Commit Graph

1022 Commits (75611ee033bc9f8ff8ae58ce5c35130c2fbb5fef)

Author SHA1 Message Date
Keisuke Kuroyanagi dc3856d758 Add LanguageModelDictContent.
This class will replace BigramDictContent and
ProbabilityDictContent.

Bug: 14425059
Change-Id: I3d15c833957e27b2f5999386db042188272bbb4b
2014-08-01 12:45:00 +09:00
Keisuke Kuroyanagi 90b7c1729f Remove DictContent.
Bug: 14425059
Change-Id: I74fa4b6ba4605447c1c87427371e4be5eb8e7ae6
2014-08-01 12:06:21 +09:00
Keisuke Kuroyanagi 0e6a1d1020 Remove getBigramsIteratorOfPtNode from structure policy.
Bug: 14425059
Change-Id: I427fbfd026c1e33b5e9cf78949fe862d8748b7ec
2014-08-01 12:06:21 +09:00
Keisuke Kuroyanagi 35c62b2cc9 Use NgramListener in MultiBigramMap.
Bug: 14425059
Change-Id: I425536290111f2a8172f31370706f858a1e07f6e
2014-08-01 02:51:07 +00:00
Keisuke Kuroyanagi de5c3a2562 Construct TrieMap from existing buffer.
Bug: 14425059
Change-Id: Ia1c58d88540f55fb38fe0bdfcd7c0a34ebc7415e
2014-07-31 09:06:22 +00:00
Keisuke Kuroyanagi e7c9415688 Merge "Remove DictContent." into lmp-dev 2014-07-24 16:04:26 +00:00
Keisuke Kuroyanagi 4a1ed92ea8 Merge "Remove getBigramsIteratorOfPtNode from structure policy." into lmp-dev 2014-07-25 11:19:20 +00:00
Keisuke Kuroyanagi 3437032a91 Merge "Add TrieMap.save()." into lmp-dev 2014-07-24 16:04:26 +00:00
Keisuke Kuroyanagi 60ae3e0be5 Add TrieMap.save().
Bug: 14425059
Change-Id: Ie0b4fe8b9a157da0401bb277c703ab084d5bf9cc
2014-07-31 06:01:52 +00:00
Keisuke Kuroyanagi 5c1decfbb9 Add entry iteration method to TrieMap.
Bug: 14425059
Change-Id: I79420b755f29f651d8eed61e7e48b6eb001d8dd2
2014-07-30 19:04:07 +09:00
Keisuke Kuroyanagi 57574b98c2 Merge "Remove doesAutoCorrectValidWord()." into lmp-dev 2014-07-25 11:19:20 +00:00
Keisuke Kuroyanagi 5b69472d56 Remove doesAutoCorrectValidWord().
Possibly offensive words are wrongly demoted when input by
gesture.

Change-Id: Ibb2eb73d4191dea6cb767643341fe96465c1f70c
2014-07-31 14:48:16 +09:00
Keisuke Kuroyanagi d3d7c31ca9 Fix native debug build.
Change-Id: I95ff6b14d7f444911ec76698381b8d2d8da94776
2014-07-31 14:36:12 +09:00
Keisuke Kuroyanagi d624091afb Merge "Fix native debug build." into lmp-dev 2014-07-25 11:19:20 +00:00
Keisuke Kuroyanagi 211370a3da Merge "Add memory view class for int and uint8_t." into lmp-dev 2014-07-24 16:04:26 +00:00
Keisuke Kuroyanagi 75d8c20d03 Add memory view class for int and uint8_t.
Bug: 16691311
Change-Id: I32eb692cdb758861e24b772fdcda57642b95ae1b
2014-08-01 12:39:31 +09:00
Keisuke Kuroyanagi f40adc97cb Merge "Add a method to iterate ngram entries." into lmp-dev 2014-07-24 16:04:26 +00:00
Keisuke Kuroyanagi 2d57b3339a Add a method to iterate ngram entries.
Bug: 14425059
Change-Id: I9962c546504288f6c22b1a7368f775decd229c62
2014-07-30 10:52:56 +09:00
Keisuke Kuroyanagi 5a7b634aaf Merge "Add entry iteration method to TrieMap." into lmp-dev 2014-07-24 16:04:26 +00:00
Keisuke Kuroyanagi a704ad254e Merge "Remove PrevWordsInfo.getBigramsIteratorForPrediction()." into lmp-dev 2014-07-25 11:19:20 +00:00
Keisuke Kuroyanagi bd1f59bda5 Remove PrevWordsInfo.getBigramsIteratorForPrediction().
This changes the behavior of predictions for some rare cases.
For example, let's consider the case of the prev word is
"ABC", "ABC" and "abc" is in the dict, "ABC" doesn't have
any bigrams, and "abc" has bigrams.
Without this change, "abc"'s bigrams are used for
prediction. But, with this change, "abc"'s bigrams are not
used for prediction. This strategy is the same as the the
current way of bigram probability computation during
traversal (for suggestion).

Bug: 14425059
Change-Id: I857ff39aa70aea265efa651451a63b759bb47d48
2014-07-30 17:26:26 +09:00
Ken Wakasa 0b1fa0c1c7 Cleanup: C++11 doesn't require a space between angle brackets
Change-Id: Id0777cf3deac94395bd63aee40b49375dff3e35a
2014-07-30 11:32:05 +09:00
Keisuke Kuroyanagi 6c3b0d3caa Merge "Add TrieMap." into lmp-dev 2014-07-25 11:19:20 +00:00
Keisuke Kuroyanagi de3121dead Add TrieMap.
Bug: 14425059
Change-Id: Ief36f808ed4f3e35bfab07837ea314d2564973ab
2014-07-29 17:00:29 +09:00
Keisuke Kuroyanagi 2e32e83b6e Merge "Get n-gram probability in structure policy." into lmp-dev 2014-07-24 16:04:26 +00:00
Keisuke Kuroyanagi 1229879e7c Get n-gram probability in structure policy.
Bug: 14425059
Change-Id: Id955a2e07dbdfb90ae55720f0c92597e0594c0d9
2014-07-29 12:01:18 +09:00
Keisuke Kuroyanagi 96d47fe745 Change getUnigramProbabilityOfPtNode to getProbabilityOfPtNode()
Bug: 14425059
Change-Id: I92403d750d2979408b41554dcd79b0ed31b41161
2014-07-29 11:35:57 +09:00
Keisuke Kuroyanagi 179f38cc8c Merge "Change getUnigramProbabilityOfPtNode to getProbabilityOfPtNode()" into lmp-dev 2014-07-25 11:19:20 +00:00
Keisuke Kuroyanagi 3225b6fe66 Add boundary check for ver2 bigram reading.
Bug: 16330528
Change-Id: I6aca6c7a735e2a652eb325572d44dff660789cff
2014-07-28 21:48:00 +09:00
Keisuke Kuroyanagi 825243bfdd Merge "Add BufferWithExtendablebufferTest" into lmp-dev 2014-07-24 15:37:54 +00:00
Keisuke Kuroyanagi d4c9d50ef0 Add BufferWithExtendablebufferTest
Bug: 13754552
Change-Id: Ib9c330dc85bdb5d4d3d5bc15cc3ce5f28faa0b12
2014-07-25 07:02:10 +09:00
Keisuke Kuroyanagi ff6ca09905 Add BufferWithExtendableBuffer::extend()
Bug: 14425059
Change-Id: I13dd8412ba4d16a1325ac2210b3852d580e71ad3
2014-07-25 07:02:05 +09:00
Keisuke Kuroyanagi dfbe2a8ca8 Merge "Add BufferWithExtendableBuffer::extend()" into lmp-dev 2014-07-24 15:35:58 +00:00
Keisuke Kuroyanagi f78eb27c7e Merge "Refactoring checkAndPrepareWriting." into lmp-dev 2014-07-24 15:35:58 +00:00
Stephen Hines 5a289ed293 Revert "Revert "Revert "Use mclinker"""
This reverts commit b48c93b3fc.

