ScriptUtils.java revision a9c12692ed11a4712ceb743f446e51a735eeabb6
1/* 2 * Copyright (C) 2012 The Android Open Source Project 3 * 4 * Licensed under the Apache License, Version 2.0 (the "License"); 5 * you may not use this file except in compliance with the License. 6 * You may obtain a copy of 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, 12 * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. 13 * See the License for the specific language governing permissions and 14 * limitations under the License. 15 */ 16 17package com.android.inputmethod.latin.utils; 18 19import java.util.Locale; 20import java.util.TreeMap; 21 22/** 23 * A class to help with handling different writing scripts. 24 */ 25public class ScriptUtils { 26 // Used for hardware keyboards 27 public static final int SCRIPT_UNKNOWN = -1; 28 // TODO: should we use ISO 15924 identifiers instead? 29 public static final int SCRIPT_LATIN = 0; 30 public static final int SCRIPT_CYRILLIC = 1; 31 public static final int SCRIPT_GREEK = 2; 32 public static final int SCRIPT_ARABIC = 3; 33 public static final int SCRIPT_HEBREW = 4; 34 public static final int SCRIPT_ARMENIAN = 5; 35 public static final int SCRIPT_GEORGIAN = 6; 36 public static final int SCRIPT_KHMER = 7; 37 public static final int SCRIPT_LAO = 8; 38 public static final TreeMap<String, Integer> mSpellCheckerLanguageToScript; 39 static { 40 // List of the supported languages and their associated script. We won't check 41 // words written in another script than the selected script, because we know we 42 // don't have those in our dictionary so we will underline everything and we 43 // will never have any suggestions, so it makes no sense checking them, and this 44 // is done in {@link #shouldFilterOut}. Also, the script is used to choose which 45 // proximity to pass to the dictionary descent algorithm. 46 // IMPORTANT: this only contains languages - do not write countries in there. 47 // Only the language is searched from the map. 48 mSpellCheckerLanguageToScript = new TreeMap<>(); 49 mSpellCheckerLanguageToScript.put("cs", SCRIPT_LATIN); 50 mSpellCheckerLanguageToScript.put("da", SCRIPT_LATIN); 51 mSpellCheckerLanguageToScript.put("de", SCRIPT_LATIN); 52 mSpellCheckerLanguageToScript.put("el", SCRIPT_GREEK); 53 mSpellCheckerLanguageToScript.put("en", SCRIPT_LATIN); 54 mSpellCheckerLanguageToScript.put("es", SCRIPT_LATIN); 55 mSpellCheckerLanguageToScript.put("fi", SCRIPT_LATIN); 56 mSpellCheckerLanguageToScript.put("fr", SCRIPT_LATIN); 57 mSpellCheckerLanguageToScript.put("hr", SCRIPT_LATIN); 58 mSpellCheckerLanguageToScript.put("it", SCRIPT_LATIN); 59 mSpellCheckerLanguageToScript.put("lt", SCRIPT_LATIN); 60 mSpellCheckerLanguageToScript.put("lv", SCRIPT_LATIN); 61 mSpellCheckerLanguageToScript.put("nb", SCRIPT_LATIN); 62 mSpellCheckerLanguageToScript.put("nl", SCRIPT_LATIN); 63 mSpellCheckerLanguageToScript.put("pt", SCRIPT_LATIN); 64 mSpellCheckerLanguageToScript.put("sl", SCRIPT_LATIN); 65 mSpellCheckerLanguageToScript.put("ru", SCRIPT_CYRILLIC); 66 } 67 /* 68 * Returns whether the code point is a letter that makes sense for the specified 69 * locale for this spell checker. 70 * The dictionaries supported by Latin IME are described in res/xml/spellchecker.xml 71 * and is limited to EFIGS languages and Russian. 72 * Hence at the moment this explicitly tests for Cyrillic characters or Latin characters 73 * as appropriate, and explicitly excludes CJK, Arabic and Hebrew characters. 