Clear wallet data from autofill when disabled.

Clear wallet data from autofill when disabled.

Re-mask all unmasked cards when the user disables autofill integration.

Delete all server cards and addresses when the user disables the sync type or stops syncing.

Fix a bug where addresses were always written to the DB on startup because profiles were being created differently from protos than they were from sqlite so wouldn't match.

Optimize autofill profile comparisons. I was offended by the number of autofill profile and card comparisons we do at startup, the small number of items in the list means std::set is operating at a worst-case. I switched to brute-force with a "give up and rewrite the data" escape valve if you have a large number of addresses.

BUG=450843
Committed: https://crrev.com/293968ff9a7b053e518eebc4e3c6233fefaf842a
Cr-Commit-Position: refs/heads/master@{#321406}

[brettw, 2015-03-16 23:58:30]

brettw@chromium.org changed reviewers:
+ estade@chromium.org

[Evan Stade, 2015-03-17 01:26:54]

nit: cl description is missing "BUG="

lgtm

https://codereview.chromium.org/1007003007/diff/40001/components/autofill/cor...
File components/autofill/core/browser/personal_data_manager.cc (right):

https://codereview.chromium.org/1007003007/diff/40001/components/autofill/cor...
components/autofill/core/browser/personal_data_manager.cc:1342:
!server_credit_cards_.empty()) {
the server_credit_cards_.empty() check seems unnecessary

https://codereview.chromium.org/1007003007/diff/40001/components/autofill/cor...
File
components/autofill/core/browser/webdata/autofill_wallet_syncable_service.cc
(right):

https://codereview.chromium.org/1007003007/diff/40001/components/autofill/cor...
components/autofill/core/browser/webdata/autofill_wallet_syncable_service.cc:154:
// set sizes, brute force O(n^2) is faster.
I suspect this is true of many places in Autofill :/

set is probably more often used for the uniqueness property than for any
performance reason. Is there a std:: or base::dumbset we can use?

https://codereview.chromium.org/1007003007/diff/40001/components/autofill/cor...
components/autofill/core/browser/webdata/autofill_wallet_syncable_service.cc:204:
// This data type is special. Normal sync data says on the client when
nit:says->stays

[brettw, 2015-03-19 16:39:39]

https://codereview.chromium.org/1007003007/diff/40001/components/autofill/cor...
File
components/autofill/core/browser/webdata/autofill_wallet_syncable_service.cc
(right):

https://codereview.chromium.org/1007003007/diff/40001/components/autofill/cor...
components/autofill/core/browser/webdata/autofill_wallet_syncable_service.cc:154:
// set sizes, brute force O(n^2) is faster.
No, there's base::SmallMap but no set version. This does brute-force up to some
limit which is kind of what we want. I got curious so checked some options more
than was necessary:

First, I tried using this using true as the "value" in the map. It generates
twice as many compares as my brute force version since each insert also checks
equality. It also generates quite a bit more code since there's a bunch of
templates to due both brute-force and tree-based searching for this small
container, and requires this shocking template declaration:
base::SmallMap<std::map<const Data*, bool, AutofillDataPtrLessThan<Data>>, 8,
AutofillDataPtrEqualTo<Data>> existing_data_set;
and an extra 

Then I tried using find_if with a lambda. This was fun to write but I find hard
to read:
 if (std::find_if(existing_data.begin(), existing_data.end(),
                       [new_data](const Data* d) { return d->Compare(new_data)
== 0; })
          == existing_data.end()) {
and still generates more machine code than my simple brute force version.

So I prefer to keep my simple brute-force version.

[brettw, 2015-03-19 18:20:54]

The CQ bit was checked by brettw@chromium.org

[brettw, 2015-03-19 18:20:55]

The patchset sent to the CQ was uploaded after l-g-t-m from estade@chromium.org
Link to the patchset: https://codereview.chromium.org/1007003007/#ps80001
(title: "merge")

[commit-bot: I haz the power, 2015-03-19 18:21:22]

CQ is trying da patch. Follow status at
 https://chromium-cq-status.appspot.com/patch-status/1007003007/80001

[commit-bot: I haz the power, 2015-03-19 19:23:21]

Committed patchset #5 (id:80001)

[commit-bot: I haz the power, 2015-03-19 19:24:28]

Patchset 5 (id:??) landed as
https://crrev.com/293968ff9a7b053e518eebc4e3c6233fefaf842a
Cr-Commit-Position: refs/heads/master@{#321406}