1cc8c8b99bd0463f5977dea82f5e2379ea1dd4e73Tadashi G. Takaoka/* 2cc8c8b99bd0463f5977dea82f5e2379ea1dd4e73Tadashi G. Takaoka * Copyright (C) 2012 The Android Open Source Project 3cc8c8b99bd0463f5977dea82f5e2379ea1dd4e73Tadashi G. Takaoka * 4cc8c8b99bd0463f5977dea82f5e2379ea1dd4e73Tadashi G. Takaoka * Licensed under the Apache License, Version 2.0 (the "License"); 5cc8c8b99bd0463f5977dea82f5e2379ea1dd4e73Tadashi G. Takaoka * you may not use this file except in compliance with the License. 6cc8c8b99bd0463f5977dea82f5e2379ea1dd4e73Tadashi G. Takaoka * You may obtain a copy of the License at 7cc8c8b99bd0463f5977dea82f5e2379ea1dd4e73Tadashi G. Takaoka * 8cc8c8b99bd0463f5977dea82f5e2379ea1dd4e73Tadashi G. Takaoka * http://www.apache.org/licenses/LICENSE-2.0 9cc8c8b99bd0463f5977dea82f5e2379ea1dd4e73Tadashi G. Takaoka * 10cc8c8b99bd0463f5977dea82f5e2379ea1dd4e73Tadashi G. Takaoka * Unless required by applicable law or agreed to in writing, software 11cc8c8b99bd0463f5977dea82f5e2379ea1dd4e73Tadashi G. Takaoka * distributed under the License is distributed on an "AS IS" BASIS, 12cc8c8b99bd0463f5977dea82f5e2379ea1dd4e73Tadashi G. Takaoka * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. 13cc8c8b99bd0463f5977dea82f5e2379ea1dd4e73Tadashi G. Takaoka * See the License for the specific language governing permissions and 14cc8c8b99bd0463f5977dea82f5e2379ea1dd4e73Tadashi G. Takaoka * limitations under the License. 15cc8c8b99bd0463f5977dea82f5e2379ea1dd4e73Tadashi G. Takaoka */ 16cc8c8b99bd0463f5977dea82f5e2379ea1dd4e73Tadashi G. Takaoka 174beeb9253a06482299e0c67467531d30436a02fcJean Chalardpackage com.android.inputmethod.latin.common; 18cc8c8b99bd0463f5977dea82f5e2379ea1dd4e73Tadashi G. Takaoka 192b39fab829ec93f7764d178167ac8d67e6e36491Satoshi Kataokaimport com.android.inputmethod.annotations.UsedForTesting; 20e28eba5074664d5716b8e58b8d0a235746b261ebKen Wakasa 21cc8c8b99bd0463f5977dea82f5e2379ea1dd4e73Tadashi G. Takaokaimport java.util.ArrayList; 22837cdd738b7ddbeac04b15230f01e44d247bd50aTadashi G. Takaokaimport java.util.Arrays; 2311d9ee742f8ff3fb31b0e3beb32ee4870c63d8e3Tadashi G. Takaokaimport java.util.Locale; 24cc8c8b99bd0463f5977dea82f5e2379ea1dd4e73Tadashi G. Takaoka 2580980574acfef74f9392da4fdbcba64d911cb66fTadashi G. Takaokaimport javax.annotation.Nonnull; 2680980574acfef74f9392da4fdbcba64d911cb66fTadashi G. Takaokaimport javax.annotation.Nullable; 2780980574acfef74f9392da4fdbcba64d911cb66fTadashi G. Takaoka 281e11c44d1b5f9ddf593c5407cb14c458be0056f2Tadashi G. Takaokapublic final class StringUtils { 297d3836d63a2eb4b79c4ad93cdae4f1f61cdb518eJean Chalard public static final int CAPITALIZE_NONE = 0; // No caps, or mixed case 307d3836d63a2eb4b79c4ad93cdae4f1f61cdb518eJean Chalard public static final int CAPITALIZE_FIRST = 1; // First only 317d3836d63a2eb4b79c4ad93cdae4f1f61cdb518eJean Chalard public static final int CAPITALIZE_ALL = 2; // All caps 327d3836d63a2eb4b79c4ad93cdae4f1f61cdb518eJean Chalard 3380980574acfef74f9392da4fdbcba64d911cb66fTadashi G. Takaoka @Nonnull 342fa3693c264a4c150ac307d9bb7f6f8f18cc4ffcKen Wakasa private static final String EMPTY_STRING = ""; 352fa3693c264a4c150ac307d9bb7f6f8f18cc4ffcKen Wakasa 365896153f931ac876ad667d73d7703289a63e9bf5Yohei Yukawa private static final char CHAR_LINE_FEED = 0X000A; 375896153f931ac876ad667d73d7703289a63e9bf5Yohei Yukawa private static final char CHAR_VERTICAL_TAB = 0X000B; 385896153f931ac876ad667d73d7703289a63e9bf5Yohei Yukawa private static final char CHAR_FORM_FEED = 0X000C; 395896153f931ac876ad667d73d7703289a63e9bf5Yohei Yukawa private static final char CHAR_CARRIAGE_RETURN = 0X000D; 405896153f931ac876ad667d73d7703289a63e9bf5Yohei Yukawa private static final char CHAR_NEXT_LINE = 0X0085; 415896153f931ac876ad667d73d7703289a63e9bf5Yohei Yukawa private static final char CHAR_LINE_SEPARATOR = 0X2028; 425896153f931ac876ad667d73d7703289a63e9bf5Yohei Yukawa private static final char CHAR_PARAGRAPH_SEPARATOR = 0X2029; 435896153f931ac876ad667d73d7703289a63e9bf5Yohei Yukawa 44cc8c8b99bd0463f5977dea82f5e2379ea1dd4e73Tadashi G. Takaoka private StringUtils() { 45cc8c8b99bd0463f5977dea82f5e2379ea1dd4e73Tadashi G. Takaoka // This utility class is not publicly instantiable. 46cc8c8b99bd0463f5977dea82f5e2379ea1dd4e73Tadashi G. Takaoka } 47cc8c8b99bd0463f5977dea82f5e2379ea1dd4e73Tadashi G. Takaoka 48ca724ef71f70044f48e2360e2627922be8ca24d5Jean Chalard // Taken from android.text.TextUtils. We are extensively using this method in many places, 49ca724ef71f70044f48e2360e2627922be8ca24d5Jean Chalard // some of which don't have the android libraries available. 50ca724ef71f70044f48e2360e2627922be8ca24d5Jean Chalard /** 51ca724ef71f70044f48e2360e2627922be8ca24d5Jean Chalard * Returns true if the string is null or 0-length. 52ca724ef71f70044f48e2360e2627922be8ca24d5Jean Chalard * @param str the string to be examined 53ca724ef71f70044f48e2360e2627922be8ca24d5Jean Chalard * @return true if str is null or zero length 54ca724ef71f70044f48e2360e2627922be8ca24d5Jean Chalard */ 5580980574acfef74f9392da4fdbcba64d911cb66fTadashi G. Takaoka public static boolean isEmpty(@Nullable final CharSequence str) { 5680980574acfef74f9392da4fdbcba64d911cb66fTadashi G. Takaoka return (str == null || str.length() == 0); 57ca724ef71f70044f48e2360e2627922be8ca24d5Jean Chalard } 58ca724ef71f70044f48e2360e2627922be8ca24d5Jean Chalard 59ca724ef71f70044f48e2360e2627922be8ca24d5Jean Chalard // Taken from android.text.TextUtils to cut the dependency to the Android framework. 60ca724ef71f70044f48e2360e2627922be8ca24d5Jean Chalard /** 61ca724ef71f70044f48e2360e2627922be8ca24d5Jean Chalard * Returns a string containing the tokens joined by delimiters. 6280980574acfef74f9392da4fdbcba64d911cb66fTadashi G. Takaoka * @param delimiter the delimiter 63ca724ef71f70044f48e2360e2627922be8ca24d5Jean Chalard * @param tokens an array objects to be joined. Strings will be formed from 64ca724ef71f70044f48e2360e2627922be8ca24d5Jean Chalard * the objects by calling object.toString(). 65ca724ef71f70044f48e2360e2627922be8ca24d5Jean Chalard */ 6680980574acfef74f9392da4fdbcba64d911cb66fTadashi G. Takaoka @Nonnull 6780980574acfef74f9392da4fdbcba64d911cb66fTadashi G. Takaoka public static String join(@Nonnull final CharSequence delimiter, 6880980574acfef74f9392da4fdbcba64d911cb66fTadashi G. Takaoka @Nonnull final Iterable<?> tokens) { 6980980574acfef74f9392da4fdbcba64d911cb66fTadashi G. Takaoka final StringBuilder sb = new StringBuilder(); 70ca724ef71f70044f48e2360e2627922be8ca24d5Jean Chalard boolean firstTime = true; 7180980574acfef74f9392da4fdbcba64d911cb66fTadashi G. Takaoka for (final Object token: tokens) { 72ca724ef71f70044f48e2360e2627922be8ca24d5Jean Chalard if (firstTime) { 73ca724ef71f70044f48e2360e2627922be8ca24d5Jean Chalard firstTime = false; 74ca724ef71f70044f48e2360e2627922be8ca24d5Jean Chalard } else { 75ca724ef71f70044f48e2360e2627922be8ca24d5Jean Chalard sb.append(delimiter); 76ca724ef71f70044f48e2360e2627922be8ca24d5Jean Chalard } 77ca724ef71f70044f48e2360e2627922be8ca24d5Jean Chalard sb.append(token); 78ca724ef71f70044f48e2360e2627922be8ca24d5Jean Chalard } 79ca724ef71f70044f48e2360e2627922be8ca24d5Jean Chalard return sb.toString(); 80ca724ef71f70044f48e2360e2627922be8ca24d5Jean Chalard } 81ca724ef71f70044f48e2360e2627922be8ca24d5Jean Chalard 82ca724ef71f70044f48e2360e2627922be8ca24d5Jean Chalard // Taken from android.text.TextUtils to cut the dependency to the Android framework. 83ca724ef71f70044f48e2360e2627922be8ca24d5Jean Chalard /** 84ca724ef71f70044f48e2360e2627922be8ca24d5Jean Chalard * Returns true if a and b are equal, including if they are both null. 85ca724ef71f70044f48e2360e2627922be8ca24d5Jean Chalard * <p><i>Note: In platform versions 1.1 and earlier, this method only worked well if 86ca724ef71f70044f48e2360e2627922be8ca24d5Jean Chalard * both the arguments were instances of String.</i></p> 87ca724ef71f70044f48e2360e2627922be8ca24d5Jean Chalard * @param a first CharSequence to check 88ca724ef71f70044f48e2360e2627922be8ca24d5Jean Chalard * @param b second CharSequence to check 89ca724ef71f70044f48e2360e2627922be8ca24d5Jean Chalard * @return true if a and b are equal 90ca724ef71f70044f48e2360e2627922be8ca24d5Jean Chalard */ 9180980574acfef74f9392da4fdbcba64d911cb66fTadashi G. Takaoka public static boolean equals(@Nullable final CharSequence a, @Nullable final CharSequence b) { 9280980574acfef74f9392da4fdbcba64d911cb66fTadashi G. Takaoka if (a == b) { 9380980574acfef74f9392da4fdbcba64d911cb66fTadashi G. Takaoka return true; 9480980574acfef74f9392da4fdbcba64d911cb66fTadashi G. Takaoka } 9580980574acfef74f9392da4fdbcba64d911cb66fTadashi G. Takaoka final int length; 96ca724ef71f70044f48e2360e2627922be8ca24d5Jean Chalard if (a != null && b != null && (length = a.length()) == b.length()) { 97ca724ef71f70044f48e2360e2627922be8ca24d5Jean Chalard if (a instanceof String && b instanceof String) { 98ca724ef71f70044f48e2360e2627922be8ca24d5Jean Chalard return a.equals(b); 9980980574acfef74f9392da4fdbcba64d911cb66fTadashi G. Takaoka } 10080980574acfef74f9392da4fdbcba64d911cb66fTadashi G. Takaoka for (int i = 0; i < length; i++) { 10180980574acfef74f9392da4fdbcba64d911cb66fTadashi G. Takaoka if (a.charAt(i) != b.charAt(i)) { 10280980574acfef74f9392da4fdbcba64d911cb66fTadashi G. Takaoka return false; 103ca724ef71f70044f48e2360e2627922be8ca24d5Jean Chalard } 104ca724ef71f70044f48e2360e2627922be8ca24d5Jean Chalard } 10580980574acfef74f9392da4fdbcba64d911cb66fTadashi G. Takaoka return true; 106ca724ef71f70044f48e2360e2627922be8ca24d5Jean Chalard } 107ca724ef71f70044f48e2360e2627922be8ca24d5Jean Chalard return false; 108ca724ef71f70044f48e2360e2627922be8ca24d5Jean Chalard } 109ca724ef71f70044f48e2360e2627922be8ca24d5Jean Chalard 11080980574acfef74f9392da4fdbcba64d911cb66fTadashi G. Takaoka public static int codePointCount(@Nullable final CharSequence text) { 11180980574acfef74f9392da4fdbcba64d911cb66fTadashi G. Takaoka if (isEmpty(text)) { 11280980574acfef74f9392da4fdbcba64d911cb66fTadashi G. Takaoka return 0; 11380980574acfef74f9392da4fdbcba64d911cb66fTadashi G. Takaoka } 114f95770354c1a0a662c4b00674f2b9029437be7eaJean Chalard return Character.codePointCount(text, 0, text.length()); 115cc8c8b99bd0463f5977dea82f5e2379ea1dd4e73Tadashi G. Takaoka } 116cc8c8b99bd0463f5977dea82f5e2379ea1dd4e73Tadashi G. Takaoka 11780980574acfef74f9392da4fdbcba64d911cb66fTadashi G. Takaoka @Nonnull 11880980574acfef74f9392da4fdbcba64d911cb66fTadashi G. Takaoka public static String newSingleCodePointString(final int codePoint) { 119c8383eda8bb05e25b22fe6d66100248f7d0e33faKen Wakasa if (Character.charCount(codePoint) == 1) { 12053fb86c447bc322c13212ae8eaa102f6a3e57604Keisuke Kuroyanagi // Optimization: avoid creating a temporary array for characters that are 121c8383eda8bb05e25b22fe6d66100248f7d0e33faKen Wakasa // represented by a single char value 122c8383eda8bb05e25b22fe6d66100248f7d0e33faKen Wakasa return String.valueOf((char) codePoint); 123c8383eda8bb05e25b22fe6d66100248f7d0e33faKen Wakasa } 124c8383eda8bb05e25b22fe6d66100248f7d0e33faKen Wakasa // For surrogate pair 125c8383eda8bb05e25b22fe6d66100248f7d0e33faKen Wakasa return new String(Character.toChars(codePoint)); 126c8383eda8bb05e25b22fe6d66100248f7d0e33faKen Wakasa } 127c8383eda8bb05e25b22fe6d66100248f7d0e33faKen Wakasa 12880980574acfef74f9392da4fdbcba64d911cb66fTadashi G. Takaoka public static boolean containsInArray(@Nonnull final String text, 12980980574acfef74f9392da4fdbcba64d911cb66fTadashi G. Takaoka @Nonnull final String[] array) { 130344af156744c6866090fb70f151efd66668c1e20Tadashi G. Takaoka for (final String element : array) { 13180980574acfef74f9392da4fdbcba64d911cb66fTadashi G. Takaoka if (text.equals(element)) { 13280980574acfef74f9392da4fdbcba64d911cb66fTadashi G. Takaoka return true; 13380980574acfef74f9392da4fdbcba64d911cb66fTadashi G. Takaoka } 134cc8c8b99bd0463f5977dea82f5e2379ea1dd4e73Tadashi G. Takaoka } 135cc8c8b99bd0463f5977dea82f5e2379ea1dd4e73Tadashi G. Takaoka return false; 136cc8c8b99bd0463f5977dea82f5e2379ea1dd4e73Tadashi G. Takaoka } 137cc8c8b99bd0463f5977dea82f5e2379ea1dd4e73Tadashi G. Takaoka 138c47bc30db84ecdf340a1faffafa2e9a9d5dd9d6aTadashi G. Takaoka /** 139c47bc30db84ecdf340a1faffafa2e9a9d5dd9d6aTadashi G. Takaoka * Comma-Splittable Text is similar to Comma-Separated Values (CSV) but has much simpler syntax. 140c47bc30db84ecdf340a1faffafa2e9a9d5dd9d6aTadashi G. Takaoka * Unlike CSV, Comma-Splittable Text has no escaping mechanism, so that the text can't contain 141c47bc30db84ecdf340a1faffafa2e9a9d5dd9d6aTadashi G. Takaoka * a comma character in it. 142c47bc30db84ecdf340a1faffafa2e9a9d5dd9d6aTadashi G. Takaoka */ 14380980574acfef74f9392da4fdbcba64d911cb66fTadashi G. Takaoka @Nonnull 144c47bc30db84ecdf340a1faffafa2e9a9d5dd9d6aTadashi G. Takaoka private static final String SEPARATOR_FOR_COMMA_SPLITTABLE_TEXT = ","; 1451b46738796eafa2dc062f4a93b0ea42023945737Tadashi G. Takaoka 14680980574acfef74f9392da4fdbcba64d911cb66fTadashi G. Takaoka public static boolean containsInCommaSplittableText(@Nonnull final String text, 14780980574acfef74f9392da4fdbcba64d911cb66fTadashi G. Takaoka @Nullable final String extraValues) { 148ca724ef71f70044f48e2360e2627922be8ca24d5Jean Chalard if (isEmpty(extraValues)) { 1491b46738796eafa2dc062f4a93b0ea42023945737Tadashi G. Takaoka return false; 1501b46738796eafa2dc062f4a93b0ea42023945737Tadashi G. Takaoka } 151c47bc30db84ecdf340a1faffafa2e9a9d5dd9d6aTadashi G. Takaoka return containsInArray(text, extraValues.split(SEPARATOR_FOR_COMMA_SPLITTABLE_TEXT)); 152344af156744c6866090fb70f151efd66668c1e20Tadashi G. Takaoka } 153344af156744c6866090fb70f151efd66668c1e20Tadashi G. Takaoka 15480980574acfef74f9392da4fdbcba64d911cb66fTadashi G. Takaoka @Nonnull 15580980574acfef74f9392da4fdbcba64d911cb66fTadashi G. Takaoka public static String removeFromCommaSplittableTextIfExists(@Nonnull final String text, 15680980574acfef74f9392da4fdbcba64d911cb66fTadashi G. Takaoka @Nullable final String extraValues) { 157ca724ef71f70044f48e2360e2627922be8ca24d5Jean Chalard if (isEmpty(extraValues)) { 1582fa3693c264a4c150ac307d9bb7f6f8f18cc4ffcKen Wakasa return EMPTY_STRING; 1591b46738796eafa2dc062f4a93b0ea42023945737Tadashi G. Takaoka } 160c47bc30db84ecdf340a1faffafa2e9a9d5dd9d6aTadashi G. Takaoka final String[] elements = extraValues.split(SEPARATOR_FOR_COMMA_SPLITTABLE_TEXT); 1611b46738796eafa2dc062f4a93b0ea42023945737Tadashi G. Takaoka if (!containsInArray(text, elements)) { 1621b46738796eafa2dc062f4a93b0ea42023945737Tadashi G. Takaoka return extraValues; 1631b46738796eafa2dc062f4a93b0ea42023945737Tadashi G. Takaoka } 164a91561aa58db1c43092c1caecc051a11fa5391c7Tadashi G. Takaoka final ArrayList<String> result = new ArrayList<>(elements.length - 1); 165344af156744c6866090fb70f151efd66668c1e20Tadashi G. Takaoka for (final String element : elements) { 1661b46738796eafa2dc062f4a93b0ea42023945737Tadashi G. Takaoka if (!text.equals(element)) { 1671b46738796eafa2dc062f4a93b0ea42023945737Tadashi G. Takaoka result.add(element); 1681b46738796eafa2dc062f4a93b0ea42023945737Tadashi G. Takaoka } 169344af156744c6866090fb70f151efd66668c1e20Tadashi G. Takaoka } 170ca724ef71f70044f48e2360e2627922be8ca24d5Jean Chalard return join(SEPARATOR_FOR_COMMA_SPLITTABLE_TEXT, result); 171cc8c8b99bd0463f5977dea82f5e2379ea1dd4e73Tadashi G. Takaoka } 172cc8c8b99bd0463f5977dea82f5e2379ea1dd4e73Tadashi G. Takaoka 173cc8c8b99bd0463f5977dea82f5e2379ea1dd4e73Tadashi G. Takaoka /** 174cc8c8b99bd0463f5977dea82f5e2379ea1dd4e73Tadashi G. Takaoka * Remove duplicates from an array of strings. 