Use sorted int[] to represent word separators

Change-Id: I4103541d99fe59bfcf12379a1298a0a690497846

3 files changed, 77 insertions, 56 deletions

diff --git a/tests/src/com/android/inputmethod/latin/RichInputConnectionAndTextRangeTests.java b/tests/src/com/android/inputmethod/latin/RichInputConnectionAndTextRangeTests.java
index f19d1853e..7f0743543 100644
--- a/tests/src/com/android/inputmethod/latin/RichInputConnectionAndTextRangeTests.java
+++ b/tests/src/com/android/inputmethod/latin/RichInputConnectionAndTextRangeTests.java
@@ -32,6 +32,7 @@ import android.view.inputmethod.InputConnectionWrapper;
 
 import com.android.inputmethod.latin.settings.SpacingAndPunctuations;
 import com.android.inputmethod.latin.utils.RunInLocale;
+import com.android.inputmethod.latin.utils.StringUtils;
 import com.android.inputmethod.latin.utils.TextRange;
 
 import java.util.Locale;
@@ -183,6 +184,12 @@ public class RichInputConnectionAndTextRangeTests extends AndroidTestCase {
     /**
      * Test logic in getting the word range at the cursor.
      */
+    private static final int[] SPACE = { Constants.CODE_SPACE };
+    static final int[] TAB = { Constants.CODE_TAB };
+    private static final int[] SPACE_TAB = StringUtils.toSortedCodePointArray(" \t");
+    // A character that needs surrogate pair to represent its code point (U+2008A).
+    private static final String SUPPLEMENTARY_CHAR = "\uD840\uDC8A";
+
     public void testGetWordRangeAtCursor() {
         ExtractedText et = new ExtractedText();
         final MockInputMethodService mockInputMethodService = new MockInputMethodService();
@@ -194,48 +201,47 @@ public class RichInputConnectionAndTextRangeTests extends AndroidTestCase {
 
         ic.beginBatchEdit();
         // basic case
-        r = ic.getWordRangeAtCursor(" ", 0);
+        r = ic.getWordRangeAtCursor(SPACE, 0);
         assertTrue(TextUtils.equals("word", r.mWord));
 
         // more than one word
-        r = ic.getWordRangeAtCursor(" ", 1);
+        r = ic.getWordRangeAtCursor(SPACE, 1);
         assertTrue(TextUtils.equals("word word", r.mWord));
         ic.endBatchEdit();
 
         // tab character instead of space
         mockInputMethodService.setInputConnection(new MockConnection("one\tword\two", "rd", et));
         ic.beginBatchEdit();
-        r = ic.getWordRangeAtCursor("\t", 1);
+        r = ic.getWordRangeAtCursor(TAB, 1);
         ic.endBatchEdit();
         assertTrue(TextUtils.equals("word\tword", r.mWord));
 
         // only one word doesn't go too far
         mockInputMethodService.setInputConnection(new MockConnection("one\tword\two", "rd", et));
         ic.beginBatchEdit();
-        r = ic.getWordRangeAtCursor("\t", 1);
+        r = ic.getWordRangeAtCursor(TAB, 1);
         ic.endBatchEdit();
         assertTrue(TextUtils.equals("word\tword", r.mWord));
 
         // tab or space
         mockInputMethodService.setInputConnection(new MockConnection("one word\two", "rd", et));
         ic.beginBatchEdit();
-        r = ic.getWordRangeAtCursor(" \t", 1);
+        r = ic.getWordRangeAtCursor(SPACE_TAB, 1);
         ic.endBatchEdit();
         assertTrue(TextUtils.equals("word\tword", r.mWord));
 
         // tab or space multiword
         mockInputMethodService.setInputConnection(new MockConnection("one word\two", "rd", et));
         ic.beginBatchEdit();
-        r = ic.getWordRangeAtCursor(" \t", 2);
+        r = ic.getWordRangeAtCursor(SPACE_TAB, 2);
         ic.endBatchEdit();
         assertTrue(TextUtils.equals("one word\tword", r.mWord));
 
         // splitting on supplementary character
-        final String supplementaryChar = "\uD840\uDC8A";
         mockInputMethodService.setInputConnection(
-                new MockConnection("one word" + supplementaryChar + "wo", "rd", et));
+                new MockConnection("one word" + SUPPLEMENTARY_CHAR + "wo", "rd", et));
         ic.beginBatchEdit();
-        r = ic.getWordRangeAtCursor(supplementaryChar, 0);
+        r = ic.getWordRangeAtCursor(StringUtils.toSortedCodePointArray(SUPPLEMENTARY_CHAR), 0);
         ic.endBatchEdit();
         assertTrue(TextUtils.equals("word", r.mWord));
     }
@@ -265,7 +271,7 @@ public class RichInputConnectionAndTextRangeTests extends AndroidTestCase {
         TextRange r;
         SuggestionSpan[] suggestions;
 
-        r = ic.getWordRangeAtCursor(" ", 0);
+        r = ic.getWordRangeAtCursor(SPACE, 0);
         suggestions = r.getSuggestionSpansAtWord();
         assertEquals(suggestions.length, 1);
         MoreAsserts.assertEquals(suggestions[0].getSuggestions(), SUGGESTIONS1);
@@ -277,7 +283,7 @@ public class RichInputConnectionAndTextRangeTests extends AndroidTestCase {
         text.setSpan(new SuggestionSpan(Locale.ENGLISH, SUGGESTIONS2, 0 /* flags */),
                 10 /* start */, 16 /* end */, 0 /* flags */);
         mockInputMethodService.setInputConnection(new MockConnection(text, cursorPos));
-        r = ic.getWordRangeAtCursor(" ", 0);
+        r = ic.getWordRangeAtCursor(SPACE, 0);
         suggestions = r.getSuggestionSpansAtWord();
         assertEquals(suggestions.length, 2);
         MoreAsserts.assertEquals(suggestions[0].getSuggestions(), SUGGESTIONS1);
@@ -290,7 +296,7 @@ public class RichInputConnectionAndTextRangeTests extends AndroidTestCase {
         text.setSpan(new SuggestionSpan(Locale.ENGLISH, SUGGESTIONS2, 0 /* flags */),
                 5 /* start */, 16 /* end */, 0 /* flags */);
         mockInputMethodService.setInputConnection(new MockConnection(text, cursorPos));
-        r = ic.getWordRangeAtCursor(" ", 0);
+        r = ic.getWordRangeAtCursor(SPACE, 0);
         suggestions = r.getSuggestionSpansAtWord();
         assertEquals(suggestions.length, 1);
         MoreAsserts.assertEquals(suggestions[0].getSuggestions(), SUGGESTIONS1);
@@ -302,7 +308,7 @@ public class RichInputConnectionAndTextRangeTests extends AndroidTestCase {
         text.setSpan(new SuggestionSpan(Locale.ENGLISH, SUGGESTIONS2, 0 /* flags */),
                 10 /* start */, 20 /* end */, 0 /* flags */);
         mockInputMethodService.setInputConnection(new MockConnection(text, cursorPos));
-        r = ic.getWordRangeAtCursor(" ", 0);
+        r = ic.getWordRangeAtCursor(SPACE, 0);
         suggestions = r.getSuggestionSpansAtWord();
         assertEquals(suggestions.length, 1);
         MoreAsserts.assertEquals(suggestions[0].getSuggestions(), SUGGESTIONS1);
@@ -314,7 +320,7 @@ public class RichInputConnectionAndTextRangeTests extends AndroidTestCase {
         text.setSpan(new SuggestionSpan(Locale.ENGLISH, SUGGESTIONS2, 0 /* flags */),
                 5 /* start */, 20 /* end */, 0 /* flags */);
         mockInputMethodService.setInputConnection(new MockConnection(text, cursorPos));
-        r = ic.getWordRangeAtCursor(" ", 0);
+        r = ic.getWordRangeAtCursor(SPACE, 0);
         suggestions = r.getSuggestionSpansAtWord();
         assertEquals(suggestions.length, 1);
         MoreAsserts.assertEquals(suggestions[0].getSuggestions(), SUGGESTIONS1);
@@ -326,7 +332,7 @@ public class RichInputConnectionAndTextRangeTests extends AndroidTestCase {
         text.setSpan(new SuggestionSpan(Locale.ENGLISH, SUGGESTIONS2, 0 /* flags */),
                 5 /* start */, 20 /* end */, 0 /* flags */);
         mockInputMethodService.setInputConnection(new MockConnection(text, cursorPos));
-        r = ic.getWordRangeAtCursor(" ", 0);
+        r = ic.getWordRangeAtCursor(SPACE, 0);
         suggestions = r.getSuggestionSpansAtWord();
         assertEquals(suggestions.length, 0);
     }
diff --git a/tests/src/com/android/inputmethod/latin/utils/RecapitalizeStatusTests.java b/tests/src/com/android/inputmethod/latin/utils/RecapitalizeStatusTests.java
index a52041264..ada80c3fa 100644
--- a/tests/src/com/android/inputmethod/latin/utils/RecapitalizeStatusTests.java
+++ b/tests/src/com/android/inputmethod/latin/utils/RecapitalizeStatusTests.java
@@ -19,31 +19,35 @@ package com.android.inputmethod.latin.utils;
 import android.test.AndroidTestCase;
 import android.test.suitebuilder.annotation.SmallTest;
 