MCLinker is not a valid option for clang 3.5.
2014-07-23 19:58:17 -07:00
Keisuke Kuroyanagi 6810e8df1e Refactoring checkAndPrepareWriting.
Bug: 14425059
Change-Id: I98940f3841a6c61b0482440b4da21f415dc843f6
2014-07-24 10:24:42 +09:00
Ken Wakasa 81a48d47d1 Add -mstackrealign to avoid crashes with SSE instructions on pre-K x86 devices.
bug: 16213554
Change-Id: Icd67ed4fd57fcc5f5bf8039b7892d0d446210247
2014-07-15 20:26:55 +09:00
Keisuke Kuroyanagi b22f95ec8a Remove isUpdatable from constructors of dict contents.
Change-Id: I2d54f477d9b341e944e265786a734f23d152bb81
2014-07-11 15:23:55 +09:00
Keisuke Kuroyanagi 2ac934296c Concatenate dict buffers other than header to a single file.
Bug: 13664080
Change-Id: I34c9d8046b339c9b855be378a5fad907382d1359
2014-07-11 15:15:47 +09:00
Keisuke Kuroyanagi 198be3a6c5 Fix: wrong bigram entry pos can be used during GC.
Bug: 14425059
Change-Id: I4e7c977c5ef184953682175b118de5473cdb64a6
2014-07-09 11:18:35 +09:00
Keisuke Kuroyanagi 3b8f2ece40 Merge "Move phrase gesture from additional features to public." 2014-07-02 20:35:42 +00:00
Keisuke Kuroyanagi b8a9479b57 Move phrase gesture from additional features to public.
Bug: 15758242
Change-Id: I703dc2703afdf9402c5d2ac27411b74ece993dba
2014-07-08 16:36:06 +09:00
Keisuke Kuroyanagi b00973952f Get bigrams iterator via dict structure policy.
Bug: 14425059
Change-Id: Ie4c6afbb6a3b707b5aa8e9e4f941c9da6e6bcdbf
2014-07-08 16:23:31 +09:00
Keisuke Kuroyanagi 804f7450fc Use linked list for bigram list.
BinaryDictionaryTests for VERSION4_DEV:
Before
Time: 36.461
After
Time: 33.031

Bug: 14425059

Change-Id: I9ca2714f450f61f713df6ebd34c953dece991cdb
2014-07-07 21:09:25 +09:00
Keisuke Kuroyanagi 6afd14dc94 Remove unused method.
Bug: 13664080
Change-Id: I47fd90f01ea9a18b11037c966150fa5fd926101b
2014-06-27 15:23:38 +09:00
Keisuke Kuroyanagi c152d7c19d Fix unit tests that use getNgramProbabilityNative().
Change-Id: I8548094bc8d1a0ea125b29e2a0ea82a6834b3e48
2014-06-26 19:34:27 +09:00
Keisuke Kuroyanagi 05b1e0d42f Extend jni method to use multiple previous words.
Bug: 14425059
Change-Id: I8150e8812ee32f8c713315a234b52e5b3a79d7a0
2014-06-26 18:47:25 +09:00
Keisuke Kuroyanagi 8681bef03c Remove BigramDictionary form native code.
Bug: 14425059
Change-Id: I48d1a4ae6a4317afaaf67748060618f8f9485d40
2014-06-26 15:15:13 +09:00
Keisuke Kuroyanagi d58a07666e Merge "Rename jni method for n-gram supporting." 2014-06-24 01:52:26 +00:00
Keisuke Kuroyanagi 83b6ee2538 Rename jni method for n-gram supporting.
Bug: 14425059
Change-Id: I242d1de5d9db37344f02d8fba9035829794a56c8
2014-06-25 14:14:40 +09:00
Keisuke Kuroyanagi 520765491c Merge "Make Beginning-of-Sentence prediction require two exposures." 2014-06-24 01:52:26 +00:00
Keisuke Kuroyanagi 0aef59746c Make Beginning-of-Sentence prediction require two exposures.
Bug: 15868192
Change-Id: I4b577ad11c992d872d6f902fd50527d9ca5cc1ba
2014-06-25 20:20:44 +09:00
Yohei Yukawa 5961f2dfac Merge "Simplify a make file" 2014-06-24 01:59:00 +00:00
Yohei Yukawa cc9b60401e Simplify a make file
This CL simplifies the makefile for host unittest.