175cc8c8b99bd0463f5977dea82f5e2379ea1dd4e73Tadashi G. Takaoka * 176a4c7733cf7b5c0f970d1a8e52ee52b6199f56031Tadashi G. Takaoka * This method will always keep the first occurrence of all strings at their position 177cc8c8b99bd0463f5977dea82f5e2379ea1dd4e73Tadashi G. Takaoka * in the array, removing the subsequent ones. 178cc8c8b99bd0463f5977dea82f5e2379ea1dd4e73Tadashi G. Takaoka */ 17980980574acfef74f9392da4fdbcba64d911cb66fTadashi G. Takaoka public static void removeDupes(@Nonnull final ArrayList<String> suggestions) { 18080980574acfef74f9392da4fdbcba64d911cb66fTadashi G. Takaoka if (suggestions.size() < 2) { 18180980574acfef74f9392da4fdbcba64d911cb66fTadashi G. Takaoka return; 18280980574acfef74f9392da4fdbcba64d911cb66fTadashi G. Takaoka } 183cc8c8b99bd0463f5977dea82f5e2379ea1dd4e73Tadashi G. Takaoka int i = 1; 184cc8c8b99bd0463f5977dea82f5e2379ea1dd4e73Tadashi G. Takaoka // Don't cache suggestions.size(), since we may be removing items 185cc8c8b99bd0463f5977dea82f5e2379ea1dd4e73Tadashi G. Takaoka while (i < suggestions.size()) { 186bc464e2952e102219f0b977fc1e9140ad5bd03e4Tadashi G. Takaoka final String cur = suggestions.get(i); 187cc8c8b99bd0463f5977dea82f5e2379ea1dd4e73Tadashi G. Takaoka // Compare each suggestion with each previous suggestion 188cc8c8b99bd0463f5977dea82f5e2379ea1dd4e73Tadashi G. Takaoka for (int j = 0; j < i; j++) { 189bc464e2952e102219f0b977fc1e9140ad5bd03e4Tadashi G. Takaoka final String previous = suggestions.get(j); 190ca724ef71f70044f48e2360e2627922be8ca24d5Jean Chalard if (equals(cur, previous)) { 191adf218eed544f2239ca5394b8a6bcc542d89a4d9Jean Chalard suggestions.remove(i); 192cc8c8b99bd0463f5977dea82f5e2379ea1dd4e73Tadashi G. Takaoka i--; 193cc8c8b99bd0463f5977dea82f5e2379ea1dd4e73Tadashi G. Takaoka break; 194cc8c8b99bd0463f5977dea82f5e2379ea1dd4e73Tadashi G. Takaoka } 195cc8c8b99bd0463f5977dea82f5e2379ea1dd4e73Tadashi G. Takaoka } 196cc8c8b99bd0463f5977dea82f5e2379ea1dd4e73Tadashi G. Takaoka i++; 197cc8c8b99bd0463f5977dea82f5e2379ea1dd4e73Tadashi G. Takaoka } 198cc8c8b99bd0463f5977dea82f5e2379ea1dd4e73Tadashi G. Takaoka } 19911d9ee742f8ff3fb31b0e3beb32ee4870c63d8e3Tadashi G. Takaoka 20080980574acfef74f9392da4fdbcba64d911cb66fTadashi G. Takaoka @Nonnull 20180980574acfef74f9392da4fdbcba64d911cb66fTadashi G. Takaoka public static String capitalizeFirstCodePoint(@Nonnull final String s, 20280980574acfef74f9392da4fdbcba64d911cb66fTadashi G. Takaoka @Nonnull final Locale locale) { 203d20c5a78e2819c21ecc5dff04bc82e9078c79dd4Tadashi G. Takaoka if (s.length() <= 1) { 2046c47403e27fd703ece844f4b1b24632721da1772Tadashi G. Takaoka return s.toUpperCase(getLocaleUsedForToTitleCase(locale)); 205d20c5a78e2819c21ecc5dff04bc82e9078c79dd4Tadashi G. Takaoka } 20699b93d17d53c2d587c45373831b327f7851ec0a8Jean Chalard // Please refer to the comment below in 20799b93d17d53c2d587c45373831b327f7851ec0a8Jean Chalard // {@link #capitalizeFirstAndDowncaseRest(String,Locale)} as this has the same shortcomings 208d20c5a78e2819c21ecc5dff04bc82e9078c79dd4Tadashi G. Takaoka final int cutoff = s.offsetByCodePoints(0, 1); 2096c47403e27fd703ece844f4b1b24632721da1772Tadashi G. Takaoka return s.substring(0, cutoff).toUpperCase(getLocaleUsedForToTitleCase(locale)) 2106c47403e27fd703ece844f4b1b24632721da1772Tadashi G. Takaoka + s.substring(cutoff); 211d20c5a78e2819c21ecc5dff04bc82e9078c79dd4Tadashi G. Takaoka } 212d20c5a78e2819c21ecc5dff04bc82e9078c79dd4Tadashi G. Takaoka 21380980574acfef74f9392da4fdbcba64d911cb66fTadashi G. Takaoka @Nonnull 21480980574acfef74f9392da4fdbcba64d911cb66fTadashi G. Takaoka public static String capitalizeFirstAndDowncaseRest(@Nonnull final String s, 21580980574acfef74f9392da4fdbcba64d911cb66fTadashi G. Takaoka @Nonnull final Locale locale) { 21611d9ee742f8ff3fb31b0e3beb32ee4870c63d8e3Tadashi G. Takaoka if (s.length() <= 1) { 2176c47403e27fd703ece844f4b1b24632721da1772Tadashi G. Takaoka return s.toUpperCase(getLocaleUsedForToTitleCase(locale)); 21811d9ee742f8ff3fb31b0e3beb32ee4870c63d8e3Tadashi G. Takaoka } 21911d9ee742f8ff3fb31b0e3beb32ee4870c63d8e3Tadashi G. Takaoka // TODO: fix the bugs below 22011d9ee742f8ff3fb31b0e3beb32ee4870c63d8e3Tadashi G. Takaoka // - It does not work for Serbian, because it fails to account for the "lj" character, 22111d9ee742f8ff3fb31b0e3beb32ee4870c63d8e3Tadashi G. Takaoka // which should be "Lj" in title case and "LJ" in upper case. 222c87449497cb54c092cfd46037320fedf1c4504b6Jean Chalard // - It does not work for Dutch, because it fails to account for the "ij" digraph when it's 223c87449497cb54c092cfd46037320fedf1c4504b6Jean Chalard // written as two separate code points. They are two different characters but both should 224c87449497cb54c092cfd46037320fedf1c4504b6Jean Chalard // be capitalized as "IJ" as if they were a single letter in most words (not all). If the 225c87449497cb54c092cfd46037320fedf1c4504b6Jean Chalard // unicode char for the ligature is used however, it works. 226c87449497cb54c092cfd46037320fedf1c4504b6Jean Chalard final int cutoff = s.offsetByCodePoints(0, 1); 2276c47403e27fd703ece844f4b1b24632721da1772Tadashi G. Takaoka return s.substring(0, cutoff).toUpperCase(getLocaleUsedForToTitleCase(locale)) 2286c47403e27fd703ece844f4b1b24632721da1772Tadashi G. Takaoka + s.substring(cutoff).toLowerCase(locale); 22911d9ee742f8ff3fb31b0e3beb32ee4870c63d8e3Tadashi G. Takaoka } 23080111f08e284655808380663f0b68547b981da72Jean Chalard 23180980574acfef74f9392da4fdbcba64d911cb66fTadashi G. Takaoka @Nonnull 23280980574acfef74f9392da4fdbcba64d911cb66fTadashi G. Takaoka public static int[] toCodePointArray(@Nonnull final CharSequence charSequence) { 2333285b68c97d966d86c08aeb837e5bf5633981357Jean Chalard return toCodePointArray(charSequence, 0, charSequence.length()); 23418638d30c059ca0f2e23674ba9ef68296f44c6bdJean Chalard } 23518638d30c059ca0f2e23674ba9ef68296f44c6bdJean Chalard 23680980574acfef74f9392da4fdbcba64d911cb66fTadashi G. Takaoka @Nonnull 23780980574acfef74f9392da4fdbcba64d911cb66fTadashi G. Takaoka private static final int[] EMPTY_CODEPOINTS = {}; 23880980574acfef74f9392da4fdbcba64d911cb66fTadashi G. Takaoka 239f16a2a36d21a2596db48bf3ba05d0f97339581bfJean Chalard /** 240f16a2a36d21a2596db48bf3ba05d0f97339581bfJean Chalard * Converts a range of a string to an array of code points. 2413285b68c97d966d86c08aeb837e5bf5633981357Jean Chalard * @param charSequence the source string. 242f16a2a36d21a2596db48bf3ba05d0f97339581bfJean Chalard * @param startIndex the start index inside the string in java chars, inclusive. 243f16a2a36d21a2596db48bf3ba05d0f97339581bfJean Chalard * @param endIndex the end index inside the string in java chars, exclusive. 244f16a2a36d21a2596db48bf3ba05d0f97339581bfJean Chalard * @return a new array of code points. At most endIndex - startIndex, but possibly less. 245f16a2a36d21a2596db48bf3ba05d0f97339581bfJean Chalard */ 24680980574acfef74f9392da4fdbcba64d911cb66fTadashi G. Takaoka @Nonnull 24780980574acfef74f9392da4fdbcba64d911cb66fTadashi G. Takaoka public static int[] toCodePointArray(@Nonnull final CharSequence charSequence, 24818638d30c059ca0f2e23674ba9ef68296f44c6bdJean Chalard final int startIndex, final int endIndex) { 2493285b68c97d966d86c08aeb837e5bf5633981357Jean Chalard final int length = charSequence.length(); 25001106f6a10ef2c6c8a338cb256e0799f7aca853bKurt Partridge if (length <= 0) { 251bc464e2952e102219f0b977fc1e9140ad5bd03e4Tadashi G. Takaoka return EMPTY_CODEPOINTS; 25201106f6a10ef2c6c8a338cb256e0799f7aca853bKurt Partridge } 2533285b68c97d966d86c08aeb837e5bf5633981357Jean Chalard final int[] codePoints = 2543285b68c97d966d86c08aeb837e5bf5633981357Jean Chalard new int[Character.codePointCount(charSequence, startIndex, endIndex)]; 2555f430e0189f6c0a1a34edb0e00809bc53a525a7bJean Chalard copyCodePointsAndReturnCodePointCount(codePoints, charSequence, startIndex, endIndex, 2565f430e0189f6c0a1a34edb0e00809bc53a525a7bJean Chalard false /* downCase */); 2575f430e0189f6c0a1a34edb0e00809bc53a525a7bJean Chalard return codePoints; 2585f430e0189f6c0a1a34edb0e00809bc53a525a7bJean Chalard } 2595f430e0189f6c0a1a34edb0e00809bc53a525a7bJean Chalard 2605f430e0189f6c0a1a34edb0e00809bc53a525a7bJean Chalard /** 2615f430e0189f6c0a1a34edb0e00809bc53a525a7bJean Chalard * Copies the codepoints in a CharSequence to an int array. 2625f430e0189f6c0a1a34edb0e00809bc53a525a7bJean Chalard * 2635f430e0189f6c0a1a34edb0e00809bc53a525a7bJean Chalard * This method assumes there is enough space in the array to store the code points. The size 2645f430e0189f6c0a1a34edb0e00809bc53a525a7bJean Chalard * can be measured with Character#codePointCount(CharSequence, int, int) before passing to this 2655f430e0189f6c0a1a34edb0e00809bc53a525a7bJean Chalard * method. If the int array is too small, an ArrayIndexOutOfBoundsException will be thrown. 