dict: 日常更新

cn_dicts/8105.dict.yaml

@@ -33,7 +33,7 @@
 # 需要注音的字词设定在 others/script/rime/需要注音.txt
 ---
 name: 8105
-version: "2023-09-03"
+version: "2023-09-08"
 sort: by_weight
 ...
 ### 按需启用
@@ -66,6 +66,8 @@ sort: by_weight
 # 觍	tian
 # 𪨊	song
 # 磺	huang
+# 屮	cao
+# 芔	cao
 
 
 ### 小时候没学过的拼音。。。
@@ -613,6 +615,8 @@ sort: by_weight
 糙	cao	12840
 槽	cao	12412
 漕	cao	2663
+屮	cao	2222
+芔	cao	1111
 艚	cao	50
 螬	cao	26
 𥕢	cao	0
@@ -3586,15 +3590,15 @@ sort: by_weight
 铹	lao	2
 络	lao	1
 𫭼	lao	0
-了	le	3481243
-乐	le	777260
-勒	le	149722
-叻	le	525
-饹	le	55
-泐	le	34
-鳓	le	22
-仂	le	15
-簕	le	8
+了	le	9
+乐	le	8
+勒	le	7
+叻	le	6
+饹	le	5
+泐	le	4
+鳓	le	4
+仂	le	3
+簕	le	2
 嘞	le	1
 雷	lei	560379
 类	lei	542775
@@ -3773,7 +3777,6 @@ sort: by_weight
 廖	liao	31067
 撩	liao	21011
 寥	liao	20857
-了	liao	20000
 僚	liao	19624
 缭	liao	16457
 燎	liao	9587
@@ -3795,6 +3798,7 @@ sort: by_weight
 憭	liao	6
 簝	liao	4
 橑	liao	1
+了	liao	0
 𪤗	liao	0
 列	lie	244324
 裂	lie	174922

cn_dicts/others.dict.yaml

@@ -8,7 +8,7 @@
 #
 ---
 name: others
-version: "2023-09-03"
+version: "2023-09-08"
 sort: by_weight
 ...
 ##### 容错词
@@ -144,6 +144,8 @@ sort: by_weight
 歇斯底里	jie si di li
 暖和	nuan huo
 暖和	nuan he
+模棱两可	mo leng liang ke
+模棱两可	mo ling liang ke
 ### 错字
 曾经	ceng jing
 曾今	ceng jin

en_dicts/en.dict.yaml

@@ -14,7 +14,7 @@
 # 转化应当大写的单词
 ---
 name: en
-version: "2023-08-15"
+version: "2023-09-08"
 sort: by_weight
 ...
 # +_+
@@ -19161,7 +19161,7 @@ toxicity	toxicity
 toxicology	toxicology
 toxins	toxins
 toy	toy
-toyota	toyota
+Toyota	Toyota
 toys	toys
 # tp	tp
 # tr	tr

en_dicts/en_ext.dict.yaml

@@ -7,7 +7,7 @@
 #
 ---
 name: en_ext
-version: "2023-09-03"
+version: "2023-09-08"
 sort: by_weight
 ...
 # 一些杂项
@@ -2263,3 +2263,5 @@ decode	decode
 KeyCastr	KeyCastr
 Laugh Tale	LaughTale
 Joy Boy	JoyBoy
+gotcha	gotcha
+wombat	wombat

lua/corrector.lua

@@ -72,6 +72,7 @@ local corrections = {
 	["tai xing shan"] = { text = "太行山", comment = "tai hang shan" },
 	["jie si di li"] = { text = "歇斯底里", comment = "xie si di li" },
 	["nuan he"] = { text = "暖和", comment = "nuan huo" },
+	["mo ling liang ke"] = { text = "模棱两可", comment = "mo leng liang ke" },
 	-- 错字
 	["ceng jin"] = { text = "曾今", comment = "曾经" },
 	["an nai"] = { text = "按耐", comment = "按捺(na)" },

others/script/rime/others.go

@@ -3,6 +3,7 @@ package rime
 import (
 	"bufio"
 	"fmt"
+	mapset "github.com/deckarep/golang-set/v2"
 	"log"
 	"os"
 	"strconv"
@@ -17,8 +18,9 @@ func Temp() {
 	// GeneratePinyinTest("都挺长的")
 	// GeneratePinyinTest("血条长")
 
-	// findP(ExtPath, "谁")
+	// findP(BasePath, "血")
 	Pinyin(ExtPath)
+	AddWeight(ExtPath, 100)
 }
 
 // 列出字表中多音字的状况：是否参与自动注音
@@ -107,6 +109,7 @@ func findP(dictPath string, ch string) {
 
 	isMark := false
 	sc := bufio.NewScanner(file)
+	set := mapset.NewSet[string]() // 去重用的
 	for sc.Scan() {
 		line := sc.Text()
 		if !isMark {
@@ -125,9 +128,10 @@ func findP(dictPath string, ch string) {
 			log.Fatalln("len(parts) != 3", line)
 		}
 		text := parts[0]
-		if strings.Contains(text, ch) && utf8.RuneCountInString(text) >= 3 {
+		if strings.Contains(text, ch) && utf8.RuneCountInString(text) >= 3 && !set.Contains(text) {
 			outFile.WriteString(line + "\n")
 		} else {
+			set.Add(text)
 			lines = append(lines, line)
 		}
 	}

others/script/rime/错别字.txt

@@ -1,4 +1,8 @@
 # 在 # -_- 上面是错词；下面是包含错词但不是错词。
+阿帕契
+德鲁依
+唐三才
+塔克拉马干
 元文件
 桃花园
 桃花缘