BUG: 13754552
Change-Id: Id09da77b45610db2747897533c18a6d0d81c59d7
2014-06-25 23:10:19 +09:00
Keisuke Kuroyanagi 88fa47a27d Support migration/dump of Beginning-of-Sentence entries.
Bug: 14119293
Change-Id: Ie975138f819794d5c34a7a547be5a6117050e084
2014-06-24 12:37:07 +09:00
Keisuke Kuroyanagi f7322b166b Make getNextWordAndNextToken output code point count.
Bug: 14425059
Change-Id: Id1c2927c8a1be0b1680206d444dbdf0c9453dceb
2014-06-23 19:08:30 +09:00
Keisuke Kuroyanagi a807663aae Merge "Remove hasBigrams/hasShortcut flags from ver4 PtNode." 2014-06-19 23:04:31 +00:00
Keisuke Kuroyanagi 24d49ee577 Remove hasBigrams/hasShortcut flags from ver4 PtNode.
Always check bigram/shortcut content instead of having
flags in each PtNode.

Bug: 14425059
Change-Id: I9eb816681bf22fc12e3ab3c92a178a7990b90e93
2014-06-23 16:49:49 +09:00
Yohei Yukawa 762a356a45 Merge "Add native unittest support on target devices for LatinIME" 2014-06-19 22:58:18 +00:00
Yohei Yukawa 7f4cca0cd1 Add native unittest support on target devices for LatinIME
This CL adds native unittest support on target devices for
LatinIME.

Note that you need to specify --target option to run-tests.sh
to run native unittest on the target device.

BUG: 13754552
Change-Id: I1dfb78ae1461163db8d47c3ba9141d4812070fd0
2014-06-23 14:41:34 +09:00
Keisuke Kuroyanagi 0fbca1ac23 Remove word from personalized dicts when it's canceled.
Bug: 15555552
Bug: 15552347
Change-Id: Ia4d8c79d079e3ce233d46d4bbad4fff1be5c5c39
2014-06-20 14:46:13 +09:00
Keisuke Kuroyanagi 89899e14df Merge "Update unit test for beginning-of-sentence." 2014-06-10 22:31:57 +00:00
Keisuke Kuroyanagi e041bfab5c Update unit test for beginning-of-sentence.
Bug: 14119293
Change-Id: I594a1ca1096f0c12c678dd30eb75de905e55c5bb
2014-06-12 16:07:39 +09:00
Keisuke Kuroyanagi 22931cd941 Enable Beginning-of-Sentence prediction for contextual dict.
Bug: 14161647
Bug: 14119293
Change-Id: I0c00f13966db88e4de85e245e7bced43c9d474b2
2014-06-12 12:26:18 +09:00
Keisuke Kuroyanagi da0ea7603b Merge "Add jni method for removeUnigramEntry()." 2014-06-10 22:31:57 +00:00
Keisuke Kuroyanagi b4d77eca55 Add jni method for removeUnigramEntry().
Bug: 15552347
Change-Id: Iadcdc73d3335d561a27b9314d3fcdab0b0c96e49
2014-06-12 12:21:44 +09:00
Keisuke Kuroyanagi f12985245c Add removeUnigramEntry() to native dictionary policy.
Bug: 15552347
Change-Id: Ic713f79ef26b5963060304df9c0a3a402c3cc6d4
2014-06-12 12:14:11 +09:00
Keisuke Kuroyanagi 86fe8081b3 Merge "Add removeUnigramEntry() to native dictionary policy." 2014-06-10 22:29:12 +00:00
Keisuke Kuroyanagi 447c19364a Merge "Enable Beginning-of-Sentence prediction for contextual dict." 2014-06-10 22:31:57 +00:00
Keisuke Kuroyanagi 59ed0c2db2 Block offensive words in native code.
Bug: 15300452

