Nutze Skyfield fuer praezisere Daemmerungsberechnung

public/py/vendor/skyfield/tests/fixes.py

@@ -0,0 +1,54 @@
+"""Helpers for making Skyfield tests stable."""
+
+import datetime as dt
+import sys
+
+from skyfield import earthlib
+import skyfield.api
+import skyfield.timelib
+
+IS_32_BIT = (sys.maxsize == 0x7fffffff)
+_real_datetime_class = dt.datetime
+
+class datetime(dt.datetime):
+    _fake_now = None
+    @staticmethod
+    def now():
+        return datetime._fake_now
+
+def setup(utc=(2015, 10, 11, 10)):
+    # Importing Pandas raises a TypeError once our fake datetime is
+    # installed.  So instead of waiting, let's import it now.
+    __import__('pandas')
+
+    fake_now = dt.datetime(*utc)
+    skyfield.api.load = skyfield.iokit.Loader('.', verbose=False)
+    skyfield.timelib.Timescale._utcnow = lambda self: fake_now
+    datetime._fake_now = fake_now
+    dt.datetime = datetime
+
+def teardown():
+    skyfield.api.load = skyfield.iokit.Loader('.')
+    skyfield.timelib.Timescale._utcnow = skyfield.timelib._utcnow
+    dt.datetime = _real_datetime_class
+
+class low_precision_ERA(object):
+    """Compute the Earth rotation angle with only a single float for UT1.
+
+    Skyfield now uses two floats ``t.whole`` and ``t.ut1_fraction`` to
+    represent the UT1 Julian date, supplying an additional 16 digits of
+    precision.  For the Earth rotation angle, which moves very quickly
+    per unit time compared to most other astronomical quantities, this
+    knocks Skyfield out of agreement with other libraries like NOVAS and
+    SOFA that round UT1 to a single float.  Various tests use this
+    context manager to make Skyfield match the lower-precision output.
+
+    """
+    def __enter__(self):
+        self.saved = earthlib.earth_rotation_angle
+        earthlib.earth_rotation_angle = self.era
+    def __exit__(self, *args):
+        earthlib.earth_rotation_angle = self.saved
+    def era(self, whole, fraction):
+        rounded_single_float = whole + fraction
+        return self.saved(rounded_single_float)