snapshot of https://github.com/go-gitea/gitea.git (main, 2026-09-15; blobs > 131072B omitted)

modules/emoji/emoji.go

@@ -0,0 +1,146 @@
+// Copyright 2020 The Gitea Authors. All rights reserved.
+// Copyright 2015 Kenneth Shaw
+// SPDX-License-Identifier: MIT
+
+package emoji
+
+import (
+	"sort"
+	"strings"
+	"sync/atomic"
+
+	"gitea.dev/modules/setting"
+	"gitea.dev/modules/util"
+)
+
+// Gemoji is a set of emoji data.
+type Gemoji []Emoji
+
+// Emoji represents a single emoji and associated data.
+type Emoji struct {
+	Emoji          string
+	Description    string
+	Aliases        []string
+	UnicodeVersion string
+	SkinTones      bool
+}
+
+type globalVarsStruct struct {
+	codeMap        map[string]int    // emoji Unicode code to its emoji data.
+	aliasMap       map[string]int    // the alias to its emoji data.
+	trie           *util.TrieNode    // trie for finding emoji positions.
+	isStartingByte [256]bool         // fast-path skip for starting bytes.
+	codeReplacer   *strings.Replacer // string replacer for emoji codes.
+	aliasReplacer  *strings.Replacer // string replacer for emoji aliases.
+}
+
+var globalVarsStore atomic.Pointer[globalVarsStruct]
+
+func globalVars() *globalVarsStruct {
+	vars := globalVarsStore.Load()
+	if vars != nil {
+		return vars
+	}
+	// although there can be concurrent calls, the result should be the same, and there is no performance problem
+	vars = &globalVarsStruct{}
+	vars.codeMap = make(map[string]int, len(GemojiData))
+	vars.aliasMap = make(map[string]int, len(GemojiData))
+	vars.trie = &util.TrieNode{}
+
+	// process emoji codes and aliases
+	codePairs := make([]string, 0)
+	aliasPairs := make([]string, 0)
+
+	// sort from largest to small so we match combined emoji first
+	sort.Slice(GemojiData, func(i, j int) bool {
+		return len(GemojiData[i].Emoji) > len(GemojiData[j].Emoji)
+	})
+
+	for idx, emoji := range GemojiData {
+		if emoji.Emoji == "" || len(emoji.Aliases) == 0 {
+			continue
+		}
+
+		// process aliases
+		firstAlias := ""
+		for _, alias := range emoji.Aliases {
+			if alias == "" {
+				continue
+			}
+			enabled := len(setting.UI.EnabledEmojisSet) == 0 || setting.UI.EnabledEmojisSet.Contains(alias)
+			if !enabled {
+				continue
+			}
+			if firstAlias == "" {
+				firstAlias = alias
+			}
+			vars.aliasMap[alias] = idx
+			aliasPairs = append(aliasPairs, ":"+alias+":", emoji.Emoji)
+		}
+
+		// process emoji code
+		if firstAlias != "" {
+			vars.codeMap[emoji.Emoji] = idx
+			codePairs = append(codePairs, emoji.Emoji, ":"+emoji.Aliases[0]+":")
+			vars.trie.Insert(emoji.Emoji)
+			vars.isStartingByte[emoji.Emoji[0]] = true
+		}
+	}
+
+	// create replacers
+	vars.codeReplacer = strings.NewReplacer(codePairs...)
+	vars.aliasReplacer = strings.NewReplacer(aliasPairs...)
+	globalVarsStore.Store(vars)
+	return vars
+}
+
+// FromCode retrieves the emoji data based on the provided Unicode code
+// e.g.: "\u2618" will return the Gemoji data for "shamrock".
+func FromCode(code string) *Emoji {
+	i, ok := globalVars().codeMap[code]
+	if !ok {
+		return nil
+	}
+
+	return &GemojiData[i]
+}
+
+// FromAlias retrieves the emoji data based on the provided alias in the form "alias" or ":alias:"
+// e.g.: "shamrock" or ":shamrock:" will return the Gemoji data for "shamrock".
+func FromAlias(alias string) *Emoji {
+	if strings.HasPrefix(alias, ":") && strings.HasSuffix(alias, ":") {
+		alias = alias[1 : len(alias)-1]
+	}
+
+	i, ok := globalVars().aliasMap[alias]
+	if !ok {
+		return nil
+	}
+
+	return &GemojiData[i]
+}
+
+// ReplaceCodes replaces all emoji codes with the first corresponding emoji alias in the form of ":alias:"
+// e.g.: "\u2618" will be converted to ":shamrock:".
+func ReplaceCodes(s string) string {
+	return globalVars().codeReplacer.Replace(s)
+}
+
+// ReplaceAliases replaces all aliases of the form ":alias:" with its corresponding Unicode value.
+func ReplaceAliases(s string) string {
+	return globalVars().aliasReplacer.Replace(s)
+}
+
+// FindEmojiSubmatchIndex returns index-pair of the first emoji in a string
+func FindEmojiSubmatchIndex(s string) []int {
+	vars := globalVars()
+	for i := 0; i < len(s); i++ {
+		if !vars.isStartingByte[s[i]] {
+			continue
+		}
+		if matchLen := vars.trie.Match(s, i); matchLen > 0 {
+			return []int{i, i + matchLen}
+		}
+	}
+	return nil
+}

