using regex in tidyverse to filter - unclear why this does not work

Question

I'm trying to filter rows containing the string "Data\\this\\way\\test". Unclear as to why this does not work. Ideally would expect to see an output like

  "D:\\Data\\this\\way\\test\\dat1",
  "D:\\Data\\this\\way\\test\\dat2",

My code:

files <- c(
  "D:\\Data\\this\\way\\test\\dat1",
  "D:\\Data\\this\\way\\test\\dat2",
  "D:\\Data\\not-this\\way\\test\\dat1",
  "D:\\Data\\not-this\\way\\test\\dat2"
)

files_filt_df <- data.frame(filenames = files, 
                             stringsAsFactors = FALSE) %>%
  filter(str_detect(filenames,"Data\\this\\way\\test"))
files_filt_df
[1] filenames
<0 rows> (or 0-length row.names)

"\\" is an escaped "\". To get two escaped "\", use "\\\\". Or "Data\\\\this\\\\way\\\\test" in your case. — dipetkov, Nov 01 '19 at 21:01

score 2 · Answer 1 · answered Nov 01 '19 at 21:01

By default str_detect expects you to pass a regular expression. Things like \w have special meaning in regular expressions. If you just want to match a literal value, the easiest way would be

files_filt_df <- data.frame(filenames = files, 
                            stringsAsFactors = FALSE) %>%
  filter(str_detect(filenames,fixed("Data\\this\\way\\test")))

Or if you want to use a regular expression, you need to add an additional level of escaping on the slashes

files_filt_df <- data.frame(filenames = files, 
                            stringsAsFactors = FALSE) %>%
  filter(str_detect(filenames,"Data\\\\this\\\\way\\\\test"))

dipetkov · Accepted Answer · 2019-11-02T18:29:50.070

Since these are filenames, you can also use the fs package to check is a file has a particular parent and let fs deal with the file separators.

library("tidyverse")

files <- c(
  "D:\\Data\\this\\way\\test\\dat1",
  "D:\\Data\\this\\way\\test\\dat2",
  "D:\\Data\\not-this\\way\\test\\dat1",
  "D:\\Data\\not-this\\way\\test\\dat2"
)

tibble(
  file = files
) %>%
  filter(map_lgl(file, ~ fs::path_has_parent(., "D:/Data/this/way")))
#> # A tibble: 2 x 1
#>   file                             
#>   <chr>                            
#> 1 "D:\\Data\\this\\way\\test\\dat1"
#> 2 "D:\\Data\\this\\way\\test\\dat2"

# Explanation:

# The `map_lgl` applies `fs::path_has_parent` to each file
# and returns TRUE/FALSE (logical = lgl) values.

# Without `map`:
fs::path_has_parent(files, "D:/Data/this/way")
#> [1] FALSE

# With `map`:
map_lgl(files, ~ fs::path_has_parent(., "D:/Data/this/way"))
#> [1]  TRUE  TRUE FALSE FALSE

# The `~` operator creates a formula.
# Here it is shorter than defining an inline function.

# Formula:
map_lgl(files, ~ fs::path_has_parent(., "D:/Data/this/way"))
#> [1]  TRUE  TRUE FALSE FALSE

# Function:
map_lgl(files, function(x) fs::path_has_parent(x, "D:/Data/this/way"))
#> [1]  TRUE  TRUE FALSE FALSE

^{Created on 2019-11-02 by the reprex package (v0.3.0)}

could you explain what map_lgl is doing and how to interpret the `~` and `.`. I'm still learning the map() function and often confused by the use of it with and without `~`. — val, Nov 02 '19 at 17:44
The use of `~` in purrr is subtle. Here is a nice explanation: https://stackoverflow.com/questions/44834446/what-is-meaning-of-first-tilde-in-purrrmap — dipetkov, Nov 02 '19 at 18:31

score 0 · Answer 3 · answered Nov 05 '19 at 05:05

0

Base R solution (find the literal pattern in files, return all values matching pattern):

data.frame(files = grep("\\Data\\this\\way\\", files, value = T, fixed = T), stringsAsFactors = F)

answered Nov 05 '19 at 05:05

hello_friend

5,682
1
11
15

using regex in tidyverse to filter - unclear why this does not work

3 Answers3