fix(itn): 货币单位精确到角和分

Author: xingchensong

itn/chinese/rules/money.py

@@ -36,12 +36,17 @@ def build_tagger(self):
             Cardinal().number_exclude_0_to_9
         # 七八美元 => $7~8
         number |= digit + insert("~") + digit
+        # 三千三百八十元五毛八分 => ¥3380.58
         tagger = (insert('value: "') + number + insert('"') +
-                  insert(' currency: "') + (code | symbol) + insert('"'))
+                  insert(' currency: "') + (code | symbol) + insert('"') +
+                  insert(' decimal: "') + (
+                      insert(".") + digit + (delete("毛") | delete("角")) + (digit + delete("分")).ques
+                  ).ques + insert('"'))
         self.tagger = self.add_tokens(tagger)
 
     def build_verbalizer(self):
         currency = delete('currency: "') + self.SIGMA + delete('"')
         value = delete(' value: "') + self.SIGMA + delete('"')
-        verbalizer = currency + value
+        decimal = delete(' decimal: "') + self.SIGMA + delete('"')
+        verbalizer = currency + value + decimal
         self.verbalizer = self.delete_tokens(verbalizer)

itn/chinese/test/data/money.txt

@@ -7,3 +7,5 @@
 四十五六新台币 => TWD45-6
 七百三四欧元 => €730-40
 七百三四十马来西亚令吉 => RM730-40
+三千三百八十元五角八分 => ¥3380.58
+二十五元三毛 => ¥25.3

runtime/processor/wetext_token_parser.cc

@@ -36,7 +36,7 @@ const std::unordered_map<std::string, std::vector<std::string>> ITN_ORDERS = {
     {"date", {"year", "month", "day"}},
     {"fraction", {"sign", "numerator", "denominator"}},
     {"measure", {"numerator", "denominator", "value"}},
-    {"money", {"currency", "value"}},
+    {"money", {"currency", "value", "decimal"}},
     {"time", {"hour", "minute", "second", "noon"}}};
 
 TokenParser::TokenParser(ParseType type) {

tn/token_parser.py

@@ -25,7 +25,7 @@
     'date': ['year', 'month', 'day'],
     'fraction': ['sign', 'numerator', 'denominator'],
     'measure': ['numerator', 'denominator', 'value'],
-    'money': ['currency', 'value'],
+    'money': ['currency', 'value', 'decimal'],
     'time': ['hour', 'minute', 'second', 'noon']}