Change-Id: Ic6c91c47453e19fb5db2c62b66659619e72b1ecf
2014-06-12 11:36:00 +09:00
Keisuke Kuroyanagi 07e1412631 Purge dict reading code for 401 and create for 402.
Change-Id: Ib0150b3306bb2db8aff443d1158fecc8464cbbdb
2014-06-10 19:22:40 +09:00
Keisuke Kuroyanagi d9b8602f48 Implement getMaxProbabilityOfExactMatches().
Bug: 13142176
Bug: 15428247
Change-Id: I5be6d683be95505412615ca7c88260de1ea05f54
2014-06-06 17:37:46 +09:00
Keisuke Kuroyanagi d6b89e17a6 Add jni method for getMaxFrequencyOfExactMatches().
Bug: 13142176
Bug: 15428247
Change-Id: I2ae25090b67ad0c6df97dec7712e2f7e0bb08c40
2014-06-05 18:16:11 +09:00
Keisuke Kuroyanagi 7852765a50 Add length check for attaching a BoS marker.
Bug: 15297872
Change-Id: I7b0a27409b81b76561d19c6caa18a5011d3e90f6
2014-05-28 20:20:39 +09:00
Keisuke Kuroyanagi f9ce867d80 Add boundary check for v4 bigram reading.
Bug: 14496386
Change-Id: Iedd3445c3222a777a2476beed7d9eb53773f406c
2014-05-27 19:29:35 +09:00
Keisuke Kuroyanagi dfca51726e Return whether the dynamic dict operation was success.
Bug: 12184250
Change-Id: Iee7e00c1e84c95551a077f4dd023c0a9b9ac9466
2014-05-27 17:30:00 +09:00
Keisuke Kuroyanagi 0088d1f668 Merge "Use whether it's exact match to detect distracters." 2014-05-26 13:53:19 +00:00
Keisuke Kuroyanagi a6278eb9c1 Use whether it's exact match to detect distracters.
Bug: 13142176
Change-Id: Id5b7286d28897931f7bfe571be45d46ffeef4adf
2014-05-26 21:56:20 +09:00
Yohei Yukawa 67e02a6aaa Specify c++_static for bundled Latin IME as well
APP_STL := c++_static has been specified for unbundled
Latin IME but somehow not for bundled build.

This CL fixes the inconsistency.