2665f430e0189f6c0a1a34edb0e00809bc53a525a7bJean Chalard * Also, this method makes no effort to be thread-safe. Do not modify the CharSequence while 2675f430e0189f6c0a1a34edb0e00809bc53a525a7bJean Chalard * this method is running, or the behavior is undefined. 2685f430e0189f6c0a1a34edb0e00809bc53a525a7bJean Chalard * This method can optionally downcase code points before copying them, but it pays no attention 2695f430e0189f6c0a1a34edb0e00809bc53a525a7bJean Chalard * to locale while doing so. 2705f430e0189f6c0a1a34edb0e00809bc53a525a7bJean Chalard * 2715f430e0189f6c0a1a34edb0e00809bc53a525a7bJean Chalard * @param destination the int array. 2725f430e0189f6c0a1a34edb0e00809bc53a525a7bJean Chalard * @param charSequence the CharSequence. 2735f430e0189f6c0a1a34edb0e00809bc53a525a7bJean Chalard * @param startIndex the start index inside the string in java chars, inclusive. 2745f430e0189f6c0a1a34edb0e00809bc53a525a7bJean Chalard * @param endIndex the end index inside the string in java chars, exclusive. 2755f430e0189f6c0a1a34edb0e00809bc53a525a7bJean Chalard * @param downCase if this is true, code points will be downcased before being copied. 2765f430e0189f6c0a1a34edb0e00809bc53a525a7bJean Chalard * @return the number of copied code points. 2775f430e0189f6c0a1a34edb0e00809bc53a525a7bJean Chalard */ 27880980574acfef74f9392da4fdbcba64d911cb66fTadashi G. Takaoka public static int copyCodePointsAndReturnCodePointCount(@Nonnull final int[] destination, 27980980574acfef74f9392da4fdbcba64d911cb66fTadashi G. Takaoka @Nonnull final CharSequence charSequence, final int startIndex, final int endIndex, 2805f430e0189f6c0a1a34edb0e00809bc53a525a7bJean Chalard final boolean downCase) { 281bc464e2952e102219f0b977fc1e9140ad5bd03e4Tadashi G. Takaoka int destIndex = 0; 28218638d30c059ca0f2e23674ba9ef68296f44c6bdJean Chalard for (int index = startIndex; index < endIndex; 2833285b68c97d966d86c08aeb837e5bf5633981357Jean Chalard index = Character.offsetByCodePoints(charSequence, index, 1)) { 2845f430e0189f6c0a1a34edb0e00809bc53a525a7bJean Chalard final int codePoint = Character.codePointAt(charSequence, index); 2855f430e0189f6c0a1a34edb0e00809bc53a525a7bJean Chalard // TODO: stop using this, as it's not aware of the locale and does not always do 2865f430e0189f6c0a1a34edb0e00809bc53a525a7bJean Chalard // the right thing. 2875f430e0189f6c0a1a34edb0e00809bc53a525a7bJean Chalard destination[destIndex] = downCase ? Character.toLowerCase(codePoint) : codePoint; 288bc464e2952e102219f0b977fc1e9140ad5bd03e4Tadashi G. Takaoka destIndex++; 28980111f08e284655808380663f0b68547b981da72Jean Chalard } 2905f430e0189f6c0a1a34edb0e00809bc53a525a7bJean Chalard return destIndex; 29180111f08e284655808380663f0b68547b981da72Jean Chalard } 2929d1c73ffd88cd1bfef3de048b0b3a9a7dfbcfa70Jean Chalard 29380980574acfef74f9392da4fdbcba64d911cb66fTadashi G. Takaoka @Nonnull 29480980574acfef74f9392da4fdbcba64d911cb66fTadashi G. Takaoka public static int[] toSortedCodePointArray(@Nonnull final String string) { 295837cdd738b7ddbeac04b15230f01e44d247bd50aTadashi G. Takaoka final int[] codePoints = toCodePointArray(string); 296837cdd738b7ddbeac04b15230f01e44d247bd50aTadashi G. Takaoka Arrays.sort(codePoints); 297837cdd738b7ddbeac04b15230f01e44d247bd50aTadashi G. Takaoka return codePoints; 298837cdd738b7ddbeac04b15230f01e44d247bd50aTadashi G. Takaoka } 299837cdd738b7ddbeac04b15230f01e44d247bd50aTadashi G. Takaoka 30053fb86c447bc322c13212ae8eaa102f6a3e57604Keisuke Kuroyanagi /** 30153fb86c447bc322c13212ae8eaa102f6a3e57604Keisuke Kuroyanagi * Construct a String from a code point array 30253fb86c447bc322c13212ae8eaa102f6a3e57604Keisuke Kuroyanagi * 30353fb86c447bc322c13212ae8eaa102f6a3e57604Keisuke Kuroyanagi * @param codePoints a code point array that is null terminated when its logical length is 30453fb86c447bc322c13212ae8eaa102f6a3e57604Keisuke Kuroyanagi * shorter than the array length. 30553fb86c447bc322c13212ae8eaa102f6a3e57604Keisuke Kuroyanagi * @return a string constructed from the code point array. 30653fb86c447bc322c13212ae8eaa102f6a3e57604Keisuke Kuroyanagi */ 30780980574acfef74f9392da4fdbcba64d911cb66fTadashi G. Takaoka @Nonnull 30880980574acfef74f9392da4fdbcba64d911cb66fTadashi G. Takaoka public static String getStringFromNullTerminatedCodePointArray( 30980980574acfef74f9392da4fdbcba64d911cb66fTadashi G. Takaoka @Nonnull final int[] codePoints) { 31053fb86c447bc322c13212ae8eaa102f6a3e57604Keisuke Kuroyanagi int stringLength = codePoints.length; 31153fb86c447bc322c13212ae8eaa102f6a3e57604Keisuke Kuroyanagi for (int i = 0; i < codePoints.length; i++) { 31253fb86c447bc322c13212ae8eaa102f6a3e57604Keisuke Kuroyanagi if (codePoints[i] == 0) { 31353fb86c447bc322c13212ae8eaa102f6a3e57604Keisuke Kuroyanagi stringLength = i; 31453fb86c447bc322c13212ae8eaa102f6a3e57604Keisuke Kuroyanagi break; 31553fb86c447bc322c13212ae8eaa102f6a3e57604Keisuke Kuroyanagi } 31653fb86c447bc322c13212ae8eaa102f6a3e57604Keisuke Kuroyanagi } 31753fb86c447bc322c13212ae8eaa102f6a3e57604Keisuke Kuroyanagi return new String(codePoints, 0 /* offset */, stringLength); 31853fb86c447bc322c13212ae8eaa102f6a3e57604Keisuke Kuroyanagi } 31953fb86c447bc322c13212ae8eaa102f6a3e57604Keisuke Kuroyanagi 320c87449497cb54c092cfd46037320fedf1c4504b6Jean Chalard // This method assumes the text is not null. For the empty string, it returns CAPITALIZE_NONE. 32180980574acfef74f9392da4fdbcba64d911cb66fTadashi G. Takaoka public static int getCapitalizationType(@Nonnull final String text) { 3227d3836d63a2eb4b79c4ad93cdae4f1f61cdb518eJean Chalard // If the first char is not uppercase, then the word is either all lower case or 3237d3836d63a2eb4b79c4ad93cdae4f1f61cdb518eJean Chalard // camel case, and in either case we return CAPITALIZE_NONE. 3247d3836d63a2eb4b79c4ad93cdae4f1f61cdb518eJean Chalard final int len = text.length(); 325c87449497cb54c092cfd46037320fedf1c4504b6Jean Chalard int index = 0; 326c87449497cb54c092cfd46037320fedf1c4504b6Jean Chalard for (; index < len; index = text.offsetByCodePoints(index, 1)) { 327c87449497cb54c092cfd46037320fedf1c4504b6Jean Chalard if (Character.isLetter(text.codePointAt(index))) { 328c87449497cb54c092cfd46037320fedf1c4504b6Jean Chalard break; 329c87449497cb54c092cfd46037320fedf1c4504b6Jean Chalard } 330c87449497cb54c092cfd46037320fedf1c4504b6Jean Chalard } 331c87449497cb54c092cfd46037320fedf1c4504b6Jean Chalard if (index == len) return CAPITALIZE_NONE; 332c87449497cb54c092cfd46037320fedf1c4504b6Jean Chalard if (!Character.isUpperCase(text.codePointAt(index))) { 333c87449497cb54c092cfd46037320fedf1c4504b6Jean Chalard return CAPITALIZE_NONE; 334c87449497cb54c092cfd46037320fedf1c4504b6Jean Chalard } 3357d3836d63a2eb4b79c4ad93cdae4f1f61cdb518eJean Chalard int capsCount = 1; 3367d3836d63a2eb4b79c4ad93cdae4f1f61cdb518eJean Chalard int letterCount = 1; 337c87449497cb54c092cfd46037320fedf1c4504b6Jean Chalard for (index = text.offsetByCodePoints(index, 1); index < len; 338c87449497cb54c092cfd46037320fedf1c4504b6Jean Chalard index = text.offsetByCodePoints(index, 1)) { 3397d3836d63a2eb4b79c4ad93cdae4f1f61cdb518eJean Chalard if (1 != capsCount && letterCount != capsCount) break; 340c87449497cb54c092cfd46037320fedf1c4504b6Jean Chalard final int codePoint = text.codePointAt(index); 3417d3836d63a2eb4b79c4ad93cdae4f1f61cdb518eJean Chalard if (Character.isUpperCase(codePoint)) { 3427d3836d63a2eb4b79c4ad93cdae4f1f61cdb518eJean Chalard ++capsCount; 3437d3836d63a2eb4b79c4ad93cdae4f1f61cdb518eJean Chalard ++letterCount; 3447d3836d63a2eb4b79c4ad93cdae4f1f61cdb518eJean Chalard } else if (Character.isLetter(codePoint)) { 3457d3836d63a2eb4b79c4ad93cdae4f1f61cdb518eJean Chalard // We need to discount non-letters since they may not be upper-case, but may 3467d3836d63a2eb4b79c4ad93cdae4f1f61cdb518eJean Chalard // still be part of a word (e.g. single quote or