feat: add more thing and parse time string

src/yuheng/plugin/stat_metadata_clustering/__main__.py

@@ -1,6 +1,8 @@
 import argparse
 import os
 import sys
+import datetime
+import zoneinfo
 
 current_dir = os.path.dirname(os.path.realpath(__file__))
 src_dir = os.path.join(current_dir, "..", "..", "..")
@@ -10,6 +12,9 @@
 from yuheng.component import Node, Way, Relation
 
 
+LOCAL_TIMEZONE = "Asia/Shanghai"
+
+
 def main(**kwargs):
     # if run in standalone cli, only support input a xml file and then parse it to Carto
     # if run by import, both parse xml or pass Carto object is acceptable.
@@ -36,15 +41,61 @@ def main(**kwargs):
             pass
 
         # reverse
-        metadata_frame = []
+        metadata_frame_raw = []
         for id, element in world.node_dict.items():
-            metadata_frame.append(("n" + str(element.id), element.timestamp))
+            metadata_frame_raw.append(
+                (
+                    "n" + str(element.id),
+                    element.timestamp,
+                    element.uid,
+                    element.changeset,
+                )
+            )
         for id, element in world.way_dict.items():
-            metadata_frame.append(("w" + str(element.id), element.timestamp))
+            metadata_frame_raw.append(
+                (
+                    "w" + str(element.id),
+                    element.timestamp,
+                    element.uid,
+                    element.changeset,
+                )
+            )
         for id, element in world.relation_dict.items():
-            metadata_frame.append(("r" + str(element.id), element.timestamp))
+            metadata_frame_raw.append(
+                (
+                    "r" + str(element.id),
+                    element.timestamp,
+                    element.uid,
+                    element.changeset,
+                )
+            )
+        logger.debug(metadata_frame_raw[0])
+        logger.debug(len(metadata_frame_raw))
+        metadata_frame = []  # 因为暂时不引入pandas所以用字典做行，用了pandas就直接行数组插入了。
+        for item in metadata_frame_raw:
+            timezone_server = datetime.timezone.utc
+            timezone_user = zoneinfo.ZoneInfo(LOCAL_TIMEZONE)
+            time_utc_str = item[1]
+            time_utc = datetime.datetime.strptime(
+                time_utc_str, "%Y-%m-%dT%H:%M:%SZ"
+            )
+            time_utc = time_utc.replace(tzinfo=timezone_server)
+            time_user = time_utc.astimezone(timezone_user)
+
+            metadata_frame.append(
+                {
+                    "id": item[0],
+                    "uid": item[2],
+                    "changeset": item[3],
+                    "time_year": time_user.year,
+                    "time_month": time_user.month,
+                    "time_day": time_user.day,
+                    "time_hour": time_user.hour,
+                    "time_minute": time_user.minute,
+                    "time_second": time_user.second,
+                }
+            )
         logger.debug(metadata_frame[0])
-        logger.debug(len(metadata_frame))
 
 
 if __name__ == "__main__":