Change-Id: Ie1f4aba27e30981d3c2aa322a16c1cd18413e920
2014-05-26 18:30:48 +09:00
Keisuke Kuroyanagi a37f374ad1 Version up dynamic dict format from 401 to 402.
Change-Id: Ibea36af905ade773ae3db3a5456f7b5a0ad7d220
2014-05-23 20:20:56 +09:00
Keisuke Kuroyanagi 1adca93381 Extend jni methods and enable Beginning-of-Sentence.
Bug: 14119293
Change-Id: I78fc877367dd0d6240eeacb750b6d2d0b93cba83
2014-05-23 19:58:58 +09:00
Keisuke Kuroyanagi 96990ca773 Support Beginning-of-Sentence in native code
Bug: 14119293
Change-Id: I0f382e33a19bf481823b23405d454de61ec835ff
2014-05-23 17:23:08 +09:00
Keisuke Kuroyanagi 6600340af5 Fix: PtNodeParams.representsNonWordInfo()
Change-Id: I7ab4fb280e8ac89f81d0b32b0663a86c9b69aa5d
2014-05-23 13:49:18 +09:00
Keisuke Kuroyanagi 2a015dcb25 Add Beginning-of-Sentence info in UnigramProperty.
Bug: 14119293
Bug: 14425059
Change-Id: I8a894352568377d32468e5563f4e89af00d22944
2014-05-23 00:07:14 +09:00
Keisuke Kuroyanagi 9f8c9a0161 Use PrevWordsInfo to add/remove n(bi)-gram in native code.
Bug: 14119293
Bug: 14425059
Change-Id: I4b9a46bfd670b35195418eaee51456d44fb91b6d
2014-05-21 18:30:34 +09:00
Keisuke Kuroyanagi fa7db65dec Support multiple previous words in DicNode.
Bug: 14425059
Change-Id: Ib8682befe4d7d9fe5122eb538e7c804f75ded463
2014-05-19 11:47:10 +09:00
Keisuke Kuroyanagi 45d1a936a7 Move prev word related logic to PrevWordsInfo.
Bug: 14119293
Bug: 14425059
Change-Id: I1bbd7ab4ace2c475f27bc468cb7b4d67e1ae2f9f
2014-05-16 22:31:45 +09:00
Keisuke Kuroyanagi b94ec1437b Introduce MAX_PREV_WORD_COUNT_FOR_N_GRAM.
Bug: 14425059
Change-Id: I587f90df026a14359d2ee452040bbfaf02c1ae51
2014-05-15 18:45:49 +09:00
Keisuke Kuroyanagi b87fffb8be Refactoring: Move prev word information into PrevWordsInfo.
Bug: 14119293
Bug: 14425059

Change-Id: I5a5f81c6b189e3ad1da093343a3121463f87c21c
2014-05-15 18:39:23 +09:00
Keisuke Kuroyanagi d6fdd58cc0 Cleanup DicTraverseSession.
Change-Id: I953665e1958e7214e9365e2bc9d3f3843016dc81
2014-05-15 17:41:23 +09:00
Keisuke Kuroyanagi e810a266fd Merge "Quit use bigram probability diff for ver4 dict." 2014-05-15 07:07:25 +00:00
Keisuke Kuroyanagi 64341927d2 Quit use bigram probability diff for ver4 dict.
Change-Id: I2cfcfbcf351877d1dff466a24974dbb05908f14e
2014-05-15 16:02:58 +09:00
Keisuke Kuroyanagi a709bc7f93 Merge "Reduce the cost for handling intentional omission." 2014-05-15 03:31:39 +00:00
Keisuke Kuroyanagi 39e5a15af0 Merge "Always keep PtNodes that represent non-word info during GC." 2014-05-15 02:38:57 +00:00
Keisuke Kuroyanagi 91d7538225 Merge "Add VERSION_4_DEV(402) in native code." 2014-05-14 13:40:10 +00:00
Keisuke Kuroyanagi 7116ea98f4 Add VERSION_4_DEV(402) in native code.
Bug: 13406708
Change-Id: I96cfacf524d670a6a5637a96a63bcd47aaf09ca0
2014-05-14 22:37:05 +09:00
Keisuke Kuroyanagi cf700695ea Always keep PtNodes that represent non-word info during GC.
Bug: 14119293
Change-Id: Ie50b341a512be6a6efe3e05b60e1d09560b01d1f
2014-05-14 22:29:21 +09:00
Keisuke Kuroyanagi 2dc8e36f8e Use nullptr instead of MmappedBufferPtr(nullptr).
Change-Id: I9bc35f723f1c105ef6155a49630b8cd492900bc8
2014-05-14 22:05:00 +09:00
Keisuke Kuroyanagi 32bdf9f344 Merge "Use JniDataUtils::putXxxToArray() to output a value." 2014-05-14 11:12:09 +00:00
Keisuke Kuroyanagi 5c00bde9bc Merge "Filter out control characters and non-unicode code points." 2014-05-14 11:07:32 +00:00