cherripick of I9c6a948331726a821bd3ccec9c1d02dec2c4703a
(forward cherrypicking this because the automerger is stuck now.)
This bug was leading to corrupted rendering of surrogate pairs in the following
scenario.
1. Type some emojis
2. Move the cursor at the beginning of the text field
3. Hit backspace even though there's nothing to delete
4. Move the cursor after some emoji
5. Hit backspace
The root cause of this issue was the out-of-sync mExpectingUpdateSelection if
handleBackspace() gets called when the cursor reaches at the beginning of the
TextView. In such case, mExpectingUpdateSelection shouldn't be set true because
there's nothing to delete, so there will be no onUpdateSelection() calls associated
with it. Due to this bug, the cache in RichInputConnection could get stale at step 4
described above. Then the following handleBackspace() that should delete a surrogate
pair was not working correctly because of the stale cache.
bug: 11181913
Change-Id: I1cbf444d8d105416e7de75c16d80b3797f470495
This is not useful because we're going to call setSelection again
with different values on the connection right away.
Also a preliminary change for
Bug: 10792236
Change-Id: I46c6ef1fbb3624086099bf81afddb0ef5ae85661
Since loadKeyboard relies on the input connection being
available to give it the auto-caps state, but also can't
be called twice in a row because it needs to save and
restore its state and invalidates it after the restore,
we need to wait until we know we have a valid input
connection to call it.
Bug: 11107229
Change-Id: I1c7baf3215682df6f6ceb357bd37254f9e7418c7
This also includes a fix that allows this code to read surrogate
pairs in this processing.
Bug: 11070482
Change-Id: If5ef8d6863938252f09128b7e99ea07ece6e7019
What's happening here is, setAlphabetKeyboard sets the
keyboard to AUTOMATIC_SHIFTED and updates the keyboard, then
restoring the keyboard old state sets it back to UNSHIFTED without
updating it. When we finally know what the correct value is,
we try to set it to UNSHIFTED, but since that's already the currently
recorded state, it skips updating the keyboard forever.
The solution is to avoid setting the shift state without updating the
keyboard.
Bug: 10948582
Change-Id: Ic8670401e378f8284e851281f91a9ad93eac8e90
This is not enough to really fix behavior with TYPE_NULL,
but it does make things a bit better.
Bug: 10949594
Change-Id: Ia359f781cdd76a2e2c5a4c9f166025d81b931174
This code is now useless.
I thought it affected 9902905, but it seems it actually does not.
Still, it's dead code.
Bug: 9902905
Change-Id: Id9c90fbf88f015af47b2368b90d173c481647093
This fixes null pointer exceptions that are caused when
getSuggestion() is called from spell checker.
Change-Id: Ifc9e7abb900e1a4646e33c91022effde439673e4
...and test it.
Also at the same time, add a facility to create a SettingsValues for
test, and some minor performance improvement to surrounding
methods.
Change-Id: I13b629ae14755c244af2a9406a7e9b4a4a16090f
Stop deleting them one after the other, it makes no sense. We
can know in advance if we should delete two of them, so let's
use that.
Bug: 6526420
Change-Id: Ifa2263cbde8fedc78bb750b84d4acd3f979774ea
We have all we know to re-evaluate auto-caps in this case.
We don't need to wait until we get the callback from TextView.
Bug: 10287972
Change-Id: I87e4e26cd193a001f91fe852ffc5a8f728db2449
The emoji switching key icon asset should be updated. The icons in
this change are just borrowed from Hangout.
Bug: 10179862
Bug: 6370846
Change-Id: Icea26d2cf2ac6a6df9beb4c98f646dbe309bdcd9
Instead of CODE_RESEARCH, special outputText ".research." is used to
invoke a research feedback dialog.
Change-Id: Ic95c956ab4f2b5f0c3f2205c9bbcdcfb67db5d6e
This frees up 1.7MB on Nexus 4 (occam), and 900KB on occam_svelte while
LatinIME goes to background.
