1package eu.siacs.conversations.utils;
2
3import java.util.ArrayList;
4import java.util.List;
5
6public class Emoticons {
7
8 private static final UnicodeRange MISC_SYMBOLS_AND_PICTOGRAPHS = new UnicodeRange(0x1F300,0x1F5FF);
9 private static final UnicodeRange SUPPLEMENTAL_SYMBOLS = new UnicodeRange(0x1F900,0x1F9FF);
10 private static final UnicodeRange EMOTICONS = new UnicodeRange(0x1F600,0x1F64F);
11 private static final UnicodeRange TRANSPORT_SYMBOLS = new UnicodeRange(0x1F680,0x1F6FF);
12 private static final UnicodeRange MISC_SYMBOLS = new UnicodeRange(0x2600,0x26FF);
13 private static final UnicodeRange DINGBATS = new UnicodeRange(0x2700,0x27BF);
14 private static final UnicodeRange ENCLOSED_ALPHANUMERIC_SUPPLEMENT = new UnicodeRange(0x1F100,0x1F1FF);
15 private static final UnicodeRange REGIONAL_INDICATORS = new UnicodeRange(0x1F1E6,0x1F1FF);
16 private static final UnicodeRange GEOMETRIC_SHAPES = new UnicodeRange(0x25A0,0x25FF);
17 private static final UnicodeRange LATIN_SUPPLEMENT = new UnicodeRange(0x80,0xFF);
18 private static final UnicodeRange MISC_TECHNICAL = new UnicodeRange(0x2300,0x23FF);
19 private static final UnicodeBlocks SYMBOLIZE = new UnicodeBlocks(GEOMETRIC_SHAPES, LATIN_SUPPLEMENT);
20 private static final UnicodeBlocks EMOJIS = new UnicodeBlocks(
21 MISC_SYMBOLS_AND_PICTOGRAPHS,
22 SUPPLEMENTAL_SYMBOLS,
23 EMOTICONS,
24 TRANSPORT_SYMBOLS,
25 MISC_SYMBOLS,
26 DINGBATS,
27 ENCLOSED_ALPHANUMERIC_SUPPLEMENT,
28 MISC_TECHNICAL);
29 private static final int ZWJ = 0x200D;
30 private static final int VARIATION_16 = 0xFE0F;
31 private static final UnicodeRange FITZPATRICK = new UnicodeRange(0x1F3FB,0x1F3FF);
32
33 private static List<Symbol> parse(String input) {
34 List<Symbol> symbols = new ArrayList<>();
35 Builder builder = new Builder();
36 boolean needsFinalBuild = false;
37 for (int cp, i = 0; i < input.length(); i += Character.charCount(cp)) {
38 cp = input.codePointAt(i);
39 if (builder.offer(cp)) {
40 needsFinalBuild = true;
41 } else {
42 symbols.add(builder.build());
43 builder = new Builder();
44 if (builder.offer(cp)) {
45 needsFinalBuild = true;
46 }
47 }
48 }
49 if (needsFinalBuild) {
50 symbols.add(builder.build());
51 }
52 return symbols;
53 }
54
55 public static boolean isEmoji(String input) {
56 List<Symbol> symbols = parse(input);
57 return symbols.size() == 1 && symbols.get(0) == Symbol.EMOJI;
58 }
59
60 public static boolean isOnlyEmoji(String input) {
61 List<Symbol> symbols = parse(input);
62 for(Symbol symbol : symbols) {
63 if (symbol == Symbol.NON_EMOJI) {
64 return false;
65 }
66 }
67 return symbols.size() > 0;
68 }
69
70 private enum Symbol {
71 EMOJI, NON_EMOJI
72 }
73
74
75 private static class Builder {
76 private final List<Integer> codepoints = new ArrayList<>();
77
78
79 public boolean offer(int codepoint) {
80 boolean add = false;
81 if (this.codepoints.size() == 0) {
82 if (SYMBOLIZE.contains(codepoint)) {
83 add = true;
84 } else if (REGIONAL_INDICATORS.contains(codepoint)) {
85 add = true;
86 } else if (EMOJIS.contains(codepoint) && !FITZPATRICK.contains(codepoint) && codepoint != ZWJ) {
87 add = true;
88 }
89 } else {
90 int previous = codepoints.get(codepoints.size() -1);
91 if (SYMBOLIZE.contains(previous)) {
92 add = codepoint == VARIATION_16;
93 } else if (REGIONAL_INDICATORS.contains(previous) && REGIONAL_INDICATORS.contains(codepoint)) {
94 add = codepoints.size() == 1;
95 } else if (previous == VARIATION_16) {
96 add = isMerger(codepoint);
97 } else if (FITZPATRICK.contains(previous)) {
98 add = codepoint == ZWJ || EMOJIS.contains(codepoint);
99 } else if (ZWJ == previous) {
100 add = EMOJIS.contains(codepoint) || FITZPATRICK.contains(codepoint);
101 } else if (isMerger(codepoint)) {
102 add = true;
103 } else if (codepoint == VARIATION_16 && EMOJIS.contains(previous)) {
104 add = true;
105 }
106 }
107 if (add) {
108 codepoints.add(codepoint);
109 return true;
110 } else {
111 return false;
112 }
113 }
114
115 private static boolean isMerger(int codepoint) {
116 return codepoint == ZWJ || FITZPATRICK.contains(codepoint);
117 }
118
119 public Symbol build() {
120 if (codepoints.size() > 0 && SYMBOLIZE.contains(codepoints.get(codepoints.size() - 1))) {
121 return Symbol.NON_EMOJI;
122 }
123 return codepoints.size() == 0 ? Symbol.NON_EMOJI : Symbol.EMOJI;
124 }
125 }
126
127 public static class UnicodeBlocks {
128 final UnicodeRange[] ranges;
129
130 public UnicodeBlocks(UnicodeRange... ranges) {
131 this.ranges = ranges;
132 }
133
134 public boolean contains(int codepoint) {
135 for(UnicodeRange range : ranges) {
136 if (range.contains(codepoint)) {
137 return true;
138 }
139 }
140 return false;
141 }
142 }
143
144
145 public static class UnicodeRange {
146
147 private final int lower;
148 private final int upper;
149
150 UnicodeRange(int lower, int upper) {
151 this.lower = lower;
152 this.upper = upper;
153 }
154
155 public boolean contains(int codePoint) {
156 return codePoint >= lower && codePoint <= upper;
157 }
158 }
159}