aloneguid / parquet-dotnet

Fully managed Apache Parquet implementation
https://aloneguid.github.io/parquet-dotnet/
MIT License
600 stars 151 forks source link

[BUG]: Incorrect logical type for TimeSpan #533

Open wldevries opened 2 months ago

wldevries commented 2 months ago

Library Version

4.24.0

OS

Windows

OS Architecture

64 bit

How to reproduce?

  1. Create a table with a DataFieldTimeSpan column
  2. Add TimeSpan values larger than 24hours
  3. Parse Parquet file using Pandas

-> ValueError: hour must be in 0..23

I've opened the parquet file using the floor tool and the column has type INT32, converted type TIME_MILIS and logical type TIME (unit: MILLIS, isAdjustedToUTC: True).

A TimeSpan is a period of time and not a time of day like the new TimeOnly type. It seems to me TIME is not the correct logical type for TimeSpan.

Failing test

No response