部分字符串匹配两列R.

Question

I have been trying to partially match two column contents based on a list of regular expressions common to both columns:

dats<-data.frame(ID=c(1:3),species=c("dog","cat","rabbit"),
species.descriptor=c("all animal dog","all animal cat","rabbit exotic"),product=c(1,2,3),
product.authorise=c("all animal dog cat rabbit","cat horse pig","dog cat"))

with the aim of achieving this:

goal<-data.frame(ID=c(1:3),species=c("dog","cat","rabbit"),
            species.descriptor=c("all animal dog","all animal cat","rabbit exotic"),
            product=c(1,2,3),product.authorise=c("all animal dog cat rabbit","cat horse pig",
            "dog cat"), authorised=c("TRUE","TRUE","FALSE"))    

So to explain further, if 'dog' appears at any point in both columns, then this would be considered 'TRUE' in $match - and this would apply for any individual species descriptor.If no matches are found, then a return of either FALSE or an na would be fine.

So far I have gotten to this point:

library(stringr)
patts<-c("dog","cat","all animal")
reg.patts<-paste(patts,collapse="|")
dats$matched<-ifelse((str_extract(dats$species.descriptor,reg.patts) == str_extract(dats$product.authorise,reg.patts)),"TRUE","FALSE")
dats
  ID species species.descriptor product         product.authorise matched
   1     dog     all animal dog       1 all animal dog cat rabbit    TRUE
   2     cat     all animal cat       2             cat horse pig   FALSE
   3  rabbit      rabbit exotic       3                   dog cat    <NA>

As you can see, this correctly identifies the first and last rows as 'all animal' appears first in both strings, and there is no match at all in the last. However, it seems to struggle (as in the second row) when the reg exp doesn't appear first in the string. I have tried str_extract_all, but have only resulted in error messages so far. I was wondering if anyone can help, please?

Answer 1

Here is a solution using dplyr for piping. The core component is using grepl for logical string matching of species in both species.descriptor and product.authorised .

library(dplyr)
dats %>%
rowwise() %>%
mutate(authorised = 
           grepl(species, species.descriptor) & 
           grepl(species, product.authorise)
       )

Source: local data frame [3 x 6]
Groups: <by row>

     ID species species.descriptor product         product.authorise authorised
  (int)  (fctr)             (fctr)   (dbl)                    (fctr)      (lgl)
1     1     dog     all animal dog       1 all animal dog cat rabbit       TRUE
2     2     cat     all animal cat       2             cat horse pig       TRUE
3     3  rabbit      rabbit exotic       3                   dog cat      FALSE

If you really like stringr you can use the str_detect function for more user friendly syntax.

library(stringr)
dats %>%
mutate(authorised = 
           str_detect(species.descriptor, species) & 
           str_detect(product.authorise, species)
       )

And if you don't like dplyr you can add the column directly

dats$authorised <- 
    with(dats, 
         str_detect(species.descriptor, species) & 
             str_detect(product.authorise, species)
         )