Getting grepl to return TRUE only if there is a match with the full string-CodePudding

I have example data as follows:

library(data.table)
dat <- fread("q1   q2 ...1  ..2  q3..1 ..1
       NA   response other  else response other
       1    4        NA   NA   1    NA")

I wanted to filter out all columns that are automatically named when reading in an Excel file with missing column names, which have names like ..x. I thought that the following piece of code would work:

grepl("\\. ", names(dat))
[1] FALSE FALSE  TRUE  TRUE  TRUE  TRUE

But it also filters out columns which have a similar structure as column q3..1.

Although I do not know why the ..x part is added to such a column (because it was not empty), I would like to adapt the grepl code, so that the outcome is TRUE, unless the structure is ONLY ..x.

How should I do this?

Desired output:

grepl("\\. ", names(dat))
[1] FALSE FALSE  TRUE  TRUE  FALSE TRUE

CodePudding user response：

Use an anchor ^ to state that the dots have to be in the start of the string:

grepl("^\\. ", names(dat))
#[1] FALSE FALSE  TRUE  TRUE FALSE  TRUE

CodePudding user response：

We may do

library(dplyr)
dat %>% 
   select(matches('^[^.] $'))
      q1       q2
   <int>   <char>
1:    NA response
2:     1        4