fix time_parser bug

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-1.4.13-green" /></a>
+        <img src="https://img.shields.io/badge/version-1.4.14-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>

README_en.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-1.4.10-green" /></a>
+        <img src="https://img.shields.io/badge/version-1.4.14-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__ = '1.4.13'
+__version__ = '1.4.14'
 
 
 import os

jionlp/algorithm/ner/time_extractor.py

@@ -87,7 +87,7 @@ def _prepare(self):
         self.four_num_year_pattern = re.compile(r'^[\d]{4}$')
         self.unit_pattern = re.compile(r'(多)?[万亿元]')  # 四数字后接单位，说明非年份
 
-        self.single_char_time = ['春', '夏', '秋', '冬']
+        self.single_char_time = set(['春', '夏', '秋', '冬'])
 
     def __call__(self, text, time_base=time.time(), with_parsing=True, ret_all=False,
                  ret_type='str', ret_future=False, period_results_num=None):

jionlp/gadget/time_parser.py

@@ -434,6 +434,12 @@ def _preprocess_regular_expression(self):
             ''.join([bracket(LIMIT_YEAR_STRING), '第', bracket(WEEK_NUM_STRING),
                      '(个)?', WEEK_STRING]))
 
+        # 1月1  此类不全的日期，缺少日
+        # 注意，此种情况只针对 日 是 阿拉伯数字的情况，若是汉字 日，如 “五月二十”，则按农历进行解析，
+        # 此时，则不存在日期的 “日” 的缺失。
+        self.num_month_num_pattern = re.compile(
+            ''.join(['^', MONTH_NUM_STRING, '月', '([12]\d|3[01]|[0]?[1-9])', '$']))
+
         # 公历固定节日
         self.year_fixed_solar_festival_pattern = re.compile(
             ''.join([bracket_absence(YEAR_STRING), FIXED_SOLAR_FESTIVAL]))
@@ -967,6 +973,21 @@ def _adjust_underlying_future_time(self, time_string):
 
         return time_string
 
+    def _compensate_num_month_num(self, time_string):
+        """ 一种特定的日期类型，“1月1”，没指明 “日”。因此需要进行补全，然后再进行处理。
+
+        Args:
+            time_string:
+
+        Returns:
+
+        """
+        matched_res = self.num_month_num_pattern.search(time_string)
+        if matched_res is not None:
+            return time_string + '日'
+        else:
+            return time_string
+
     def parse_time_span_point(self, time_string):
         # 按照 “从 …… 至 ……” 进行解析
         first_time_string, second_time_string = self.parse_span_2_2_point(time_string)
@@ -976,6 +997,7 @@ def parse_time_span_point(self, time_string):
             old_time_base_handler = self.time_base_handler
             try:
                 if first_time_string is not None and second_time_string is None:
+                    first_time_string = self._compensate_num_month_num(first_time_string)
 
                     first_full_time_handler, _, _, blur_time = self.parse_time_point(
                         first_time_string, self.time_base_handler)
@@ -990,6 +1012,8 @@ def parse_time_span_point(self, time_string):
                         second_full_time_handler = self.time_base_handler
                 elif first_time_string is not None and second_time_string is not None:
 
+                    first_time_string = self._compensate_num_month_num(first_time_string)
+                    second_time_string = self._compensate_num_month_num(second_time_string)
                     first_time_string, second_time_string = self._compensate_string(
                         time_string, first_time_string, second_time_string)
 
@@ -1013,6 +1037,8 @@ def parse_time_span_point(self, time_string):
                                 second_full_time_handler[4:] = [0, 0]
 
                 elif first_time_string is None and second_time_string is not None:
+                    second_time_string = self._compensate_num_month_num(second_time_string)
+
                     _, second_full_time_handler, _, blur_time = self.parse_time_point(
                         second_time_string, self.time_base_handler)
 
@@ -1037,6 +1063,8 @@ def parse_time_span_point(self, time_string):
                         time_string, self.time_base_handler)
         else:
             # 非 time span，按 time_point 解析
+            time_string = self._compensate_num_month_num(time_string)
+
             first_full_time_handler, second_full_time_handler, time_type, \
                 blur_time = self.parse_time_point(
                     time_string, self.time_base_handler)

test/test_time_extractor.py

@@ -24,6 +24,8 @@ def test_time_extractor(self):
             ['有十分之一的概率，股票赔钱了。', []],
             ['住在南京网2021-09-21热度 578瞰地', [{'text': '2021-09-21', 'offset': [5, 15], 'type': 'time_point'}]],
             ['根据财税2016 36号文', [{'text': '2016', 'offset': [4, 8], 'type': 'time_span'}]],
+            ['他在10月22出生', [{'text': '10月22', 'offset': [2, 7], 'type': 'time_point'}]],
+            ['1月3至2月10', [{'text': '1月3至2月10', 'offset': [0, 8], 'type': 'time_span'}]],
         ]
 
         for item in text_string_list:

test/test_time_parser.py

@@ -311,6 +311,12 @@ def test_time_parser(self):
             ['上一个月', _ts_1,
              {'type': 'time_point', 'definition': 'blur', 'time': ['2021-05-01 00:00:00', '2021-05-31 23:59:59']}],
 
+            # 残缺 月、日，
+            ['1月3', _ts_1,
+             {'type': 'time_point', 'definition': 'accurate', 'time': ['2021-01-03 00:00:00', '2021-01-03 23:59:59']}],
+            ['十月31', _ts_1,
+             {'type': 'time_point', 'definition': 'accurate', 'time': ['2021-10-31 00:00:00', '2021-10-31 23:59:59']}],
+
             # 限定月、模糊日
             ['本月初', _ts_1,
              {'type': 'time_point', 'definition': 'accurate', 'time': ['2021-06-01 00:00:00', '2021-06-05 23:59:59']}],
@@ -460,6 +466,10 @@ def test_time_parser(self):
             ['今晚八点以后', _ts_1,
              {'type': 'time_span', 'definition': 'accurate', 'time': ['2021-06-14 20:00:00', 'inf']}],
 
+            # 残缺型 ……至……
+            ['1月3至2月10', _ts_1,
+             {'type': 'time_span', 'definition': 'accurate', 'time': ['2021-01-03 00:00:00', '2021-02-10 23:59:59']}],
+
             # time_span，limit 型
             ['前天中午到明天晚上', _ts_1,
              {'type': 'time_span', 'definition': 'blur', 'time': ['2021-06-12 12:00:00', '2021-06-15 23:59:59']}],