fix money re string bug, some chars should not be at the beginning of the money string

Author: dongrixinyu

README.md

@@ -11,7 +11,7 @@
     <a alt="Downloads">
         <img src="https://pepy.tech/badge/jionlp/month" /></a>
     <a alt="Version">
-        <img src="https://img.shields.io/badge/version-1.4.19-green" /></a>
+        <img src="https://img.shields.io/badge/version-1.4.20-green" /></a>
     <a href="https://github.com/dongrixinyu/JioNLP/pulse" alt="Activity">
         <img src="https://img.shields.io/github/commit-activity/m/dongrixinyu/JioNLP?color=blue" /></a>
 </p>

jionlp/__init__.py

@@ -9,7 +9,7 @@
 # website: www.jionlp.com
 """
 
-__version__ = '1.4.19'
+__version__ = '1.4.20'
 
 
 import os

jionlp/dictionary/__init__.py

@@ -4,7 +4,8 @@
 # license: Apache License 2.0
 # Email: [email protected]
 # github: https://github.com/dongrixinyu/JioNLP
-# description: Preprocessing tool for Chinese NLP
+# description: Preprocessing & Parsing tool for Chinese NLP
+# website: www.jionlp.com
 
 
 from .dictionary_loader import char_distribution_loader

jionlp/gadget/money_parser.py

@@ -4,7 +4,9 @@
 # license: Apache License 2.0
 # Email: [email protected]
 # github: https://github.com/dongrixinyu/JioNLP
-# description: Preprocessing tool for Chinese NLP
+# description: Preprocessing & Parsing tool for Chinese NLP
+# website: www.jionlp.com
+
 
 """
 TODO:

jionlp/rule/rule_pattern.py

@@ -438,7 +438,9 @@
 # MONEY_SPAN_GAP_MIDDLE = r'(\~+|\-+|～+|－+|至(?!少)|(?<![达不])到)'
 
 # 该金额抽取正则将会包含一些错例，例如“2019-07-18”此类字符串也将被抽取出，即范围覆盖面过大，且还能被正确解析，因此需要进行错例删除正则进行处理
-MONEY_CHAR_STRING = r'((将)?近|只有|仅|(大)?约(莫|合)?|大概|至少(要)?|不(到|足|超过)?|逾|高于|(高)?达(到)?|^上|(超)?过|超|' \
+# (?! ) 指的是字符串首部不应该包含的字符
+MONEY_CHAR_STRING = r'(?!(余|多))' \
+                    r'((将)?近|只有|仅|(大)?约(莫|合)?|大概|至少(要)?|不(到|足|超过)?|逾|高于|(高)?达(到)?|^上|(超)?过|超|' \
                     r'以上|以下|左右|上下|港币|人民币|(新)?台币|(分|角|毛|块|元)钱?|(人民|港|日|澳|(新)?台)币|圆(整)?|英镑|' \
                     r'美(金|分|刀)|马克|法郎|卢布|泰铢|元((人民|港|日|澳|韩|(新)?台)币)?|(美|港|澳门|日|韩|缅|马|新加坡|欧|' \
                     r'加|加拿大|新西兰|澳|澳大利亚)元|(越(南)?)盾|雷亚尔|' \

test/test_money_extractor.py

@@ -37,6 +37,8 @@ def test_money_extractor(self):
              []],
             ['到了次日的凌晨2许时就发现车被盗了，该车价值2000余元',
              [{'text': '2000余元', 'offset': [22, 28], 'type': 'money'}]],
+            ['年末结转和结余10.56亿元。',
+             [{'text': '10.56亿元', 'offset': [7, 14], 'type': 'money'}]],
         ]
 
         for item in money_string_list: