[TTS] Add some initial signals for Tap in content.

[TTS] Add some initial signals for Tap in content.

Adds a few simple signals for a Tap gesture relative to the content
that was tapped.  These signals include word length and tap offset.
Logs UMA for CTR when these signals are present.

Updates the CS Context with analysis of the content where the Tap
gesture occurred.

One signal determines if the user tapped near the edge of a word
or not. The rest look at very short words, or relatively long words.

All signals use the existing CSHeuristics framework to do the checking and logging.

Updates the CSHeuristics framework to make the CSContext available
to heuristics so they can inspect the text tapped before deciding
whether to suppress or not.

Adds Field Trial params to enable actual suppression based on
these new signals (likely to be used only for interactive-testing
and demonstration purposes).

BUG=723194
Review-Url: https://codereview.chromium.org/2906763002
Cr-Commit-Position: refs/heads/master@{#476485}
Committed: https://chromium.googlesource.com/chromium/src/+/4a22f717453dcd99cd92689dc7b91116387872bf

[Donn Denman, 2017-05-25 19:44:18]

The CQ bit was checked by donnd@chromium.org to run a CQ dry run

[commit-bot: I haz the power, 2017-05-25 19:45:03]

Dry run: CQ is trying da patch.

Follow status at:
https://chromium-cq-status.appspot.com/v2/patch-status/codereview.chromium.or...

[Donn Denman, 2017-05-25 19:45:36]

donnd@chromium.org changed reviewers:
+ holte@google.com, tedchoc@chromium.org

[Donn Denman, 2017-05-25 19:45:37]

Steven, PTAL at histograms.xml.

Ted, PTAL at everything.  I suggest starting with
TapWordNotLongEnoughSuppression, then the Short word file, then the word, edge
file, then the rest.

[Donn Denman, 2017-05-25 19:56:47]

https://codereview.chromium.org/2906763002/diff/1/chrome/android/java/src/org...
File
chrome/android/java/src/org/chromium/chrome/browser/contextualsearch/ContextualSearchSelectionController.java
(right):

https://codereview.chromium.org/2906763002/diff/1/chrome/android/java/src/org...
chrome/android/java/src/org/chromium/chrome/browser/contextualsearch/ContextualSearchSelectionController.java:208:
resetSelectionStates();
This looks like a bug that encountered when doing some other debugging... that
is orthogonal to this CL:
Seems like clearSelection() should only clear the "selection" states, not all
states.

[Donn Denman, 2017-05-26 17:41:32]

donnd@chromium.org changed reviewers:
+ rkaplow@chromium.org
- holte@google.com

[Donn Denman, 2017-05-26 17:41:33]

Robert, PTAL at histograms.xml.

tedchoc@  friendly ping! :-)
Ted PTAL at everything.

Thanks!

[Ted C, 2017-05-26 18:19:05]

a few general comments and some style nits

https://codereview.chromium.org/2906763002/diff/20001/chrome/android/java/src...
File
chrome/android/java/src/org/chromium/chrome/browser/contextualsearch/ContextualSearchContext.java
(right):

https://codereview.chromium.org/2906763002/diff/20001/chrome/android/java/src...
chrome/android/java/src/org/chromium/chrome/browser/contextualsearch/ContextualSearchContext.java:24:
static final int SOFT_HYPHEN = 173;
I would leave off the German Wikipedia pages part.  This seems like it could be
used in any context.  From wikipedia:
"a code point reserved in some coded character sets for the purpose of breaking
words across lines by inserting visible hyphens."

I think if we want to keep the name as is, we should use the char version
'\u00AD'

If we want to compare the codePoint, I would call this SOFT_HYPHEN_CODE_POINT as
I think using codePoint is more uncommon from what I've seen.

https://codereview.chromium.org/2906763002/diff/20001/chrome/android/java/src...
chrome/android/java/src/org/chromium/chrome/browser/contextualsearch/ContextualSearchContext.java:103:
mTapStartOffset = startOffset;
why store two values if we know they're always the same?

https://codereview.chromium.org/2906763002/diff/20001/chrome/android/java/src...
chrome/android/java/src/org/chromium/chrome/browser/contextualsearch/ContextualSearchContext.java:199:
private boolean haveValidTappedText() {
nit here and below

s/have/has

https://codereview.chromium.org/2906763002/diff/20001/chrome/android/java/src...
chrome/android/java/src/org/chromium/chrome/browser/contextualsearch/ContextualSearchContext.java:222:
assert haveValidTappedText();
instead of having this lazily fetch the value, can this be updated at
setSurroundingText (or whatever is valid) so the getter here and below don't
have any side effects?