dash, as in "IT'S" or "FULL-TIME") 3477d3836d63a2eb4b79c4ad93cdae4f1f61cdb518eJean Chalard ++letterCount; 3487d3836d63a2eb4b79c4ad93cdae4f1f61cdb518eJean Chalard } 3497d3836d63a2eb4b79c4ad93cdae4f1f61cdb518eJean Chalard } 3507d3836d63a2eb4b79c4ad93cdae4f1f61cdb518eJean Chalard // We know the first char is upper case. So we want to test if either every letter other 3517d3836d63a2eb4b79c4ad93cdae4f1f61cdb518eJean Chalard // than the first is lower case, or if they are all upper case. If the string is exactly 3527d3836d63a2eb4b79c4ad93cdae4f1f61cdb518eJean Chalard // one char long, then we will arrive here with letterCount 1, and this is correct, too. 3537d3836d63a2eb4b79c4ad93cdae4f1f61cdb518eJean Chalard if (1 == capsCount) return CAPITALIZE_FIRST; 3547d3836d63a2eb4b79c4ad93cdae4f1f61cdb518eJean Chalard return (letterCount == capsCount ? CAPITALIZE_ALL : CAPITALIZE_NONE); 3557d3836d63a2eb4b79c4ad93cdae4f1f61cdb518eJean Chalard } 3561cbb0b8f2a745acf8d792e6ceafbe111816869bdJean Chalard 35780980574acfef74f9392da4fdbcba64d911cb66fTadashi G. Takaoka public static boolean isIdenticalAfterUpcase(@Nonnull final String text) { 358af5fbe70ac9eff6979f444c3d938d28a0ec76ebdJean Chalard final int length = text.length(); 359af5fbe70ac9eff6979f444c3d938d28a0ec76ebdJean Chalard int i = 0; 360af5fbe70ac9eff6979f444c3d938d28a0ec76ebdJean Chalard while (i < length) { 3611cbb0b8f2a745acf8d792e6ceafbe111816869bdJean Chalard final int codePoint = text.codePointAt(i); 3621cbb0b8f2a745acf8d792e6ceafbe111816869bdJean Chalard if (Character.isLetter(codePoint) && !Character.isUpperCase(codePoint)) { 3631cbb0b8f2a745acf8d792e6ceafbe111816869bdJean Chalard return false; 3641cbb0b8f2a745acf8d792e6ceafbe111816869bdJean Chalard } 365af5fbe70ac9eff6979f444c3d938d28a0ec76ebdJean Chalard i += Character.charCount(codePoint); 3661cbb0b8f2a745acf8d792e6ceafbe111816869bdJean Chalard } 3671cbb0b8f2a745acf8d792e6ceafbe111816869bdJean Chalard return true; 3681cbb0b8f2a745acf8d792e6ceafbe111816869bdJean Chalard } 3691cbb0b8f2a745acf8d792e6ceafbe111816869bdJean Chalard 37080980574acfef74f9392da4fdbcba64d911cb66fTadashi G. Takaoka public static boolean isIdenticalAfterDowncase(@Nonnull final String text) { 371af5fbe70ac9eff6979f444c3d938d28a0ec76ebdJean Chalard final int length = text.length(); 372af5fbe70ac9eff6979f444c3d938d28a0ec76ebdJean Chalard int i = 0; 373af5fbe70ac9eff6979f444c3d938d28a0ec76ebdJean Chalard while (i < length) { 3741cbb0b8f2a745acf8d792e6ceafbe111816869bdJean Chalard final int codePoint = text.codePointAt(i); 3751cbb0b8f2a745acf8d792e6ceafbe111816869bdJean Chalard if (Character.isLetter(codePoint) && !Character.isLowerCase(codePoint)) { 3761cbb0b8f2a745acf8d792e6ceafbe111816869bdJean Chalard return false; 3771cbb0b8f2a745acf8d792e6ceafbe111816869bdJean Chalard } 378af5fbe70ac9eff6979f444c3d938d28a0ec76ebdJean Chalard i += Character.charCount(codePoint); 3791cbb0b8f2a745acf8d792e6ceafbe111816869bdJean Chalard } 3801cbb0b8f2a745acf8d792e6ceafbe111816869bdJean Chalard return true; 3811cbb0b8f2a745acf8d792e6ceafbe111816869bdJean Chalard } 3827cd7cf73f4ce6f0e577d6382eb0fc25f60dc63e1Jean Chalard 38380980574acfef74f9392da4fdbcba64d911cb66fTadashi G. Takaoka public static boolean isIdenticalAfterCapitalizeEachWord(@Nonnull final String text, 38480980574acfef74f9392da4fdbcba64d911cb66fTadashi G. Takaoka @Nonnull final int[] sortedSeparators) { 385c93cf1c398fbea8bde4b568dae1fbe2f8d9b4180Tadashi G. Takaoka boolean needsCapsNext = true; 3861cbb0b8f2a745acf8d792e6ceafbe111816869bdJean Chalard final int len = text.length(); 3871cbb0b8f2a745acf8d792e6ceafbe111816869bdJean Chalard for (int i = 0; i < len; i = text.offsetByCodePoints(i, 1)) { 3881cbb0b8f2a745acf8d792e6ceafbe111816869bdJean Chalard final int codePoint = text.codePointAt(i); 3891cbb0b8f2a745acf8d792e6ceafbe111816869bdJean Chalard if (Character.isLetter(codePoint)) { 390c93cf1c398fbea8bde4b568dae1fbe2f8d9b4180Tadashi G. Takaoka if ((needsCapsNext && !Character.isUpperCase(codePoint)) 391c93cf1c398fbea8bde4b568dae1fbe2f8d9b4180Tadashi G. Takaoka || (!needsCapsNext && !Character.isLowerCase(codePoint))) { 3921cbb0b8f2a745acf8d792e6ceafbe111816869bdJean Chalard return false; 3931cbb0b8f2a745acf8d792e6ceafbe111816869bdJean Chalard } 3941cbb0b8f2a745acf8d792e6ceafbe111816869bdJean Chalard } 3951cbb0b8f2a745acf8d792e6ceafbe111816869bdJean Chalard // We need a capital letter next if this is a separator. 396c93cf1c398fbea8bde4b568dae1fbe2f8d9b4180Tadashi G. Takaoka needsCapsNext = (Arrays.binarySearch(sortedSeparators, codePoint) >= 0); 3971cbb0b8f2a745acf8d792e6ceafbe111816869bdJean Chalard } 3981cbb0b8f2a745acf8d792e6ceafbe111816869bdJean Chalard return true; 3991cbb0b8f2a745acf8d792e6ceafbe111816869bdJean Chalard } 4001cbb0b8f2a745acf8d792e6ceafbe111816869bdJean Chalard 4011cbb0b8f2a745acf8d792e6ceafbe111816869bdJean Chalard // TODO: like capitalizeFirst*, this does not work perfectly for Dutch because of the IJ digraph 4021cbb0b8f2a745acf8d792e6ceafbe111816869bdJean Chalard // which should be capitalized together in *some* cases. 40380980574acfef74f9392da4fdbcba64d911cb66fTadashi G. Takaoka @Nonnull 40480980574acfef74f9392da4fdbcba64d911cb66fTadashi G. Takaoka public static String capitalizeEachWord(@Nonnull final String text, 40580980574acfef74f9392da4fdbcba64d911cb66fTadashi G. Takaoka @Nonnull final int[] sortedSeparators, @Nonnull final Locale locale) { 4061cbb0b8f2a745acf8d792e6ceafbe111816869bdJean Chalard final StringBuilder builder = new StringBuilder(); 407c93cf1c398fbea8bde4b568dae1fbe2f8d9b4180Tadashi G. Takaoka boolean needsCapsNext = true; 4081cbb0b8f2a745acf8d792e6ceafbe111816869bdJean Chalard final int len = text.length(); 4091cbb0b8f2a745acf8d792e6ceafbe111816869bdJean Chalard for (int i = 0; i < len; i = text.offsetByCodePoints(i, 1)) { 4101cbb0b8f2a745acf8d792e6ceafbe111816869bdJean Chalard final String nextChar = text.substring(i, text.offsetByCodePoints(i, 1)); 411c93cf1c398fbea8bde4b568dae1fbe2f8d9b4180Tadashi G. Takaoka if (needsCapsNext) { 4121cbb0b8f2a745acf8d792e6ceafbe111816869bdJean Chalard builder.append(nextChar.toUpperCase(locale)); 4131cbb0b8f2a745acf8d792e6ceafbe111816869bdJean Chalard } else { 4141cbb0b8f2a745acf8d792e6ceafbe111816869bdJean Chalard builder.append(nextChar.toLowerCase(locale)); 4151cbb0b8f2a745acf8d792e6ceafbe111816869bdJean Chalard } 4161cbb0b8f2a745acf8d792e6ceafbe111816869bdJean Chalard // We need a capital letter next if this is a separator. 417c93cf1c398fbea8bde4b568dae1fbe2f8d9b4180Tadashi G. Takaoka needsCapsNext = (Arrays.binarySearch(sortedSeparators, nextChar.codePointAt(0)) >= 0); 4181cbb0b8f2a745acf8d792e6ceafbe111816869bdJean Chalard } 4191cbb0b8f2a745acf8d792e6ceafbe111816869bdJean Chalard return builder.toString(); 4201cbb0b8f2a745acf8d792e6ceafbe111816869bdJean Chalard } 42173ec85b8ad3102ce1c7e6013be73afe83475e589Jean Chalard 42273ec85b8ad3102ce1c7e6013be73afe83475e589Jean Chalard /** 42373ec85b8ad3102ce1c7e6013be73afe83475e589Jean Chalard * Approximates whether the text before the cursor looks like a URL. 42473ec85b8ad3102ce1c7e6013be73afe83475e589Jean Chalard * 42573ec85b8ad3102ce1c7e6013be73afe83475e589Jean Chalard * This is not foolproof, but it should work well in the practice. 42673ec85b8ad3102ce1c7e6013be73afe83475e589Jean Chalard * Essentially it walks backward from the cursor until it finds something that's not a letter, 42773ec85b8ad3102ce1c7e6013be73afe83475e589Jean Chalard * digit, or common URL symbol like underscore. If it hasn't found a period yet, then it 42873ec85b8ad3102ce1c7e6013be73afe83475e589Jean Chalard * does not look like a URL. 42973ec85b8ad3102ce1c7e6013be73afe83475e589Jean Chalard * If the text: 43073ec85b8ad3102ce1c7e6013be73afe83475e589Jean Chalard * - starts with www and contains a period 43173ec85b8ad3102ce1c7e6013be73afe83475e589Jean Chalard * - starts with a slash preceded by either a slash, whitespace, or start-of-string 43273ec85b8ad3102ce1c7e6013be73afe83475e589Jean Chalard * Then it looks like a URL and we return true. Otherwise, we return false. 