74 */ 75 public static boolean isLetterPartOfScript(final int codePoint, final int scriptId) { 76 switch (scriptId) { 77 case SCRIPT_LATIN: 78 // Our supported latin script dictionaries (EFIGS) at the moment only include 79 // characters in the C0, C1, Latin Extended A and B, IPA extensions unicode 80 // blocks. As it happens, those are back-to-back in the code range 0x40 to 0x2AF, 81 // so the below is a very efficient way to test for it. As for the 0-0x3F, it's 82 // excluded from isLetter anyway. 83 return codePoint <= 0x2AF && Character.isLetter(codePoint); 84 case SCRIPT_CYRILLIC: 85 // All Cyrillic characters are in the 400~52F block. There are some in the upper 86 // Unicode range, but they are archaic characters that are not used in modern 87 // Russian and are not used by our dictionary. 88 return codePoint >= 0x400 && codePoint <= 0x52F && Character.isLetter(codePoint); 89 case SCRIPT_GREEK: 90 // Greek letters are either in the 370~3FF range (Greek & Coptic), or in the 91 // 1F00~1FFF range (Greek extended). Our dictionary contains both sort of characters. 92 // Our dictionary also contains a few words with 0xF2; it would be best to check 93 // if that's correct, but a web search does return results for these words so 94 // they are probably okay. 95 return (codePoint >= 0x370 && codePoint <= 0x3FF) 96 || (codePoint >= 0x1F00 && codePoint <= 0x1FFF) 97 || codePoint == 0xF2; 98 case SCRIPT_ARABIC: 99 // Arabic letters can be in any of the following blocks: 100 // Arabic U+0600..U+06FF 101 // Arabic Supplement U+0750..U+077F 102 // Arabic Extended-A U+08A0..U+08FF 103 // Arabic Presentation Forms-A U+FB50..U+FDFF 104 // Arabic Presentation Forms-B U+FE70..U+FEFF 105 return (codePoint >= 0x600 && codePoint <= 0x6FF) 106 || (codePoint >= 0x750 && codePoint <= 0x77F) 107 || (codePoint >= 0x8A0 && codePoint <= 0x8FF) 108 || (codePoint >= 0xFB50 && codePoint <= 0xFDFF) 109 || (codePoint >= 0xFE70 && codePoint <= 0xFEFF); 110 case SCRIPT_HEBREW: 111 // Hebrew letters are in the Hebrew unicode block, which spans from U+0590 to U+05FF, 112 // or in the Alphabetic Presentation Forms block, U+FB00..U+FB4F, but only in the 113 // Hebrew part of that block, which is U+FB1D..U+FB4F. 114 return (codePoint >= 0x590 && codePoint <= 0x5FF 115 || codePoint >= 0xFB1D && codePoint <= 0xFB4F); 116 case SCRIPT_ARMENIAN: 117 // Armenian letters are in the Armenian unicode block, U+0530..U+058F and 118 // Alphabetic Presentation Forms block, U+FB00..U+FB4F, but only in the Armenian part 119 // of that block, which is U+FB13..U+FB17. 120 return (codePoint >= 0x530 && codePoint <= 0x58F 121 || codePoint >= 0xFB13 && codePoint <= 0xFB17); 122 case SCRIPT_GEORGIAN: 123 // Georgian letters are in the Georgian unicode block, U+10A0..U+10FF, 124 // or Georgian supplement block, U+2D00..U+2D2F 125 return (codePoint >= 0x10A0 && codePoint <= 0x10FF 126 || codePoint >= 0x2D00 && codePoint <= 0x2D2F); 127 case SCRIPT_KHMER: 128 // Khmer letters are in unicode block U+1780..U+17FF, and the Khmer symbols block 129 // is U+19E0..U+19FF 130 return (codePoint >= 0x1780 && codePoint <= 0x17FF 131 || codePoint >= 0x19E0 && codePoint <= 0x19FF); 132 case SCRIPT_LAO: 133 // The Lao block is U+0E80..U+0EFF 134 return (codePoint >= 0xE80 && codePoint <= 0xEFF); 135 case SCRIPT_UNKNOWN: 136 return true; 137 default: 138 // Should never come here 139 throw new RuntimeException("Impossible value of script: " + scriptId); 140 } 141 } 142 143 public static int getScriptFromSpellCheckerLocale(final Locale locale) { 144 final Integer script = mSpellCheckerLanguageToScript.get(locale.getLanguage()); 145 if (null == script) { 146 throw new RuntimeException("We have been called with an unsupported language: \"" 147 + locale.getLanguage() + "\". Framework bug?"); 148 } 149 return script; 150 } 151} 152