The keyboard bring-up speed feels acceptable on both devices.
bug: 8967766
Change-Id: I5bc980c4dc41bc304be3c190934c1f7c55c9d244
This change also eliminates a reference of
AudioAndHapticFeedbackManager from KeyboardSwitcher and MainKeyboard.
Bug: 6522943
Change-Id: Iac42ec8ff00c66deb76a660ffc07477923a58959
This is much better interface-wise. It eliminates all blinking
of the line in the practice.
Bug: 8874148
Bug: 8864306
Change-Id: I87754e44784327c2e9c8b162d598d145e20668e8
I548d899b introduced a new method to fix a sync miss between
the cursor position and the cached cursor position, but did not
take into account that it should also update the cached text
before and after the cursor in this case and that there was
already a method for doing this.
Change-Id: I31bd741893207c822827304e77791b1159774e1a
in the hope of avoiding UnsatisfiedLinkError. This is definitely not a
confident fix as we still don't know the root cause of the issue.
bug: 9325110
Change-Id: I1abf8031b98509f63fce6a2d069d497895ea712c
SpannableStrings are not the only type of Spanned, they are
only the mutable kind. SpannedString is the immutable one.
Spanned is the correct class to use there.
Bug: 8839763
Bug: 8862327
Change-Id: Ic5e6199a51f22368914f2748ac2d0d1ca6a33f78
This changes how the Range class stores its data, but not its
functionality. It also improves encapsulation a bit.
Bug: 8839763
Bug: 8862327
Change-Id: I5bd583b3fc96a99b93a2632882d8fd587c03ab76
The documentation for setComposingRegion states explicitly
that it does not move the cursor. This is just a bug.
This does not have any ill effects right now, but it will have
in later changes if not fixed.
As for the selection handling, the specific test that this code
removes used to serve a purpose, but it does not any more because
the code using the value has been much sanitized. Now the variable
can just take the obvious value, and become so self-explanatory
that the comments are unnecessary.
Change-Id: I548d899b38776bd3ab5f5361aab0d89d98f12e73
This will spare a lot of IPC for Latin IME at the cost of very little
retained memory.
This improves the loading by potentially a lot - between 15 and 30%
when the layout is cached (which should now be the case almost every
time), and half that if it's not. More importantly, it makes the
load time less sensitive to high device load, which is one of the
sore points.
Bug: 8689779
Change-Id: I2e07736f1a92c38eed0e203bc690761a181da8b9
This change utilizes the no panel auto more key feature to implement
long press shift key for shift lock.
Change-Id: I3995d25dc35aea3c67b5aa29299815462eff9cad
Now that separators have their own LogUnits, they must be uncommitted
from the LogBuffer when backspacing over them.
Bug: 9088919
Change-Id: Ib36cc94939b93abe18850a06bced17caf8aaa5b9
Currently when the user reverts a batch input, a LogUnit is uncommitted
from the LogBuffer. It should not be, because the LogUnit containing the
batch input is never committed in the first place (it is only committed
to the LogBuffer when a key is pressed or a new batch input is entered).
Bug: 9088919
Change-Id: I323af453ce082437a663ccae977b21b775a964bc
This should be done in onFinishInput rather than onFinishInputView
but for an unknown reason onFinishInput never comes.
Bug: 9113167
Change-Id: Iafb40846a3ed8fad78bfc160d17cd431f81fe891
Punctuation is currently put into the same LogUnit as the
following word. This is counter-intuitive and makes
post-processing more complicated.
This change causes punctuation to be put into its own
LogUnit.
A tricky case is the phantom space. A phantom space is
inserted right before other text, whose MotionEvents have
already been collected and inserted into mCurrentLogBuffer.
Phantom spaces are handled by creating their own LogUnit and
inserting it into the LogBuffer right before
mCurrentLogBuffer is shifted in.
