Bug: 11584525
Change-Id: I76cc3e4ee21d62fbd56042adcf085efd5cafb53f

This returns the wrong string, but since it's used for getting the
previous word for bigrams, it only results in slightly worse
suggestions quality.

Bug: 11273655
Change-Id: I6ce5de2f76effc453ca691a654ab6bf17445b9e7

Bug: 11311002
Change-Id: I62955e364c9ffc75322cf05fa3ad7985f1d09259

- Suggestions form user history can contain invalid words.
- isValidWord always returns false.

Bug: 11139426
Change-Id: I6075b275603332ddb00f4a9284afcaa82d824270

This allows testing for suggestion freshness in an asynchronous
suggestions world.

In-advance cherrypick of Ic76cd17568598d8534aec81e037f9e37f52eb6b4
because there's a merge conflict.

Bug: 11301597
Change-Id: I4aec765a975298fcac30a48dede73d2622224fe5

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 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: I9c6a948331726a821bd3ccec9c1d02dec2c4703a

Small optimization on generating a String instance from a single code
point too.

bug: 11181913

Change-Id: I0f905e4dc6ec7841092bb4d3d940daf3b2303f5b

Bug: 11163495
Change-Id: I23d12c430125cc2a66a6e00715a4c609bb9e2bb1

Bug: 11163495
Change-Id: I3247b8d1bbd3406b29a30a25aebd932c63431943

The available space was not computed correctly. This makes the
decision to ellipsize/expand the middle suggestion correct.

Also, the scale was not applied correctly in some cases.

Finally, for some reason ellipsize does not work as we expect.
I'm not sure how to get it to work like we want it to, but
this works around the problem in a simple manner.

Bug: 11158757
Change-Id: I083e7ac56e157f100923733e67b92a6368a526c9

Bug: 11029983
Change-Id: I5547910c7b5dd7974292fc075af33568940f4e81

Bug: 11029983
Change-Id: I1bec2c3c6b547de68562a818a8d6c9fef83787b7

Bug: 10920255
Change-Id: I26d2cce3c322a4ff39a614f8615f43fb7bd3baed

In new format, each bigram entry has flags (1 byte), a terminal id (3 byte),
a time-stamp (4 byte), a counter (1 byte) and a level (1 byte).

Bug: 10920255
Bug: 10920165
Change-Id: I0f7fc125a6178e6d25a07e8462afc41a7f57e3e1

Bug: 11158604
Change-Id: If8c94ef91ed58bb5028f51be7c8d9beb677436a4

This reverts commit 220ce8f2.

Sorry, but we are now in the week before the final cut for the next launch. We shouldn't be taking new features like this at this point. Let me revert this change for now. I'll re-revert this after the cut next week.

Change-Id: Ia8b00680c992ec526e6a858945f2380cbcdfdd55

There used to be two "Swiss" keyboard layouts, labelled fr_CH and de_QY,
available in LatinIME. They were, however, not actually Swiss layouts
but instead work-arounds to be able to get French or German with qwertz
or qwerty (instead of azerty and qwertz) layouts, respectively. These
were thus removed by commit f6972561
replacing them with the generic fr-qwertz and de-qwerty layouts instead.

Here, now, is a proper Swiss keyboard layout, following Swiss standards
for people accustomed to typing on Swiss keyboards, with the expected
tailorings for French and German.

Change-Id: I9feed752053ecc694c84db713284f69bc1daf155

Bug: 10118761
Change-Id: I63501d6c2b5f561d7ab8b7362498665d805d5e1e

Bug: 11136982

Change-Id: I9d022a178c47bad3566be4c0bd8ffbce0b14896d

Bug: 11073222
Change-Id: I680e4304c6ab701e9aa132e1e140d789097a53ae

Bug: 10118761
Change-Id: I63501d6c2b5f561d7ab8b7362498665d805d5e1e

Bug: 11136982

Change-Id: I9d022a178c47bad3566be4c0bd8ffbce0b14896d

Workaround for framework bug:
Bug: 10792236

Change-Id: I3706b9eed5223889791840e30660f8d17625cb70

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

Bug: 10902556
Change-Id: I3c8da35a601de2bfb248292c7b59235fc1120988

Bug: 11107229
Change-Id: I0b36341d60b634a860eb13fafd0dc69fe734bdeb

This will help handing correctly the armenian full stop.

Bug: 10082781
Change-Id: Id7bb219ebd89daba203216eab362d1cc26a65a36

We want to use StringUtils here, but it's full of references to
stuff not accessible host-side like JsonReader and TextUtils
and SettingsValues :/

Bug: 11061476
Change-Id: I3c0194979833ede283b4f9190335dba5376fe6fc

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

Bug: 11077905
Change-Id: I9495f13a20edca0e6153cd1a7a0ac6891f707c59

Bug: 11076722
Change-Id: I0de10ab24d33ec31f45664fe92d80d5a7bec9c20

14 is the right value.

Bug: 11076722
Change-Id: I95d404b540f7fbe4932d1f8498cde23f1df0314f

- GC gets failure when the dictionary become empty.
- Useless unigrams are sometimes not removed.

Bug: 10197478
Change-Id: I8d1479c01efba61a81f03bc077da6bcb4797a940

This is not the right way to do it, but it's simple and will kill the
regression. We should investigate why the return values are
wrong in this case.

Bug: 10780091
Change-Id: I808727242c9cb1598df00a7426358dc3a658add4

This is a very rare corner case.

Bug: 11072132
Change-Id: Iad2aa69511f7dc99105284a049c63f2f997b8ef0

This also includes a fix that allows this code to read surrogate
pairs in this processing.

Bug: 11070482
Change-Id: If5ef8d6863938252f09128b7e99ea07ece6e7019

Change-Id: I5f85a7a0f94ea9ecbe0c4a8caebcf551fa9c9669

Change-Id: I986ab26faf535fc4bc98443053f534eced9d048f

Change-Id: I2208378b33038771b460abb33f9a690872e998e2