12fa3693c264a4c150ac307d9bb7f6f8f18cc4ffcKen Wakasa/*
22fa3693c264a4c150ac307d9bb7f6f8f18cc4ffcKen Wakasa * Copyright (C) 2013, The Android Open Source Project
32fa3693c264a4c150ac307d9bb7f6f8f18cc4ffcKen Wakasa *
42fa3693c264a4c150ac307d9bb7f6f8f18cc4ffcKen Wakasa * Licensed under the Apache License, Version 2.0 (the "License");
52fa3693c264a4c150ac307d9bb7f6f8f18cc4ffcKen Wakasa * you may not use this file except in compliance with the License.
62fa3693c264a4c150ac307d9bb7f6f8f18cc4ffcKen Wakasa * You may obtain a copy of the License at
72fa3693c264a4c150ac307d9bb7f6f8f18cc4ffcKen Wakasa *
82fa3693c264a4c150ac307d9bb7f6f8f18cc4ffcKen Wakasa *     http://www.apache.org/licenses/LICENSE-2.0
92fa3693c264a4c150ac307d9bb7f6f8f18cc4ffcKen Wakasa *
102fa3693c264a4c150ac307d9bb7f6f8f18cc4ffcKen Wakasa * Unless required by applicable law or agreed to in writing, software
112fa3693c264a4c150ac307d9bb7f6f8f18cc4ffcKen Wakasa * distributed under the License is distributed on an "AS IS" BASIS,
122fa3693c264a4c150ac307d9bb7f6f8f18cc4ffcKen Wakasa * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
132fa3693c264a4c150ac307d9bb7f6f8f18cc4ffcKen Wakasa * See the License for the specific language governing permissions and
142fa3693c264a4c150ac307d9bb7f6f8f18cc4ffcKen Wakasa * limitations under the License.
152fa3693c264a4c150ac307d9bb7f6f8f18cc4ffcKen Wakasa */
162fa3693c264a4c150ac307d9bb7f6f8f18cc4ffcKen Wakasa
1788bc312ad34321fb3e81be2dc939a889d065f4a7Keisuke Kuroyanagi#include "dictionary/structure/v4/ver4_dict_constants.h"
182fa3693c264a4c150ac307d9bb7f6f8f18cc4ffcKen Wakasa
192fa3693c264a4c150ac307d9bb7f6f8f18cc4ffcKen Wakasanamespace latinime {
202fa3693c264a4c150ac307d9bb7f6f8f18cc4ffcKen Wakasa
212ac934296c0571ea252f3fb5a23fba29eb89c666Keisuke Kuroyanagiconst char *const Ver4DictConstants::BODY_FILE_EXTENSION = ".body";
222fa3693c264a4c150ac307d9bb7f6f8f18cc4ffcKen Wakasaconst char *const Ver4DictConstants::HEADER_FILE_EXTENSION = ".header";
232fa3693c264a4c150ac307d9bb7f6f8f18cc4ffcKen Wakasa
242fa3693c264a4c150ac307d9bb7f6f8f18cc4ffcKen Wakasa// Version 4 dictionary size is implicitly limited to 8MB due to 3-byte offsets.
25a245d15da5d295af21ead9a01583c64796a31ad7Jean Chalardconst int Ver4DictConstants::MAX_DICTIONARY_SIZE = 8 * 1024 * 1024;
262fa3693c264a4c150ac307d9bb7f6f8f18cc4ffcKen Wakasa// Extended region size, which is not GCed region size in dict file + additional buffer size, is
272fa3693c264a4c150ac307d9bb7f6f8f18cc4ffcKen Wakasa// limited to 1MB to prevent from inefficient traversing.
282fa3693c264a4c150ac307d9bb7f6f8f18cc4ffcKen Wakasaconst int Ver4DictConstants::MAX_DICT_EXTENDED_REGION_SIZE = 1 * 1024 * 1024;
292fa3693c264a4c150ac307d9bb7f6f8f18cc4ffcKen Wakasa
30851e0458fe460526b1f953e39a1e406a21ab4647Keisuke Kuroyanagi// NUM_OF_BUFFERS_FOR_SINGLE_DICT_CONTENT for Trie and TerminalAddressLookupTable.
