Wip

.gitignore

@@ -1,3 +1,4 @@
 *~
 *.pyc
-home-assistant-core
+home-assistant-core
+.venv

homeassistant2influxdb.py

@@ -1,15 +1,22 @@
-#!/usr/bin/python3
+#!.venv/bin/python3
 # -*- coding: utf-8 -*-
 
 import argparse
 import json
 import yaml
 
 # MySQL / MariaDB
-from MySQLdb import connect as mysql_connect, cursors
+try:
+    from MySQLdb import connect as mysql_connect, cursors
+except:
+    print("Warning: Could not load Mysql driver, might not be a problem if you intend to use sqlite")
 
 # SQLite (not tested)
-#import sqlite3
+try:
+    import sqlite3
+except:
+    print("Warning: Could not load sqlite3 driver, might not be a problem if you intend to use Mysql")
+
 
 # progress bar
 from tqdm import tqdm
@@ -22,6 +29,7 @@
 sys.path.append("home-assistant-core")
 from homeassistant.helpers import location
 from homeassistant.core import Event, State
+from homeassistant.util import dt
 from homeassistant.components.influxdb import get_influx_connection, _generate_event_to_json, INFLUX_SCHEMA
 from homeassistant.exceptions import InvalidEntityFormatError
 
@@ -63,25 +71,38 @@ def main():
     """
 
     parser = argparse.ArgumentParser()
+    parser.add_argument('--type', '-t',
+                        dest='type', action='store', required=True,
+                        help='Database type: MySQL, MariaDB or SQLite')
     parser.add_argument('--user', '-u',
-                        dest='user', action='store', required=True,
+                        dest='user', action='store', required=False,
                         help='MySQL/MariaDB username')
     parser.add_argument('--password', "-p",
-                        dest='password', action='store',
+                        dest='password', action='store', required=False,
                         help='MySQL/MariaDB password')
     parser.add_argument('--host', '-s',
-                        dest='host', action='store', required=True,
+                        dest='host', action='store', required=False,
                         help='MySQL/MariaDB host')
     parser.add_argument('--database', '-d',
                         dest='database', action='store', required=True,
-                        help='MySQL/MariaDB database')
+                        help='MySQL/MariaDB database or SQLite databasefile')
     parser.add_argument('--count', '-c',
                         dest='row_count', action='store', required=False, type=int, default=0,
                         help='If 0 (default), determine upper bound of number of rows by querying database, '
                              'otherwise use this number (used for progress bar only)')
+    parser.add_argument('--table', '-x',
+                        dest='table', action='store', required=True,
+                        help='Source Table is either states or statistics'
+                             'Home Assistant keeps 10 days of states by default and keeps statistics forever for some entities')
+    parser.add_argument('--dry-run', '-y',
+                        dest='dry', action='store_true', required=False,
+                        help='do all work except writing to InfluxDB')
 
     args = parser.parse_args()
 
+    if (args.dry) : 
+        print("option --dry-run was given, nothing will be writen on InfluxDB") 
+
     # load InfluxDB configuration file (the one from Home Assistant) (without using !secrets)
     with open("influxdb.yaml") as config_file:
         influx_config = yaml.load(config_file, Loader=yaml.FullLoader)
@@ -94,12 +115,14 @@ def main():
     influx = get_influx_connection(influx_config, test_write=True, test_read=True)
     converter = _generate_event_to_json(influx_config)
 
-    # connect to MySQL/MariaDB database
-    connection = mysql_connect(host=args.host, user=args.user, password=args.password, database=args.database, cursorclass=cursors.SSCursor, charset="utf8")
-    cursor = connection.cursor()
+    if (args.type == "MySQL" or args.type == "MariaDB"):
+        # connect to MySQL/MariaDB database
+        connection = mysql_connect(host=args.host, user=args.user, password=args.password, database=args.database, cursorclass=cursors.SSCursor, charset="utf8")
+    else:
+        # connect to SQLite file instead
+        connection = sqlite3.connect(args.database)
 
-    # untested: connect to SQLite file instead (you need to get rid of the first three `add_argument` calls above)
-    #connection = sqlite3.connect('home_assistant_v2.db')
+    cursor = connection.cursor()
 
     if args.row_count == 0:
         # query number of rows in states table - this will be more than the number of rows we
@@ -110,7 +133,13 @@ def main():
         total = args.row_count
 
     # select the values we are interested in
-    cursor.execute("select states.entity_id, states.state, states.attributes, events.event_type, events.time_fired from states, events where events.event_id = states.event_id")
+    if (args.table == "states"):
+        cursor.execute("select states.entity_id, states.state, state_attributes.shared_attrs, events.event_type, events.time_fired from states, events, state_attributes where events.event_id = states.event_id and states.attributes_id=state_attributes.attributes_id")
+    elif (args.table == "statistics"):
+        cursor.execute("select statistic_id,mean,state_attributes.shared_attrs,'state_changed',created from statistics_meta,statistics,state_attributes where metadata_id= statistics_meta.id and mean!=0 and state_attributes.attributes_id = (select states.attributes_id from states where states.entity_id = statistic_id limit 1 )")
+    else:
+        print("ERROR: argument --table should be \"states\" or \"statistics\"");
+        exit
 
     # map to count names and number of measurements for each entity
     statistics = {}
@@ -129,7 +158,7 @@ def main():
                 _attributes_raw = row[2]
                 _attributes = rename_friendly_name(json.loads(_attributes_raw))
                 _event_type = row[3]
-                _time_fired = row[4]
+                _time_fired = dt.parse_datetime(row[4])
             except Exception as e:
                 print("Failed extracting data from %s: %s.\nAttributes: %s" % (row, e, _attributes_raw))
                 continue
@@ -169,11 +198,13 @@ def main():
                 batch_size_cur += 1
 
                 if batch_size_cur >= batch_size_max:
-                    influx.write(batch_json)
+                    if (not args.dry) : 
+                        influx.write(batch_json)
                     batch_json = []
                     batch_size_cur = 0
 
-    influx.write(batch_json)
+    if (not args.dry) : 
+        influx.write(batch_json)
     influx.close()
 
     # print statistics - ideally you have one friendly name per entity_id