From a250d3830914532cd02da11d2dd5ca10922d2862 Mon Sep 17 00:00:00 2001 From: Stephen Paul Weber Date: Wed, 5 Jun 2024 14:25:35 -0500 Subject: [PATCH] Fix emoji issues by using a maintained library No need to hand-roll everything --- build.gradle | 1 + .../siacs/conversations/utils/Emoticons.java | 247 +----------------- 2 files changed, 7 insertions(+), 241 deletions(-) diff --git a/build.gradle b/build.gradle index 885b5057838c7602fc4bd170c5ca867cda5c208a..7b23257613f9c0723fcc202cbcdc99b65e17b3aa 100644 --- a/build.gradle +++ b/build.gradle @@ -120,6 +120,7 @@ dependencies { implementation 'com.tbuonomo:dotsindicator:4.2' implementation 'com.github.Priyansh-Kedia:OpenGraphParser:2.5.6' implementation 'me.xdrop:fuzzywuzzy:1.4.0' + implementation 'net.fellbaum:jemoji:1.4.1' } ext { diff --git a/src/main/java/eu/siacs/conversations/utils/Emoticons.java b/src/main/java/eu/siacs/conversations/utils/Emoticons.java index 8f2001ce3eed00c14afb647dba63ad158b057b1c..8ab891a68d57e5abe9c942dc576544054a8b95c3 100644 --- a/src/main/java/eu/siacs/conversations/utils/Emoticons.java +++ b/src/main/java/eu/siacs/conversations/utils/Emoticons.java @@ -31,83 +31,16 @@ package eu.siacs.conversations.utils; import android.util.LruCache; -import androidx.annotation.NonNull; - -import java.util.ArrayList; -import java.util.Arrays; import java.util.HashSet; -import java.util.List; import java.util.regex.Pattern; -public class Emoticons { - - private static final UnicodeRange MISC_SYMBOLS_AND_PICTOGRAPHS = new UnicodeRange(0x1F300, 0x1F5FF); - private static final UnicodeRange SUPPLEMENTAL_SYMBOLS = new UnicodeRange(0x1F900, 0x1F9FF); - private static final UnicodeRange EMOTICONS = new UnicodeRange(0x1F600, 0x1FAF6); - //private static final UnicodeRange TRANSPORT_SYMBOLS = new UnicodeRange(0x1F680, 0x1F6FF); - private static final UnicodeRange MISC_SYMBOLS = new UnicodeRange(0x2600, 0x26FF); - private static final UnicodeRange DINGBATS = new UnicodeRange(0x2700, 0x27BF); - private static final UnicodeRange ENCLOSED_ALPHANUMERIC_SUPPLEMENT = new UnicodeRange(0x1F100, 0x1F1FF); - private static final UnicodeRange ENCLOSED_IDEOGRAPHIC_SUPPLEMENT = new UnicodeRange(0x1F200, 0x1F2FF); - private static final UnicodeRange REGIONAL_INDICATORS = new UnicodeRange(0x1F1E6, 0x1F1FF); - private static final UnicodeRange GEOMETRIC_SHAPES = new UnicodeRange(0x25A0, 0x25FF); - private static final UnicodeRange LATIN_SUPPLEMENT = new UnicodeRange(0x80, 0xFF); - private static final UnicodeRange MISC_TECHNICAL = new UnicodeRange(0x2300, 0x23FF); - private static final UnicodeRange TAGS = new UnicodeRange(0xE0020, 0xE007F); - private static final UnicodeList CYK_SYMBOLS_AND_PUNCTUATION = new UnicodeList(0x3030, 0x303D); - private static final UnicodeList LETTERLIKE_SYMBOLS = new UnicodeList(0x2122, 0x2139); +import net.fellbaum.jemoji.EmojiManager; - private static final UnicodeBlocks KEYCAP_COMBINEABLE = new UnicodeBlocks(new UnicodeList(0x23), new UnicodeList(0x2A), new UnicodeRange(0x30, 0x39)); - - private static final UnicodeBlocks SYMBOLIZE = new UnicodeBlocks( - GEOMETRIC_SHAPES, - LATIN_SUPPLEMENT, - CYK_SYMBOLS_AND_PUNCTUATION, - LETTERLIKE_SYMBOLS, - KEYCAP_COMBINEABLE); - private static final UnicodeBlocks EMOJIS = new UnicodeBlocks( - MISC_SYMBOLS_AND_PICTOGRAPHS, - SUPPLEMENTAL_SYMBOLS, - EMOTICONS, - //TRANSPORT_SYMBOLS, - MISC_SYMBOLS, - DINGBATS, - ENCLOSED_ALPHANUMERIC_SUPPLEMENT, - ENCLOSED_IDEOGRAPHIC_SUPPLEMENT, - MISC_TECHNICAL); +public class Emoticons { private static final int MAX_EMOIJS = 42; - - private static final int ZWJ = 0x200D; - private static final int VARIATION_16 = 0xFE0F; - private static final int COMBINING_ENCLOSING_KEYCAP = 0x20E3; - private static final int BLACK_FLAG = 0x1F3F4; - private static final UnicodeRange FITZPATRICK = new UnicodeRange(0x1F3FB, 0x1F3FF); - private static final LruCache CACHE = new LruCache<>(256); - private static List parse(String input) { - List symbols = new ArrayList<>(); - Builder builder = new Builder(); - boolean needsFinalBuild = false; - for (int cp, i = 0; i < input.length(); i += Character.charCount(cp)) { - cp = input.codePointAt(i); - if (builder.offer(cp)) { - needsFinalBuild = true; - } else { - symbols.add(builder.build()); - builder = new Builder(); - if (builder.offer(cp)) { - needsFinalBuild = true; - } - } - } - if (needsFinalBuild) { - symbols.add(builder.build()); - } - return symbols; - } - public static Pattern getEmojiPattern(final CharSequence input) { Pattern pattern = CACHE.get(input); if (pattern == null) { @@ -119,14 +52,8 @@ public class Emoticons { private static Pattern generatePattern(CharSequence