43373ec85b8ad3102ce1c7e6013be73afe83475e589Jean Chalard * 43473ec85b8ad3102ce1c7e6013be73afe83475e589Jean Chalard * Note: this method is called quite often, and should be fast. 43573ec85b8ad3102ce1c7e6013be73afe83475e589Jean Chalard * 43673ec85b8ad3102ce1c7e6013be73afe83475e589Jean Chalard * TODO: This will return that "abc./def" and ".abc/def" look like URLs to keep down the 43773ec85b8ad3102ce1c7e6013be73afe83475e589Jean Chalard * code complexity, but ideally it should not. It's acceptable for now. 43873ec85b8ad3102ce1c7e6013be73afe83475e589Jean Chalard */ 43980980574acfef74f9392da4fdbcba64d911cb66fTadashi G. Takaoka public static boolean lastPartLooksLikeURL(@Nonnull final CharSequence text) { 44073ec85b8ad3102ce1c7e6013be73afe83475e589Jean Chalard int i = text.length(); 44180980574acfef74f9392da4fdbcba64d911cb66fTadashi G. Takaoka if (0 == i) { 44280980574acfef74f9392da4fdbcba64d911cb66fTadashi G. Takaoka return false; 44380980574acfef74f9392da4fdbcba64d911cb66fTadashi G. Takaoka } 44473ec85b8ad3102ce1c7e6013be73afe83475e589Jean Chalard int wCount = 0; 44573ec85b8ad3102ce1c7e6013be73afe83475e589Jean Chalard int slashCount = 0; 44673ec85b8ad3102ce1c7e6013be73afe83475e589Jean Chalard boolean hasSlash = false; 44773ec85b8ad3102ce1c7e6013be73afe83475e589Jean Chalard boolean hasPeriod = false; 44873ec85b8ad3102ce1c7e6013be73afe83475e589Jean Chalard int codePoint = 0; 44973ec85b8ad3102ce1c7e6013be73afe83475e589Jean Chalard while (i > 0) { 450c7ef305bbc119b820fd619d3ed205198d4f98c3fJean Chalard codePoint = Character.codePointBefore(text, i); 45173ec85b8ad3102ce1c7e6013be73afe83475e589Jean Chalard if (codePoint < Constants.CODE_PERIOD || codePoint > 'z') { 45273ec85b8ad3102ce1c7e6013be73afe83475e589Jean Chalard // Handwavy heuristic to see if that's a URL character. Anything between period 45373ec85b8ad3102ce1c7e6013be73afe83475e589Jean Chalard // and z. This includes all lower- and upper-case ascii letters, period, 45473ec85b8ad3102ce1c7e6013be73afe83475e589Jean Chalard // underscore, arrobase, question mark, equal sign. It excludes spaces, exclamation 45573ec85b8ad3102ce1c7e6013be73afe83475e589Jean Chalard // marks, double quotes... 45673ec85b8ad3102ce1c7e6013be73afe83475e589Jean Chalard // Anything that's not a URL-like character causes us to break from here and 45773ec85b8ad3102ce1c7e6013be73afe83475e589Jean Chalard // evaluate normally. 45873ec85b8ad3102ce1c7e6013be73afe83475e589Jean Chalard break; 45973ec85b8ad3102ce1c7e6013be73afe83475e589Jean Chalard } 46073ec85b8ad3102ce1c7e6013be73afe83475e589Jean Chalard if (Constants.CODE_PERIOD == codePoint) { 46173ec85b8ad3102ce1c7e6013be73afe83475e589Jean Chalard hasPeriod = true; 46273ec85b8ad3102ce1c7e6013be73afe83475e589Jean Chalard } 46373ec85b8ad3102ce1c7e6013be73afe83475e589Jean Chalard if (Constants.CODE_SLASH == codePoint) { 46473ec85b8ad3102ce1c7e6013be73afe83475e589Jean Chalard hasSlash = true; 46573ec85b8ad3102ce1c7e6013be73afe83475e589Jean Chalard if (2 == ++slashCount) { 46673ec85b8ad3102ce1c7e6013be73afe83475e589Jean Chalard return true; 46773ec85b8ad3102ce1c7e6013be73afe83475e589Jean Chalard } 46873ec85b8ad3102ce1c7e6013be73afe83475e589Jean Chalard } else { 46973ec85b8ad3102ce1c7e6013be73afe83475e589Jean Chalard slashCount = 0; 47073ec85b8ad3102ce1c7e6013be73afe83475e589Jean Chalard } 47173ec85b8ad3102ce1c7e6013be73afe83475e589Jean Chalard if ('w' == codePoint) { 47273ec85b8ad3102ce1c7e6013be73afe83475e589Jean Chalard ++wCount; 47373ec85b8ad3102ce1c7e6013be73afe83475e589Jean Chalard } else { 47473ec85b8ad3102ce1c7e6013be73afe83475e589Jean Chalard wCount = 0; 47573ec85b8ad3102ce1c7e6013be73afe83475e589Jean Chalard } 47673ec85b8ad3102ce1c7e6013be73afe83475e589Jean Chalard i = Character.offsetByCodePoints(text, i, -1); 47773ec85b8ad3102ce1c7e6013be73afe83475e589Jean Chalard } 47873ec85b8ad3102ce1c7e6013be73afe83475e589Jean Chalard // End of the text run. 47973ec85b8ad3102ce1c7e6013be73afe83475e589Jean Chalard // If it starts with www and includes a period, then it looks like a URL. 48080980574acfef74f9392da4fdbcba64d911cb66fTadashi G. Takaoka if (wCount >= 3 && hasPeriod) { 48180980574acfef74f9392da4fdbcba64d911cb66fTadashi G. Takaoka return true; 48280980574acfef74f9392da4fdbcba64d911cb66fTadashi G. Takaoka } 48373ec85b8ad3102ce1c7e6013be73afe83475e589Jean Chalard // If it starts with a slash, and the code point before is whitespace, it looks like an URL. 48480980574acfef74f9392da4fdbcba64d911cb66fTadashi G. Takaoka if (1 == slashCount && (0 == i || Character.isWhitespace(codePoint))) { 48580980574acfef74f9392da4fdbcba64d911cb66fTadashi G. Takaoka return true; 48680980574acfef74f9392da4fdbcba64d911cb66fTadashi G. Takaoka } 48773ec85b8ad3102ce1c7e6013be73afe83475e589Jean Chalard // If it has both a period and a slash, it looks like an URL. 48880980574acfef74f9392da4fdbcba64d911cb66fTadashi G. Takaoka if (hasPeriod && hasSlash) { 48980980574acfef74f9392da4fdbcba64d911cb66fTadashi G. Takaoka return true; 49080980574acfef74f9392da4fdbcba64d911cb66fTadashi G. Takaoka } 49173ec85b8ad3102ce1c7e6013be73afe83475e589Jean Chalard // Otherwise, it doesn't look like an URL. 49273ec85b8ad3102ce1c7e6013be73afe83475e589Jean Chalard return false; 49373ec85b8ad3102ce1c7e6013be73afe83475e589Jean Chalard } 4944ee953125af53d8b49e958e50f0002f002776d36Satoshi Kataoka 495c7ef305bbc119b820fd619d3ed205198d4f98c3fJean Chalard /** 496c7ef305bbc119b820fd619d3ed205198d4f98c3fJean Chalard * Examines the string and returns whether we're inside a double quote. 497c7ef305bbc119b820fd619d3ed205198d4f98c3fJean Chalard * 498c7ef305bbc119b820fd619d3ed205198d4f98c3fJean Chalard * This is used to decide whether we should put an automatic space before or after a double 499c7ef305bbc119b820fd619d3ed205198d4f98c3fJean Chalard * quote character. If we're inside a quotation, then we want to close it, so we want a space 500c7ef305bbc119b820fd619d3ed205198d4f98c3fJean Chalard * after and not before. Otherwise, we want to open the quotation, so we want a space before 501c7ef305bbc119b820fd619d3ed205198d4f98c3fJean Chalard * and not after. Exception: after a digit, we never want a space because the "inch" or 502c7ef305bbc119b820fd619d3ed205198d4f98c3fJean Chalard * "minutes" use cases is dominant after digits. 503c7ef305bbc119b820fd619d3ed205198d4f98c3fJean Chalard * In the practice, we determine whether we are in a quotation or not by finding the previous 504c7ef305bbc119b820fd619d3ed205198d4f98c3fJean Chalard * double quote character, and looking at whether it's followed by whitespace. If so, that 505c7ef305bbc119b820fd619d3ed205198d4f98c3fJean Chalard * was a closing quotation mark, so we're not inside a double quote. If it's not followed 506c7ef305bbc119b820fd619d3ed205198d4f98c3fJean Chalard * by whitespace, then it was an opening quotation mark, and we're inside a quotation. 507c7ef305bbc119b820fd619d3ed205198d4f98c3fJean Chalard * 508c7ef305bbc119b820fd619d3ed205198d4f98c3fJean Chalard * @param text the text to examine. 509c7ef305bbc119b820fd619d3ed205198d4f98c3fJean Chalard * @return whether we're inside a double quote. 510c7ef305bbc119b820fd619d3ed205198d4f98c3fJean Chalard */ 51180980574acfef74f9392da4fdbcba64d911cb66fTadashi G. Takaoka public static boolean isInsideDoubleQuoteOrAfterDigit(@Nonnull final CharSequence text) { 512c7ef305bbc119b820fd619d3ed205198d4f98c3fJean Chalard int i = text.length(); 51380980574acfef74f9392da4fdbcba64d911cb66fTadashi G. Takaoka if (0 == i) { 51480980574acfef74f9392da4fdbcba64d911cb66fTadashi G. Takaoka return false; 51580980574acfef74f9392da4fdbcba64d911cb66fTadashi G. Takaoka } 516c7ef305bbc119b820fd619d3ed205198d4f98c3fJean Chalard int codePoint = Character.codePointBefore(text, i); 51780980574acfef74f9392da4fdbcba64d911cb66fTadashi G. Takaoka if (Character.isDigit(codePoint)) { 51880980574acfef74f9392da4fdbcba64d911cb66fTadashi G. Takaoka return true; 51980980574acfef74f9392da4fdbcba64d911cb66fTadashi G. Takaoka } 520c7ef305bbc119b820fd619d3ed205198d4f98c3fJean Chalard int prevCodePoint = 0; 521c7ef305bbc119b820fd619d3ed205198d4f98c3fJean Chalard while (i > 0) { 522c7ef305bbc119b820fd619d3ed205198d4f98c3fJean Chalard codePoint = Character.codePointBefore(text, i); 523c7ef305bbc119b820fd619d3ed205198d4f98c3fJean Chalard if (Constants.CODE_DOUBLE_QUOTE == codePoint) { 524c7ef305bbc119b820fd619d3ed205198d4f98c3fJean Chalard // If we see a double quote followed by whitespace, then that 525c7ef305bbc119b820fd619d3ed205198d4f98c3fJean Chalard // was a closing quote. 