1 /* 2 * Copyright (C) 2012 The Android Open Source Project 3 * 4 * Licensed under the Apache License, Version 2.0 (the "License"); you may not 5 * use this file except in compliance with the License. You may obtain a copy of 6 * the License at 7 * 8 * http://www.apache.org/licenses/LICENSE-2.0 9 * 10 * Unless required by applicable law or agreed to in writing, software 11 * distributed under the License is distributed on an "AS IS" BASIS, WITHOUT 12 * WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. See the 13 * License for the specific language governing permissions and limitations under 14 * the License. 15 */ 16 17 package com.android.inputmethod.latin.makedict; 18 19 import com.android.inputmethod.latin.Constants; 20 import com.android.inputmethod.latin.makedict.FusionDictionary.DictionaryOptions; 21 22 /** 23 * Dictionary File Format Specification. 24 */ 25 public final class FormatSpec { 26 27 /* 28 * Array of Node(FusionDictionary.Node) layout is as follows: 29 * 30 * g | 31 * r | the number of groups, 1 or 2 bytes. 32 * o | 1 byte = bbbbbbbb match 33 * u | case 1xxxxxxx => xxxxxxx << 8 + next byte 34 * p | otherwise => bbbbbbbb 35 * c | 36 * ount 37 * 38 * g | 39 * r | sequence of groups, 40 * o | the layout of each group is described below. 41 * u | 42 * ps 43 * 44 * f | 45 * o | IF SUPPORTS_DYNAMIC_UPDATE (defined in the file header) 46 * r | forward link address, 3byte 47 * w | 1 byte = bbbbbbbb match 48 * a | case 1xxxxxxx => -((xxxxxxx << 16) + (next byte << 8) + next byte) 49 * r | otherwise => (xxxxxxx << 16) + (next byte << 8) + next byte 50 * d | 51 * linkaddress 52 */ 53 54 /* Node(CharGroup) layout is as follows: 55 * | IF !SUPPORTS_DYNAMIC_UPDATE 56 * | addressType xx : mask with MASK_GROUP_ADDRESS_TYPE 57 * | 2 bits, 00 = no children : FLAG_GROUP_ADDRESS_TYPE_NOADDRESS 58 * f | 01 = 1 byte : FLAG_GROUP_ADDRESS_TYPE_ONEBYTE 59 * l | 10 = 2 bytes : FLAG_GROUP_ADDRESS_TYPE_TWOBYTES 60 * a | 11 = 3 bytes : FLAG_GROUP_ADDRESS_TYPE_THREEBYTES 61 * g | ELSE 62 * s | is moved ? 2 bits, 11 = no 63 * | 01 = yes 64 * | the new address is stored in the same place as the parent address 65 * | has several chars ? 1 bit, 1 = yes, 0 = no : FLAG_HAS_MULTIPLE_CHARS 66 * | has a terminal ? 1 bit, 1 = yes, 0 = no : FLAG_IS_TERMINAL 67 * | has shortcut targets ? 1 bit, 1 = yes, 0 = no : FLAG_HAS_SHORTCUT_TARGETS 68 * | has bigrams ? 1 bit, 1 = yes, 0 = no : FLAG_HAS_BIGRAMS 69 * | is not a word ? 1 bit, 1 = yes, 0 = no : FLAG_IS_NOT_A_WORD 70 * | is blacklisted ? 1 bit, 1 = yes, 0 = no : FLAG_IS_BLACKLISTED 71 * 72 * p | 73 * a | IF SUPPORTS_DYNAMIC_UPDATE (defined in the file header) 74 * r | parent address, 3byte 75 * e | 1 byte = bbbbbbbb match 76 * n | case 1xxxxxxx => -((0xxxxxxx << 16) + (next byte << 8) + next byte) 77 * t | otherwise => (bbbbbbbb << 16) + (next byte << 8) + next byte 78 * a | 79 * ddress 80 * 81 * c | IF FLAG_HAS_MULTIPLE_CHARS 82 * h | char, char, char, char n * (1 or 3 bytes) : use CharGroupInfo for i/o helpers 83 * a | end 1 byte, = 0 84 * r | ELSE 85 * s | char 1 or 3 bytes 86 * | END 87 * 88 * f | 89 * r | IF FLAG_IS_TERMINAL 90 * e | frequency 1 byte 91 * q | 92 * 93 * c | IF 00 = FLAG_GROUP_ADDRESS_TYPE_NOADDRESS = addressType 94 * h | // nothing 95 * i | ELSIF 01 = FLAG_GROUP_ADDRESS_TYPE_ONEBYTE == addressType 96 * l | children address, 1 byte 97 * d | ELSIF 10 = FLAG_GROUP_ADDRESS_TYPE_TWOBYTES == addressType 98 * r | children address, 2 bytes 99 * e | ELSE // 11 = FLAG_GROUP_ADDRESS_TYPE_THREEBYTES = addressType 100 * n | children address, 3 bytes 101 * A | END 102 * d 103 * dress 104 * 105 * | IF FLAG_IS_TERMINAL && FLAG_HAS_SHORTCUT_TARGETS 106 * | shortcut string list 107 * | IF FLAG_IS_TERMINAL && FLAG_HAS_BIGRAMS 108 * | bigrams address list 109 * 110 * Char format is: 111 * 1 byte = bbbbbbbb match 112 * case 000xxxxx: xxxxx << 16 + next byte << 8 + next byte 113 * else: if 00011111 (= 0x1F) : this is the terminator. This is a relevant choice because 114 * unicode code points range from 0 to 0x10FFFF, so any 3-byte value starting with 115 * 00011111 would be outside unicode. 116 * else: iso-latin-1 code 117 * This allows for the whole unicode range to be encoded, including chars outside of 118 * the BMP. Also everything in the iso-latin-1 charset is only 1 byte, except control 119 * characters which should never happen anyway (and still work, but take 3 bytes). 120 * 121 * bigram address list is: 122 * <flags> = | hasNext = 1 bit, 1 = yes, 0 = no : FLAG_ATTRIBUTE_HAS_NEXT 123 * | addressSign = 1 bit, : FLAG_ATTRIBUTE_OFFSET_NEGATIVE 124 * | 1 = must take -address, 0 = must take +address 125 * | xx : mask with MASK_ATTRIBUTE_ADDRESS_TYPE 126 * | addressFormat = 2 bits, 00 = unused : FLAG_ATTRIBUTE_ADDRESS_TYPE_ONEBYTE 127 * | 01 = 1 byte : FLAG_ATTRIBUTE_ADDRESS_TYPE_ONEBYTE 128 * | 10 = 2 bytes : FLAG_ATTRIBUTE_ADDRESS_TYPE_TWOBYTES 129 * | 11 = 3 bytes : FLAG_ATTRIBUTE_ADDRESS_TYPE_THREEBYTES 130 * | 4 bits : frequency : mask with FLAG_ATTRIBUTE_FREQUENCY 131 * <address> | IF (01 == FLAG_ATTRIBUTE_ADDRESS_TYPE_ONEBYTE == addressFormat) 132 * | read 1 byte, add top 4 bits 133 * | ELSIF (10 == FLAG_ATTRIBUTE_ADDRESS_TYPE_TWOBYTES == addressFormat) 134 * | read 2 bytes, add top 4 bits 135 * | ELSE // 11 == FLAG_ATTRIBUTE_ADDRESS_TYPE_THREEBYTES == addressFormat 136 * | read 3 bytes, add top 4 bits 137 * | END 138 * | if (FLAG_ATTRIBUTE_OFFSET_NEGATIVE) then address = -address 139 * if (FLAG_ATTRIBUTE_HAS_NEXT) goto bigram_and_shortcut_address_list_is 140 * 141 * shortcut string list is: 142 * <byte size> = GROUP_SHORTCUT_LIST_SIZE_SIZE bytes, big-endian: size of the list, in bytes. 143 * <flags> = | hasNext = 1 bit, 1 = yes, 0 = no : FLAG_ATTRIBUTE_HAS_NEXT 144 * | reserved = 3 bits, must be 0 145 * | 4 bits : frequency : mask with FLAG_ATTRIBUTE_FREQUENCY 146 * <shortcut> = | string of characters at the char format described above, with the terminator 147 * | used to signal the end of the string. 148 * if (FLAG_ATTRIBUTE_HAS_NEXT goto flags 149 */ 150 151 static final int VERSION_1_MAGIC_NUMBER = 0x78B1; 152 public static final int VERSION_2_MAGIC_NUMBER = 0x9BC13AFE; 153 static final int MINIMUM_SUPPORTED_VERSION = 1; 154 static final int MAXIMUM_SUPPORTED_VERSION = 3; 155 static final int NOT_A_VERSION_NUMBER = -1; 156 static final int FIRST_VERSION_WITH_HEADER_SIZE = 2; 157 static final int FIRST_VERSION_WITH_DYNAMIC_UPDATE = 3; 158 159 // These options need to be the same numeric values as the one in the native reading code. 160 static final int GERMAN_UMLAUT_PROCESSING_FLAG = 0x1; 161 // TODO: Make the native reading code read this variable. 