modules/emoji/emoji_test.go

@@ -0,0 +1,125 @@
+// Copyright 2020 The Gitea Authors. All rights reserved.
+// Copyright 2015 Kenneth Shaw
+// SPDX-License-Identifier: MIT
+
+package emoji
+
+import (
+	"testing"
+
+	"gitea.dev/modules/container"
+	"gitea.dev/modules/setting"
+	"gitea.dev/modules/test"
+
+	"github.com/stretchr/testify/assert"
+)
+
+func TestLookup(t *testing.T) {
+	a := FromCode("\U0001f37a")
+	b := FromCode("🍺")
+	c := FromAlias(":beer:")
+	d := FromAlias("beer")
+
+	assert.Equal(t, a, b)
+	assert.Equal(t, b, c)
+	assert.Equal(t, c, d)
+
+	m := FromCode("\U0001f44d")
+	n := FromAlias(":thumbsup:")
+	o := FromAlias("+1")
+
+	assert.Equal(t, m, n)
+	assert.Equal(t, m, o)
+
+	defer test.MockVariableValue(&setting.UI.EnabledEmojisSet, container.SetOf("thumbsup"))()
+	defer globalVarsStore.Store(nil)
+	globalVarsStore.Store(nil)
+	a = FromCode("\U0001f37a")
+	c = FromAlias(":beer:")
+	m = FromCode("\U0001f44d")
+	n = FromAlias(":thumbsup:")
+	o = FromAlias("+1")
+	assert.Nil(t, a)
+	assert.Nil(t, c)
+	assert.NotNil(t, m)
+	assert.NotNil(t, n)
+	assert.Nil(t, o)
+}
+
+func TestReplacers(t *testing.T) {
+	tests := []struct {
+		f      func(string) string
+		v, exp string
+	}{
+		{ReplaceCodes, ":thumbsup: +1 for \U0001f37a! 🍺 \U0001f44d", ":thumbsup: +1 for :beer:! :beer: :+1:"},
+		{ReplaceAliases, ":thumbsup: +1 :+1: :beer:", "\U0001f44d +1 \U0001f44d \U0001f37a"},
+	}
+
+	for i, x := range tests {
+		s := x.f(x.v)
+		assert.Equalf(t, x.exp, s, "test %d `%s` expected `%s`, got: `%s`", i, x.v, x.exp, s)
+	}
+}
+
+const (
+	testInputWithEmojis = "This is a test string containing some emojis like \U0001f44d and \U0001f37a and some text in between."
+	testInputNoEmojis   = "This is a test string containing no emojis at all, just plain old ASCII text, which should ideally be scanned very quickly by our trie implementation."
+)
+
+func TestFindEmojiSubmatchIndex(t *testing.T) {
+	type testcase struct {
+		input    string
+		expected []int
+	}
+
+	testCases := []testcase{
+		{
+			"\U0001f44d",
+			[]int{0, len("\U0001f44d")},
+		},
+		{
+			"\U0001f44d +1 \U0001f44d \U0001f37a",
+			[]int{0, 4},
+		},
+		{
+			" \U0001f44d",
+			[]int{1, 1 + len("\U0001f44d")},
+		},
+		{
+			"\u0001\U0001f44d",
+			[]int{1, 1 + len("\U0001f44d")},
+		},
+		{
+			// This package can handle keycap emoji if it is registered in the emoji data.
+			// However, many other places (e.g.: markup rendering) also might not handle such cases correctly.
+			// For example: how is "**{U+FE0F}{U+20E3}**" rendered in Markdown/Markup?
+			"a 8\U0000fe0f\U000020e3 b", // keycap emoji "8\ufe0f\u20e3" in emoji data
+			[]int{2, 2 + len("8\U0000fe0f\U000020e3")},
+		},
+		{
+			testInputWithEmojis,
+			[]int{50, 54},
+		},
+		{
+			testInputNoEmojis,
+			nil,
+		},
+	}
+
+	for _, tc := range testCases {
+		actual := FindEmojiSubmatchIndex(tc.input)
+		assert.Equal(t, tc.expected, actual)
+	}
+}
+
+func BenchmarkFindEmojiSubmatchIndex(b *testing.B) {
+	for i := 0; i < b.N; i++ {
+		_ = FindEmojiSubmatchIndex(testInputWithEmojis)
+	}
+}
+
+func BenchmarkFindEmojiSubmatchIndexNoMatch(b *testing.B) {
+	for i := 0; i < b.N; i++ {
+		_ = FindEmojiSubmatchIndex(testInputNoEmojis)
+	}
+}