fix emojis not rendering correctly with trailing variant selector. fixes #3819

This commit is contained in:
Daniel Gultsch 2020-07-09 17:43:30 +02:00
parent 14bb8b0cf1
commit 71a56002fe
1 changed files with 228 additions and 227 deletions

View File

@ -29,34 +29,34 @@
package eu.siacs.conversations.utils;
import android.support.annotation.NonNull;
import android.util.LruCache;
import java.util.ArrayList;
import java.util.Arrays;
import java.util.HashMap;
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(0x1F600,0x1F64F);
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);
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, 0x1F64F);
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);
private static final UnicodeBlocks KEYCAP_COMBINEABLE = new UnicodeBlocks(new UnicodeList(0x23),new UnicodeList(0x2A),new UnicodeRange(0x30,0x39));
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,
@ -81,9 +81,9 @@ public class Emoticons {
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 UnicodeRange FITZPATRICK = new UnicodeRange(0x1F3FB, 0x1F3FF);
private static final LruCache<CharSequence,Pattern> CACHE = new LruCache<>(256);
private static final LruCache<CharSequence, Pattern> CACHE = new LruCache<>(256);
private static List<Symbol> parse(String input) {
List<Symbol> symbols = new ArrayList<>();
@ -107,7 +107,7 @@ public class Emoticons {
return symbols;
}
public static Pattern getEmojiPattern(CharSequence input) {
public static Pattern getEmojiPattern(final CharSequence input) {
Pattern pattern = CACHE.get(input);
if (pattern == null) {
pattern = generatePattern(input);
@ -119,7 +119,7 @@ public class Emoticons {
private static Pattern generatePattern(CharSequence input) {
final HashSet<String> emojis = new HashSet<>();
int i = 0;
for(Symbol symbol : parse(input.toString())) {
for (final Symbol symbol : parse(input.toString())) {
if (symbol instanceof Emoji) {
emojis.add(symbol.toString());
if (++i >= MAX_EMOIJS) {
@ -128,7 +128,7 @@ public class Emoticons {
}
}
final StringBuilder pattern = new StringBuilder();
for(String emoji : emojis) {
for (String emoji : emojis) {
if (pattern.length() != 0) {
pattern.append('|');
}
@ -144,7 +144,7 @@ public class Emoticons {
public static boolean isOnlyEmoji(String input) {
List<Symbol> symbols = parse(input);
for(Symbol symbol : symbols) {
for (Symbol symbol : symbols) {
if (!symbol.isEmoji()) {
return false;
}
@ -156,9 +156,9 @@ public class Emoticons {
private final String value;
public Symbol(List<Integer> codepoints) {
StringBuilder builder = new StringBuilder();
for(Integer codepoint : codepoints) {
Symbol(List<Integer> codepoints) {
final StringBuilder builder = new StringBuilder();
for (final Integer codepoint : codepoints) {
builder.appendCodePoint(codepoint);
}
this.value = builder.toString();
@ -166,6 +166,7 @@ public class Emoticons {
abstract boolean isEmoji();
@NonNull
@Override
public String toString() {
return value;
@ -174,7 +175,7 @@ public class Emoticons {
public static class Emoji extends Symbol {
public Emoji(List<Integer> codepoints) {
Emoji(List<Integer> codepoints) {
super(codepoints);
}
@ -211,7 +212,7 @@ public class Emoticons {
add = true;
}
} else {
int previous = codepoints.get(codepoints.size() -1);
int previous = codepoints.get(codepoints.size() - 1);
if (codepoints.get(0) == BLACK_FLAG) {
add = TAGS.contains(codepoint);
} else if (COMBINING_ENCLOSING_KEYCAP == codepoint) {
@ -221,7 +222,7 @@ public class Emoticons {
} else if (REGIONAL_INDICATORS.contains(previous) && REGIONAL_INDICATORS.contains(codepoint)) {
add = codepoints.size() == 1;
} else if (previous == VARIATION_16) {
add = isMerger(codepoint);
add = isMerger(codepoint) || codepoint == VARIATION_16;
} else if (FITZPATRICK.contains(previous)) {
add = codepoint == ZWJ;
} else if (ZWJ == previous) {
@ -250,20 +251,20 @@ public class Emoticons {
} 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);
return codepoints.size() == 0 ? new Other(codepoints) : new Emoji(codepoints);
}
}
public static class UnicodeBlocks implements UnicodeSet {
final UnicodeSet[] unicodeSets;
public UnicodeBlocks(UnicodeSet... sets) {
UnicodeBlocks(final UnicodeSet... sets) {
this.unicodeSets = sets;
}
@Override
public boolean contains(int codepoint) {
for(UnicodeSet unicodeSet : unicodeSets) {
for (UnicodeSet unicodeSet : unicodeSets) {
if (unicodeSet.contains(codepoint)) {
return true;
}
@ -280,7 +281,7 @@ public class Emoticons {
private final List<Integer> list;
public UnicodeList(Integer... codes) {
UnicodeList(final Integer... codes) {
this.list = Arrays.asList(codes);
}