Interpolation or retrieval does not work at subsecond level

[valeriu-balaban]

When trying to extract the value from TGeomPointSeq using the value_at_timestamp method, and the timestamp provided has a non-zero microseconds part, the microseconds would not be considered for interpolation or retrieval.

To Reproduce
Let tgps be equal to TGeomPointSeq([POINT(295.9722900390625 654.239990234375)@2024-05-07 14:47:07-07, POINT(343.66021728515625 633.1279296875)@2024-05-07 14:47:07.1001-07, POINT(419.7856750488281 599.8984985351562)@2024-05-07 14:47:07.2002-07, POINT(483.23553466796875 565.5811767578125)@2024-05-07 14:47:07.3003-07, POINT(547.3091430664062 539.0184326171875)@2024-05-07 14:47:07.4004-07, POINT(607.9834594726562 513.4208374023438)@2024-05-07 14:47:07.5005-07, POINT(662.676513671875 490.9242858886719)@2024-05-07 14:47:07.6006-07, POINT(714.170654296875 475.9928894042969)@2024-05-07 14:47:07.7007-07, POINT(764.0394897460938 462.255126953125)@2024-05-07 14:47:07.8008-07, POINT(807.018310546875 448.92486572265625)@2024-05-07 14:47:07.9009-07, POINT(846.0618896484375 436.96124267578125)@2024-05-07 14:47:08.001-07]), then calling tgps.value_at_timestamp(datetime.datetime(2024, 5, 7, 14, 47, 7, 400400)) would return <POINT (295.972 654.24)> (the first point in the sequence) instead of the expected <POINT (547.30 539.01)> (the value of the 5th instant in the sequence).

Desktop (please complete the following information):

• OS: macOS
• Architecture: x64
• Environment: conda
• Version 1.1.4

[Diviloper]

This was an issue in the pymeos_cffi library. It should be fixed in the new 1.1.2 version (MobilityDB/PyMEOS-CFFI@7f4afb4).
Can you update it and try again to confirm it works now as expected?
I'll close the issue afterwards.

[valeriu-balaban]

Many thanks for the quick reply.

I updated the cffi package. There were no errors on Linux, but on MacOS, there was a compilation error during installation. I will create a separate issue for that on the cffi repo.

Now, when running the code below, it throws an error:

import datetime

import pymeos
import pymeos_cffi

pymeos.pymeos_initialize()

print(pymeos_cffi.__version__)

t = pymeos.TGeomPointSeq(
  "[POINT(295.9722900390625 654.239990234375)@2024-05-07 14:47:07-07, POINT(343.66021728515625 633.1279296875)@2024-05-07 14:47:07.1001-07, POINT(419.7856750488281 599.8984985351562)@2024-05-07 14:47:07.2002-07, POINT(483.23553466796875 565.5811767578125)@2024-05-07 14:47:07.3003-07, POINT(547.3091430664062 539.0184326171875)@2024-05-07 14:47:07.4004-07, POINT(607.9834594726562 513.4208374023438)@2024-05-07 14:47:07.5005-07, POINT(662.676513671875 490.9242858886719)@2024-05-07 14:47:07.6006-07, POINT(714.170654296875 475.9928894042969)@2024-05-07 14:47:07.7007-07, POINT(764.0394897460938 462.255126953125)@2024-05-07 14:47:07.8008-07, POINT(807.018310546875 448.92486572265625)@2024-05-07 14:47:07.9009-07, POINT(846.0618896484375 436.96124267578125)@2024-05-07 14:47:08.001-07]"
)

t.value_at_timestamp(datetime.datetime(2024, 5, 7, 14, 47, 7, 400400))

I get the following output

1.1.2
---------------------------------------------------------------------------
TypeError                                 Traceback (most recent call last)
 print(pymeos_cffi.__version__)
 t = pymeos.TGeomPointSeq(
   "[POINT(295.9722900390625 654.239990234375)@2024-05-07 14:47:07-07, POINT(343.66021728515625 633.1279296875)@2024-05-07 14:47:07.1001-07, POINT(419.7856750488281 599.8984985351562)@2024-05-07 14:47:07.2002-07, POINT(483.23553466796875 565.5811767578125)@2024-05-07 14:47:07.3003-07, POINT(547.3091430664062 539.0184326171875)@2024-05-07 14:47:07.4004-07, POINT(607.9834594726562 513.4208374023438)@2024-05-07 14:47:07.5005-07, POINT(662.676513671875 490.9242858886719)@2024-05-07 14:47:07.6006-07, POINT(714.170654296875 475.9928894042969)@2024-05-07 14:47:07.7007-07, POINT(764.0394897460938 462.255126953125)@2024-05-07 14:47:07.8008-07, POINT(807.018310546875 448.92486572265625)@2024-05-07 14:47:07.9009-07, POINT(846.0618896484375 436.96124267578125)@2024-05-07 14:47:08.001-07]"
) )
--->  t.value_at_timestamp(datetime.datetime(2024, 5, 7, 14, 47, 7, 400400))

File ~/study-venv/lib/python3.10/site-packages/pymeos/main/tpoint.py:227, in TPoint.value_at_timestamp(self, timestamp, precision)
 def value_at_timestamp(self, timestamp: datetime, precision: int = 15) -> shp.Point:
     """
   Returns the value of the temporal point at the given timestamp.
   (...)
        tpoint_value_at_timestamp
     """
     return gserialized_to_shapely_point(
         tpoint_value_at_timestamptz(
             self._inner, datetime_to_timestamptz(timestamp), True
         )[0],
         precision,
     )

TypeError: 'NoneType' object is not subscriptable

[Diviloper]

I think there are two separate issues here. One is that it crashes when the function returns None. I I'll fix this as soon as I can.
However, in this case, the intention was to use a timestamp in the sequence, so it shouldn't be returning a None. It's probably due to the timezones. Can you try removing the timezone from the sequence input, or adding it to the used timestamp and see if it still crashes?

Many thanks!

[valeriu-balaban]

Yes, many thanks. It now interpolates at the subsecond level.

import datetime
import pymeos

pymeos.pymeos_initialize()


t = pymeos.TGeomPointSeq(
  "[POINT(295.9722900390625 654.239990234375)@2024-05-07 14:47:07-07, POINT(343.66021728515625 633.1279296875)@2024-05-07 14:47:07.1001-07]"
)

print(t.value_at_timestamp(t.timestamps()[0] + datetime.timedelta(seconds=0.05)))

prints POINT (319.7924335186298 643.6945054457261) as expected.

[Diviloper]

Great!
I'll leave the issue open until I fix the None bug.