https://codereview.chromium.org/2906763002/diff/20001/chrome/android/java/src...
chrome/android/java/src/org/chromium/chrome/browser/contextualsearch/ContextualSearchContext.java:246:
* @return The start of the word that contains the given initial offset, within
{@code text}.
is this the bit of code meant to mimic the behavior in blink?  If so, I would
consider mentioning it and potentially adding a point to the code in blink that
does it (or the ipc name or something)

https://codereview.chromium.org/2906763002/diff/20001/chrome/android/java/src...
chrome/android/java/src/org/chromium/chrome/browser/contextualsearch/ContextualSearchContext.java:266:
for (int offset = initial; offset <= text.length(); offset++) {
not initial+1?

I guess if you tap on a space, findStart above would return initial and so would
this, so maybe that is right

https://codereview.chromium.org/2906763002/diff/20001/chrome/android/java/src...
File
chrome/android/java/src/org/chromium/chrome/browser/contextualsearch/ContextualSearchFieldTrial.java
(right):

https://codereview.chromium.org/2906763002/diff/20001/chrome/android/java/src...
chrome/android/java/src/org/chromium/chrome/browser/contextualsearch/ContextualSearchFieldTrial.java:266:
static boolean isWordEdgeSuppressionEnabled() {
tangent, but you might one day consider combining all of these boolean checks
into something like:

Map<String, Boolean> sEnableState = new HashMap<>();

static boolean isEnabled(String feature) {
    Boolean enabled = sEnableState.get(feature);
    if (enabled == null) {
        enabled = getBooleanParam(feature);
        sEnableState.put(feature, enabled);
    }
    return enabled;
}

But this has gotten you thus far, so no need to rock the boat (and definitely
not in this change)

https://codereview.chromium.org/2906763002/diff/20001/chrome/android/java/src...
File
chrome/android/java/src/org/chromium/chrome/browser/contextualsearch/ContextualSearchUma.java
(right):

https://codereview.chromium.org/2906763002/diff/20001/chrome/android/java/src...
chrome/android/java/src/org/chromium/chrome/browser/contextualsearch/ContextualSearchUma.java:812:
if (!isNearWordEdge) {
for these, I would do

if (!isNearWordEdge) return;

Then the rest can be less indented.

https://codereview.chromium.org/2906763002/diff/20001/chrome/android/java/src...
File
chrome/android/java/src/org/chromium/chrome/browser/contextualsearch/TapShortWordSuppression.java
(right):

https://codereview.chromium.org/2906763002/diff/20001/chrome/android/java/src...
chrome/android/java/src/org/chromium/chrome/browser/contextualsearch/TapShortWordSuppression.java:1:
// Copyright 2016 The Chromium Authors. All rights reserved.
2017

https://codereview.chromium.org/2906763002/diff/20001/chrome/android/java/src...
File
chrome/android/java/src/org/chromium/chrome/browser/contextualsearch/TapSuppressionHeuristics.java
(right):

https://codereview.chromium.org/2906763002/diff/20001/chrome/android/java/src...
chrome/android/java/src/org/chromium/chrome/browser/contextualsearch/TapSuppressionHeuristics.java:35:
mHeuristics.add(tapWordEdgeSuppression);
if these fit, I think this could be shorter as

mHeuristics.add(new TapWordEdgeSuppression(contextualSearchContext));

no need for the local vars IMO

https://codereview.chromium.org/2906763002/diff/20001/chrome/android/java/src...
File
chrome/android/java/src/org/chromium/chrome/browser/contextualsearch/TapWordEdgeSuppression.java
(right):

https://codereview.chromium.org/2906763002/diff/20001/chrome/android/java/src...
chrome/android/java/src/org/chromium/chrome/browser/contextualsearch/TapWordEdgeSuppression.java:1:
// Copyright 2016 The Chromium Authors. All rights reserved.
2017 (all the places)

