Revert "Straighten the dead char implementation."

This reverts commit 5ef447aeea97c755008785b36468715922b7ccc1.

Reverting because this broke a bunch of unit tests -- https://sponge.corp.google.com/target?id=b6439742-099e-46cb-8f00-d0c602a4d9bb&target=com.google.android.inputmethod.latin.tests&searchFor=&show=FAILED&sortBy=STATUS

Change-Id: Ibbd9e11372993f62e630fd9da37158ffcf47c17a

1 files changed, 22 insertions, 243 deletions

diff --git a/java/src/com/android/inputmethod/event/DeadKeyCombiner.java b/java/src/com/android/inputmethod/event/DeadKeyCombiner.java
index 88c70630d..4f3f4d25f 100644
--- a/java/src/com/android/inputmethod/event/DeadKeyCombiner.java
+++ b/java/src/com/android/inputmethod/event/DeadKeyCombiner.java
@@ -17,12 +17,10 @@
 package com.android.inputmethod.event;
 
 import android.text.TextUtils;
-import android.util.SparseIntArray;
 import android.view.KeyCharacterMap;
 
 import com.android.inputmethod.latin.Constants;
 
-import java.text.Normalizer;
 import java.util.ArrayList;
 
 import javax.annotation.Nonnull;
@@ -31,209 +29,9 @@ import javax.annotation.Nonnull;
  * A combiner that handles dead keys.
  */
 public class DeadKeyCombiner implements Combiner {
-
-    private static class Data {
-        // This class data taken from KeyCharacterMap.java.
-
-        /* Characters used to display placeholders for dead keys. */
-        private static final int ACCENT_ACUTE = '\u00B4';
-        private static final int ACCENT_BREVE = '\u02D8';
-        private static final int ACCENT_CARON = '\u02C7';
-        private static final int ACCENT_CEDILLA = '\u00B8';
-        private static final int ACCENT_CIRCUMFLEX = '\u02C6';
-        private static final int ACCENT_COMMA_ABOVE = '\u1FBD';
-        private static final int ACCENT_COMMA_ABOVE_RIGHT = '\u02BC';
-        private static final int ACCENT_DOT_ABOVE = '\u02D9';
-        private static final int ACCENT_DOT_BELOW = Constants.CODE_PERIOD; // approximate
-        private static final int ACCENT_DOUBLE_ACUTE = '\u02DD';
-        private static final int ACCENT_GRAVE = '\u02CB';
-        private static final int ACCENT_HOOK_ABOVE = '\u02C0';
-        private static final int ACCENT_HORN = Constants.CODE_SINGLE_QUOTE; // approximate
-        private static final int ACCENT_MACRON = '\u00AF';
-        private static final int ACCENT_MACRON_BELOW = '\u02CD';
-        private static final int ACCENT_OGONEK = '\u02DB';
-        private static final int ACCENT_REVERSED_COMMA_ABOVE = '\u02BD';
-        private static final int ACCENT_RING_ABOVE = '\u02DA';
-        private static final int ACCENT_STROKE = Constants.CODE_DASH; // approximate
-        private static final int ACCENT_TILDE = '\u02DC';
-        private static final int ACCENT_TURNED_COMMA_ABOVE = '\u02BB';
-        private static final int ACCENT_UMLAUT = '\u00A8';
-        private static final int ACCENT_VERTICAL_LINE_ABOVE = '\u02C8';
-        private static final int ACCENT_VERTICAL_LINE_BELOW = '\u02CC';
-
-        /* Legacy dead key display characters used in previous versions of the API (before L)
-         * We still support these characters by mapping them to their non-legacy version. */
-        private static final int ACCENT_GRAVE_LEGACY = Constants.CODE_GRAVE_ACCENT;
-        private static final int ACCENT_CIRCUMFLEX_LEGACY = Constants.CODE_CIRCUMFLEX_ACCENT;
-        private static final int ACCENT_TILDE_LEGACY = Constants.CODE_TILDE;
-
-        /**
-         * Maps Unicode combining diacritical to display-form dead key.
-         */
-        private static final SparseIntArray sCombiningToAccent = new SparseIntArray();
-        private static final SparseIntArray sAccentToCombining = new SparseIntArray();
-        static {
-            // U+0300: COMBINING GRAVE ACCENT
-            addCombining('\u0300', ACCENT_GRAVE);