+import com.android.inputmethod.latin.Constants;
+
 import java.util.Locale;
 
 @SmallTest
 public class RecapitalizeStatusTests extends AndroidTestCase {
+    private static final int[] SPACE = { Constants.CODE_SPACE };
+
     public void testTrim() {
         final RecapitalizeStatus status = new RecapitalizeStatus();
-        status.initialize(30, 40, "abcdefghij", Locale.ENGLISH, " ");
+        status.initialize(30, 40, "abcdefghij", Locale.ENGLISH, SPACE);
         status.trim();
         assertEquals("abcdefghij", status.getRecapitalizedString());
         assertEquals(30, status.getNewCursorStart());
         assertEquals(40, status.getNewCursorEnd());
 
-        status.initialize(30, 44, "    abcdefghij", Locale.ENGLISH, " ");
+        status.initialize(30, 44, "    abcdefghij", Locale.ENGLISH, SPACE);
         status.trim();
         assertEquals("abcdefghij", status.getRecapitalizedString());
         assertEquals(34, status.getNewCursorStart());
         assertEquals(44, status.getNewCursorEnd());
 
-        status.initialize(30, 40, "abcdefgh  ", Locale.ENGLISH, " ");
+        status.initialize(30, 40, "abcdefgh  ", Locale.ENGLISH, SPACE);
         status.trim();
         assertEquals("abcdefgh", status.getRecapitalizedString());
         assertEquals(30, status.getNewCursorStart());
         assertEquals(38, status.getNewCursorEnd());
 
-        status.initialize(30, 45, "   abcdefghij  ", Locale.ENGLISH, " ");
+        status.initialize(30, 45, "   abcdefghij  ", Locale.ENGLISH, SPACE);
         status.trim();
         assertEquals("abcdefghij", status.getRecapitalizedString());
         assertEquals(33, status.getNewCursorStart());
@@ -52,7 +56,7 @@ public class RecapitalizeStatusTests extends AndroidTestCase {
 
     public void testRotate() {
         final RecapitalizeStatus status = new RecapitalizeStatus();
-        status.initialize(29, 40, "abcd efghij", Locale.ENGLISH, " ");
+        status.initialize(29, 40, "abcd efghij", Locale.ENGLISH, SPACE);
         status.rotate();
         assertEquals("Abcd Efghij", status.getRecapitalizedString());
         assertEquals(29, status.getNewCursorStart());
@@ -64,7 +68,7 @@ public class RecapitalizeStatusTests extends AndroidTestCase {
         status.rotate();
         assertEquals("Abcd Efghij", status.getRecapitalizedString());
 