31851e0458fe460526b1f953e39a1e406a21ab4647Keisuke Kuroyanagi// NUM_OF_BUFFERS_FOR_LANGUAGE_MODEL_DICT_CONTENT for language model.
32fe395232d69df0887863c1cbabe63def2586d29eKeisuke Kuroyanagi// NUM_OF_BUFFERS_FOR_SPARSE_TABLE_DICT_CONTENT for shortcut.
332ac934296c0571ea252f3fb5a23fba29eb89c666Keisuke Kuroyanagiconst size_t Ver4DictConstants::NUM_OF_CONTENT_BUFFERS_IN_BODY_FILE =
34851e0458fe460526b1f953e39a1e406a21ab4647Keisuke Kuroyanagi        NUM_OF_BUFFERS_FOR_SINGLE_DICT_CONTENT * 2
35c4696b2eb6b25eea4d5c869683104ab99aec0421Keisuke Kuroyanagi                + NUM_OF_BUFFERS_FOR_LANGUAGE_MODEL_DICT_CONTENT
36fe395232d69df0887863c1cbabe63def2586d29eKeisuke Kuroyanagi                + NUM_OF_BUFFERS_FOR_SPARSE_TABLE_DICT_CONTENT;
372ac934296c0571ea252f3fb5a23fba29eb89c666Keisuke Kuroyanagiconst int Ver4DictConstants::TRIE_BUFFER_INDEX = 0;
382ac934296c0571ea252f3fb5a23fba29eb89c666Keisuke Kuroyanagiconst int Ver4DictConstants::TERMINAL_ADDRESS_LOOKUP_TABLE_BUFFER_INDEX =
392ac934296c0571ea252f3fb5a23fba29eb89c666Keisuke Kuroyanagi        TRIE_BUFFER_INDEX + NUM_OF_BUFFERS_FOR_SINGLE_DICT_CONTENT;
40c4696b2eb6b25eea4d5c869683104ab99aec0421Keisuke Kuroyanagiconst int Ver4DictConstants::LANGUAGE_MODEL_BUFFER_INDEX =
41851e0458fe460526b1f953e39a1e406a21ab4647Keisuke Kuroyanagi        TERMINAL_ADDRESS_LOOKUP_TABLE_BUFFER_INDEX + NUM_OF_BUFFERS_FOR_SINGLE_DICT_CONTENT;
422ac934296c0571ea252f3fb5a23fba29eb89c666Keisuke Kuroyanagiconst int Ver4DictConstants::SHORTCUT_BUFFERS_INDEX =
43fe395232d69df0887863c1cbabe63def2586d29eKeisuke Kuroyanagi        LANGUAGE_MODEL_BUFFER_INDEX + NUM_OF_BUFFERS_FOR_LANGUAGE_MODEL_DICT_CONTENT;
442ac934296c0571ea252f3fb5a23fba29eb89c666Keisuke Kuroyanagi
452fa3693c264a4c150ac307d9bb7f6f8f18cc4ffcKen Wakasaconst int Ver4DictConstants::NOT_A_TERMINAL_ID = -1;
462fa3693c264a4c150ac307d9bb7f6f8f18cc4ffcKen Wakasaconst int Ver4DictConstants::PROBABILITY_SIZE = 1;
479a23f0fba25137760a60e9bfaf6bf20a5889648cKeisuke Kuroyanagiconst int Ver4DictConstants::FLAGS_IN_LANGUAGE_MODEL_SIZE = 1;
482fa3693c264a4c150ac307d9bb7f6f8f18cc4ffcKen Wakasaconst int Ver4DictConstants::TERMINAL_ADDRESS_TABLE_ADDRESS_SIZE = 3;
492fa3693c264a4c150ac307d9bb7f6f8f18cc4ffcKen Wakasaconst int Ver4DictConstants::NOT_A_TERMINAL_ADDRESS = 0;
502fa3693c264a4c150ac307d9bb7f6f8f18cc4ffcKen Wakasaconst int Ver4DictConstants::TERMINAL_ID_FIELD_SIZE = 4;