-            // U+0301: COMBINING ACUTE ACCENT
-            addCombining('\u0301', ACCENT_ACUTE);
-            // U+0302: COMBINING CIRCUMFLEX ACCENT
-            addCombining('\u0302', ACCENT_CIRCUMFLEX);
-            // U+0303: COMBINING TILDE
-            addCombining('\u0303', ACCENT_TILDE);
-            // U+0304: COMBINING MACRON
-            addCombining('\u0304', ACCENT_MACRON);
-            // U+0306: COMBINING BREVE
-            addCombining('\u0306', ACCENT_BREVE);
-            // U+0307: COMBINING DOT ABOVE
-            addCombining('\u0307', ACCENT_DOT_ABOVE);
-            // U+0308: COMBINING DIAERESIS
-            addCombining('\u0308', ACCENT_UMLAUT);
-            // U+0309: COMBINING HOOK ABOVE
-            addCombining('\u0309', ACCENT_HOOK_ABOVE);
-            // U+030A: COMBINING RING ABOVE
-            addCombining('\u030A', ACCENT_RING_ABOVE);
-            // U+030B: COMBINING DOUBLE ACUTE ACCENT
-            addCombining('\u030B', ACCENT_DOUBLE_ACUTE);
-            // U+030C: COMBINING CARON
-            addCombining('\u030C', ACCENT_CARON);
-            // U+030D: COMBINING VERTICAL LINE ABOVE
-            addCombining('\u030D', ACCENT_VERTICAL_LINE_ABOVE);
-            // U+030E: COMBINING DOUBLE VERTICAL LINE ABOVE
-            //addCombining('\u030E', ACCENT_DOUBLE_VERTICAL_LINE_ABOVE);
-            // U+030F: COMBINING DOUBLE GRAVE ACCENT
-            //addCombining('\u030F', ACCENT_DOUBLE_GRAVE);
-            // U+0310: COMBINING CANDRABINDU
-            //addCombining('\u0310', ACCENT_CANDRABINDU);
-            // U+0311: COMBINING INVERTED BREVE
-            //addCombining('\u0311', ACCENT_INVERTED_BREVE);
-            // U+0312: COMBINING TURNED COMMA ABOVE
-            addCombining('\u0312', ACCENT_TURNED_COMMA_ABOVE);
-            // U+0313: COMBINING COMMA ABOVE
-            addCombining('\u0313', ACCENT_COMMA_ABOVE);
-            // U+0314: COMBINING REVERSED COMMA ABOVE
-            addCombining('\u0314', ACCENT_REVERSED_COMMA_ABOVE);
-            // U+0315: COMBINING COMMA ABOVE RIGHT
-            addCombining('\u0315', ACCENT_COMMA_ABOVE_RIGHT);
-            // U+031B: COMBINING HORN
-            addCombining('\u031B', ACCENT_HORN);
-            // U+0323: COMBINING DOT BELOW
-            addCombining('\u0323', ACCENT_DOT_BELOW);
-            // U+0326: COMBINING COMMA BELOW
-            //addCombining('\u0326', ACCENT_COMMA_BELOW);
-            // U+0327: COMBINING CEDILLA
-            addCombining('\u0327', ACCENT_CEDILLA);
-            // U+0328: COMBINING OGONEK
-            addCombining('\u0328', ACCENT_OGONEK);
-            // U+0329: COMBINING VERTICAL LINE BELOW
-            addCombining('\u0329', ACCENT_VERTICAL_LINE_BELOW);
-            // U+0331: COMBINING MACRON BELOW
-            addCombining('\u0331', ACCENT_MACRON_BELOW);
-            // U+0335: COMBINING SHORT STROKE OVERLAY
-            addCombining('\u0335', ACCENT_STROKE);
-            // U+0342: COMBINING GREEK PERISPOMENI
-            //addCombining('\u0342', ACCENT_PERISPOMENI);
-            // U+0344: COMBINING GREEK DIALYTIKA TONOS
-            //addCombining('\u0344', ACCENT_DIALYTIKA_TONOS);
-            // U+0345: COMBINING GREEK YPOGEGRAMMENI
-            //addCombining('\u0345', ACCENT_YPOGEGRAMMENI);
-
-            // One-way mappings to equivalent preferred accents.
-            // U+0340: COMBINING GRAVE TONE MARK
-            sCombiningToAccent.append('\u0340', ACCENT_GRAVE);
-            // U+0341: COMBINING ACUTE TONE MARK
-            sCombiningToAccent.append('\u0341', ACCENT_ACUTE);
-            // U+0343: COMBINING GREEK KORONIS
-            sCombiningToAccent.append('\u0343', ACCENT_COMMA_ABOVE);
-
-            // One-way legacy mappings to preserve compatibility with older applications.
