Add system

system/process_time.py

@@ -114,50 +114,69 @@ def extract_columns_to_dict(file_path, delimiter='\t'):
     return data_dict
 
 
+import datetime
+import re
+
 def parse_date(date_str):
     if not date_str:
         return None, date_str
+    date_str = date_str.strip()
+
+    # Case 1: YYYY-MM-DD
     try:
         return datetime.datetime.strptime(date_str, "%Y-%m-%d"), date_str
     except ValueError:
-        match = re.search(r'(.*) \(relative to (\d{4}-\d{2}-\d{2})\)', date_str)
-        if match:
-            reference = datetime.datetime.strptime(match.group(2), "%Y-%m-%d")
-            if "Last month" in match.group(1):
-                return reference - datetime.timedelta(days=30), date_str
-            elif "Yesterday" in match.group(1):
-                return reference - datetime.timedelta(days=1), date_str
-            elif "Last week" in match.group(1):
-                return reference - datetime.timedelta(days=7), date_str
-            elif "This week" in match.group(1):
-                return reference, date_str
-        
-        # 处理不同格式的日期
-        match = re.fullmatch(r'\d{4}', date_str)  # 处理年份格式: '2014'
-        if match:
-            return datetime.datetime(int(date_str), 1, 1), date_str
-        
-        match = re.fullmatch(r'(\w+) (\d{4})', date_str)  # 处理月份+年份格式: 'November 2023'
-        if match:
-            try:
-                return datetime.datetime.strptime(date_str, "%B %Y"), date_str
-            except ValueError:
-                return None, date_str
-        
-        match = re.fullmatch(r'(\d{4})-Q(\d)', date_str)  # 处理季度格式: '2024-Q1'
-        if match:
-            year, quarter = int(match.group(1)), int(match.group(2))
-            month = (quarter - 1) * 3 + 1
-            return datetime.datetime(year, month, 1), date_str
-        
-        match = re.fullmatch(r'(\d{4}) (Spring|Summer|Autumn|Fall|Winter)', date_str, re.IGNORECASE)  # 处理季度名称格式: '2023 Autumn' 或 '2023 Fall'
-        if match:
-            year = int(match.group(1))
-            season_map = {"Spring": 3, "Summer": 6, "Autumn": 9, "Fall": 9, "Winter": 12}
-            month = season_map[match.group(2).capitalize()]
-            return datetime.datetime(year, month, 1), date_str
-        
-        return None, date_str
+        pass
+
+    # Case 2: Relative date
+    match = re.search(r'(.*) \(relative to (\d{4}-\d{2}-\d{2})\)', date_str)
+    if match:
+        reference = datetime.datetime.strptime(match.group(2), "%Y-%m-%d")
+        relative_term = match.group(1).strip().lower()
+        if relative_term == "last month":
+            target_date = reference - datetime.timedelta(days=30)
+        elif relative_term == "yesterday":
+            target_date = reference - datetime.timedelta(days=1)
+        elif relative_term == "last week":
+            target_date = reference - datetime.timedelta(days=7)
+        elif relative_term == "this week":
+            target_date = reference
+        else:
+            return None, date_str
+        return target_date, date_str  # ✅ 返回 datetime 对象，不是字符串
+
+    # Case 3: YYYY
+    match = re.fullmatch(r'(\d{4})', date_str)
+    if match:
+        year = int(match.group(1))
+        return datetime.datetime(year, 1, 1), date_str
+
+    # Case 4: Month YYYY
+    match = re.fullmatch(r'(\w+) (\d{4})', date_str)
+    if match:
+        try:
+            target_date = datetime.datetime.strptime(date_str, "%B %Y")
+            return target_date, date_str
+        except ValueError:
+            return None, date_str
+
+    # Case 5: YYYY-QX
+    match = re.fullmatch(r'(\d{4})-Q(\d)', date_str)
+    if match:
+        year, quarter = int(match.group(1)), int(match.group(2))
+        month = (quarter - 1) * 3 + 1
+        return datetime.datetime(year, month, 1), date_str
+
+    # Case 6: YYYY Season
+    match = re.fullmatch(r'(\d{4}) (Spring|Summer|Autumn|Fall|Winter)', date_str, re.IGNORECASE)
+    if match:
+        year = int(match.group(1))
+        season_map = {"spring": 3, "summer": 6, "autumn": 9, "fall": 9, "winter": 12}
+        month = season_map[match.group(2).lower()]
+        return datetime.datetime(year, month, 1), date_str
+
+    return None, date_str
+
 
 def extract_and_sort_events(data_dir, pledge_date, pledge_author, claim, suggestion_meta):
 
@@ -220,10 +239,10 @@ def extract_and_sort_events(data_dir, pledge_date, pledge_author, claim, suggest
         for event in doc.get("output", {}).get("events", []):
             parsed_date, original_date = parse_date(event["date"])
             if parsed_date:
-                if mete_date!= parsed_date:
-                    event_date_and_pub_date = original_date+f" ({mete_date})"
-                else:
-                    event_date_and_pub_date = original_date
+                # if mete_date!= parsed_date:
+                event_date_and_pub_date = parsed_date+f" ({mete_date})"
+                # else:
+                # event_date_and_pub_date = original_date
                 
                 test_instance = f"Pledge: {pledge} (Speaker: {pledge_author}; Pledge Date: {pledge_date})\nEvent Summary: {event['event']} (Event Date: {original_date})\nIs this event summary useful to track the fulfilment of this pledge"
 
@@ -238,10 +257,9 @@ def extract_and_sort_events(data_dir, pledge_date, pledge_author, claim, suggest
                     "event": event["event"],
                     "url": URL,
                     "label": label,
-                    "confident": score
+                    "confident": score,
                 })
     
-    # 按时间排序
     events.sort(key=lambda x: parse_date(x["date"])[0], reverse=True)
     return events