512fa3693c264a4c150ac307d9bb7f6f8f18cc4ffcKen Wakasaconst int Ver4DictConstants::TIME_STAMP_FIELD_SIZE = 4;
522383575d2d695efcca093e69ed2daa88aec58862Keisuke Kuroyanagiconst int Ver4DictConstants::WORD_LEVEL_FIELD_SIZE = 0;
532383575d2d695efcca093e69ed2daa88aec58862Keisuke Kuroyanagiconst int Ver4DictConstants::WORD_COUNT_FIELD_SIZE = 2;
542fa3693c264a4c150ac307d9bb7f6f8f18cc4ffcKen Wakasa
55623067a183caf62fbe33223675430a246b5ae13dKeisuke Kuroyanagiconst uint8_t Ver4DictConstants::FLAG_REPRESENTS_BEGINNING_OF_SENTENCE = 0x1;
564926b90ec530ba1e247b7a0f6edd719b2b01870bKeisuke Kuroyanagiconst uint8_t Ver4DictConstants::FLAG_NOT_A_VALID_ENTRY = 0x2;
577d911d6f91af56586fbca40672bfb77b494ee871Keisuke Kuroyanagiconst uint8_t Ver4DictConstants::FLAG_NOT_A_WORD = 0x4;
587d911d6f91af56586fbca40672bfb77b494ee871Keisuke Kuroyanagiconst uint8_t Ver4DictConstants::FLAG_BLACKLISTED = 0x8;
597d911d6f91af56586fbca40672bfb77b494ee871Keisuke Kuroyanagiconst uint8_t Ver4DictConstants::FLAG_POSSIBLY_OFFENSIVE = 0x10;
60623067a183caf62fbe33223675430a246b5ae13dKeisuke Kuroyanagi
61c1163c85187bf4deddb54c64f05e2f5ec97e08b4Keisuke Kuroyanagiconst int Ver4DictConstants::SHORTCUT_ADDRESS_TABLE_BLOCK_SIZE = 64;
622fa3693c264a4c150ac307d9bb7f6f8f18cc4ffcKen Wakasaconst int Ver4DictConstants::SHORTCUT_ADDRESS_TABLE_DATA_SIZE = 4;
632fa3693c264a4c150ac307d9bb7f6f8f18cc4ffcKen Wakasa
642fa3693c264a4c150ac307d9bb7f6f8f18cc4ffcKen Wakasaconst int Ver4DictConstants::SHORTCUT_FLAGS_FIELD_SIZE = 1;
652fa3693c264a4c150ac307d9bb7f6f8f18cc4ffcKen Wakasaconst int Ver4DictConstants::SHORTCUT_PROBABILITY_MASK = 0x0F;
662fa3693c264a4c150ac307d9bb7f6f8f18cc4ffcKen Wakasaconst int Ver4DictConstants::SHORTCUT_HAS_NEXT_MASK = 0x80;
672fa3693c264a4c150ac307d9bb7f6f8f18cc4ffcKen Wakasa
682ac934296c0571ea252f3fb5a23fba29eb89c666Keisuke Kuroyanagiconst size_t Ver4DictConstants::NUM_OF_BUFFERS_FOR_SINGLE_DICT_CONTENT = 1;
692ac934296c0571ea252f3fb5a23fba29eb89c666Keisuke Kuroyanagiconst size_t Ver4DictConstants::NUM_OF_BUFFERS_FOR_SPARSE_TABLE_DICT_CONTENT = 3;
706b0561f9d26215209e8e8895f5c35982af5158f0Keisuke Kuroyanagiconst size_t Ver4DictConstants::NUM_OF_BUFFERS_FOR_LANGUAGE_MODEL_DICT_CONTENT = 2;
712ac934296c0571ea252f3fb5a23fba29eb89c666Keisuke Kuroyanagi
722fa3693c264a4c150ac307d9bb7f6f8f18cc4ffcKen Wakasa} // namespace latinime
73