feat: 不派生中英混输词库的大些形式； close #344

iDvel · Jun 13, 2023 · 6f51bdd · 6f51bdd
1 parent 67979e9
commit 6f51bdd
Show file tree

Hide file tree

Showing 8 changed files with 2,755 additions and 2,741 deletions.
diff --git a/en_dicts/cn_en.dict.yaml b/en_dicts/cn_en.dict.yaml
diff --git a/en_dicts/cn_en_double_pinyin.dict.yaml b/en_dicts/cn_en_double_pinyin.dict.yaml
diff --git a/en_dicts/cn_en_double_pinyin_abc.dict.yaml b/en_dicts/cn_en_double_pinyin_abc.dict.yaml
diff --git a/en_dicts/cn_en_double_pinyin_flypy.dict.yaml b/en_dicts/cn_en_double_pinyin_flypy.dict.yaml
diff --git a/en_dicts/cn_en_double_pinyin_mspy.dict.yaml b/en_dicts/cn_en_double_pinyin_mspy.dict.yaml
diff --git a/en_dicts/cn_en_double_pinyin_ziguang.dict.yaml b/en_dicts/cn_en_double_pinyin_ziguang.dict.yaml
diff --git a/melt_eng.schema.yaml b/melt_eng.schema.yaml
@@ -86,16 +86,19 @@ speller:
       - derive/[@]//
       - derive/[-]//
       - derive/[_]//
+
+      # 下面带 ⓘ 的，是为了中英混输词库不派生出纯大写
+
       # 派生无任何非字母数字字符的拼写
-      - derive/[^a-zA-Z0-9]//
+      - derive/[^a-zA-Z0-9ⓘ]//
       # 禁用非英文、数字开头的编码
-      - erase/^[\W|_].+$/
+      - erase/^[^a-zA-Z0-9ⓘ].+$/
       # 全小写
       - derive/^.+$/\L$0/
       # 全大写
-      - derive/^.+$/\U$0/
+      - derive/^(?!ⓘ).+$/\U$0/
       # 首字母大写
-      - derive/^./\U$0/
+      - derive/^(?!ⓘ)./\U$0/
       # 前 2~10 个字母大写
       - derive/^([a-z]{2})/\U$1/
       - derive/^([a-z]{3})/\U$1/
@@ -106,6 +109,9 @@ speller:
       - derive/^([a-z]{8})/\U$1/
       - derive/^([a-z]{9})/\U$1/
       - derive/^([a-z]{10})/\U$1/
+      # 派生中英混输词库（词库编码以ⓘ开头）
+      - derive/^ⓘ(.+)$/$1/
+      - derive/^ⓘ(.+)$/\L$1/
 
 translator:
   dictionary: melt_eng

diff --git a/others/script/rime/cn_en.go b/others/script/rime/cn_en.go
@@ -65,10 +65,11 @@ var polyphones = map[string]string{
 	"AB血型 > 血":      "xue",
 	"O型血 > 血":       "xue",
 	"O血型 > 血":       "xue",
-	"没Bug > 没":      "mei",
-	"没有Bug > 没":     "mei",
-	"卡Bug > 卡":      "ka",
-	"提Bug > 提":      "ti",
+	"没bug > 没":      "mei",
+	"没有bug > 没":     "mei",
+	"卡bug > 卡":      "ka",
+	"查bug > 查":      "cha",
+	"提bug > 提":      "ti",
 	"CT检查 > 查":      "cha",
 	"N卡 > 卡":        "ka",
 	"A卡 > 卡":        "ka",
@@ -412,7 +413,7 @@ func CnEn() {
 		uniq.Add(line)
 		for _, schema := range schemas {
 			code := textToPinyin(line, schema)
-			_, err := schema.file.WriteString(line + "\t" + code + "\n")
+			_, err := schema.file.WriteString(line + "\t" + "ⓘ" + code + "\n")
 			if err != nil {
 				log.Fatalln(err)
 			}
@@ -432,8 +433,9 @@ func writePrefix(s schema) {
 #
 # https://github.com/iDvel/rime-ice
 # ------- 中英混输词库 for %s -------
-# 由 others/cn_en.txt 生成
-#
+# 由 others/cn_en.txt 自动生成
+# 编码前的 ⓘ 符号是为了防止英文方案拼写派生时派生出全大写字母
+# 示例：输入 txu 得到 T恤；输入 Txu 得到 T恤； 输入 TXU 则只会得到 TXU
 ---
 name: %s
 version: "1"