From 7f73bb3d4de0f0647d8358461009d0bc1c96203e Mon Sep 17 00:00:00 2001
From: 兔子 <i@b612.me>
Date: Wed, 12 Apr 2023 16:47:40 +0800
Subject: [PATCH] add emoji search support

---
 search/search.go      | 54 ++++++++++++++++++++++++++++++++++++++++++-
 search/search_test.go | 16 +++++++++++++
 2 files changed, 69 insertions(+), 1 deletion(-)
 create mode 100644 search/search_test.go

diff --git a/search/search.go b/search/search.go
index 26a847e..3ade3a2 100644
--- a/search/search.go
+++ b/search/search.go
@@ -5,22 +5,25 @@ import (
 	"b612.me/starlog"
 	"b612.me/startext"
 	"bufio"
+	"errors"
 	"fmt"
 	"github.com/spf13/cobra"
 	"io/ioutil"
 	"os"
 	"path/filepath"
 	"regexp"
+	"strconv"
 	"strings"
 )
 
 var stFolder string
 var stNum, stMax, stMin int
-var stautoGBK bool
+var stautoGBK, allowEmoji bool
 
 func init() {
 	Cmd.Flags().StringVarP(&stFolder, "folder", "f", "./", "搜索的文件夹")
 	Cmd.Flags().IntVarP(&stNum, "thread-num", "n", 5, "并发搜寻协程数")
+	Cmd.Flags().BoolVarP(&allowEmoji, "allow-emoji", "e", false, "使用\\U输入Emoji")
 	Cmd.Flags().BoolVarP(&stautoGBK, "autogbk", "g", true, "自动GBK识别")
 	Cmd.Flags().IntVar(&stMax, "max", 0, "行最大字数")
 	Cmd.Flags().IntVar(&stMin, "min", 0, "行最小字数")
@@ -35,6 +38,9 @@ var Cmd = &cobra.Command{
 			starlog.Errorln("应当传入两个参数，搜寻文件后缀和搜寻文本")
 			os.Exit(1)
 		}
+		if allowEmoji {
+			args[1], _ = replaceUnicodeEmoji(args[1])
+		}
 		err := searchText(stFolder, args[0], args[1], stNum, stautoGBK, stMax, stMin)
 		if err != nil {
 			os.Exit(2)
@@ -43,6 +49,52 @@ var Cmd = &cobra.Command{
 	},
 }
 
+func replaceUnicodeEmoji(text string) (string, error) {
+	// 查找含有 \U 开头的代码点
+	re := regexp.MustCompile(`\\U([0-9A-Fa-f]{1,8})`)
+	matches := re.FindAllStringSubmatch(text, -1)
+
+	// 如果没有匹配到任何内容，则直接返回原始文本
+	if matches == nil {
+		return text, nil
+	}
+
+	// 将代码点替换为相应的表情符号
+	for _, match := range matches {
+		emoji, err := unicodeToEmoji(match[1])
+		if err != nil {
+			return "", err
+		}
+		text = strings.Replace(text, match[0], emoji, -1)
+	}
+
+	return text, nil
+}
+
+func unicodeToEmoji(codepoint string) (string, error) {
+	// 将16进制字符串转换为 uint32 类型
+	hexcode, err := strconv.ParseUint(codepoint, 16, 32)
+	if err != nil {
+		return "", err
+	}
+
+	// 检查代码点是否位于Unicode BMP（基本多文本平面）中
+	if hexcode > 0x10FFFF {
+		return "", errors.New("invalid Unicode code point")
+	}
+
+	// 将 uint32 类型的代码点转换为 rune 类型
+	r := rune(hexcode)
+
+	// 将 rune 格式化为 8 位宽度的 16 进制数值并前置 0 填充
+	emoji := fmt.Sprintf("%08X", hexcode)
+
+	// 将 rune 类型的字符转换为对应的字符串表情符号
+	emoji = string(r)
+
+	return emoji, nil
+}
+
 func searchText(folder string, filematch string, text string, thread int, autoGBK bool, max, min int) error {
 	data, err := ioutil.ReadDir(folder)
 	if err != nil {
diff --git a/search/search_test.go b/search/search_test.go
new file mode 100644
index 0000000..49b1a1e
--- /dev/null
+++ b/search/search_test.go
@@ -0,0 +1,16 @@
+package search
+
+import (
+	"fmt"
+	"testing"
+)
+
+func TestEmoji(t *testing.T) {
+	fmt.Println(replaceUnicodeEmoji("\\U1F441\\UFE0F"))
+	fmt.Println("👁️")
+	p, _ := replaceUnicodeEmoji("\\U1F441\\UFE0F")
+	fmt.Println("👁️" == p)
+	for _, v := range p {
+		fmt.Printf("%U", v)
+	}
+}