52680980574acfef74f9392da4fdbcba64d911cb66fTadashi G. Takaoka if (Character.isWhitespace(prevCodePoint)) { 52780980574acfef74f9392da4fdbcba64d911cb66fTadashi G. Takaoka return false; 52880980574acfef74f9392da4fdbcba64d911cb66fTadashi G. Takaoka } 529c7ef305bbc119b820fd619d3ed205198d4f98c3fJean Chalard } 530c7ef305bbc119b820fd619d3ed205198d4f98c3fJean Chalard if (Character.isWhitespace(codePoint) && Constants.CODE_DOUBLE_QUOTE == prevCodePoint) { 531c7ef305bbc119b820fd619d3ed205198d4f98c3fJean Chalard // If we see a double quote preceded by whitespace, then that 532c7ef305bbc119b820fd619d3ed205198d4f98c3fJean Chalard // was an opening quote. No need to continue seeking. 533c7ef305bbc119b820fd619d3ed205198d4f98c3fJean Chalard return true; 534c7ef305bbc119b820fd619d3ed205198d4f98c3fJean Chalard } 535c7ef305bbc119b820fd619d3ed205198d4f98c3fJean Chalard i -= Character.charCount(codePoint); 536c7ef305bbc119b820fd619d3ed205198d4f98c3fJean Chalard prevCodePoint = codePoint; 537c7ef305bbc119b820fd619d3ed205198d4f98c3fJean Chalard } 538c7ef305bbc119b820fd619d3ed205198d4f98c3fJean Chalard // We reached the start of text. If the first char is a double quote, then we're inside 539c7ef305bbc119b820fd619d3ed205198d4f98c3fJean Chalard // a double quote. Otherwise we're not. 540c7ef305bbc119b820fd619d3ed205198d4f98c3fJean Chalard return Constants.CODE_DOUBLE_QUOTE == codePoint; 541c7ef305bbc119b820fd619d3ed205198d4f98c3fJean Chalard } 542c7ef305bbc119b820fd619d3ed205198d4f98c3fJean Chalard 54380980574acfef74f9392da4fdbcba64d911cb66fTadashi G. Takaoka public static boolean isEmptyStringOrWhiteSpaces(@Nonnull final String s) { 5444ee953125af53d8b49e958e50f0002f002776d36Satoshi Kataoka final int N = codePointCount(s); 5454ee953125af53d8b49e958e50f0002f002776d36Satoshi Kataoka for (int i = 0; i < N; ++i) { 5464ee953125af53d8b49e958e50f0002f002776d36Satoshi Kataoka if (!Character.isWhitespace(s.codePointAt(i))) { 5474ee953125af53d8b49e958e50f0002f002776d36Satoshi Kataoka return false; 5484ee953125af53d8b49e958e50f0002f002776d36Satoshi Kataoka } 5494ee953125af53d8b49e958e50f0002f002776d36Satoshi Kataoka } 5504ee953125af53d8b49e958e50f0002f002776d36Satoshi Kataoka return true; 5514ee953125af53d8b49e958e50f0002f002776d36Satoshi Kataoka } 552283cf9cfc9ea8aaf437849ca469f13f029de3ed5Satoshi Kataoka 553283cf9cfc9ea8aaf437849ca469f13f029de3ed5Satoshi Kataoka @UsedForTesting 55480980574acfef74f9392da4fdbcba64d911cb66fTadashi G. Takaoka @Nonnull 55580980574acfef74f9392da4fdbcba64d911cb66fTadashi G. Takaoka public static String byteArrayToHexString(@Nullable final byte[] bytes) { 556283cf9cfc9ea8aaf437849ca469f13f029de3ed5Satoshi Kataoka if (bytes == null || bytes.length == 0) { 5572fa3693c264a4c150ac307d9bb7f6f8f18cc4ffcKen Wakasa return EMPTY_STRING; 558283cf9cfc9ea8aaf437849ca469f13f029de3ed5Satoshi Kataoka } 559283cf9cfc9ea8aaf437849ca469f13f029de3ed5Satoshi Kataoka final StringBuilder sb = new StringBuilder(); 56080980574acfef74f9392da4fdbcba64d911cb66fTadashi G. Takaoka for (final byte b : bytes) { 561283cf9cfc9ea8aaf437849ca469f13f029de3ed5Satoshi Kataoka sb.append(String.format("%02x", b & 0xff)); 562283cf9cfc9ea8aaf437849ca469f13f029de3ed5Satoshi Kataoka } 563283cf9cfc9ea8aaf437849ca469f13f029de3ed5Satoshi Kataoka return sb.toString(); 564283cf9cfc9ea8aaf437849ca469f13f029de3ed5Satoshi Kataoka } 565283cf9cfc9ea8aaf437849ca469f13f029de3ed5Satoshi Kataoka 5663894a5599baf217f036ddfb9f4ac2ac0b4d7a2deSatoshi Kataoka /** 5673894a5599baf217f036ddfb9f4ac2ac0b4d7a2deSatoshi Kataoka * Convert hex string to byte array. The string length must be an even number. 5683894a5599baf217f036ddfb9f4ac2ac0b4d7a2deSatoshi Kataoka */ 569283cf9cfc9ea8aaf437849ca469f13f029de3ed5Satoshi Kataoka @UsedForTesting 57080980574acfef74f9392da4fdbcba64d911cb66fTadashi G. Takaoka @Nullable 57180980574acfef74f9392da4fdbcba64d911cb66fTadashi G. Takaoka public static byte[] hexStringToByteArray(@Nullable final String hexString) { 572ca724ef71f70044f48e2360e2627922be8ca24d5Jean Chalard if (isEmpty(hexString)) { 573283cf9cfc9ea8aaf437849ca469f13f029de3ed5Satoshi Kataoka return null; 574283cf9cfc9ea8aaf437849ca469f13f029de3ed5Satoshi Kataoka } 575283cf9cfc9ea8aaf437849ca469f13f029de3ed5Satoshi Kataoka final int N = hexString.length(); 5763894a5599baf217f036ddfb9f4ac2ac0b4d7a2deSatoshi Kataoka if (N % 2 != 0) { 5773894a5599baf217f036ddfb9f4ac2ac0b4d7a2deSatoshi Kataoka throw new NumberFormatException("Input hex string length must be an even number." 5783894a5599baf217f036ddfb9f4ac2ac0b4d7a2deSatoshi Kataoka + " Length = " + N); 5793894a5599baf217f036ddfb9f4ac2ac0b4d7a2deSatoshi Kataoka } 580283cf9cfc9ea8aaf437849ca469f13f029de3ed5Satoshi Kataoka final byte[] bytes = new byte[N / 2]; 581283cf9cfc9ea8aaf437849ca469f13f029de3ed5Satoshi Kataoka for (int i = 0; i < N; i += 2) { 582283cf9cfc9ea8aaf437849ca469f13f029de3ed5Satoshi Kataoka bytes[i / 2] = (byte) ((Character.digit(hexString.charAt(i), 16) << 4) 583283cf9cfc9ea8aaf437849ca469f13f029de3ed5Satoshi Kataoka + Character.digit(hexString.charAt(i + 1), 16)); 584283cf9cfc9ea8aaf437849ca469f13f029de3ed5Satoshi Kataoka } 585283cf9cfc9ea8aaf437849ca469f13f029de3ed5Satoshi Kataoka return bytes; 586283cf9cfc9ea8aaf437849ca469f13f029de3ed5Satoshi Kataoka } 587639bf62e4fc12c4f75ed0491512f6ed9d2fc2432Tadashi G. Takaoka 588f62b5d633d17b94a7ea46c968e073fdaa3fcbe15Tadashi G. Takaoka private static final String LANGUAGE_GREEK = "el"; 589f62b5d633d17b94a7ea46c968e073fdaa3fcbe15Tadashi G. Takaoka 590f62b5d633d17b94a7ea46c968e073fdaa3fcbe15Tadashi G. Takaoka @Nonnull 591f62b5d633d17b94a7ea46c968e073fdaa3fcbe15Tadashi G. Takaoka private static Locale getLocaleUsedForToTitleCase(@Nonnull final Locale locale) { 592f62b5d633d17b94a7ea46c968e073fdaa3fcbe15Tadashi G. Takaoka // In Greek locale {@link String#toUpperCase(Locale)} eliminates accents from its result. 593f62b5d633d17b94a7ea46c968e073fdaa3fcbe15Tadashi G. Takaoka // In order to get accented upper case letter, {@link Locale#ROOT} should be used. 594f62b5d633d17b94a7ea46c968e073fdaa3fcbe15Tadashi G. Takaoka if (LANGUAGE_GREEK.equals(locale.getLanguage())) { 595f62b5d633d17b94a7ea46c968e073fdaa3fcbe15Tadashi G. Takaoka return Locale.ROOT; 596f62b5d633d17b94a7ea46c968e073fdaa3fcbe15Tadashi G. Takaoka } 597f62b5d633d17b94a7ea46c968e073fdaa3fcbe15Tadashi G. Takaoka return locale; 598f62b5d633d17b94a7ea46c968e073fdaa3fcbe15Tadashi G. Takaoka } 599f62b5d633d17b94a7ea46c968e073fdaa3fcbe15Tadashi G. Takaoka 60080980574acfef74f9392da4fdbcba64d911cb66fTadashi G. Takaoka @Nullable 6016c47403e27fd703ece844f4b1b24632721da1772Tadashi G. Takaoka public static String toTitleCaseOfKeyLabel(@Nullable final String label, 6026c47403e27fd703ece844f4b1b24632721da1772Tadashi G. Takaoka @Nonnull final Locale locale) { 6036c47403e27fd703ece844f4b1b24632721da1772Tadashi G. Takaoka if (label == null) { 6046c47403e27fd703ece844f4b1b24632721da1772Tadashi G. Takaoka return label; 60580980574acfef74f9392da4fdbcba64d911cb66fTadashi G. Takaoka } 6066c47403e27fd703ece844f4b1b24632721da1772Tadashi G. Takaoka return label.toUpperCase(getLocaleUsedForToTitleCase(locale)); 607639bf62e4fc12c4f75ed0491512f6ed9d2fc2432Tadashi G. Takaoka } 608639bf62e4fc12c4f75ed0491512f6ed9d2fc2432Tadashi G. Takaoka 6096c47403e27fd703ece844f4b1b24632721da1772Tadashi G. Takaoka public static int toTitleCaseOfKeyCode(final int code, @Nonnull final Locale locale) { 6106c47403e27fd703ece844f4b1b24632721da1772Tadashi G. Takaoka if (!Constants.isLetterCode(code)) { 61180980574acfef74f9392da4fdbcba64d911cb66fTadashi G. Takaoka return code; 