input) { final HashSet emojis = new HashSet<>(); - int i = 0; - for (final Symbol symbol : parse(input.toString())) { - if (symbol instanceof Emoji) { - emojis.add(symbol.toString()); - if (++i >= MAX_EMOIJS) { - return Pattern.compile(""); - } - } + for (final var emoji : EmojiManager.extractEmojisInOrder(input.toString())) { + emojis.add(emoji.getUnicode()); } final StringBuilder pattern = new StringBuilder(); for (String emoji : emojis) { @@ -139,172 +66,10 @@ public class Emoticons { } public static boolean isEmoji(String input) { - List symbols = parse(input); - return symbols.size() == 1 && symbols.get(0).isEmoji(); + return EmojiManager.isEmoji(input); } public static boolean isOnlyEmoji(String input) { - List symbols = parse(input); - for (Symbol symbol : symbols) { - if (!symbol.isEmoji()) { - return false; - } - } - return symbols.size() > 0; - } - - private static abstract class Symbol { - - private final String value; - - Symbol(List codepoints) { - final StringBuilder builder = new StringBuilder(); - for (final Integer codepoint : codepoints) { - builder.appendCodePoint(codepoint); - } - this.value = builder.toString(); - } - - abstract boolean isEmoji(); - - @NonNull - @Override - public String toString() { - return value; - } - } - - public static class Emoji extends Symbol { - - Emoji(List codepoints) { - super(codepoints); - } - - @Override - boolean isEmoji() { - return true; - } - } - - public static class Other extends Symbol { - - public Other(List codepoints) { - super(codepoints); - } - - @Override - boolean isEmoji() { - return false; - } - } - - private static class Builder { - private final List codepoints = new ArrayList<>(); - - - public boolean offer(int codepoint) { - boolean add = false; - if (this.codepoints.size() == 0) { - if (SYMBOLIZE.contains(codepoint)) { - add = true; - } else if (REGIONAL_INDICATORS.contains(codepoint)) { - add = true; - } else if (EMOJIS.contains(codepoint) && !FITZPATRICK.contains(codepoint) && codepoint != ZWJ) { - add = true; - } - } else { - int previous = codepoints.get(codepoints.size() - 1); - if (codepoints.get(0) == BLACK_FLAG) { - add = TAGS.contains(codepoint); - } else if (COMBINING_ENCLOSING_KEYCAP == codepoint) { - add = KEYCAP_COMBINEABLE.contains(previous) || previous == VARIATION_16; - } else if (SYMBOLIZE.contains(previous)) { - add = codepoint == VARIATION_16; - } else if (REGIONAL_INDICATORS.contains(previous) && REGIONAL_INDICATORS.contains(codepoint)) { - add = codepoints.size() == 1; - } else if (previous == VARIATION_16) { - add = isMerger(codepoint) || codepoint == VARIATION_16; - } else if (FITZPATRICK.contains(previous)) { - add = codepoint == ZWJ; - } else if (ZWJ == previous) { - add = EMOJIS.contains(codepoint); - } else if (isMerger(codepoint)) { - add = true; - } else if (codepoint == VARIATION_16 && EMOJIS.contains(previous)) { - add = true; - } - } - if (add) { - codepoints.add(codepoint); - return true; - } else { - return false; - } - } - - private static boolean isMerger(int codepoint) { - return codepoint == ZWJ || FITZPATRICK.contains(codepoint); - } - - public Symbol build() { - if (codepoints.size() > 0 && SYMBOLIZE.contains(codepoints.get(codepoints.size() - 1))) { - return new Other(codepoints); - } else if (codepoints.size() > 1 && KEYCAP_COMBINEABLE.contains(codepoints.get(0)) && codepoints.get(codepoints.size() - 1) != COMBINING_ENCLOSING_KEYCAP) { - return new Other(codepoints); - } - return codepoints.size() == 0 ? new Other(codepoints) : new Emoji(codepoints); - } - } - - public static class UnicodeBlocks implements UnicodeSet { - final UnicodeSet[] unicodeSets; - - UnicodeBlocks(final UnicodeSet... sets) { - this.unicodeSets = sets; - } - - @Override - public boolean contains(int codepoint) { - for (UnicodeSet unicodeSet : unicodeSets) { - if (unicodeSet.contains(codepoint)) { - return true; - } - } - return false; - } - } - - public interface UnicodeSet { - boolean contains(int codepoint); - } - - public static class UnicodeList implements UnicodeSet { - - private final List list; - - UnicodeList(final Integer... codes) { - this.list = Arrays.asList(codes); - } - - @Override - public boolean contains(int codepoint) { - return this.list.contains(codepoint); - } - } - - - public static class UnicodeRange implements UnicodeSet { - - private final int lower; - private final int upper; - - UnicodeRange(int lower, int upper) { - this.lower = lower; - this.upper = upper; - } - - public boolean contains(int codePoint) { - return codePoint >= lower && codePoint <= upper; - } + return EmojiManager.removeAllEmojis(input).trim().length() == 0; } }