-        status.initialize(29, 40, "Abcd Efghij", Locale.ENGLISH, " ");
+        status.initialize(29, 40, "Abcd Efghij", Locale.ENGLISH, SPACE);
         status.rotate();
         assertEquals("ABCD EFGHIJ", status.getRecapitalizedString());
         assertEquals(29, status.getNewCursorStart());
@@ -76,7 +80,7 @@ public class RecapitalizeStatusTests extends AndroidTestCase {
         status.rotate();
         assertEquals("ABCD EFGHIJ", status.getRecapitalizedString());
 
-        status.initialize(29, 40, "ABCD EFGHIJ", Locale.ENGLISH, " ");
+        status.initialize(29, 40, "ABCD EFGHIJ", Locale.ENGLISH, SPACE);
         status.rotate();
         assertEquals("abcd efghij", status.getRecapitalizedString());
         assertEquals(29, status.getNewCursorStart());
@@ -88,7 +92,7 @@ public class RecapitalizeStatusTests extends AndroidTestCase {
         status.rotate();
         assertEquals("abcd efghij", status.getRecapitalizedString());
 
-        status.initialize(29, 39, "AbCDefghij", Locale.ENGLISH, " ");
+        status.initialize(29, 39, "AbCDefghij", Locale.ENGLISH, SPACE);
         status.rotate();
         assertEquals("abcdefghij", status.getRecapitalizedString());
         assertEquals(29, status.getNewCursorStart());
@@ -102,7 +106,7 @@ public class RecapitalizeStatusTests extends AndroidTestCase {
         status.rotate();
         assertEquals("abcdefghij", status.getRecapitalizedString());
 
-        status.initialize(29, 40, "Abcd efghij", Locale.ENGLISH, " ");
+        status.initialize(29, 40, "Abcd efghij", Locale.ENGLISH, SPACE);
         status.rotate();
         assertEquals("abcd efghij", status.getRecapitalizedString());
         assertEquals(29, status.getNewCursorStart());
@@ -116,7 +120,8 @@ public class RecapitalizeStatusTests extends AndroidTestCase {
         status.rotate();
         assertEquals("abcd efghij", status.getRecapitalizedString());
 
-        status.initialize(30, 34, "grüß", Locale.GERMAN, " "); status.rotate();
+        status.initialize(30, 34, "grüß", Locale.GERMAN, SPACE);
+        status.rotate();
         assertEquals("Grüß", status.getRecapitalizedString());
         assertEquals(30, status.getNewCursorStart());
         assertEquals(34, status.getNewCursorEnd());
@@ -133,7 +138,8 @@ public class RecapitalizeStatusTests extends AndroidTestCase {
         assertEquals(30, status.getNewCursorStart());
         assertEquals(34, status.getNewCursorEnd());
 
-        status.initialize(30, 33, "œuf", Locale.FRENCH, " "); status.rotate();
+        status.initialize(30, 33, "œuf", Locale.FRENCH, SPACE);
+        status.rotate();
         assertEquals("Œuf", status.getRecapitalizedString());
         assertEquals(30, status.getNewCursorStart());
         assertEquals(33, status.getNewCursorEnd());
@@ -150,7 +156,8 @@ public class RecapitalizeStatusTests extends AndroidTestCase {
         assertEquals(30, status.getNewCursorStart());
         assertEquals(33, status.getNewCursorEnd());
 
-        status.initialize(30, 33, "œUf", Locale.FRENCH, " "); status.rotate();
+        status.initialize(30, 33, "œUf", Locale.FRENCH, SPACE);
+        status.rotate();
         assertEquals("œuf", status.getRecapitalizedString());
         assertEquals(30, status.getNewCursorStart());
         assertEquals(33, status.getNewCursorEnd());
@@ -171,7 +178,8 @@ public class RecapitalizeStatusTests extends AndroidTestCase {
         assertEquals(30, status.getNewCursorStart());
         assertEquals(33, status.getNewCursorEnd());
 
