Skip to content

TableTools.readCsv should be able to type detect ISO 8601 format for timestamps as DBDateTime. #1570

@jcferretti

Description

@jcferretti

A file generated in python pandas using .to_csv(...) for a dataframe that was loaded from a parquet file with a timestamp(nanos) type column:

cfs@erke 16:55:20 ~/dh/oss1/deephaven-core/data
$ head -n 10 /a1/tmp/workqueue-no-nulls-100m-no-index.csv
workqueue_id,user_id,timestamp
1,1001,2021-11-07 14:00:00.011463807+00:00
2,1003,2021-11-07 14:00:00.080748962+00:00
3,1002,2021-11-07 14:00:00.114976256+00:00
4,1004,2021-11-07 14:00:00.118275772+00:00
5,1003,2021-11-07 14:00:00.131173174+00:00
6,1001,2021-11-07 14:00:00.174901899+00:00
7,1001,2021-11-07 14:00:00.194499326+00:00
8,1002,2021-11-07 14:00:00.201754097+00:00
9,1004,2021-11-07 14:00:00.237870788+00:00

The timestamp column formatted as above is not recognized by DHC when using TableTools.readCsv; the column is read as String.

More generally, we should accept valid ISO 8601 strings as timestamps, any of them.

Metadata

Metadata

Assignees

Labels

bugSomething isn't workingtriage

Type

No type

Projects

No projects

Relationships

None yet

Development

No branches or pull requests

Issue actions