Change-Id: I102d7cd93982ad37ba0abb6c1ac0f66b85b5f95d
When the user reverts a commit, this causes the committed
word to be changed. The user usually then goes back to edit
that word. Going back to edit causes an uncommit, so that
the editing actions are added onto the LogUnit containing
the operations used to construct the word.
But currently, reverting also involves its own
commit/uncommit pair, with the caller performing the
uncommit, and the commit happening inside the revert
command. This may have been necessary in the past, when
revert was called in different situations, but is
unnecessary now. Furthermore, the guarding conditions are
currently such that the uncommit doesn't always undo the
effects of the commit, corrupting the log output.
Change-Id: I74af41f4f1db2fcabfa496dcc4a2d7bd0af19b3a
- Don't call finishComposingText when useless.
- Add safeguards against calling setComposingRegion when the
data returned by the editor is inconsistent.
- Cancel pending recorrection messages when new messages arrive.
Bug: 8842941
Bug: 8845001
Change-Id: I939701033cf5c2bbd85871ecf83e329021ddeb91
Also, theoretically, we don't want to retrieve older suggestions if
there are no results for a recorrection - that would look random.
This bug addresses this too.
Bug: 8657919
Change-Id: I44f36d34a708a968ab71fa0592da57f4c97d4b7f
This is about as ad-hoc as it gets, but then again, what we want
is probably as ad-hoc as it gets.
All URL boxes I know of double as search bars, and not adding
automatic spaces there sucks (e.g. in Chrome URL bar).
And in other boxes actually you don't want to add a space if
it looks like a URL. QSB isn't even a search box, and it behaves
like this.
So I think this is actually the right answer to the problem.
Bug: 7062925
Change-Id: Ib09472b34644fd5bf2dc84bb97cedeeba28bcd02
The only place where it's used is checked for nullity.
Also, it's possible, also difficult, to match a different
recapitalize with the old code, triggering a bug that
this fixes.
Change-Id: I717d6df489025c75d1caca290a9086c3b39a9306
Upon pressing Shift, if there is currently a selected string, have
Latin IME change its capitalization.
This does not yet have the keyboard mode follow the mode - the change
is complicated enough as is.
Bug: 7657025
Change-Id: I54fe8485f44e04efd72c71ac9feee5ce21ba06f2
If the user gestures a word, then hits backspace in
disapproval, and gestures about the same thing again,
make sure that we don't suggest the same thing again.
Bug: 7549311
Change-Id: I793bc4df7c3841fa8f2f4146707c26e873f374c1
This uses the old suggestions. It does not try to recompute
new suggestions if there are no old suggestions yet: this is
coming in a later change.
If there are no suggestions, this shows the word itself
as a suggestion.
Bug: 8084810
Change-Id: I4c2e25df0ff3673be1825f57a0c19a9d23d47a48
Previously handleSeparator() caused the ResearchLogger to mark the time at which a LogUnit should be
broken. However, this causes the motion data associated with a separator to be associated with the
LogUnit of the previous word. This change corrects this bug.
Change-Id: I8b4d4fa6de2a013de9e2a28bb668c446a07f1957
Previously, ResearchLogger#onWordFinished() was called with an outdated parameter value for
isBatchMode, causing it to report false even for gestures. This changes fixes this problem.
Change-Id: Ifcabee236ba5fe20376ad882155d3f3142cd7613
Previously, ResearchLogger#stop() was called both in
LatinIME#onFinishInputView() and in LatinIME#onWindowHidden(). This
resulted in multiple logs being written.
Since onFinishInputView is the more reliable of the two (it is called
in InputMethodService#onDestroy; onWindowHidden is not), the code now
uses onFinishInputView as a stopping signal.
Change-Id: Iae4b8c3bdab226027624eeab19b3737367e4a108
This unifies the software and hardware keyboard code
under a single decision process that works.
Bug: 8129303
Bug: 8152758
Change-Id: I7574c563d5f957d57bfe62fe5e3eec59a519d335
- Add support for replaying log files to the ResearchLogger. This will let
users preview data that they choose to upload.
