Switch to local TZ lookup

tapiriik/database/__init__.py

@@ -3,3 +3,4 @@
 _connection = MongoClient(host=MONGO_HOST)
 db = _connection["tapiriik"]
 cachedb = _connection["tapiriik_cache"]
+tzdb = _connection["tapiriik_tz"]

tapiriik/database/tz.py

@@ -0,0 +1,12 @@
+from tapiriik.database import tzdb
+from bson.son import SON
+
+def TZLookup(lat, lng):
+	pt = [lng, lat]
+	res = tzdb.boundaries.find_one({"Boundary": {"$geoIntersects": {"$geometry": {"type":"Point", "coordinates": pt}}}}, {"TZID": True})
+	if not res:
+		res = tzdb.boundaries.find_one({"Boundary": SON([("$near", {"$geometry": {"type": "Point", "coordinates": pt}}), ("$maxDistance", 200000)])}, {"TZID": True})
+	res = res["TZID"] if res else None
+	if not res or res == "uninhabited":
+		res = round(lng / 15)
+	return res

tapiriik/services/interchange.py

@@ -1,9 +1,8 @@
 from datetime import timedelta, datetime
 from tapiriik.database import cachedb
-import requests
+from tapiriik.database.tz import TZLookup
 import hashlib
 import pytz
-import warnings
 
 
 class ActivityType:  # taken from RK API docs. The text values have no meaning except for debugging
@@ -107,19 +106,12 @@ def CalculateTZ(self, loc=None):
                     break
             if loc is None:
                 raise Exception("Can't find TZ without a waypoint with a location")
+
+        # I guess at some point it will be faster to perform a full lookup than digging through this table.
         cachedTzData = cachedb.tz_cache.find_one({"Latitude": loc.Latitude, "Longitude": loc.Longitude})
         if cachedTzData is None:
-            warnings.filterwarnings("ignore", "the 'strict' argument")
-            warnings.filterwarnings("ignore", "unclosed <socket")
-            resp = requests.get("http://api.geonames.org/timezoneJSON?username=tapiriik&radius=0.5&lat=" + str(loc.Latitude) + "&lng=" + str(loc.Longitude))
-            data = resp.json()
-            cachedTzData = {}
-            if "timezoneId" in data:
-                cachedTzData["TZ"] = data["timezoneId"]
-            else:
-                cachedTzData["TZ"] = data["rawOffset"]
-            cachedTzData["Latitude"] = loc.Latitude
-            cachedTzData["Longitude"] = loc.Longitude
+            res = TZLookup(loc.Latitude, loc.Longitude)
+            cachedTzData = {"TZ": res, "Latitude": loc.Latitude, "Longitude": loc.Longitude}
             cachedb.tz_cache.insert(cachedTzData)
 
         if type(cachedTzData["TZ"]) != str:

tz_ingest.py

@@ -0,0 +1,32 @@
+# This file isn't called in normal operation, just to update the TZ boundary DB.
+# Should be called with `tz_world.*` files from http://efele.net/maps/tz/world/ in the working directory.
+# Requires pyshp and shapely for py3k (from https://github.com/mwtoews/shapely/tree/py3)
+
+import shapefile
+from shapely.geometry import Polygon, mapping
+import pymongo
+from tapiriik.database import tzdb
+
+print("Dropping boundaries collection")
+tzdb.drop_collection("boundaries")
+
+print("Setting up index")
+tzdb.boundaries.ensure_index([("Boundary", pymongo.GEOSPHERE)])
+
+print("Reading shapefile")
+records = []
+sf = shapefile.Reader("tz_world.shp")
+shapeRecs = sf.shapeRecords()
+
+ct = 0
+total = len(shapeRecs)
+for shape in shapeRecs:
+	tzid = shape.record[0]
+	print("%3d%% %s" % (round(ct * 100 / total), tzid))
+	ct += 1
+	polygon = Polygon(list(shape.shape.points))
+	if not polygon.is_valid:
+		polygon = polygon.buffer(0) # Resolves issues with most self-intersecting geometry
+		assert polygon.is_valid
+	record = {"TZID": tzid, "Boundary": mapping(polygon)}
+	tzdb.boundaries.insert(record) # Would be bulk insert, but that makes it a pain to debug geometry issues