https://codereview.chromium.org/2906763002/diff/20001/chrome/android/java/src...
chrome/android/java/src/org/chromium/chrome/browser/contextualsearch/TapWordEdgeSuppression.java:16:
private static final double MIN_WORD_START_RATIO = 0.25;
I wonder if a separate experiment could be to look at number of characters
inset.  the percentage will have bias towards word length where looking at
number of characters in could provide a bit different perspective.  just a
thought and nothing at all wrong with this approach

https://codereview.chromium.org/2906763002/diff/20001/chrome/android/java/src...
File
chrome/android/java/src/org/chromium/chrome/browser/contextualsearch/TapWordNotLongEnoughSuppression.java
(right):

https://codereview.chromium.org/2906763002/diff/20001/chrome/android/java/src...
chrome/android/java/src/org/chromium/chrome/browser/contextualsearch/TapWordNotLongEnoughSuppression.java:13:
class TapWordNotLongEnoughSuppression extends ContextualSearchHeuristic {
I think it's a bit confusing to have ShortWord and NotLongEnough Suppression
branches.  They like they could/should be the same thing.  I wonder if we should
keep the same naming would it address this.

Something like
<3 = VeryShortWord
<9 = ShortWord

Or
<3 = ShortWordLength
<9 = MediumWordLength

[rkaplow, 2017-05-29 19:18:53]

lgtm

histogram lg

[Donn Denman, 2017-05-30 16:42:59]

Ted, PTAL.  Thanks Robert!

https://codereview.chromium.org/2906763002/diff/20001/chrome/android/java/src...
File
chrome/android/java/src/org/chromium/chrome/browser/contextualsearch/ContextualSearchContext.java
(right):

https://codereview.chromium.org/2906763002/diff/20001/chrome/android/java/src...
chrome/android/java/src/org/chromium/chrome/browser/contextualsearch/ContextualSearchContext.java:24:
static final int SOFT_HYPHEN = 173;
On 2017/05/26 18:19:04, Ted C wrote:
> I would leave off the German Wikipedia pages part.  This seems like it could
be
> used in any context.  From wikipedia:
> "a code point reserved in some coded character sets for the purpose of
breaking
> words across lines by inserting visible hyphens."
> 
> I think if we want to keep the name as is, we should use the char version
> '\u00AD'
> 
> If we want to compare the codePoint, I would call this SOFT_HYPHEN_CODE_POINT
as
> I think using codePoint is more uncommon from what I've seen.

Done.

https://codereview.chromium.org/2906763002/diff/20001/chrome/android/java/src...
chrome/android/java/src/org/chromium/chrome/browser/contextualsearch/ContextualSearchContext.java:103:
mTapStartOffset = startOffset;
On 2017/05/26 18:19:04, Ted C wrote:
> why store two values if we know they're always the same?

Oops, evolved that way.  Done!

https://codereview.chromium.org/2906763002/diff/20001/chrome/android/java/src...
chrome/android/java/src/org/chromium/chrome/browser/contextualsearch/ContextualSearchContext.java:199:
private boolean haveValidTappedText() {
On 2017/05/26 18:19:04, Ted C wrote:
> nit here and below
> 
> s/have/has

Done.

https://codereview.chromium.org/2906763002/diff/20001/chrome/android/java/src...
chrome/android/java/src/org/chromium/chrome/browser/contextualsearch/ContextualSearchContext.java:222:
assert haveValidTappedText();
On 2017/05/26 18:19:04, Ted C wrote:
> instead of having this lazily fetch the value, can this be updated at
> setSurroundingText (or whatever is valid) so the getter here and below don't
> have any side effects?
Done: those side-effects were bothering me, thanks for the suggestion.

https://codereview.chromium.org/2906763002/diff/20001/chrome/android/java/src...
chrome/android/java/src/org/chromium/chrome/browser/contextualsearch/ContextualSearchContext.java:246:
* @return The start of the word that contains the given initial offset, within
{@code text}.
On 2017/05/26 18:19:04, Ted C wrote:
> is this the bit of code meant to mimic the behavior in blink?  If so, I would
> consider mentioning it and potentially adding a point to the code in blink
that
> does it (or the ipc name or something)

Done: Added a brief explanation and caveat.

https://codereview.chromium.org/2906763002/diff/20001/chrome/android/java/src...
chrome/android/java/src/org/chromium/chrome/browser/contextualsearch/ContextualSearchContext.java:266:
for (int offset = initial; offset <= text.length(); offset++) {
On 2017/05/26 18:19:04, Ted C wrote:
> not initial+1?
> 
> I guess if you tap on a space, findStart above would return initial and so
would
> this, so maybe that is right