-        status.initialize(30, 35, "école", Locale.FRENCH, " "); status.rotate();
+        status.initialize(30, 35, "école", Locale.FRENCH, SPACE);
+        status.rotate();
         assertEquals("École", status.getRecapitalizedString());
         assertEquals(30, status.getNewCursorStart());
         assertEquals(35, status.getNewCursorEnd());
diff --git a/tests/src/com/android/inputmethod/latin/utils/StringAndJsonUtilsTests.java b/tests/src/com/android/inputmethod/latin/utils/StringAndJsonUtilsTests.java
index 71966124f..e55c32bd0 100644
--- a/tests/src/com/android/inputmethod/latin/utils/StringAndJsonUtilsTests.java
+++ b/tests/src/com/android/inputmethod/latin/utils/StringAndJsonUtilsTests.java
@@ -19,6 +19,8 @@ package com.android.inputmethod.latin.utils;
 import android.test.AndroidTestCase;
 import android.test.suitebuilder.annotation.SmallTest;
 
+import com.android.inputmethod.latin.Constants;
+
 import java.util.Arrays;
 import java.util.List;
 import java.util.Locale;
@@ -206,42 +208,47 @@ public class StringAndJsonUtilsTests extends AndroidTestCase {
         assertTrue(StringUtils.isIdenticalAfterDowncase(""));
     }
 
-    private static void checkCapitalize(final String src, final String dst, final String separators,
-            final Locale locale) {
-        assertEquals(dst, StringUtils.capitalizeEachWord(src, separators, locale));
+    private static void checkCapitalize(final String src, final String dst,
+            final int[] sortedSeparators, final Locale locale) {
+        assertEquals(dst, StringUtils.capitalizeEachWord(src, sortedSeparators, locale));
         assert(src.equals(dst)
-                == StringUtils.isIdenticalAfterCapitalizeEachWord(src, separators));
+                == StringUtils.isIdenticalAfterCapitalizeEachWord(src, sortedSeparators));
     }
 
