完善时间范围解析

kusen-alpha · Nov 28, 2023 · 2f9c02d · 2f9c02d
1 parent 9a63c78
commit 2f9c02d
Show file tree

Hide file tree

Showing 3 changed files with 38 additions and 12 deletions.
diff --git a/README.md b/README.md
@@ -79,7 +79,7 @@
    ......
 ```
 
-4. 指定formats
+4. 指定formats解析
 
 ```python
 import gggdtparser
@@ -88,7 +88,7 @@ dt = gggdtparser.parse("2023-02-20", format_list=["%Y-%m-%d"])
 print(dt)
 ```
 
-5. 指定正则
+5. 指定正则解析
 
 正则使用有名分组形式，对应关系如下：
 
@@ -150,6 +150,17 @@ print(dt)
 
 ```
 
+7. 时间范围解析
+```python
+
+import datetime
+from gggdtparser import parse_frame
+
+print(parse_frame('2022年10月1日至2023年10月1日', regex_list=[None, ('(?P<am>\d+)\s*(个)?月',)], base_datetime=datetime.datetime(year=2023, month=1, day=1)))
+print(parse_frame('10个月', regex_list=[None, ('(?P<am>\d+)\s*(个)?月',)], base_datetime=datetime.datetime(year=2023, month=1, day=1)))
+
+```
+
 ## 待完善
 
 1. 兼容更多语言

diff --git a/gggdtparser/__init__.py b/gggdtparser/__init__.py
@@ -8,3 +8,4 @@
 from .dtparser import check
 from .dtparser import parse_by_format
 from .dtparser import parse_by_regex
+from .dtframe import parse as parse_frame
diff --git a/gggdtparser/dtframe.py b/gggdtparser/dtframe.py
@@ -10,25 +10,36 @@
 from . import dtparser
 
 SEPS = (
-    '至'
+    '至',
+    '-'
 )
 
 
 def _parse(s, sep, start_format_list, end_format_list, start_regex_list,
-           end_regex_list):
+           end_regex_list, base_datetime):
     start, end = None, None
     try:
         left, right = s.split(sep)
     except ValueError:
+        if base_datetime:
+            start = base_datetime
+            end = dtparser.parse(
+                s, end_format_list, end_regex_list,
+                base_datetime=base_datetime)
         return start, end
     if left:
-        start = dtparser.parse(left, start_format_list, start_regex_list)
+        start = dtparser.parse(
+            left, start_format_list, start_regex_list,
+            base_datetime=base_datetime)
     if right:
-        end = dtparser.parse(right, end_format_list, end_regex_list)
+        end = dtparser.parse(
+            right, end_format_list, end_regex_list,
+            base_datetime=base_datetime)
     return start, end
 
 
-def parse(s, seps=None, format_list=None, regex_list=None):
+def parse(s, seps=None, format_list=None, regex_list=None,
+          base_datetime=None, ):
     """
 
     :param s:
@@ -37,28 +48,31 @@ def parse(s, seps=None, format_list=None, regex_list=None):
         [(start, start, ....), (end, end, ....)]
     :param regex_list: 正则表达式
         [(start, start, ....), (end, end, ....)]
+    :param base_datetime: 相对时间的基准，当起始时间没有解析到时，当做时间的范围的起始
+
     :return:
     """
     if not seps:
         seps = []
     start_format_list, end_format_list = format_list if format_list else (
         None, None)
-    start_regex_list, end_regex_list = regex_list if regex_list else [], []
+
+    start_regex_list, end_regex_list = regex_list if regex_list else (
+        None, None)
     for sep in seps:
         start, end = _parse(
             s, sep, start_format_list, end_format_list,
-            start_regex_list, end_regex_list)
+            start_regex_list, end_regex_list, base_datetime)
         if any([start, end]):
             return start, end
     for sep in SEPS:
         start, end = _parse(
             s, sep, start_format_list, end_format_list,
-            start_regex_list, end_regex_list)
+            start_regex_list, end_regex_list, base_datetime)
         if any([start, end]):
             return start, end
     return None, None
 
 
 if __name__ == '__main__':
-    print(parse('2022年10月1日至2023年10月1日',
-                format_list=[('%Y年%m月%d日',), None]))
+    print(parse('2022年10月1日至2023年10月1日',))