Begging for feedback!

[moodymudskipper]

I received positive comments when I released the package but it's REALLY HARD to get specific feedback, so If you end up here, have 5 min to spare, and would like to make me happy, please share :

• For which type of task do you use unglue ?
• Does it work as you expect ?
• What would you like to do with unglue that you can't, or think you can't ?
• • any feature request and criticism

[tmastny]

Love the package!

One thing that would be cool would be to have a shortcut to ignore whitespace. Here's an example:

library(unglue)
library(dplyr)
#> 
#> Attaching package: 'dplyr'
#> The following objects are masked from 'package:stats':
#> 
#>     filter, lag
#> The following objects are masked from 'package:base':
#> 
#>     intersect, setdiff, setequal, union
library(stringr)

# common string format
example_text <- c("20-20-32    1 file_name", "20-20-33   23 file_name2")
cat(example_text[1], example_text[2], sep = '\n')
#> 20-20-32    1 file_name
#> 20-20-33   23 file_name2

# how I do it today
example_text %>%
  unglue_data("{date} {size} {file}") %>%
  mutate(unglued = unglue(str_trim(file), "{bytes} {name}")) %>%
  tidyr::unnest(unglued)
#> # A tibble: 2 x 5
#>   date     size  file             bytes name      
#>   <chr>    <lgl> <chr>            <int> <chr>     
#> 1 20-20-32 NA    "  1 file_name"      1 file_name 
#> 2 20-20-33 NA    " 23 file_name2"    23 file_name2

# I'd like to do this to get the same answer
example_text %>%
  unglue_data("{date} {size} {file}")
#>       date size           file
#> 1 20-20-32   NA    1 file_name
#> 2 20-20-33   NA  23 file_name2

^{Created on 2020-02-19 by the reprex package (v0.3.0)}

[moodymudskipper]

Hi @tmastny , thanks for the kind words!

I believe you can get what you want by running :

example_text %>%
  unglue_data("{date}{=\\s+}{size}{=\\s+}{file}")

Where the {=\\s+} will match any number of space and won't assign it to any variable.

I understand that there might be value in something more obvious though but I can't make it the default, I'll think about it as I don't have any idea now,

[tmastny]

Thanks for the tip!

I definitely agree, ignoring whitespace shouldn't be the default. I was thinking like a function argument, unglue(...., ignore_whitespace = TRUE). But I think {=\\s+} makes a more sense, and is consistent with the rest of glue.

I was originally thinking something along the lines of this issue: #19
so you don't need any regex (even something like \\s+).

One reason I like unglue so much is that it is intuitive and I can figure out the parsing without any regex.

[moodymudskipper]

I had forgotten this wild experiment in #19! I was hesitant to implement this as I've tried to make unglue "tidy compliant" and I don't think they'd approve this weird feature.

Do you feel that the following is intuitive ?

example_text %>%
  unglue_data("{date}{~space(s)}{size}{~space(s)}{file}")

example_text %>%
  unglue_data("{date}{~one or more spaces}{size}{~one or more spaces}{file}")

Or is it just weird and mildly interesting ? :)

ignore_space = TRUE seems ambiguous to me, I'm not sure what it means here exactly.

Note that you can also do (still using regex):

example_text %>%
  gsub("\\s+", " ", .) %>% 
  unglue_data("{date} {size} {file}")

To avoid regex, if it's a task common enough, we can define a helper function

merge_multiple_spaces <- function(x) gsub("\\s+", " ", x) 

example_text %>%
  merge_multiple_spaces() %>% 
  unglue_data("{date} {size} {file}")

Or use stringr::str_squish(), which does just that

example_text %>%
  stringr::str_squish() %>% 
  unglue_data("{date} {size} {file}")
#>       date size       file
#> 1 20-20-32    1  file_name
#> 2 20-20-33   23 file_name2

Actually the latter is now my official recommended solution for this case if you use tidyverse tools in your workflow :).

[ymer]

I would like to use it with a tidyverse tibble in a simple way.

For example, let's say we start with this tibble:
a <- tibble(l = c("so_word1", "so_word2"))

Then I would like to run a command like this:
a %>% unglue(l, "so_{word}")

To get a tibble like this:
tibble(word = c("word1", "word2"))

Maybe this is already simple to do, but I have a hard time understanding how from the vignette.

[moodymudskipper]

Hi @ymer, I believe you want unglue_unnest(), it will do just that. I'll try to clarify the doc. Tell me if you still have issues and I ll run a reprex when I m in front of my computer.

[moodymudskipper]

leaving this pinned, feedback is always welcome, but closing, please open new issues!

[github-actions]

This old thread has been automatically locked. If you think you have found something related to this, please open a new issue and link to this old issue if necessary.