+    private static final int[] SPACE = { Constants.CODE_SPACE };
+    private static final int[] SPACE_PERIOD = StringUtils.toSortedCodePointArray(" .");
+    private static final int[] SENTENCE_SEPARATORS =
+            StringUtils.toSortedCodePointArray(" \n.!?*()&");
+    private static final int[] WORD_SEPARATORS = StringUtils.toSortedCodePointArray(" \n.!?*,();&");
+
     public void testCapitalizeEachWord() {
-        checkCapitalize("", "", " ", Locale.ENGLISH);
-        checkCapitalize("test", "Test", " ", Locale.ENGLISH);
-        checkCapitalize("    test", "    Test", " ", Locale.ENGLISH);
-        checkCapitalize("Test", "Test", " ", Locale.ENGLISH);
-        checkCapitalize("    Test", "    Test", " ", Locale.ENGLISH);
-        checkCapitalize(".Test", ".test", " ", Locale.ENGLISH);
-        checkCapitalize(".Test", ".Test", " .", Locale.ENGLISH);
-        checkCapitalize(".Test", ".Test", ". ", Locale.ENGLISH);
-        checkCapitalize("test and retest", "Test And Retest", " .", Locale.ENGLISH);
-        checkCapitalize("Test and retest", "Test And Retest", " .", Locale.ENGLISH);
-        checkCapitalize("Test And Retest", "Test And Retest", " .", Locale.ENGLISH);
-        checkCapitalize("Test And.Retest  ", "Test And.Retest  ", " .", Locale.ENGLISH);
-        checkCapitalize("Test And.retest  ", "Test And.Retest  ", " .", Locale.ENGLISH);
-        checkCapitalize("Test And.retest  ", "Test And.retest  ", " ", Locale.ENGLISH);
-        checkCapitalize("Test And.Retest  ", "Test And.retest  ", " ", Locale.ENGLISH);
-        checkCapitalize("test and ietest", "Test And İetest", " .", new Locale("tr"));
-        checkCapitalize("test and ietest", "Test And Ietest", " .", Locale.ENGLISH);
-        checkCapitalize("Test&Retest", "Test&Retest", " \n.!?*()&", Locale.ENGLISH);
-        checkCapitalize("Test&retest", "Test&Retest", " \n.!?*()&", Locale.ENGLISH);
-        checkCapitalize("test&Retest", "Test&Retest", " \n.!?*()&", Locale.ENGLISH);
+        checkCapitalize("", "", SPACE, Locale.ENGLISH);
+        checkCapitalize("test", "Test", SPACE, Locale.ENGLISH);
+        checkCapitalize("    test", "    Test", SPACE, Locale.ENGLISH);
+        checkCapitalize("Test", "Test", SPACE, Locale.ENGLISH);
+        checkCapitalize("    Test", "    Test", SPACE, Locale.ENGLISH);
+        checkCapitalize(".Test", ".test", SPACE, Locale.ENGLISH);
+        checkCapitalize(".Test", ".Test", SPACE_PERIOD, Locale.ENGLISH);
+        checkCapitalize("test and retest", "Test And Retest", SPACE_PERIOD, Locale.ENGLISH);
+        checkCapitalize("Test and retest", "Test And Retest", SPACE_PERIOD, Locale.ENGLISH);
+        checkCapitalize("Test And Retest", "Test And Retest", SPACE_PERIOD, Locale.ENGLISH);
+        checkCapitalize("Test And.Retest  ", "Test And.Retest  ", SPACE_PERIOD, Locale.ENGLISH);
+        checkCapitalize("Test And.retest  ", "Test And.Retest  ", SPACE_PERIOD, Locale.ENGLISH);
+        checkCapitalize("Test And.retest  ", "Test And.retest  ", SPACE, Locale.ENGLISH);
+        checkCapitalize("Test And.Retest  ", "Test And.retest  ", SPACE, Locale.ENGLISH);
+        checkCapitalize("test and ietest", "Test And İetest", SPACE_PERIOD, new Locale("tr"));
+        checkCapitalize("test and ietest", "Test And Ietest", SPACE_PERIOD, Locale.ENGLISH);
+        checkCapitalize("Test&Retest", "Test&Retest", SENTENCE_SEPARATORS, Locale.ENGLISH);
+        checkCapitalize("Test&retest", "Test&Retest", SENTENCE_SEPARATORS, Locale.ENGLISH);
+        checkCapitalize("test&Retest", "Test&Retest", SENTENCE_SEPARATORS, Locale.ENGLISH);
         checkCapitalize("rest\nrecreation! And in the end...",
-                "Rest\nRecreation! And In The End...", " \n.!?*,();&", Locale.ENGLISH);
+                "Rest\nRecreation! And In The End...", WORD_SEPARATORS, Locale.ENGLISH);
         checkCapitalize("lorem ipsum dolor sit amet", "Lorem Ipsum Dolor Sit Amet",
-                " \n.,!?*()&;", Locale.ENGLISH);
+                WORD_SEPARATORS, Locale.ENGLISH);
         checkCapitalize("Lorem!Ipsum (Dolor) Sit * Amet", "Lorem!Ipsum (Dolor) Sit * Amet",
-                " \n,.;!?*()&", Locale.ENGLISH);
+                WORD_SEPARATORS, Locale.ENGLISH);
         checkCapitalize("Lorem!Ipsum (dolor) Sit * Amet", "Lorem!Ipsum (Dolor) Sit * Amet",
-                " \n,.;!?*()&", Locale.ENGLISH);
+                WORD_SEPARATORS, Locale.ENGLISH);
     }
 
     public void testLooksLikeURL() {