resample/interpolate doesn't handle columns with dot in name
ISSUE
While maybe not the best practice, databricks does allow columns to have "." in the name. When doing a resample with interporlate, this results in a cannot resolve column name error
How to reproduce
-
Create a TSDF with columns that include a "."
-
Attempt to resample and interporlate with
resample_tsdf = base_tsdf.resample(freq="30 seconds", func="mean").interpolate(method="ffill")
An error is produced
AnalysisException: Cannot resolve column name "Bundler.Status.CurMachSpeed" among (site, line, ts, Bundler.Status.CurMachSpeed, Bundler.Status.MachSpeed, agg_key); did you mean to quote the Bundler.Status.CurMachSpeed column?
Workaround
Rename the columns before resampling / interpolate
Thanks for bringing this to our attention @ghormann - we'll look into it and see if we can get a fix out soon