- When the user explicitly requests that the system record their action, it
will record everything up to, and including, the motion involved in shutting
off the recording. This change also removes the stop-recording motion
commands.
Change-Id: Ib1df383bbf1881512cb111fab9f6749c25e436ba
This change adds the following flags to the intent to lauch settings activity.
- Intent.FLAG_ACTIVITY_RESET_TASK_IF_NEEDED
- Intent.FLAG_ACTIVITY_CLEAR_TOP
Bug: 8055342
Change-Id: I85284409bbba3fd284442e5ac4c7a55e58ad15ec
Heuristic to determine whether a word was a typo correction or a complete
replacement by examining a correction to see if it falls within the list of
suggested words of the original.
Change-Id: Ieec4861a811e96aef0d14622e662b589ef8b4772
The implementation is not really ready yet, so we want to
disable it for normal builds for now.
This goes together with Ic1edc6ed
Bug: 7987492
Change-Id: Id81f417e200c118f24424943823665f0b8bd6849
It would insert a space when it shouldn't when composing a
word ending in either dash or apostrophe and gesturing right
away.
Bug: 7889078
Change-Id: Ib3e2f6f5e551888cc081b464a641e2deac6f5ef7
ResearchLogger.latinIME_handleSeparator wasn't doing anything really specific to
the separator, so its implementation has been renamed according to its role.
Change-Id: I24e8691c7dc440dd067e96d23e50463683e17cfc
- Normal delete events are recorded
- Includes number of characters deleted for both batch and normal delete
Change-Id: I422da3ddc94fa0ccd6c7586eaacf14c62dd16679
- Fix function calls to happen at more resilient times
- In the case of phantom spaces, ensure that the logStatement goes to the
correct logUnit
Change-Id: Ida8f6eba1e4f33d9f9b4735316e3c6316ed316c0
This change moves dismiss timer of gesture floating preview text from
PreviewPlacerView to KeyboardView.
Bug: 7967461
Change-Id: I0ca5beddc93cb4bc4a405f914d217d37b997402b
It's useless to setAutoCorrection(getTypedWord()). Every time the
contents of the word composer are altered, the auto-correction is
reset, and at use time if it's null then the typed word is used
anyway.
Change-Id: I0870657a1ab3f456f376995b27e70703f7a5d23a
Essentially this does activate auto-correction with a hardware
keyboard, although a lot of things are still left to implement.
No proximity is used yet which means only missing and excessive
letters are considered. Dead keys are not handled. No combiner
is supported. No suggestions are displayed. Resuming suggestions
does not work correctly with a hardware key (because the view
holds a temporary hardware event 'onKeyPreIme' and the event
from the IME won't be handled until this is handled which won't
happen until after the IME said that it did handle the event).
Bug: 5037589
Change-Id: Idcb5c7b26d56717ed772d53c062362807f11cdae
- count how many times a gesture is deleted
- count num chars entered by gesture
- change name of fns to reflect capturing gesture information when setting composing text
Change-Id: I1b8ebf9cfcf304e0efd827c8b8e28a4010199728
When the user edits a word before adding it to the user
dictionary, the keyboard should replace whatever was
committed before with the amended version.
Bug: 7725834
Change-Id: I1a417be6c5a86d6a96bc2c76aca314ad8f1202a9
Eliminate use of slower popup window for rendering the more keys menu and more suggestions menu.
These menus are now drawn directly on the preview placer view.
Change-Id: I40638724a6ffde2be4b5d0341d2a5db7ff7be930
Correctly check the variations to determine if we should
add automatic spaces (in manual picks and gestures) or not.
Bug: 7164981
Change-Id: I19e25af839b4062c217113cd804b1473912f22a5
This test is seriously wrong : although we do have a method that
does exactly the right thing, this looks at a very specific, unrelated
variable state in hope it happens to be set only in the relevant
cases by the play of logic.
