How to avoid overwriting a variable by a tibble column name?-CodePudding

I observed the behavior of the filter() function which bothered me a bit.

I wan to select rows from a tibble that match simple criterion, they have some value in a particular column. The value is stored in a variable. Super easy stuff. The problem is that a name of my variable is this same as a column name of a tibble, which produces behaviour that i didn't expect.

>block 
[1] 41
>msg %>% filter(text == "trial_run" & block == block)
# A tibble: 42 × 4
    ...1 block    time text     
   <dbl> <dbl>   <dbl> <chr>    
 1    14     1 1149175 trial_run
 2    30     2 1164422 trial_run
 3    46     3 1193408 trial_run
 4    62     4 1199713 trial_run
 5    78     5 1211763 trial_run
 6    94     6 1218312 trial_run
 7   110     7 1222947 trial_run
 8   126     8 1236795 trial_run
 9   142     9 1247513 trial_run
10   158    10 1254297 trial_run
# … with 32 more rows

I assume that block == block produces a tautology like 1 == 1 thus my block criterion wasn't applied. A simple solution is just to change the name of a block variable.

> blk <- block
> msg %>% filter(text == "trial_run" & block == blk)
# A tibble: 1 × 4
   ...1 block    time text     
  <dbl> <dbl>   <dbl> <chr>    
1   654    41 1513347 trial_run

But I feel that this situation may lead me to a hard-to-track bug in the future. What am I doing wrong? How can I avoid this problem in the future (aside from making unique variable and column names)?

CodePudding user response：

The filter command has .data and .env "pronouns" that you can use to make it clear where variables are coming from. Try

msg %>% filter(text == "trial_run" & .data$block == .env$block)

The .data means it's coming from the data.frame, and .env means it's a value from the environment. You can read more about in the rlang help. There is an example just like this on that page.

CodePudding user response：

We can either escape !!

library(dplyr)
block <- 4
msg %>% 
    filter(text == "trial_run" & block == !!block)

-output

...1 block    time      text
4   62     4 1199713 trial_run

or may access the variable from the globalenv

msg %>% 
    filter(text == "trial_run" & block == .GlobalEnv$block)

-output

...1 block    time      text
4   62     4 1199713 trial_run

data

msg <- structure(list(...1 = c(14L, 30L, 46L, 62L, 78L, 94L, 110L, 126L, 
142L, 158L), block = 1:10, time = c(1149175L, 1164422L, 1193408L, 
1199713L, 1211763L, 1218312L, 1222947L, 1236795L, 1247513L, 1254297L
), text = c("trial_run", "trial_run", "trial_run", "trial_run", 
"trial_run", "trial_run", "trial_run", "trial_run", "trial_run", 
"trial_run")), class = "data.frame", row.names = c("1", "2", 
"3", "4", "5", "6", "7", "8", "9", "10"))