I added more detailed comments about how these scanners work.  When you tap on a
space the word before it is selected and the space itself is at the offset of
the end of the word. This is why the scan for the word start starts one to the
left and returns the char just past the separator.

https://codereview.chromium.org/2906763002/diff/20001/chrome/android/java/src...
File
chrome/android/java/src/org/chromium/chrome/browser/contextualsearch/ContextualSearchFieldTrial.java
(right):

https://codereview.chromium.org/2906763002/diff/20001/chrome/android/java/src...
chrome/android/java/src/org/chromium/chrome/browser/contextualsearch/ContextualSearchFieldTrial.java:266:
static boolean isWordEdgeSuppressionEnabled() {
On 2017/05/26 18:19:05, Ted C wrote:
> tangent, but you might one day consider combining all of these boolean checks
> into something like:
> 
> Map<String, Boolean> sEnableState = new HashMap<>();
> 
> static boolean isEnabled(String feature) {
>     Boolean enabled = sEnableState.get(feature);
>     if (enabled == null) {
>         enabled = getBooleanParam(feature);
>         sEnableState.put(feature, enabled);
>     }
>     return enabled;
> }
> 
> But this has gotten you thus far, so no need to rock the boat (and definitely
> not in this change)

Good idea, I've added a TODO here to consider this.

https://codereview.chromium.org/2906763002/diff/20001/chrome/android/java/src...
File
chrome/android/java/src/org/chromium/chrome/browser/contextualsearch/ContextualSearchUma.java
(right):

https://codereview.chromium.org/2906763002/diff/20001/chrome/android/java/src...
chrome/android/java/src/org/chromium/chrome/browser/contextualsearch/ContextualSearchUma.java:812:
if (!isNearWordEdge) {
On 2017/05/26 18:19:05, Ted C wrote:
> for these, I would do
> 
> if (!isNearWordEdge) return;
> 
> Then the rest can be less indented.

Done.

https://codereview.chromium.org/2906763002/diff/20001/chrome/android/java/src...
File
chrome/android/java/src/org/chromium/chrome/browser/contextualsearch/TapShortWordSuppression.java
(right):

https://codereview.chromium.org/2906763002/diff/20001/chrome/android/java/src...
chrome/android/java/src/org/chromium/chrome/browser/contextualsearch/TapShortWordSuppression.java:1:
// Copyright 2016 The Chromium Authors. All rights reserved.
On 2017/05/26 18:19:05, Ted C wrote:
> 2017

Done.

https://codereview.chromium.org/2906763002/diff/20001/chrome/android/java/src...
File
chrome/android/java/src/org/chromium/chrome/browser/contextualsearch/TapSuppressionHeuristics.java
(right):

https://codereview.chromium.org/2906763002/diff/20001/chrome/android/java/src...
chrome/android/java/src/org/chromium/chrome/browser/contextualsearch/TapSuppressionHeuristics.java:35:
mHeuristics.add(tapWordEdgeSuppression);
On 2017/05/26 18:19:05, Ted C wrote:
> if these fit, I think this could be shorter as
> 
> mHeuristics.add(new TapWordEdgeSuppression(contextualSearchContext));
> 
> no need for the local vars IMO

Done.

https://codereview.chromium.org/2906763002/diff/20001/chrome/android/java/src...
File
chrome/android/java/src/org/chromium/chrome/browser/contextualsearch/TapWordEdgeSuppression.java
(right):

https://codereview.chromium.org/2906763002/diff/20001/chrome/android/java/src...
chrome/android/java/src/org/chromium/chrome/browser/contextualsearch/TapWordEdgeSuppression.java:1:
// Copyright 2016 The Chromium Authors. All rights reserved.
On 2017/05/26 18:19:05, Ted C wrote:
> 2017 (all the places)

Done.

https://codereview.chromium.org/2906763002/diff/20001/chrome/android/java/src...
chrome/android/java/src/org/chromium/chrome/browser/contextualsearch/TapWordEdgeSuppression.java:16:
private static final double MIN_WORD_START_RATIO = 0.25;
On 2017/05/26 18:19:05, Ted C wrote:
> I wonder if a separate experiment could be to look at number of characters
> inset.  the percentage will have bias towards word length where looking at
> number of characters in could provide a bit different perspective.  just a
> thought and nothing at all wrong with this approach

Yep, seems like a reasonable alternative.  Thanks!