I don't think this changes anything concretely, as the assumption
was probably true in the practice ; still, this code is very hard to
read and prone to unexpected breakage.
Change-Id: I77ab576f90762429fcc752df31ed30e1c81871dc
Specifically, reset the KeyboardState to the main Alphabet layout in
LatinIME.onStartInputViewInternal() if a keyboard layout set doesn't get
reloaded in the method.
Please note TextView.setText() calls up resetInput(), then the IME should
reset with it.
bug: 7482086
Note that bug: 6851364 needs to be revisited.
Change-Id: I5d448c10963d3dd952dd13cb587085ec7b014e69
This change also
* Introduces adaptive gesture detecting threshold for time domain in addition to length domain.
* Tunes the parameters for detecting gesture after fast typing.
* Fixes a bug in dismissing gesture floating preview text.
* Cleanup debug messages
Bug: 7218902
Change-Id: Iafccd872c6efe0c3b5ae65fa40b04c80d9f139c7
All this code is done inside the StringUtils or RichInputConnection
classes, where it's more logical. Since the IPC call is entirely
gone now, it's no use having these guards inside Latin IME.
Change-Id: Id324567e7458aed12b6f21d7114be3bd39725f15
A space would be inserted after an open parenthesis if
it's typed after a manual pick.
Bug: 7167687
Change-Id: Ic49d09c34747117ccfbdbb8c66f634d88d4a45b0
We decided to revert this UI decision after negative feedback.
Indeed, if you use to gesture but tap just one word, chances
are it's because you know you do NOT want a dictionary word.
Bug: 6965241
Change-Id: I85fdfab3c1cf1d3c0fd3e95908e01bdf56bccbba
This should have on effect at all on behavior,
except an increase in performance.
Bug: 4967874
Bug: 6950087
Change-Id: Ie2b51efefe84ca767f5dc8e3b80bfef7e1faab3d
This is pretty much as strong as it gets. It should be
impossible to get false positives and nearly impossible to
get true negatives with this new code.
Bug: 6981089
Change-Id: Ia32ab62f89c5943f0be169b979abab652e67bf5b
...if punctuation suggestions are requested.
Fixes the symptom, but not the root cause of
Bug: 6981089
Change-Id: I683a7451c99457829cb4e5a762f43548ff89a535
This change makes Latin IME behave consistently with regards
to other auto-correction cancellations in cases of auto-correction
cancellation after smiley-triggered auto-correction. That is,
pressing the smiley key when the keyboard signals it's about to
auto-correct will get the auto-correction there plus a smiley,
and pressing backspace will cancel the auto-correction, and
pressing backspace again will delete the smiley.
Bug: 7067593
Change-Id: Ia7eef70a5d06b8b9afa1f1fbb0ed1dbc21a3059f
- Captures complete motion data for all words in an n-gram.
- Also filters n-grams properly; if any word in the n-gram is not
in the dictionary, it is not included.
- Simplify ResearchLog to not require explicit state
- Added LogBuffer class MainLogBuffer class to allow n-gram-level decisions
about privacy.
- Moved LogUnit out from ResearchLogger
multi-project change with Ic70ccb6c2e11eb02d887821434b44daa3eb7a3e2
Bug: 6188932
Change-Id: I731d6cff6539e997874f723b68bdb0d9b651b937
Move many ResearchLogger data collection points to RichInputConnection.
By collecting data here, developers do not have to remember to keep the
ResearchLog consistent with calls to the RichInputConnection.
In addition, some unnecessary log points were removed, and the ResearchLogger
is now independent of LatinImeLogger.
multi-project change with I05496cfd762e9a41c42c50099143f1efc0a3d8a2
Bug: 6188932
Change-Id: I424abb2e437901262a9620255493916b7c3ff74b
DO NOT MERGE
- Captures complete motion data for all words in an n-gram.
- Also filters n-grams properly; if any word in the n-gram is not
in the dictionary, it is not included.
