str_replace_all: unexpected behaviour

[shaggycamel]

Hello,

I'm experiencing some unexpected behaviour from str_replace_all.

Here is my problem:

# Create dataframe
df <- tibble(competitor_id = c(1, 5))

# Named vector used in string replacements
id_rpl <- c("1" = "Philly 5 star", "5" = "senor_cactus")

# Mutate dataframe / new column
mutate(df, competitor_name = str_replace_all(competitor_id, id_rpl))

This produces the output

  competitor_id competitor_name             
          <dbl> <chr>                       
             1 Philly senor_cactus star
             5 senor_cactus  

My problem with the output is "senor_cactus" has replaced the "5" within the "Philly 5 star" string. The replace all description is taken quite literally, almost recursively. And, replaces the original "5" within the competitor_id column, but also replaces the "5" within the string that replaces "1".

Is this intended behaviour? - Just thought i'd mention it, because it doesn't seem right.

Thanks

[werkstattcodes]

can't speak for the package author(s), but a solution could be to add some metacharacters to your patterns.

id_rpl <- c("1" = "Philly 5 star", "^5$" = "senor_cactus")

  competitor_id competitor_name
          <dbl> <chr>
1             1 Philly 5 star  
2             5 senor_cactus

[shaggycamel]

Thanks @werkstattcodes. Interesting blog, btw.

[hadley]

Yeah, the replacements happen serially so you'll either need to use a more specific regexp like @werkstattcodes suggests, or change the order of the replacements.