add emojis for history for next-word suggestions

app/src/main/java/org/dslul/openboard/inputmethod/latin/common/StringUtils.java

@@ -700,23 +700,24 @@ public final class StringUtils {
         return false;
     }
 
-    public static boolean probablyContainsEmoji(final String s) {
+    public static boolean mightBeEmoji(final String s) {
         int offset = 0;
         final int length = s.length();
         while (offset < length) {
             int c = Character.codePointAt(s, offset);
-            if (probablyIsEmojiCodePoint(c))
+            if (mightBeEmoji(c))
                 return true;
             offset += Character.charCount(c);
         }
         return false;
     }
 
-    // seemingly arbitrary ranges taken from "somewhere on the internet"
+    // unicode blocks that contain emojis
-    public static boolean probablyIsEmojiCodePoint(final int c) {
+    // very fast check, but there are very few blocks that exclusively contain emojis,
-        return (0x200D <= c && c <= 0x3299) // ??
+    public static boolean mightBeEmoji(final int c) {
-                || (0x1F004 <= c && c <= 0x1F251) // ??
+        return (0x200D <= c && c <= 0x2BFF) // unicode blocks from General Punctuation to Miscellaneous Symbols and Arrows
-                || (0x1F300 <= c && c <= 0x1FFFF) // ??
+                || (0x1F104 <= c && c <= 0x1FAFF) // unicode blocks from Mahjong Tiles to Symbols and Pictographs Extended-A
+                || (0xE0000 <= c && c <= 0xE007F) // unicode block Tags
                 || c == 0xFE0F; // variation selector emoji with color
     }
 

app/src/main/java/org/dslul/openboard/inputmethod/latin/inputlogic/InputLogic.java

@@ -8,7 +8,6 @@ package org.dslul.openboard.inputmethod.latin.inputlogic;
 
 import android.graphics.Color;
 import android.os.SystemClock;
-import android.text.InputType;
 import android.text.SpannableString;
 import android.text.Spanned;
 import android.text.TextUtils;
@@ -42,6 +41,7 @@ import org.dslul.openboard.inputmethod.latin.WordComposer;
 import org.dslul.openboard.inputmethod.latin.common.Constants;
 import org.dslul.openboard.inputmethod.latin.common.InputPointers;
 import org.dslul.openboard.inputmethod.latin.common.StringUtils;
+import org.dslul.openboard.inputmethod.latin.common.StringUtilsKt;
 import org.dslul.openboard.inputmethod.latin.define.DebugFlags;
 import org.dslul.openboard.inputmethod.latin.settings.SettingsValues;
 import org.dslul.openboard.inputmethod.latin.settings.SpacingAndPunctuations;
@@ -231,7 +231,9 @@ public final class InputLogic {
         mConnection.beginBatchEdit();
         if (mWordComposer.isComposingWord()) {
             commitCurrentAutoCorrection(settingsValues, rawText, handler);
+            addToHistoryIfEmoji(rawText, settingsValues); // add emoji after committing text
         } else {
+            addToHistoryIfEmoji(rawText, settingsValues); // add emoji before resetting, otherwise lastComposedWord is empty
             resetComposingState(true /* alsoResetLastComposedWord */);
         }
         handler.postUpdateSuggestionStrip(SuggestedWords.INPUT_STYLE_TYPING);
@@ -841,6 +843,7 @@ public final class InputLogic {
         //  until the next character is entered, and the word is added to history
         //  -> the changing selection would be confusing, and adding partial URLs to history is probably bad
         if (Character.getType(codePoint) == Character.OTHER_SYMBOL
+                || (Character.getType(codePoint) == Character.UNASSIGNED && StringUtils.mightBeEmoji(codePoint)) // outdated java doesn't detect some emojis
                 || (sv.isWordSeparator(codePoint)
                     && (Character.isWhitespace(codePoint) // whitespace is always a separator
                         || !textBeforeCursorMayBeUrlOrSimilar(sv, false) // if text before is not URL or similar, it's a separator
@@ -849,6 +852,7 @@ public final class InputLogic {
                 )
         ) {
             handleSeparatorEvent(event, inputTransaction, handler);
+            addToHistoryIfEmoji(StringUtils.newSingleCodePointString(codePoint), sv);
         } else {
             if (SpaceState.PHANTOM == inputTransaction.getMSpaceState()) {
                 if (mWordComposer.isCursorFrontOrMiddleOfComposingWord()) {
@@ -866,6 +870,25 @@ public final class InputLogic {
         }
     }
 
+    private void addToHistoryIfEmoji(final String text, final SettingsValues settingsValues) {
+        if (mLastComposedWord == LastComposedWord.NOT_A_COMPOSED_WORD // we want a last composed word, also to avoid storing consecutive emojis
+                || mWordComposer.isComposingWord() // emoji will be part of the word in this case, better do nothing
+                || !settingsValues.mAutoCorrectionEnabledPerUserSettings
+                || settingsValues.mIncognitoModeEnabled // add nothing
+                || !StringUtilsKt.isEmoji(text) // obviously we need an emoji
+        ) return;
+        mLastComposedWord = LastComposedWord.NOT_A_COMPOSED_WORD; // avoid storing consecutive emojis
+
+        // commit emoji to dictionary, so it ends up in history and can be suggested as next word
+        mDictionaryFacilitator.addToUserHistory(
+                text,
+                false,
+                mConnection.getNgramContextFromNthPreviousWord(settingsValues.mSpacingAndPunctuations, 2),
+                (int) TimeUnit.MILLISECONDS.toSeconds(System.currentTimeMillis()),
+                settingsValues.mBlockPotentiallyOffensive
+        );
+    }
+
     /**
      * Handle a non-separator.
      * @param event The event to handle.
@@ -1017,7 +1040,7 @@ public final class InputLogic {
                         mConnection.getCodePointBeforeCursor())) {
             needsPrecedingSpace = false;
         } else {
-            needsPrecedingSpace = settingsValues.isUsuallyPrecededBySpace(codePoint);
+            needsPrecedingSpace = settingsValues.isUsuallyPrecededBySpace(codePoint) || StringUtilsKt.isEmoji(codePoint);
         }
 