[Donn Denman, 2017-06-01 00:07:16]

donnd@chromium.org changed reviewers:
- rkaplow@chromium.org

[Donn Denman, 2017-06-01 00:07:17]

Ted, PTAL or I could send this to Theresa if you're busy.

[Ted C, 2017-06-01 00:09:56]

Sorry...last few comments that I wrote but didn't send earlier in the week.

https://codereview.chromium.org/2906763002/diff/100001/chrome/android/java/sr...
File
chrome/android/java/src/org/chromium/chrome/browser/contextualsearch/ContextualSearchContext.java
(right):

https://codereview.chromium.org/2906763002/diff/100001/chrome/android/java/sr...
chrome/android/java/src/org/chromium/chrome/browser/contextualsearch/ContextualSearchContext.java:24:
static final int SOFT_HYPHEN_CODE_POINT = '\u00AD';
:-P

This is why code review isn't always the best.  I think if we leave it as a
number it should be _CODE_POINT.  But now that you have converted to the
character definition, I would name this SOFT_HYPHEN_CHAR

https://codereview.chromium.org/2906763002/diff/100001/chrome/android/java/sr...
chrome/android/java/src/org/chromium/chrome/browser/contextualsearch/ContextualSearchContext.java:247:
mTappedWord = mSurroundingText.substring(wordStartOffset, wordEndOffset);
should we be setting mTappedWord = null and mTappedWordStartOffset =
INVALID_OFFSET at the top of this function?

Or should that go into setSurroundingText?

Maybe this function should be called unconditionally and updates all the
necessary state?  In which case, maybe this should be refreshTappedWord?  or
refreshTappedWordState?

https://codereview.chromium.org/2906763002/diff/100001/chrome/android/java/sr...
chrome/android/java/src/org/chromium/chrome/browser/contextualsearch/ContextualSearchContext.java:248:
mTappedWordStartOffset = wordStartOffset;
mTappedWordStartOffset is only used above and always subtracted from mTapOffset.
 I would just store that computed value here and remove the subtraction above
(unless mTapOffset can change out of band of mTappedWordStartOffset).

But it looks like it all gets set at the same time.  I do think we should
potentially have this function set mTapOffset as well and do all the state
setting in a single place.

[Donn Denman, 2017-06-01 01:21:59]

Ted, PTAL.  Thanks for the great review!

https://codereview.chromium.org/2906763002/diff/100001/chrome/android/java/sr...
File
chrome/android/java/src/org/chromium/chrome/browser/contextualsearch/ContextualSearchContext.java
(right):

https://codereview.chromium.org/2906763002/diff/100001/chrome/android/java/sr...
chrome/android/java/src/org/chromium/chrome/browser/contextualsearch/ContextualSearchContext.java:24:
static final int SOFT_HYPHEN_CODE_POINT = '\u00AD';
On 2017/06/01 00:09:55, Ted C wrote:
> :-P
> 
> This is why code review isn't always the best.  I think if we leave it as a
> number it should be _CODE_POINT.  But now that you have converted to the
> character definition, I would name this SOFT_HYPHEN_CHAR

Done.  Also made it private.

https://codereview.chromium.org/2906763002/diff/100001/chrome/android/java/sr...
chrome/android/java/src/org/chromium/chrome/browser/contextualsearch/ContextualSearchContext.java:248:
mTappedWordStartOffset = wordStartOffset;
On 2017/06/01 00:09:55, Ted C wrote:
> mTappedWordStartOffset is only used above and always subtracted from
mTapOffset.
>  I would just store that computed value here and remove the subtraction above
> (unless mTapOffset can change out of band of mTappedWordStartOffset).
> 
> But it looks like it all gets set at the same time.  I do think we should
> potentially have this function set mTapOffset as well and do all the state
> setting in a single place.

Done.

I also renamed this method to analyzeTap because in my next CL I'll look for
words before and after the tapped word.

[Ted C, 2017-06-01 14:03:15]

lgtm

Could the suppression metrics have tests?  We should look at writing some tests
for the analyzeTap logic as well.