-            // U+0300: COMBINING GRAVE ACCENT
-            sAccentToCombining.append(ACCENT_GRAVE_LEGACY, '\u0300');
-            // U+0302: COMBINING CIRCUMFLEX ACCENT
-            sAccentToCombining.append(ACCENT_CIRCUMFLEX_LEGACY, '\u0302');
-            // U+0303: COMBINING TILDE
-            sAccentToCombining.append(ACCENT_TILDE_LEGACY, '\u0303');
-        }
-
-        private static void addCombining(int combining, int accent) {
-            sCombiningToAccent.append(combining, accent);
-            sAccentToCombining.append(accent, combining);
-        }
-
-        // Caution! This may only contain chars, not supplementary code points. It's unlikely
-        // it will ever need to, but if it does we'll have to change this
-        private static final SparseIntArray sNonstandardDeadCombinations = new SparseIntArray();
-        static {
-            // Non-standard decompositions.
-            // Stroke modifier for Finnish multilingual keyboard and others.
-            // U+0110: LATIN CAPITAL LETTER D WITH STROKE
-            addNonStandardDeadCombination(ACCENT_STROKE, 'D', '\u0110');
-            // U+01E4: LATIN CAPITAL LETTER G WITH STROKE
-            addNonStandardDeadCombination(ACCENT_STROKE, 'G', '\u01e4');
-            // U+0126: LATIN CAPITAL LETTER H WITH STROKE
-            addNonStandardDeadCombination(ACCENT_STROKE, 'H', '\u0126');
-            // U+0197: LATIN CAPITAL LETTER I WITH STROKE
-            addNonStandardDeadCombination(ACCENT_STROKE, 'I', '\u0197');
-            // U+0141: LATIN CAPITAL LETTER L WITH STROKE
-            addNonStandardDeadCombination(ACCENT_STROKE, 'L', '\u0141');
-            // U+00D8: LATIN CAPITAL LETTER O WITH STROKE
-            addNonStandardDeadCombination(ACCENT_STROKE, 'O', '\u00d8');
-            // U+0166: LATIN CAPITAL LETTER T WITH STROKE
-            addNonStandardDeadCombination(ACCENT_STROKE, 'T', '\u0166');
-            // U+0111: LATIN SMALL LETTER D WITH STROKE
-            addNonStandardDeadCombination(ACCENT_STROKE, 'd', '\u0111');
-            // U+01E5: LATIN SMALL LETTER G WITH STROKE
-            addNonStandardDeadCombination(ACCENT_STROKE, 'g', '\u01e5');
-            // U+0127: LATIN SMALL LETTER H WITH STROKE
-            addNonStandardDeadCombination(ACCENT_STROKE, 'h', '\u0127');
-            // U+0268: LATIN SMALL LETTER I WITH STROKE
-            addNonStandardDeadCombination(ACCENT_STROKE, 'i', '\u0268');
-            // U+0142: LATIN SMALL LETTER L WITH STROKE
-            addNonStandardDeadCombination(ACCENT_STROKE, 'l', '\u0142');
-            // U+00F8: LATIN SMALL LETTER O WITH STROKE
-            addNonStandardDeadCombination(ACCENT_STROKE, 'o', '\u00f8');
-            // U+0167: LATIN SMALL LETTER T WITH STROKE
-            addNonStandardDeadCombination(ACCENT_STROKE, 't', '\u0167');
-        }
-
-        private static void addNonStandardDeadCombination(final int deadCodePoint,
-                final int spacingCodePoint, final int result) {
-            final int combination = (deadCodePoint << 16) | spacingCodePoint;
-            sNonstandardDeadCombinations.put(combination, result);
-        }
-
-        public static final int NOT_A_CHAR = 0;
-        public static final int BITS_TO_SHIFT_DEAD_CODE_POINT_FOR_NON_STANDARD_COMBINATION = 16;
-        // Get a non-standard combination
-        public static char getNonstandardCombination(final int deadCodePoint,
-                final int spacingCodePoint) {
-            final int combination = spacingCodePoint |
-                    (deadCodePoint << BITS_TO_SHIFT_DEAD_CODE_POINT_FOR_NON_STANDARD_COMBINATION);
-            return (char)sNonstandardDeadCombinations.get(combination, NOT_A_CHAR);
-        }
-    }
-
     // TODO: make this a list of events instead
     final StringBuilder mDeadSequence = new StringBuilder();
 
