Add ability to parse multiple recognition results and send them to WebKit.

Add ability to parse multiple recognition results and send them to WebKit.
The relevant Webkit change was submitted earlier: http://bugs.webkit.org/show_bug.cgi?id=48068
The unit tests have been updated for the multi-results case as well.

BUG=none
TEST=unit_tests --gtest_filter=Speech*

Committed: http://src.chromium.org/viewvc/chrome?view=rev&revision=63891

[bulach, 2010-10-26 10:56:01]

LGTM

I have one suggestion that may clarify "results": from my point of view, the
recognition result is singular, and it contains N recognition elements.. let me
know if it makes sense!


few nits below:

http://codereview.chromium.org/4119004/diff/1/5
File chrome/browser/speech/speech_input_manager.cc (right):

http://codereview.chromium.org/4119004/diff/1/5#newcode38
chrome/browser/speech/speech_input_manager.cc:38: const SpeechInputResultArray&
results);
ditto

http://codereview.chromium.org/4119004/diff/1/5#newcode154
chrome/browser/speech/speech_input_manager.cc:154: int caller_id, bool error,
const SpeechInputResultArray& results) {
ditto

http://codereview.chromium.org/4119004/diff/1/5#newcode156
chrome/browser/speech/speech_input_manager.cc:156:
GetDelegate(caller_id)->SetRecognitionResult(caller_id, results);
ditto

http://codereview.chromium.org/4119004/diff/1/6
File chrome/browser/speech/speech_input_manager.h (right):

http://codereview.chromium.org/4119004/diff/1/6#newcode27
chrome/browser/speech/speech_input_manager.h:27: const SpeechInputResultArray&
results) = 0;
nit: if we consider the "recognition result" as a singular, then
s/results/result/
otherwhise, it's probably better to use plural everywhere (method name, IPC
message, etc..)

http://codereview.chromium.org/4119004/diff/1/7
File chrome/browser/speech/speech_input_result.h (right):

http://codereview.chromium.org/4119004/diff/1/7#newcode8
chrome/browser/speech/speech_input_result.h:8: #include <vector>
\n

http://codereview.chromium.org/4119004/diff/1/7#newcode14
chrome/browser/speech/speech_input_result.h:14: struct SpeechInputResult {
suggestion: there's a bit of a mix between result / results.

perhaps: rename this to SpeechInputElement, and the vector to be
SpeechInputResult, that is, a single result contains multiple elements?
it may disambiguate between result/results, what do you think?

http://codereview.chromium.org/4119004/diff/1/8
File chrome/browser/speech/speech_recognition_request.cc (right):

http://codereview.chromium.org/4119004/diff/1/8#newcode22
chrome/browser/speech/speech_recognition_request.cc:22:
speech_input::SpeechInputResultArray& results) {
ditto, and also it needs to be a * rather than non-const&

http://codereview.chromium.org/4119004/diff/1/8#newcode87
chrome/browser/speech/speech_recognition_request.cc:87: double confidence = 0;
nit: 0.0

http://codereview.chromium.org/4119004/diff/1/8#newcode154
chrome/browser/speech/speech_recognition_request.cc:154:
delegate_->SetRecognitionResult(error, results);
as above, perhaps:
SpeechInputResult result;
...Parse(..., &result);
...
SetRecognitionResult(error, result);

http://codereview.chromium.org/4119004/diff/1/9
File chrome/browser/speech/speech_recognition_request.h (right):

http://codereview.chromium.org/4119004/diff/1/9#newcode32
chrome/browser/speech/speech_recognition_request.h:32: bool error, const
SpeechInputResultArray& results) = 0;
ditto

[Satish, 2010-10-26 11:25:22]

http://codereview.chromium.org/4119004/diff/1/7
File chrome/browser/speech/speech_input_result.h (right):

http://codereview.chromium.org/4119004/diff/1/7#newcode14
chrome/browser/speech/speech_input_result.h:14: struct SpeechInputResult {
On 2010/10/26 10:56:01, bulach wrote:
> suggestion: there's a bit of a mix between result / results.
> 
> perhaps: rename this to SpeechInputElement, and the vector to be
> SpeechInputResult, that is, a single result contains multiple elements?
> it may disambiguate between result/results, what do you think?

Sure, would "SpeechInputResultItem" be ok for the struct and
"SpeechInputResultArray" can be maintained as is? With that, there would be no
SpeechInputResult which seems to be the main cause of confusion..

[bulach, 2010-10-26 12:02:54]

http://codereview.chromium.org/4119004/diff/1/7
File chrome/browser/speech/speech_input_result.h (right):

http://codereview.chromium.org/4119004/diff/1/7#newcode14
chrome/browser/speech/speech_input_result.h:14: struct SpeechInputResult {
On 2010/10/26 11:25:22, Satish wrote:
> On 2010/10/26 10:56:01, bulach wrote:
> > suggestion: there's a bit of a mix between result / results.
> > 
> > perhaps: rename this to SpeechInputElement, and the vector to be
> > SpeechInputResult, that is, a single result contains multiple elements?
> > it may disambiguate between result/results, what do you think?
> 
> Sure, would "SpeechInputResultItem" be ok for the struct and
> "SpeechInputResultArray" can be maintained as is? With that, there would be no
> SpeechInputResult which seems to be the main cause of confusion..

I think my problem was with various SetResult(..results)..
I guess it'd be clearer to treat "SetResult" as having a "result" which in turn
contains various "items".

the mismatch seem that SpeechInputResultArray type is sometimes associated with
a "results" param / variable in a "SetResult" method.. in this case,
"SpeechInputResult result" in a "SetResult" method that would then traverse as
many items as it wants may be clearer..

but as I said, this is just a suggestion, I don't feel strongly about it either
way. :)

[Satish, 2010-10-26 14:38:37]

I have made the naming changes that we discussed in person and uploaded a new
patch, passing in try bots fine now. Will land shortly. Thanks for the quick
review.