Have read_* functions parse non-traditional logicals

Functions like read_delim() and read_csv() do not recognize columns with logical data coded as "0/1" or "true/false", but it does recognize "T/F". This behavior is different than parse_logical() which accepts all of the examples listed above. Is it possible to have these functions parse similar to parse_logical()?

read_delim() Behavior

lines <- "a,b
A,true
B,true
C,false
"
read_delim(lines, ",")
# A tibble: 3 x 2
  a     b    
  <chr> <chr>
1 A     true 
2 B     true 
3 C     false

Desired Behavior

lines <- "a,b
A,true
B,true
C,false
"
read_delim(lines, ",")
# A tibble: 3 x 2
  a     b    
  <chr> <lgl>
1 A     TRUE 
2 B     TRUE 
3 C     FALSE

parse_logical() Behavior

readr::parse_logical(c("T", "F", "true", "false", 0, 1))
[1]  TRUE FALSE  TRUE FALSE FALSE  TRUE

You need to specify the columns are logical types when you read the data.

library(readr)
lines <- "a,b
A,true
B,true
C,false
"
read_delim(lines, ",", col_type = "cl")
#> # A tibble: 3 x 2
#>   a     b    
#>   <chr> <lgl>
#> 1 A     TRUE 
#> 2 B     TRUE 
#> 3 C     FALSE

Created on 2018-03-16 by the reprex package (v0.2.0).

@jimhester

Thanks, @jimhester, but I think you mistook my feature request for a question. I know about read_delim(lines, ",", col_type = "cl") (that's how I created the desired behavior block of code). I want to know why functions like read_delim are not smart enough to recognize non-traditional logicals on their own; however parse_logical() can. My feature request/enhancement is to replicate the parse_logical() behavior inside of read_delim() so it is more robust to different types of logicals.

The guesser currently does this

readr/src/CollectorGuess.cpp

Line 35 in 31dfea0

return x == "T" || x == "F" || x == "TRUE" || x == "FALSE";

But the parser does an insensitive compare

readr/src/Collector.cpp

Line 294 in 31dfea0

if (strncasecmp(string.first, "true", 4) == 0) {

So yes we could change the guesser to do the same. If you would like to open a PR to do this it would be great, if not we will get to it eventually, but it may take some time.

@jimhester

Thanks for taking a second look, @jimhester. No rush on implementing.

Here's how this is done in readxl, in case there is interest in doing same. It delegates the comparison back to R:

https://github.com/tidyverse/readxl/blob/ed499fb299934b2c3355e239644d118498edbdab/src/utils.h#L116-L126

@jennybc

Thanks @jennybc, using Rf_StringTrue seems like a good idea

@jimhester

@jimhester It was probably your idea 😁

Oh yeah (tidyverse/readxl#277 (comment)), still thank you for reminding me about it and using it for this!

This old issue has been automatically locked. If you believe you have found a related problem, please file a new issue (with reprex) and link to this issue. https://reprex.tidyverse.org/

Navigation Menu

Repository navigation

Have read_* functions parse non-traditional logicals #818

Metadata

Assignees

Labels

Type

Projects

Milestone

Relationships

Development

Have read_* functions parse non-traditional logicals #818

Description

Activity

jimhester commented on Mar 16, 2018

StevenMMortimer commented on Mar 16, 2018

jimhester commented on Mar 16, 2018

StevenMMortimer commented on Mar 16, 2018

jennybc commented on Mar 16, 2018

jimhester commented on Mar 16, 2018

jennybc commented on Mar 16, 2018

jimhester commented on Mar 16, 2018

lock commented on May 15, 2019

Metadata

Metadata

Assignees

Labels

Type

Projects

Milestone

Relationships

Development

Issue actions