rfuzzy-comparisonagrep

agrep output approximate macthing


Having

 agrep('timothy', c('timo','tim','timoth', 'timothys'), max.distance = 0.01, value=TRUE)

I want to output the original string and all possible results together in a data frame as below.

Original Replace1 Replace2
timothy  timoth   timothys

Is this possible or is there a better function to use?


Solution

  • I'd personally keep it in "long" format vs wide (you can always transformit later):

    data.frame(
      original = "timothy",
      replacement = agrep('timothy', c('timo','tim','timoth', 'timothys'), max.distance = 0.01, value=TRUE),
      stringsAsFactors=FALSE
    )
    ##   original replacement
    ## 1  timothy      timoth
    ## 2  timothy    timothys
    

    You likely want to do this more than once, so I'd make that a function. And, since the output of agrep() can be character(0), we need to handle that so we'll add in a helper function, too:

    `%|l0%` <- function(x, y) if (length(x) == 0) y else x
    
    agrep_to_data_frame <- function(pattern, x, max.distance=0.01, costs=NULL) {
      data.frame(
        original = pattern,
        replacement = agrep(pattern, x, max.distance = max.distance, value=TRUE) %|l0% NA_character_,
        stringsAsFactors=FALSE
      )
    } 
    

    And, now it's a single call which you can use in purrr::map2() or mapply(), etc.

    agrep_to_data_frame('timothy', c('timo','tim','timoth', 'timothys'))
    ##   original replacement
    ## 1  timothy      timoth
    ## 2  timothy    timothys