-    @Nonnull
-    private Event createEventChainFromSequence(final @Nonnull CharSequence text,
-            final Event originalEvent) {
-        if (text.length() <= 0) {
-            return originalEvent;
-        } else {
-            Event lastEvent = null;
-            int codePoint = 0;
-            for (int i = text.length(); i > 0; i -= Character.charCount(codePoint)) {
-                codePoint = Character.codePointBefore(text, i);
-                final Event thisEvent = Event.createHardwareKeypressEvent(codePoint,
-                        originalEvent.mKeyCode, lastEvent, false /* isKeyRepeat */);
-                lastEvent = thisEvent;
-            }
-            return lastEvent;
-        }
-    }
-
     @Override
     @Nonnull
     public Event processEvent(final ArrayList<Event> previousEvents, final Event event) {
@@ -249,48 +47,29 @@ public class DeadKeyCombiner implements Combiner {
             // simply returns the event as is. The majority of events will go through this path.
             return event;
         } else {
-            if (Character.isWhitespace(event.mCodePoint)
-                    || event.mCodePoint == mDeadSequence.codePointBefore(mDeadSequence.length())) {
-                // When whitespace or twice the same dead key, we should output the dead sequence
-                // as is.
-                final Event resultEvent = createEventChainFromSequence(mDeadSequence.toString(),
-                        event);
-                mDeadSequence.setLength(0);
-                return resultEvent;
-            } else if (event.isFunctionalKeyEvent()) {
-                if (Constants.CODE_DELETE == event.mKeyCode) {
-                    // Remove the last code point
-                    final int trimIndex = mDeadSequence.length() - Character.charCount(
-                            mDeadSequence.codePointBefore(mDeadSequence.length()));
-                    mDeadSequence.setLength(trimIndex);
-                    return Event.createConsumedEvent(event);
-                } else {
-                    return event;
-                }
-            } else if (event.isDead()) {
-                mDeadSequence.appendCodePoint(event.mCodePoint);
-                return Event.createConsumedEvent(event);
+            // TODO: Allow combining for several dead chars rather than only the first one.
+            // The framework doesn't know how to do this now.
+            final int deadCodePoint = mDeadSequence.codePointAt(0);
+            mDeadSequence.setLength(0);
+            final int resultingCodePoint =
+                    KeyCharacterMap.getDeadChar(deadCodePoint, event.mCodePoint);
+            if (0 == resultingCodePoint) {
+                // We can't combine both characters. We need to commit the dead key as a separate
+                // character, and the next char too unless it's a space (because as a special case,
+                // dead key + space should result in only the dead key being committed - that's
+                // how dead keys work).
+                // If the event is a space, we should commit the dead char alone, but if it's
+                // not, we need to commit both.
+                // TODO: this is not necessarily triggered by hardware key events, so it's not
+                // a good idea to masquerade as one. This should be typed as a software
+                // composite event or something.
+                return Event.createHardwareKeypressEvent(deadCodePoint, event.mKeyCode,
+                        Constants.CODE_SPACE == event.mCodePoint ? null : event /* next */,
+                        false /* isKeyRepeat */);
             } else {
-                // Combine normally.
-                final StringBuilder sb = new StringBuilder();
-                sb.appendCodePoint(event.mCodePoint);
-                int codePointIndex = 0;
-                while (codePointIndex < mDeadSequence.length()) {
-                    final int deadCodePoint = mDeadSequence.codePointAt(codePointIndex);
-                    final char replacementSpacingChar =
-                            Data.getNonstandardCombination(deadCodePoint, event.mCodePoint);
-                    if (Data.NOT_A_CHAR != replacementSpacingChar) {
-                        sb.setCharAt(0, replacementSpacingChar);
-                    } else {
-                        final int combining = Data.sAccentToCombining.get(deadCodePoint);
-                        sb.appendCodePoint(0 == combining ? deadCodePoint : combining);
-                    }
-                    codePointIndex += Character.isSupplementaryCodePoint(deadCodePoint) ? 2 : 1;
-                }
-                final String normalizedString = Normalizer.normalize(sb, Normalizer.Form.NFC);
-                final Event resultEvent = createEventChainFromSequence(normalizedString, event);
-                mDeadSequence.setLength(0);
-                return resultEvent;
+                // We could combine the characters.
+                return Event.createHardwareKeypressEvent(resultingCodePoint, event.mKeyCode,
+                        null /* next */, false /* isKeyRepeat */);
             }
         }
     }