions/api/activity_log_private.json (right):

https://codereview.chromium.org/16061002/diff/66001/chrome/common/extensions/...
chrome/common/extensions/api/activity_log_private.json:1: // Copyright 2013 The
Chromium Authors. All rights reserved.
This might be easier to read as an .idl file. Up to you.

https://codereview.chromium.org/16061002/diff/66001/chrome/common/extensions/...
chrome/common/extensions/api/activity_log_private.json:14: "domActivityType":
{"type": "string", "enum": ["GETTER", "SETTER", "METHOD", "INSERTED", "XHR",
"WEBREQUEST", "MODIFIED"], "optional": true},
prefer lower_case style for enums.

[felt, 2013-06-05 06:04:59]

Hi Matt, made changes as requested.

https://chromiumcodereview.appspot.com/16061002/diff/66001/chrome/browser/ext...
File chrome/browser/extensions/activity_log/activity_actions.h (right):

https://chromiumcodereview.appspot.com/16061002/diff/66001/chrome/browser/ext...
chrome/browser/extensions/activity_log/activity_actions.h:32: virtual
std::map<std::string, std::string> GetFieldsAsMap() = 0;
On 2013/06/05 01:24:47, Matt Perry wrote:
> Why not have this allocate and return an ExtensionActivity instead?

Done.

https://chromiumcodereview.appspot.com/16061002/diff/66001/chrome/browser/ext...
chrome/browser/extensions/activity_log/activity_actions.h:40: const { return
activity_type_; }
On 2013/06/05 01:24:47, Matt Perry wrote:
> nit: wrap before the function name

Done.

https://chromiumcodereview.appspot.com/16061002/diff/66001/chrome/browser/ext...
File
chrome/browser/extensions/api/activity_log_private/activity_log_private_api.cc
(right):

https://chromiumcodereview.appspot.com/16061002/diff/66001/chrome/browser/ext...
chrome/browser/extensions/api/activity_log_private/activity_log_private_api.cc:54:
extension_id == kActivityLogTestExtensionId)
On 2013/06/05 01:24:47, Matt Perry wrote:
> nit: just return the result of this if clause

Done.

https://chromiumcodereview.appspot.com/16061002/diff/66001/chrome/browser/ext...
chrome/browser/extensions/api/activity_log_private/activity_log_private_api.cc:92:
DispatchEventToExtension(extension_id, event.Pass());
whoops! done

On 2013/06/05 01:24:47, Matt Perry wrote:
> indent to 4 when wrapping

https://chromiumcodereview.appspot.com/16061002/diff/66001/chrome/browser/ext...
chrome/browser/extensions/api/activity_log_private/activity_log_private_api.cc:92:
DispatchEventToExtension(extension_id, event.Pass());
On 2013/06/05 01:24:47, Matt Perry wrote:
> Just use BroadcastEvent instead of DispatchEventToExtension. That way you
don't
> need to know the extension IDs. The EventRouter and other code should ensure
> unprivileged extensions don't listen to these events.

Done.

https://chromiumcodereview.appspot.com/16061002/diff/66001/chrome/browser/ext...
File
chrome/browser/extensions/api/activity_log_private/activity_log_private_api.h
(right):

https://chromiumcodereview.appspot.com/16061002/diff/66001/chrome/browser/ext...
chrome/browser/extensions/api/activity_log_private/activity_log_private_api.h:56:
const char extension_id[]);
On 2013/06/05 01:24:47, Matt Perry wrote:
> pass a const std::string& here.

Done.

https://chromiumcodereview.appspot.com/16061002/diff/66001/chrome/common/exte...
File chrome/common/extensions/api/activity_log_private.json (right):

https://chromiumcodereview.appspot.com/16061002/diff/66001/chrome/common/exte...
chrome/common/extensions/api/activity_log_private.json:1: // Copyright 2013 The
Chromium Authors. All rights reserved.
On 2013/06/05 01:24:47, Matt Perry wrote:
> This might be easier to read as an .idl file. Up to you.

Since I already have it working as a json file, I'll keep it as a json file
unless you strongly object.

https://chromiumcodereview.appspot.com/16061002/diff/66001/chrome/common/exte...
chrome/common/extensions/api/activity_log_private.json:14: "domActivityType":
{"type": "string", "enum": ["GETTER", "SETTER", "METHOD", "INSERTED", "XHR",
"WEBREQUEST", "MODIFIED"], "optional": true},
On 2013/06/05 01:24:47, Matt Perry wrote:
> prefer lower_case style for enums.

Done.

[Ilya Sherman, 2013-06-05 09:05:25]

histograms.xml LGTM

[felt, 2013-06-05 18:59:16]

I uploaded a new patchset that makes a change to the ActivityLogAPI constructor
and destructor. I noticed that during unit testing conditions, the event_router
might not be available (causing a crash if I then try to reference it). Also,
calling ActivityLog::GetInstance seems to cause a problem in unit tests.

[felt, 2013-06-06 00:30:42]

Hi Miranda,

Can you please review the 2-line change to
chrome/browser/profiles/chrome_browser_main_extra_parts_profiles.cc?

Thanks,
Adrienne

[Miranda Callahan, 2013-06-06 12:11:56]

On 2013/06/06 00:30:42, felt wrote:
> Hi Miranda,
> 
> Can you please review the 2-line change to
> chrome/browser/profiles/chrome_browser_main_extra_parts_profiles.cc?
> 
> Thanks,
> Adrienne

Hey, Adrienne -- sure. LGTM profiles/*.

[Matt Perry, 2013-06-06 21:06:47]

lgtm

https://chromiumcodereview.appspot.com/16061002/diff/103001/chrome/browser/ex...
File chrome/browser/extensions/activity_log/activity_actions.h (right):

https://chromiumcodereview.appspot.com/16061002/diff/103001/chrome/browser/ex...
chrome/browser/extensions/activity_log/activity_actions.h:32: virtual
api::activity_log_private::ExtensionActivity*
Consider returning a scoped_ptr to make ownership transfer explicit.

[felt, 2013-06-07 04:49:06]

Thanks! Made the changes, and removed the documentation since it is a private
API.

https://chromiumcodereview.appspot.com/16061002/diff/103001/chrome/browser/ex...
File chrome/browser/extensions/activity_log/activity_actions.h (right):

https://chromiumcodereview.appspot.com/16061002/diff/103001/chrome/browser/ex...
chrome/browser/extensions/activity_log/activity_actions.h:32: virtual
api::activity_log_private::ExtensionActivity*
On 2013/06/06 21:06:48, Matt Perry wrote:
> Consider returning a scoped_ptr to make ownership transfer explicit.

Done.

[commit-bot: I haz the power, 2013-06-07 06:54:37]

CQ is trying da patch. Follow status at
https://chromium-status.appspot.com/cq/[email protected]/16061002/154001

[commit-bot: I haz the power, 2013-06-07 07:03:42]

CQ is trying da patch. Follow status at
https://chromium-status.appspot.com/cq/[email protected]/16061002/163001

[commit-bot: I haz the power, 2013-06-07 07:21:28]

Step "update" is always a major failure.
Look at the try server FAQ for more details.
http://build.chromium.org/p/tryserver.chromium/buildstatus?builder=win&number...

[commit-bot: I haz the power, 2013-06-07 07:37:18]

CQ is trying da patch. Follow status at
https://chromium-status.appspot.com/cq/[email protected]/16061002/163001

[commit-bot: I haz the power, 2013-06-07 12:51:22]

Change committed as 204796