Local_Time Conversion error

[kiimot]

Hi! I'm doing some tests, and when I try to convert a String format variable to LocalTime, when you insert it into the table it looks correct, but if you view it from the dashboard it doesn't.

However, when the variable is read from a CSV file, it is displayed correctly in both the table and the dashboard.

Is it possible that I am doing the process of passing from String to LocalTime wrong? It inserts it correctly into the table, but it doesn't work correctly

Steps to reproduce

If from Atoti, I read a CSV that contains a LocalTime field and when reading the CSV I indicate that this field is LocalTime, it works correctly

If I do a logs.head() it works correctly and in the dashboard too...

But if I try to pass to an endpoint, a String to LocalTime transforms it correctly and inserts it correctly in the Logs table:

But if I check the dashboard, it doesn't show the same as the table.

Actual Result

Expected Result

Environment

• atoti: 0.7.2
• Python: 3.10.9
• Operating System: Windows 10 Enterprise

Logs (if relevant)

[tibdex]

Hi,

In your append_rows_to_table() endpoint, can you try printing the whole rows to show that not only the first row is good but all the other ones too?

Instead of Table.append() can you try to use Table.load_pandas() passing it a DataFrame where you made sure that the HMS_INI and HMS_FIN columns both have the good type (time and not string or something else)?

[kiimot]

Hi,

Of course, I put a screenshot of before and after the conversion

And reference to the issue of changing the append to load_pandas, it still doesn't work

The result remains the same for both changes

Thanks for your time

[tibdex]

Hi,

I can reproduce the error with this simpler code:

from datetime import datetime, date, time

import atoti as tt

session = tt.Session()

table = session.create_table("test", keys=["id"], types={
    "id": tt.INT,
    "datetime": tt.LOCAL_DATE_TIME,
    "date": tt.LOCAL_DATE,
    "time": tt.LOCAL_TIME
})
cube = session.create_cube(table)
l, m = cube.levels, cube.measures

table += (1, datetime(2023, 1, 24, 16, 57, 0), date(2023, 1, 24), time(16, 57, 0))

table.head()
#               datetime       date            time
# id                                               
# 1  2023-01-24 16:57:00 2023-01-24 0 days 16:57:00

When I look at the Arrow table created to transfer the row from Python to the atoti server, I get this:

pyarrow.Table
id: int32
datetime: timestamp[ns]
date: date32[day]
time: time64[ns]
----
id: [[1]]
datetime: [[2023-01-24 16:57:00.000000000]]
date: [[2023-01-24]]
time: [[16:57:00.000000000]]

so the types are right but something goes wrong after since I also get wrong data from the server in the pivot table:

The issue also occurs with a query launched from Python:

cube.query(m["contributors.COUNT"], levels=[l["datetime"], l["time"]])
#                                contributors.COUNT
# datetime         time                             
# 2023-01-24T16:57 61020000000000                  1

This is tracked in an internal ticket, I'll let you know when this is fixed.