ARROW-16439: [R] Implement binding for lubridate::fast_strptime

[dragosmg]

This PR adds:

• a binding emulating lubridate's fast_strptime functionality

This PR does not support the following arguments:

• lt = TRUE - this returns a POSIXlt object (a list) which cannot be easily used in a dplyr pipeline (currently it actually errors) => we have a different default lt = FALSE for the Arrow binding, and
• cutoff_2000 = 68L - for the y% format two-digit numbers smaller or equal to cutoff_2000 are parsed as though starting with 20, otherwise parsed as though starting with 19. It would be nice to have this, so I raised ARROW-16596. We can always suggest users they manipulate the strings before parsing, so I don't think this is crucial functionality.

The following code will be possible once the PR is merged:

library(lubridate, warn.conflicts = FALSE)
library(dplyr, warn.conflicts = FALSE)
library(arrow, warn.conflicts = FALSE)

dates_table <- tibble(
  string_with_short_year = c("68-05-17", "69-05-17", "55-05-17")
)

dates_table %>% 
  mutate(
    date = fast_strptime(
      string_with_short_year, 
      format = "%y-%m-%d",
      lt = FALSE
    )
  )
#> # A tibble: 3 × 2
#>   string_with_short_year date               
#>   <chr>                  <dttm>             
#> 1 68-05-17               2068-05-17 00:00:00
#> 2 69-05-17               1969-05-17 00:00:00
#> 3 55-05-17               2055-05-17 00:00:00

dates_table %>% 
  arrow_table() %>% 
  mutate(
    date = fast_strptime(
      string_with_short_year, 
      format = "%y-%m-%d",
      lt = FALSE
    )
  ) %>%
  collect()
#> # A tibble: 3 × 2
#>   string_with_short_year date               
#>   <chr>                  <dttm>             
#> 1 68-05-17               2068-05-17 00:00:00
#> 2 69-05-17               1969-05-17 00:00:00
#> 3 55-05-17               2055-05-17 00:00:00

^{Created on 2022-05-18 by the reprex package (v2.0.1)}

[github-actions]

https://issues.apache.org/jira/browse/ARROW-16439

[github-actions]

⚠️ Ticket has not been started in JIRA, please click 'Start Progress'.

[thisisnic]

Looks good to me!

[dragosmg]

There is a slight problem though - in the way the fast_strptime binding parses a string like "68-12-12 12:34:56" when the format is "%Y..." (incorrect) and not "%y..." (correct).
base R and lubridate behaviour (currently the fast_strptime binding behaves like base R - i.e. doesn't error, but parses a short year xy to 00xy):

b <- "68-10-07 19:04:0"
strptime(b, format = "%Y-%m-%d %H:%M:%S")
#> [1] "0068-10-07 19:04:00 LMT"
strptime(b, format = "%y-%m-%d %H:%M:%S")
#> [1] "2068-10-07 19:04:00 BST"

lubridate::fast_strptime(b, format = "%Y-%m-%d %H:%M:%S")
#> [1] NA
lubridate::fast_strptime(b, format = "%y-%m-%d %H:%M:%S")
#> [1] "2068-10-07 19:04:00 UTC"

^{Created on 2022-05-19 by the reprex package (v2.0.1)}

This could trip users up (mostly when they pass multiple formats):

library(arrow, warn.conflicts = FALSE)
library(lubridate, warn.conflicts = FALSE)
library(dplyr, warn.conflicts = FALSE)

dates_table <- tibble(
  string_with_short_year = c("68-05-17", "69-05-17", "55-05-17")
)

dates_table %>% 
  mutate(
    date = fast_strptime(
      string_with_short_year, 
      format = c("%Y-%m-%d", "%y-%m-%d"),
      lt = TRUE
    )
  )
#> # A tibble: 3 × 2
#>   string_with_short_year date               
#>   <chr>                  <dttm>             
#> 1 68-05-17               2068-05-17 00:00:00
#> 2 69-05-17               1969-05-17 00:00:00
#> 3 55-05-17               2055-05-17 00:00:00

dates_table %>% 
  arrow_table() %>% 
  mutate(
    date = fast_strptime(
      string_with_short_year, 
      format = c("%Y-%m-%d", "%y-%m-%d"),
      lt = FALSE
    )
  ) %>% 
  collect()
#> # A tibble: 3 × 2
#>   string_with_short_year date               
#>   <chr>                  <dttm>             
#> 1 68-05-17               0068-05-17 00:00:00
#> 2 69-05-17               0069-05-17 00:00:00
#> 3 55-05-17               0055-05-17 00:00:00

^{Created on 2022-05-19 by the reprex package (v2.0.1)}

What happens here 👆🏻 is that lubridate::fast_strptime() fails on the first format and moves on the the second, while the arrow binding doesn't.

I flagged this as part of ARROW-16596. Not quite sure how to go about handling that only on the R side though. arrow's strptime defers to the platform's strptime so we might see different behaviours on different platforms too.

[jonkeane]

One last cleanup

[jonkeane]

We should remove these commented lines, yeah?

[dragosmg]

I yes, sorry. Forgot about those. Do I open a minor PR?

[dragosmg]

Or should I just do it in one of the other PR I have going?

[dragosmg]

@jonkeane I removed the tests in dragosmg@bde85ce

[ursabot]

Benchmark runs are scheduled for baseline = 8394571 and contender = 7f31c9d. 7f31c9d is a master commit associated with this PR. Results will be available as each benchmark for each run completes.
Conbench compare runs links:
[Finished ⬇️0.0% ⬆️0.0%] ec2-t3-xlarge-us-east-2
[Failed ⬇️0.74% ⬆️0.0%] test-mac-arm
[Failed ⬇️0.0% ⬆️0.0%] ursa-i9-9960x
[Finished ⬇️0.04% ⬆️0.0%] ursa-thinkcentre-m75q
Buildkite builds:
[Finished] 7f31c9d2 ec2-t3-xlarge-us-east-2
[Failed] 7f31c9d2 test-mac-arm
[Failed] 7f31c9d2 ursa-i9-9960x
[Finished] 7f31c9d2 ursa-thinkcentre-m75q
[Finished] 83945714 ec2-t3-xlarge-us-east-2
[Failed] 83945714 test-mac-arm
[Failed] 83945714 ursa-i9-9960x
[Finished] 83945714 ursa-thinkcentre-m75q
Supported benchmarks:
ec2-t3-xlarge-us-east-2: Supported benchmark langs: Python, R. Runs only benchmarks with cloud = True
test-mac-arm: Supported benchmark langs: C++, Python, R
ursa-i9-9960x: Supported benchmark langs: Python, R, JavaScript
ursa-thinkcentre-m75q: Supported benchmark langs: C++, Java