ScriptUtils.java revision 0dab3171d442a4d0acc87cc0019bfcbd4ea4123f 
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 TreeMap<String, Integer> mSpellCheckerLanguageToScript;
37    static {
38        // List of the supported languages and their associated script. We won't check
39        // words written in another script than the selected script, because we know we
40        // don't have those in our dictionary so we will underline everything and we
41        // will never have any suggestions, so it makes no sense checking them, and this
42        // is done in {@link #shouldFilterOut}. Also, the script is used to choose which
43        // proximity to pass to the dictionary descent algorithm.
44        // IMPORTANT: this only contains languages - do not write countries in there.
45        // Only the language is searched from the map.
46        mSpellCheckerLanguageToScript = new TreeMap<>();
47        mSpellCheckerLanguageToScript.put("cs", SCRIPT_LATIN);
48        mSpellCheckerLanguageToScript.put("da", SCRIPT_LATIN);
49        mSpellCheckerLanguageToScript.put("de", SCRIPT_LATIN);
50        mSpellCheckerLanguageToScript.put("el", SCRIPT_GREEK);
51        mSpellCheckerLanguageToScript.put("en", SCRIPT_LATIN);
52        mSpellCheckerLanguageToScript.put("es", SCRIPT_LATIN);
53        mSpellCheckerLanguageToScript.put("fi", SCRIPT_LATIN);
54        mSpellCheckerLanguageToScript.put("fr", SCRIPT_LATIN);
55        mSpellCheckerLanguageToScript.put("hr", SCRIPT_LATIN);
56        mSpellCheckerLanguageToScript.put("it", SCRIPT_LATIN);
57        mSpellCheckerLanguageToScript.put("lt", SCRIPT_LATIN);
58        mSpellCheckerLanguageToScript.put("lv", SCRIPT_LATIN);
59        mSpellCheckerLanguageToScript.put("nb", SCRIPT_LATIN);
60        mSpellCheckerLanguageToScript.put("nl", SCRIPT_LATIN);
61        mSpellCheckerLanguageToScript.put("pt", SCRIPT_LATIN);
62        mSpellCheckerLanguageToScript.put("sl", SCRIPT_LATIN);
63        mSpellCheckerLanguageToScript.put("ru", SCRIPT_CYRILLIC);
64    }
65    /*
66     * Returns whether the code point is a letter that makes sense for the specified
67     * locale for this spell checker.
68     * The dictionaries supported by Latin IME are described in res/xml/spellchecker.xml
69     * and is limited to EFIGS languages and Russian.
70     * Hence at the moment this explicitly tests for Cyrillic characters or Latin characters
71     * as appropriate, and explicitly excludes CJK, Arabic and Hebrew characters.
72     */
73    public static boolean isLetterPartOfScript(final int codePoint, final int scriptId) {
74        switch (scriptId) {
75        case SCRIPT_LATIN:
76            // Our supported latin script dictionaries (EFIGS) at the moment only include
77            // characters in the C0, C1, Latin Extended A and B, IPA extensions unicode
78            // blocks. As it happens, those are back-to-back in the code range 0x40 to 0x2AF,
79            // so the below is a very efficient way to test for it. As for the 0-0x3F, it's
80            // excluded from isLetter anyway.
81            return codePoint <= 0x2AF && Character.isLetter(codePoint);
82        case SCRIPT_CYRILLIC:
83            // All Cyrillic characters are in the 400~52F block. There are some in the upper
84            // Unicode range, but they are archaic characters that are not used in modern
85            // Russian and are not used by our dictionary.
86            return codePoint >= 0x400 && codePoint <= 0x52F && Character.isLetter(codePoint);
87        case SCRIPT_GREEK:
88            // Greek letters are either in the 370~3FF range (Greek & Coptic), or in the
89            // 1F00~1FFF range (Greek extended). Our dictionary contains both sort of characters.
90            // Our dictionary also contains a few words with 0xF2; it would be best to check
91            // if that's correct, but a web search does return results for these words so
92            // they are probably okay.
93            return (codePoint >= 0x370 && codePoint <= 0x3FF)
94                    || (codePoint >= 0x1F00 && codePoint <= 0x1FFF)
95                    || codePoint == 0xF2;
96        case SCRIPT_ARABIC:
97            // Arabic letters can be in any of the following blocks:
98            // Arabic U+0600..U+06FF
99            // Arabic Supplement U+0750..U+077F
100            // Arabic Extended-A U+08A0..U+08FF
101            // Arabic Presentation Forms-A U+FB50..U+FDFF
102            // Arabic Presentation Forms-B U+FE70..U+FEFF
103            return (codePoint >= 0x600 && codePoint <= 0x6FF)
104                    || (codePoint >= 0x750 && codePoint <= 0x77F)
105                    || (codePoint >= 0x8A0 && codePoint <= 0x8FF)
106                    || (codePoint >= 0xFB50 && codePoint <= 0xFDFF)
107                    || (codePoint >= 0xFE70 && codePoint <= 0xFEFF);
108        case SCRIPT_HEBREW:
109            // Hebrew letters are in the Hebrew unicode block, which spans from U+0590 to U+05FF,
110            // or in the Alphabetic Presentation Forms block, U+FB00..U+FB4F, but only in the
111            // Hebrew part of that block, which is U+FB1D..U+FB4F.
112            return (codePoint >= 0x590 && codePoint <= 0x5FF
113                    || codePoint >= 0xFB1D && codePoint <= 0xFB4F);
114        case SCRIPT_ARMENIAN:
115            // Armenian letters are in the Armenian unicode block, U+0530..U+058F and
116            // Alphabetic Presentation Forms block, U+FB00..U+FB4F, but only in the Armenian part
117            // of that block, which is U+FB13..U+FB17.
118            return (codePoint >= 0x530 && codePoint <= 0x58F
119                    || codePoint >= 0xFB13 && codePoint <= 0xFB17);
120        case SCRIPT_GEORGIAN:
121            // Georgian letters are in the Georgian unicode block, U+10A0..U+10FF,
122            // or Georgian supplement block, U+2D00..U+2D2F
123            return (codePoint >= 0x10A0 && codePoint <= 0x10FF
124                    || codePoint >= 0x2D00 && codePoint <= 0x2D2F);
125        case SCRIPT_UNKNOWN:
126            return true;
127        default:
128            // Should never come here
129            throw new RuntimeException("Impossible value of script: " + scriptId);
130        }
131    }
132
133    public static int getScriptFromSpellCheckerLocale(final Locale locale) {
134        final Integer script = mSpellCheckerLanguageToScript.get(locale.getLanguage());
135        if (null == script) {
136            throw new RuntimeException("We have been called with an unsupported language: \""
137                    + locale.getLanguage() + "\". Framework bug?");
138        }
139        return script;
140    }
141}
142