location: add all.py, using takeout/gpslogger/ip

my/time/tz/via_location.py

@@ -7,27 +7,34 @@ REQUIRES = [
 ]
 
 
+from my.config import time
+from my.core import dataclass
+
+
+@dataclass
+class config(time.tz.via_location):
+    # less precise, but faster
+    fast: bool = True
+
+    # if the accuracy for the location is more than 5km, don't use
+    require_accuracy: float = 5_000
+
+
 from collections import Counter
 from datetime import date, datetime
 from functools import lru_cache
 from itertools import groupby
-from typing import Iterator, NamedTuple, Optional
+from typing import Iterator, NamedTuple, Optional, Tuple, Any, List
 
 from more_itertools import seekable
 import pytz
 
-from ...core.common import LazyLogger, mcachew, tzdatetime
-from ...core.cachew import cache_dir
-from ...location.google import locations
+from my.core.common import LazyLogger, mcachew, tzdatetime
 
+logger = LazyLogger(__name__, level='warning')
 
-logger = LazyLogger(__name__, level='debug')
-
-
-# todo should move to config? not sure
-_FASTER: bool = True
 @lru_cache(2)
-def _timezone_finder(fast: bool):
+def _timezone_finder(fast: bool) -> Any:
     if fast:
         # less precise, but faster
         from timezonefinder import TimezoneFinderL as Finder  # type: ignore
@@ -46,20 +53,40 @@ class DayWithZone(NamedTuple):
     zone: Zone
 
 
-def _iter_local_dates(start=0, stop=None) -> Iterator[DayWithZone]:
-    finder = _timezone_finder(fast=_FASTER) # rely on the default
+from my.location.common import LatLon
+
+# for backwards compatibility
+def _locations() -> Iterator[Tuple[LatLon, datetime]]:
+    try:
+        import my.location.all
+        for loc in my.location.all.locations():
+            yield ((loc.lat, loc.lon), loc.dt)
+
+    except Exception as e:
+        from my.core.warnings import high
+        logger.exception("Could not setup via_location using my.location.all provider, falling back to legacy google implemetation", exc_info=e)
+        high("Setup my.google.takeout.parser, then my.location.all for better google takeout/location data")
+
+        import my.location.google
+
+        for loc in my.location.google.locations():
+            yield ((loc.lat, loc.lon), loc.dt)
+
+
+def _iter_local_dates() -> Iterator[DayWithZone]:
+    finder = _timezone_finder(fast=config.fast) # rely on the default
     pdt = None
     warnings = []
     # todo allow to skip if not noo many errors in row?
-    for l in locations(start=start, stop=stop):
+    for (lat, lon), dt in _locations():
         # TODO right. its _very_ slow...
-        zone = finder.timezone_at(lng=l.lon, lat=l.lat)
+        zone = finder.timezone_at(lat=lat, lng=lon)
         if zone is None:
-            warnings.append(f"Couldn't figure out tz for {l}")
+            warnings.append(f"Couldn't figure out tz for {lat}, {lon}")
             continue
         tz = pytz.timezone(zone)
         # TODO this is probably a bit expensive... test & benchmark
-        ldt = l.dt.astimezone(tz)
+        ldt = dt.astimezone(tz)
         ndate = ldt.date()
         if pdt is not None and ndate < pdt.date():
             # TODO for now just drop and collect the stats
@@ -71,12 +98,13 @@ def _iter_local_dates(start=0, stop=None) -> Iterator[DayWithZone]:
         yield DayWithZone(day=ndate, zone=z)
 
 
-def most_common(l):
-    res, count = Counter(l).most_common(1)[0] # type: ignore[var-annotated]
+def most_common(lst: List[DayWithZone]) -> DayWithZone:
+    res, _ = Counter(lst).most_common(1)[0]  # type: ignore[var-annotated]
     return res
 
 
-@mcachew(cache_path=cache_dir())
+# refresh _iter_tzs once per day -- don't think a better depends_on is possible dynamically
+@mcachew(logger=logger, depends_on=lambda: str(date.today()))
 def _iter_tzs() -> Iterator[DayWithZone]:
     for d, gr in groupby(_iter_local_dates(), key=lambda p: p.day):
         logger.info('processed %s', d)
@@ -106,6 +134,7 @@ def _get_day_tz(d: date) -> Optional[pytz.BaseTzInfo]:
             break
     return None if zone is None else pytz.timezone(zone)
 
+
 # ok to cache, there are only a few home locations?
 @lru_cache(maxsize=None)
 def _get_home_tz(loc) -> Optional[pytz.BaseTzInfo]:
@@ -119,8 +148,10 @@ def _get_home_tz(loc) -> Optional[pytz.BaseTzInfo]:
         return pytz.timezone(zone)
 
 
-# TODO expose? to main as well?
 def _get_tz(dt: datetime) -> Optional[pytz.BaseTzInfo]:
+    '''
+    Given a datetime, returns the timezone for that date.
+    '''
     res = _get_day_tz(d=dt.date())
     if res is not None:
         return res
@@ -129,6 +160,9 @@ def _get_tz(dt: datetime) -> Optional[pytz.BaseTzInfo]:
     loc = home.get_location(dt)
     return _get_home_tz(loc=loc)
 
+# expose as 'public' function
+get_tz = _get_tz
+
 
 def localize(dt: datetime) -> tzdatetime:
     tz = _get_tz(dt)