- Simplify ResearchLog to not require explicit state
- Added LogBuffer class MainLogBuffer class to allow n-gram-level decisions
about privacy.
- Moved LogUnit out from ResearchLogger
multi-project change with Ie2bc79fd7fe6b951b24771e94b8d4ca21989af65
Bug: 6188932
Change-Id: I568c90d4af07e7c759c1e7fc64b716bd8c7b4ae5
Move many ResearchLogger data collection points to RichInputConnection.
By collecting data here, developers do not have to remember to keep the
ResearchLog consistent with calls to the RichInputConnection.
In addition, some unnecessary log points were removed, and the ResearchLogger
is now independent of LatinImeLogger.
multi-project change with Ib71f841044ec1072610ab5638a5edfce29b7c05b
DO NOT MERGE
Bug: 6188932
Change-Id: I9ec7500a5b18bfe1a5a5bb1e5bf21c43351fc59e
For some reason we get a "restarting" flag from the browser
some times when the input field is different. The attributes
however are not the same, so we can detect that and reload
appropriately.
Bug: 6946793
Change-Id: I6762dae6f41db690497b026a707d9cc89c840b34
The commitTyped() called updateSuggestionStrip(), and that led to call
Suggest.getSuggestedWordsForTypingInput() even after a batch input was
started because WordComposer.commitWord() reset the batch input state.
Bug: 6932442
Change-Id: If645b1ef0343aa1e21ca71a9275b40c6c3be13df
The Italian dictionary has been amended to contain words that
include apostrophes, and accordingly the apostrophe is now
excluded from word separators in Italian.
This code was only used for Italian, and in the general case
it does the wrong thing; it was only needed by limitations of
the old dictionary.
Bug: 6929108
Change-Id: I7ff4615e01899a1cebed2c4c51679cd0746740fe
Not having a main dictionary is not sufficient reason to display
the hint. If the user already added the word to the dictionary then
we shouldn't be displaying it again, even if we don't have a main
dictionary for the current language.
Bug: 6829798
Change-Id: I6b081f51c82ef395fe2e394fd3167363e743ea6f
- also cleaned up RLog menu
multi-project commit with If0fd4fef89d390073e6939d5188ed5696866cb33
Bug: 6188932
Change-Id: I4f66f13bd366b4e8bde742ccd0704f812c6d33f9
This change also removes the reference of SuggestedWords from
GestureTracker and KeyboardActionListener.
Change-Id: I25ef8756007986abf99a931afd665bbfe6fa387f
showSuggestions is only ever called by updateSuggestionsOrPredictions.
It only feels natural that, when called with a 0-sized or null
suggestion list, it clears the suggestions.
Change-Id: I5b52bc9151afca1bb6c6f0a7f81e3255add92177
- separate ResearchLog from ResearchLogger
- fix abort menu to explicitly indicate re-enabling of logging
- auto flush entries every 5s
- conform to new suggestedWords class
multi-project change with I9fe1f8fe59d5891daa4ca84af157f36fb1bb17dc
Bug: 6188932
Change-Id: I7e99390f934481c6ed9cea2b259944583edda5bb
Many methods were public but could have been private: change them
to private. Also, add a comment above public methods that don't
come from the IMF to make it clearer why they are necessary.
Change-Id: I574154b015d09c0f8542b81763f497691afdc3c1
If !mWordComposer.isComposingWord(), then mWordComposer.getTypedWord()
will always return an empty string.
Change-Id: Ife66d0abc44c743cbc30d31724e833cda168fd5c
This is not exactly the same logically speaking, because it's
theoretically possible that the composing state changed in between
the message enqueueing and it's retrieval. However in the practice,
if the composing state changed the message *must* have been
cancelled and resent, else the behavior breaks. So this actually
is more robust, and removes some obscure requirements on the
calling code.
In the practice, it should also make the cancelUpdateSuggestionStrip
message useless, although this change does not yet remove it.
Change-Id: I75141920ce64e38e2f92e9c02b6c979936eee9a9