[Donn Denman, 2017-06-01 16:43:49]

We've been light on tests because this is probably going to be throw-away code
that validates usefulness of Content features.  However it could end up getting
used for non-ML suppression, and some basic tests are a good idea.  I'll add
them in the next CL!

Thanks for the review Ted!

[Donn Denman, 2017-06-01 16:43:55]

The CQ bit was checked by donnd@chromium.org

[Donn Denman, 2017-06-01 16:43:55]

The patchset sent to the CQ was uploaded after l-g-t-m from rkaplow@chromium.org
Link to the patchset: https://codereview.chromium.org/2906763002/#ps160001
(title: "Move a misplaced assert.")

[commit-bot: I haz the power, 2017-06-01 16:44:22]

CQ is trying da patch.

Follow status at:
https://chromium-cq-status.appspot.com/v2/patch-status/codereview.chromium.or...

[commit-bot: I haz the power, 2017-06-01 19:18:53]

The CQ bit was unchecked by commit-bot@chromium.org

[commit-bot: I haz the power, 2017-06-01 19:18:55]

Try jobs failed on following builders:
  linux_android_rel_ng on master.tryserver.chromium.android (JOB_FAILED,
https://build.chromium.org/p/tryserver.chromium.android/builders/linux_androi...)

[Donn Denman, 2017-06-01 21:55:47]

The CQ bit was checked by donnd@chromium.org

[Donn Denman, 2017-06-01 21:55:48]

The patchset sent to the CQ was uploaded after l-g-t-m from
tedchoc@chromium.org, rkaplow@chromium.org
Link to the patchset: https://codereview.chromium.org/2906763002/#ps180001
(title: "Just fix an off-by-one bug on an index bounds found by a failing
test.")

[commit-bot: I haz the power, 2017-06-01 21:56:29]

CQ is trying da patch.

Follow status at:
https://chromium-cq-status.appspot.com/v2/patch-status/codereview.chromium.or...

[commit-bot: I haz the power, 2017-06-01 23:47:35]

CQ is committing da patch.

Bot data: {"patchset_id": 180001, "attempt_start_ts": 1496354147493330,
"parent_rev": "9aecc2e11e14c10c5d3be0e526e46d22927e1ed0", "commit_rev":
"4a22f717453dcd99cd92689dc7b91116387872bf"}

[commit-bot: I haz the power, 2017-06-01 23:47:47]

Description was changed from

==========
[TTS] Add some initial signals for Tap in content.

Adds a few simple signals for a Tap gesture relative to the content
that was tapped.  These signals include word length and tap offset.
Logs UMA for CTR when these signals are present.

Updates the CS Context with analysis of the content where the Tap
gesture occurred.

One signal determines if the user tapped near the edge of a word
or not. The rest look at very short words, or relatively long words.

All signals use the existing CSHeuristics framework to do the checking and
logging.

Updates the CSHeuristics framework to make the CSContext available
to heuristics so they can inspect the text tapped before deciding
whether to suppress or not.

Adds Field Trial params to enable actual suppression based on
these new signals (likely to be used only for interactive-testing
and demonstration purposes).

BUG=723194
==========

to

==========
[TTS] Add some initial signals for Tap in content.

Adds a few simple signals for a Tap gesture relative to the content
that was tapped.  These signals include word length and tap offset.
Logs UMA for CTR when these signals are present.

Updates the CS Context with analysis of the content where the Tap
gesture occurred.

One signal determines if the user tapped near the edge of a word
or not. The rest look at very short words, or relatively long words.

All signals use the existing CSHeuristics framework to do the checking and
logging.

Updates the CSHeuristics framework to make the CSContext available
to heuristics so they can inspect the text tapped before deciding
whether to suppress or not.

Adds Field Trial params to enable actual suppression based on
these new signals (likely to be used only for interactive-testing
and demonstration purposes).

BUG=723194

Review-Url: https://codereview.chromium.org/2906763002
Cr-Commit-Position: refs/heads/master@{#476485}
Committed:
https://chromium.googlesource.com/chromium/src/+/4a22f717453dcd99cd92689dc7b9...
==========

[commit-bot: I haz the power, 2017-06-01 23:47:48]

Committed patchset #10 (id:180001) as
https://chromium.googlesource.com/chromium/src/+/4a22f717453dcd99cd92689dc7b9...