         if (needsPrecedingSpace) {
@@ -1246,9 +1269,9 @@ public final class InputLogic {
                     }
                     final int lengthToDelete =
                             Character.isSupplementaryCodePoint(codePointBeforeCursor) ? 2 : 1;
-                    if (StringUtils.probablyIsEmojiCodePoint(codePointBeforeCursor)) {
+                    if (StringUtils.mightBeEmoji(codePointBeforeCursor)) {
                         // emoji length varies, so we'd need to find out length to delete correctly
-                        // this is not optimal, but a reasonable workaround for issues when trying to delete emojis
+                        // the solution is not optimal, but a reasonable workaround for issues when trying to delete emojis
                         sendDownUpKeyEvent(KeyEvent.KEYCODE_DEL);
                     } else {
                         mConnection.deleteTextBeforeCursor(lengthToDelete);
@@ -1532,10 +1555,8 @@ public final class InputLogic {
         }
 
         if (TextUtils.isEmpty(suggestion)) return;
-        final boolean wasAutoCapitalized =
+        final boolean wasAutoCapitalized = mWordComposer.wasAutoCapitalized() && !mWordComposer.isMostlyCaps();
-                mWordComposer.wasAutoCapitalized() && !mWordComposer.isMostlyCaps();
+        final int timeStampInSeconds = (int)TimeUnit.MILLISECONDS.toSeconds(System.currentTimeMillis());
-        final int timeStampInSeconds = (int)TimeUnit.MILLISECONDS.toSeconds(
-                System.currentTimeMillis());
         mDictionaryFacilitator.addToUserHistory(stripWordSeparatorsFromEnd(suggestion, settingsValues), wasAutoCapitalized,
                 ngramContext, timeStampInSeconds, settingsValues.mBlockPotentiallyOffensive);
     }

app/src/test/java/org/dslul/openboard/inputmethod/latin/InputLogicTest.kt

@@ -463,6 +463,37 @@ class InputLogicTest {
         assertEquals("b", composingText)
     }
 
+    @Test fun `emoji is added to dictionary`() {
+        // check both text and codepoint input
+        reset()
+        chainInput("hello ")
+        input(0x1F36D)
+        assertEquals(StringUtils.newSingleCodePointString(0x1F36D), lastAddedWord)
+        reset()
+        chainInput("hello ")
+        input("🤗")
+        assertEquals("\uD83E\uDD17", lastAddedWord)
+
+        reset()
+        chainInput("hello ")
+        input("why 🤗 ") // not added because it's not only emoji (input can come from pasting)
+        assertEquals("hello", lastAddedWord)
+    }
+
+    @Test fun `emoji uses phantom space`() {
+        // check both text and codepoint input
+        reset()
+        pickSuggestion("hi")
+        input("🤗")
+        assertEquals("\uD83E\uDD17", lastAddedWord)
+        assertEquals("hi \uD83E\uDD17", text)
+        reset()
+        pickSuggestion("hi")
+        input(0x1F36D)
+        assertEquals(StringUtils.newSingleCodePointString(0x1F36D), lastAddedWord)
+        assertEquals("hi ${StringUtils.newSingleCodePointString(0x1F36D)}", text)
+    }
+
     // ------- helper functions ---------
 
     // should be called before every test, so the same state is guaranteed
@@ -495,7 +526,7 @@ class InputLogicTest {
         handleMessages()
 
         if (phantomSpaceToInsert.isEmpty())
-            assertEquals(oldBefore + phantomSpaceToInsert + insert, textBeforeCursor)
+            assertEquals(oldBefore + insert, textBeforeCursor)
         else // in some cases autospace might be suppressed
             assert(oldBefore + phantomSpaceToInsert + insert == textBeforeCursor || oldBefore + insert == textBeforeCursor)
         assertEquals(oldAfter, textAfterCursor)
@@ -514,11 +545,16 @@ class InputLogicTest {
     private fun input(insert: String) {
         val oldBefore = textBeforeCursor
         val oldAfter = textAfterCursor
+        val phantomSpaceToInsert = if (spaceState == SpaceState.PHANTOM) " " else ""
 
         latinIME.onTextInput(insert)
         handleMessages()
 
+        if (phantomSpaceToInsert.isEmpty())
             assertEquals(oldBefore + insert, textBeforeCursor)
+        else // in some cases autospace might be suppressed
+            assert(oldBefore + phantomSpaceToInsert + insert == textBeforeCursor || oldBefore + insert == textBeforeCursor)
+        assert(oldBefore + insert == textBeforeCursor || "$oldBefore $insert" == textBeforeCursor)
         assertEquals(oldAfter, textAfterCursor)
         assertEquals(textBeforeCursor + textAfterCursor, getText())
         checkConnectionConsistency()