162 static final int SUPPORTS_DYNAMIC_UPDATE = 0x2; 163 static final int FRENCH_LIGATURE_PROCESSING_FLAG = 0x4; 164 static final int CONTAINS_BIGRAMS_FLAG = 0x8; 165 166 // TODO: Make this value adaptative to content data, store it in the header, and 167 // use it in the reading code. 168 static final int MAX_WORD_LENGTH = Constants.Dictionary.MAX_WORD_LENGTH; 169 170 static final int PARENT_ADDRESS_SIZE = 3; 171 static final int FORWARD_LINK_ADDRESS_SIZE = 3; 172 173 static final int MASK_GROUP_ADDRESS_TYPE = 0xC0; 174 static final int FLAG_GROUP_ADDRESS_TYPE_NOADDRESS = 0x00; 175 static final int FLAG_GROUP_ADDRESS_TYPE_ONEBYTE = 0x40; 176 static final int FLAG_GROUP_ADDRESS_TYPE_TWOBYTES = 0x80; 177 static final int FLAG_GROUP_ADDRESS_TYPE_THREEBYTES = 0xC0; 178 179 static final int FLAG_HAS_MULTIPLE_CHARS = 0x20; 180 181 static final int FLAG_IS_TERMINAL = 0x10; 182 static final int FLAG_HAS_SHORTCUT_TARGETS = 0x08; 183 static final int FLAG_HAS_BIGRAMS = 0x04; 184 static final int FLAG_IS_NOT_A_WORD = 0x02; 185 static final int FLAG_IS_BLACKLISTED = 0x01; 186 static final int FLAG_IS_MOVED = 0x40; 187 188 static final int FLAG_ATTRIBUTE_HAS_NEXT = 0x80; 189 static final int FLAG_ATTRIBUTE_OFFSET_NEGATIVE = 0x40; 190 static final int MASK_ATTRIBUTE_ADDRESS_TYPE = 0x30; 191 static final int FLAG_ATTRIBUTE_ADDRESS_TYPE_ONEBYTE = 0x10; 192 static final int FLAG_ATTRIBUTE_ADDRESS_TYPE_TWOBYTES = 0x20; 193 static final int FLAG_ATTRIBUTE_ADDRESS_TYPE_THREEBYTES = 0x30; 194 static final int FLAG_ATTRIBUTE_FREQUENCY = 0x0F; 195 196 static final int GROUP_CHARACTERS_TERMINATOR = 0x1F; 197 198 static final int GROUP_TERMINATOR_SIZE = 1; 199 static final int GROUP_FLAGS_SIZE = 1; 200 static final int GROUP_FREQUENCY_SIZE = 1; 201 static final int GROUP_MAX_ADDRESS_SIZE = 3; 202 static final int GROUP_ATTRIBUTE_FLAGS_SIZE = 1; 203 static final int GROUP_ATTRIBUTE_MAX_ADDRESS_SIZE = 3; 204 static final int GROUP_SHORTCUT_LIST_SIZE_SIZE = 2; 205 206 static final int NO_CHILDREN_ADDRESS = Integer.MIN_VALUE; 207 static final int NO_PARENT_ADDRESS = 0; 208 static final int NO_FORWARD_LINK_ADDRESS = 0; 209 static final int INVALID_CHARACTER = -1; 210 211 static final int MAX_CHARGROUPS_FOR_ONE_BYTE_CHARGROUP_COUNT = 0x7F; // 127 212 static final int MAX_CHARGROUPS_IN_A_NODE = 0x7FFF; // 32767 213 214 static final int MAX_TERMINAL_FREQUENCY = 255; 215 static final int MAX_BIGRAM_FREQUENCY = 15; 216 217 // This option needs to be the same numeric value as the one in binary_format.h. 218 static final int NOT_VALID_WORD = -99; 219 static final int SIGNED_CHILDREN_ADDRESS_SIZE = 3; 220 221 /** 222 * Options about file format. 223 */ 224 public static final class FormatOptions { 225 public final int mVersion; 226 public final boolean mSupportsDynamicUpdate; 227 public FormatOptions(final int version) { 228 this(version, false); 229 } 230 public FormatOptions(final int version, final boolean supportsDynamicUpdate) { 231 mVersion = version; 232 if (version < FIRST_VERSION_WITH_DYNAMIC_UPDATE && supportsDynamicUpdate) { 233 throw new RuntimeException("Dynamic updates are only supported with versions " 234 + FIRST_VERSION_WITH_DYNAMIC_UPDATE + " and ulterior."); 235 } 236 mSupportsDynamicUpdate = supportsDynamicUpdate; 237 } 238 } 239 240 /** 241 * Class representing file header. 242 */ 243 static final class FileHeader { 244 public final int mHeaderSize; 245 public final DictionaryOptions mDictionaryOptions; 246 public final FormatOptions mFormatOptions; 247 public FileHeader(final int headerSize, final DictionaryOptions dictionaryOptions, 248 final FormatOptions formatOptions) { 249 mHeaderSize = headerSize; 250 mDictionaryOptions = dictionaryOptions; 251 mFormatOptions = formatOptions; 252 } 253 } 254 255 private FormatSpec() { 256 // This utility class is not publicly instantiable. 257 } 258 } 259