YYYY week into a date

[Bidek56]

Polars version checks

• I have checked that this issue has not already been reported.

• I have confirmed this bug exists on the latest version of Polars.

Issue description

Attempting to parse YYYY week into date results in an error: ComputeError: strict conversion to dates failed, maybe set strict=False

import polars as pl
pl.DataFrame(
{
 "week": [201901, 201902, 201903, 201942, 201943, 201944]
}).with_columns(pl.col('week').cast(pl.Utf8).str.strptime(pl.Date, fmt='%Y%U').alias("date"))

According to the Rust Chrono specifiers, %U is supported.

Pandas can read it: df['date'] = pd.to_datetime(df.week.astype(str) + '0', format='%Y%W%w')

Thanks to SO, I have found a workaround but I think it's a bug.

Reproducible example

import polars as pl
pl.DataFrame(
{
 "week": [201901, 201902, 201903, 201942, 201943, 201944]
}).with_columns(pl.col('week').cast(pl.Utf8).str.strptime(pl.Date, fmt='%Y%U').alias("date"))

Expected behavior

date

2019-01-01
2019-01-08
2019-01-15
2019-10-15
2019-10-22
2019-10-29

Installed versions

---Version info---
Polars: 0.16.3
Index type: UInt32
Platform: Windows-10-10.0.22621-SP0
Python: 3.11.2 (tags/v3.11.2:878ead1, Feb  7 2023, 16:38:35) [MSC v.1934 64 bit (AMD64)]
---Optional dependencies---
pyarrow: 11.0.0
pandas: 1.5.3
numpy: 1.24.2
fsspec: 2023.1.0
connectorx: <not installed>
xlsx2csv: 0.8.1
deltalake: <not installed>
matplotlib: 3.6.3

[MarcoGorelli]

Hi @Bidek56

Looks like this doesn't quite work in chrono:

use chrono::{NaiveDate};

fn main() {
    let date_str = "201901";
    let naive_date = NaiveDate::parse_from_str(date_str, "%Y%U").unwrap();
    println!("{:?}", naive_date);
}

thread 'main' panicked at 'called `Result::unwrap()` on an `Err` value: ParseError(NotEnough)', src/main.rs:5:66
note: [run with `RUST_BACKTRACE=1` environment variable to display a backtrace](https://play.rust-lang.org/?version=stable&mode=debug&edition=2021#)

I think you need to also pass the day of the week, e.g.

    let date_str = "201901Mon";
    let naive_date = NaiveDate::parse_from_str(date_str, "%Y%U%a").unwrap();

[Bidek56]

Does the code below work for you? b/c "%Y%U%a" works but not "%Y%W%a" but the docs says that it should.

    let date_str = "201901Mon";
    let naive_date = NaiveDate::parse_from_str(date_str, "%Y%W%a").unwrap();

[FObersteiner]

I think YYYY-WW is just not a precise date. For example, take the ISO week date definition

A precise date is specified by the ISO week-numbering year in the format YYYY, a week number in the format ww prefixed by the letter 'W', and the weekday number, a digit d from 1 through 7, beginning with Monday and ending with Sunday.

That's probably why you need to specify the day of the week - although one might think that with the week number only, it should default to a certain day of week (but which one?!).

So I'd conclude this is not a bug but a feature to avoid ambiguity. Only the chrono docs could be a bit more specific, IMHO.

[Bidek56]

Hmmm, I wonder why "%Y%W%a" works for you in Rust but it does not work in Python.

[FObersteiner]

@Bidek56 sorry, that was a bit unclear. %W doesn't work for me either. Here's a playground example.

In Python, both %U and %W work, but give the same result if you set the weekday to 1 ?!

from datetime import datetime

print(datetime.strptime("2019-01-0", "%Y-%U-%w"))
# 2019-01-06 00:00:00
print(datetime.strptime("2019-01-0", "%Y-%W-%w"))
# 2019-01-13 00:00:00

print(datetime.strptime("2019-01-1", "%Y-%U-%w"))
# 2019-01-07 00:00:00
print(datetime.strptime("2019-01-1", "%Y-%W-%w"))
# 2019-01-07 00:00:00

It seem something strange is going on here; chrono errors out, and Python gives results that are unexpected for me. ISO week date parsing directives ("%G-%V-%u") give consistent results though. But I still don't fully understand what's going on here, so I put up a question on stackoverflow.

[Bidek56]

 let date = NaiveDate::parse_from_str("2019-01-0", "%Y-%W-%w")?;

Returns: Error: ParseError(Impossible)

But

let date = NaiveDate::parse_from_str("2020-01-0", "%Y-%W-%w")?;

Returns: 2020-01-12

[FObersteiner]

 let date = NaiveDate::parse_from_str("2019-01-0", "%Y-%W-%w")?;

Returns: Error: ParseError(Impossible)

But

let date = NaiveDate::parse_from_str("2020-01-0", "%Y-%W-%w")?;

Returns: 2020-01-12

Exactly. This seems somehow related to year 2019. But this polars issue #6796 is the wrong place to discuss I think. We should open an issue at chrono or a question on stackoverflow.

[Bidek56]

@FObersteiner I will close this ticket and submit a Chrono issue instead. Thanks for your help.