61280980574acfef74f9392da4fdbcba64d911cb66fTadashi G. Takaoka } 6136c47403e27fd703ece844f4b1b24632721da1772Tadashi G. Takaoka final String label = newSingleCodePointString(code); 6146c47403e27fd703ece844f4b1b24632721da1772Tadashi G. Takaoka final String titleCaseLabel = toTitleCaseOfKeyLabel(label, locale); 6156c47403e27fd703ece844f4b1b24632721da1772Tadashi G. Takaoka return codePointCount(titleCaseLabel) == 1 6166c47403e27fd703ece844f4b1b24632721da1772Tadashi G. Takaoka ? titleCaseLabel.codePointAt(0) : Constants.CODE_UNSPECIFIED; 617639bf62e4fc12c4f75ed0491512f6ed9d2fc2432Tadashi G. Takaoka } 618ff8405cdfbd575657a6f615a1ac4d86eb1b07f74Tadashi G. Takaoka 61980980574acfef74f9392da4fdbcba64d911cb66fTadashi G. Takaoka public static int getTrailingSingleQuotesCount(@Nonnull final CharSequence charSequence) { 62034873a66f03e0b9945474213fa2bc48cc272a7caKeisuke Kuroyanagi final int lastIndex = charSequence.length() - 1; 62134873a66f03e0b9945474213fa2bc48cc272a7caKeisuke Kuroyanagi int i = lastIndex; 62234873a66f03e0b9945474213fa2bc48cc272a7caKeisuke Kuroyanagi while (i >= 0 && charSequence.charAt(i) == Constants.CODE_SINGLE_QUOTE) { 62334873a66f03e0b9945474213fa2bc48cc272a7caKeisuke Kuroyanagi --i; 62434873a66f03e0b9945474213fa2bc48cc272a7caKeisuke Kuroyanagi } 62534873a66f03e0b9945474213fa2bc48cc272a7caKeisuke Kuroyanagi return lastIndex - i; 62634873a66f03e0b9945474213fa2bc48cc272a7caKeisuke Kuroyanagi } 62734873a66f03e0b9945474213fa2bc48cc272a7caKeisuke Kuroyanagi 62880980574acfef74f9392da4fdbcba64d911cb66fTadashi G. Takaoka @UsedForTesting 629ff8405cdfbd575657a6f615a1ac4d86eb1b07f74Tadashi G. Takaoka public static class Stringizer<E> { 63080980574acfef74f9392da4fdbcba64d911cb66fTadashi G. Takaoka @Nonnull 63180980574acfef74f9392da4fdbcba64d911cb66fTadashi G. Takaoka private static final String[] EMPTY_STRING_ARRAY = new String[0]; 63280980574acfef74f9392da4fdbcba64d911cb66fTadashi G. Takaoka 63380980574acfef74f9392da4fdbcba64d911cb66fTadashi G. Takaoka @UsedForTesting 63480980574acfef74f9392da4fdbcba64d911cb66fTadashi G. Takaoka @Nonnull 63580980574acfef74f9392da4fdbcba64d911cb66fTadashi G. Takaoka public String stringize(@Nullable final E element) { 63680980574acfef74f9392da4fdbcba64d911cb66fTadashi G. Takaoka if (element == null) { 63780980574acfef74f9392da4fdbcba64d911cb66fTadashi G. Takaoka return "null"; 63880980574acfef74f9392da4fdbcba64d911cb66fTadashi G. Takaoka } 63980980574acfef74f9392da4fdbcba64d911cb66fTadashi G. Takaoka return element.toString(); 640ff8405cdfbd575657a6f615a1ac4d86eb1b07f74Tadashi G. Takaoka } 641ff8405cdfbd575657a6f615a1ac4d86eb1b07f74Tadashi G. Takaoka 64280980574acfef74f9392da4fdbcba64d911cb66fTadashi G. Takaoka @UsedForTesting 64380980574acfef74f9392da4fdbcba64d911cb66fTadashi G. Takaoka @Nonnull 64480980574acfef74f9392da4fdbcba64d911cb66fTadashi G. Takaoka public final String join(@Nullable final E[] array) { 645ff8405cdfbd575657a6f615a1ac4d86eb1b07f74Tadashi G. Takaoka return joinStringArray(toStringArray(array), null /* delimiter */); 646ff8405cdfbd575657a6f615a1ac4d86eb1b07f74Tadashi G. Takaoka } 647ff8405cdfbd575657a6f615a1ac4d86eb1b07f74Tadashi G. Takaoka 64880980574acfef74f9392da4fdbcba64d911cb66fTadashi G. Takaoka @UsedForTesting 64980980574acfef74f9392da4fdbcba64d911cb66fTadashi G. Takaoka public final String join(@Nullable final E[] array, @Nullable final String delimiter) { 650ff8405cdfbd575657a6f615a1ac4d86eb1b07f74Tadashi G. Takaoka return joinStringArray(toStringArray(array), delimiter); 651ff8405cdfbd575657a6f615a1ac4d86eb1b07f74Tadashi G. Takaoka } 652ff8405cdfbd575657a6f615a1ac4d86eb1b07f74Tadashi G. Takaoka 65380980574acfef74f9392da4fdbcba64d911cb66fTadashi G. Takaoka @Nonnull 65480980574acfef74f9392da4fdbcba64d911cb66fTadashi G. Takaoka protected String[] toStringArray(@Nullable final E[] array) { 65580980574acfef74f9392da4fdbcba64d911cb66fTadashi G. Takaoka if (array == null) { 65680980574acfef74f9392da4fdbcba64d911cb66fTadashi G. Takaoka return EMPTY_STRING_ARRAY; 65780980574acfef74f9392da4fdbcba64d911cb66fTadashi G. Takaoka } 658ff8405cdfbd575657a6f615a1ac4d86eb1b07f74Tadashi G. Takaoka final String[] stringArray = new String[array.length]; 659ff8405cdfbd575657a6f615a1ac4d86eb1b07f74Tadashi G. Takaoka for (int index = 0; index < array.length; index++) { 660ff8405cdfbd575657a6f615a1ac4d86eb1b07f74Tadashi G. Takaoka stringArray[index] = stringize(array[index]); 661ff8405cdfbd575657a6f615a1ac4d86eb1b07f74Tadashi G. Takaoka } 662ff8405cdfbd575657a6f615a1ac4d86eb1b07f74Tadashi G. Takaoka return stringArray; 663ff8405cdfbd575657a6f615a1ac4d86eb1b07f74Tadashi G. Takaoka } 664ff8405cdfbd575657a6f615a1ac4d86eb1b07f74Tadashi G. Takaoka 66580980574acfef74f9392da4fdbcba64d911cb66fTadashi G. Takaoka @Nonnull 66680980574acfef74f9392da4fdbcba64d911cb66fTadashi G. Takaoka protected String joinStringArray(@Nonnull final String[] stringArray, 66780980574acfef74f9392da4fdbcba64d911cb66fTadashi G. Takaoka @Nullable final String delimiter) { 668ff8405cdfbd575657a6f615a1ac4d86eb1b07f74Tadashi G. Takaoka if (delimiter == null) { 669ff8405cdfbd575657a6f615a1ac4d86eb1b07f74Tadashi G. Takaoka return Arrays.toString(stringArray); 670ff8405cdfbd575657a6f615a1ac4d86eb1b07f74Tadashi G. Takaoka } 671ff8405cdfbd575657a6f615a1ac4d86eb1b07f74Tadashi G. Takaoka final StringBuilder sb = new StringBuilder(); 672ff8405cdfbd575657a6f615a1ac4d86eb1b07f74Tadashi G. Takaoka for (int index = 0; index < stringArray.length; index++) { 673ff8405cdfbd575657a6f615a1ac4d86eb1b07f74Tadashi G. Takaoka sb.append(index == 0 ? "[" : delimiter); 674ff8405cdfbd575657a6f615a1ac4d86eb1b07f74Tadashi G. Takaoka sb.append(stringArray[index]); 675ff8405cdfbd575657a6f615a1ac4d86eb1b07f74Tadashi G. Takaoka } 676ff8405cdfbd575657a6f615a1ac4d86eb1b07f74Tadashi G. Takaoka return sb + "]"; 677ff8405cdfbd575657a6f615a1ac4d86eb1b07f74Tadashi G. Takaoka } 678ff8405cdfbd575657a6f615a1ac4d86eb1b07f74Tadashi G. Takaoka } 6795896153f931ac876ad667d73d7703289a63e9bf5Yohei Yukawa 6805896153f931ac876ad667d73d7703289a63e9bf5Yohei Yukawa /** 6815896153f931ac876ad667d73d7703289a63e9bf5Yohei Yukawa * Returns whether the last composed word contains line-breaking character (e.g. CR or LF). 6825896153f931ac876ad667d73d7703289a63e9bf5Yohei Yukawa * @param text the text to be examined. 6835896153f931ac876ad667d73d7703289a63e9bf5Yohei Yukawa * @return {@code true} if the last composed word contains line-breaking separator. 6845896153f931ac876ad667d73d7703289a63e9bf5Yohei Yukawa */ 68580980574acfef74f9392da4fdbcba64d911cb66fTadashi G. Takaoka public static boolean hasLineBreakCharacter(@Nullable final String text) { 686ca724ef71f70044f48e2360e2627922be8ca24d5Jean Chalard if (isEmpty(text)) { 6875896153f931ac876ad667d73d7703289a63e9bf5Yohei Yukawa return false; 6885896153f931ac876ad667d73d7703289a63e9bf5Yohei Yukawa } 6895896153f931ac876ad667d73d7703289a63e9bf5Yohei Yukawa for (int i = text.length() - 1; i >= 0; --i) { 6905896153f931ac876ad667d73d7703289a63e9bf5Yohei Yukawa final char c = text.charAt(i); 6915896153f931ac876ad667d73d7703289a63e9bf5Yohei Yukawa switch (c) { 6925896153f931ac876ad667d73d7703289a63e9bf5Yohei Yukawa case CHAR_LINE_FEED: 6935896153f931ac876ad667d73d7703289a63e9bf5Yohei Yukawa case CHAR_VERTICAL_TAB: 6945896153f931ac876ad667d73d7703289a63e9bf5Yohei Yukawa case CHAR_FORM_FEED: 6955896153f931ac876ad667d73d7703289a63e9bf5Yohei Yukawa case CHAR_CARRIAGE_RETURN: 6965896153f931ac876ad667d73d7703289a63e9bf5Yohei Yukawa case CHAR_NEXT_LINE: 6975896153f931ac876ad667d73d7703289a63e9bf5Yohei Yukawa case CHAR_LINE_SEPARATOR: 6985896153f931ac876ad667d73d7703289a63e9bf5Yohei Yukawa case CHAR_PARAGRAPH_SEPARATOR: 6995896153f931ac876ad667d73d7703289a63e9bf5Yohei Yukawa return true; 7005896153f931ac876ad667d73d7703289a63e9bf5Yohei Yukawa } 7015896153f931ac876ad667d73d7703289a63e9bf5Yohei Yukawa } 7025896153f931ac876ad667d73d7703289a63e9bf5Yohei Yukawa return false; 7035896153f931ac876ad667d73d7703289a63e9bf5Yohei Yukawa } 704cc8c8b99bd0463f5977dea82f5e2379ea1dd4e73Tadashi G. Takaoka} 705