Handle timezones correctly

attrlogimport.py

@@ -3,8 +3,7 @@ import os
 import re
 import sys
 from datetime import datetime
-from zoneinfo import ZoneInfo
-from pytz import timezone
+from pytz import timezone, AmbiguousTimeError, NonExistentTimeError
 
 import psycopg
 
@@ -45,24 +44,59 @@ def create_or_find_device_id(name):
 
 
 def parse_attrlog_file(filename, device_id=None):
-    tz = timezone("Europe/Tallinn")
+    tz_dst = timezone("Europe/Tallinn")
     utc = timezone("UTC")
+    # This is a date when smartmontools switched from UTC time to local time
+    # That change was done in commit b75b99551368da1a8623cd76b3c67bdd3aaceddc
+    smartmontools_update_date = datetime(2021, 9, 16, 18, 00, 00)
+    dst = None  # Is daylight saving time in effect?
+    prev_time = None
+
     fd = open(filename)
     file_size = fd.seek(0, os.SEEK_END)
     fd.seek(0)
+
     while fd.tell() != file_size:
         line = fd.readline()
         line_parts = [p for p in line.strip().split(";") if p.strip()]
-        #dt = line_parts.pop(0)+"+0200"
-        dt = datetime.strptime(line_parts.pop(0), "%Y-%m-%d %H:%M:%S")#.replace(tzinfo=tz)
-        dtz = dt.replace(tzinfo=tz)
+        plain_dt = datetime.strptime(line_parts.pop(0), "%Y-%m-%d %H:%M:%S")
+
+        # Debian 10 to 11 upgrade added timezone and dst support to smartmontools timestamps
+        if plain_dt > smartmontools_update_date:
+            tz = tz_dst
+        else:
+            tz = utc
+
+        try:
+            dt = tz.localize(plain_dt, is_dst=None)
+
+            # We are currently in normal time
+            cur_dst = bool(dt.dst())
+            if dst != cur_dst:
+                dst = cur_dst
+        except (AmbiguousTimeError, NonExistentTimeError):
+            # We are in Ambiguous time where localtime cant be translated to UTC
+            # Hack around it by tracking previous DST and time values
+            dt = tz.localize(plain_dt, is_dst=dst)
+            if prev_time and prev_time > dt:
+                dt = tz.localize(plain_dt, is_dst=not dst)
+
+        prev_time = dt
+        cur_dst = bool(dt.dst())
+
         dtu = dt.astimezone(utc)
-        print(dt, dtz, dtu, sep='; ')
+        #print(plain_dt, dt, dtu, dst, cur_dst, sep='; ')
+
+        if dst is None:
+            dst = cur_dst
+
         while line_parts:
             id = int(line_parts.pop(0))
             norm = int(line_parts.pop(0))
             raw = int(line_parts.pop(0))
-            yield dtu, id, norm, raw, device_id
+            yield str(dtu), id, norm, raw, device_id
+
+            # Pretty progress indicator
             if fd.tell() % 100 == 0:
                 print(f"{int((fd.tell() / file_size)*100):>5}%", end='\r')
     print()
@@ -74,15 +108,16 @@ def import_attrlog_file(filename):
     device_id = create_or_find_device_id(drive_name)
     with conn.cursor() as cur:
         cur.execute("START TRANSACTION")
-        for row in parse_attrlog_file(filename, device_id):
-            cur.execute("""
-            INSERT INTO attrlog (time, id, norm, raw, device_id)
-            VALUES (%s, %s, %s, %s, %s)
-            """, row)
-        # with cur.copy("copy attrlog (time, id, norm, raw, device_id) FROM STDIN") as copy:
-        #     for row in parse_attrlog_file(filename, device_id):
-        #         #print(row)
-        #         copy.write_row(row)
+        # for row in parse_attrlog_file(filename, device_id):
+        #     cur.execute("""
+        #     INSERT INTO attrlog (time, id, norm, raw, device_id)
+        #     VALUES (%s, %s, %s, %s, %s)
+        #     """, row)
+        with cur.copy("copy attrlog (time, id, norm, raw, device_id) FROM STDIN") as copy:
+            for row in parse_attrlog_file(filename, device_id):
+                #print(row)
+                copy.write_row(row)
+        #cur.execute("ROLLBACK")
         cur.execute("COMMIT")
 
 

requirements.txt

@@ -0,0 +1,3 @@
+psycopg==3.0.1
+psycopg-binary==3.0.1
+pytz==2021.3