将字符串分成两列

Question

I am working with someone else's data with a column for which possible values are "short" and "long". Unfortunately, the data creator also added letters and question marks after those words to annotate certain things, that I want to split into a separate column. Here's some fake data to work with:

vars <- c('long','short','longG','short?','short?F','long?G')
species <- c('sp1','sp2','sp3','sp4','sp5','sp6')
testdf <- cbind(vars, species)

I would like to split the vars column into the actual value long or short , and a new column with just the annotated characters. I can get halfway there with the following, which correctly produces a new column with just the annotating characters:

testdf %>% mutate(notes = gsub('long|short',"",vars)

But I can't figure out how to split or subset var such that I get a column that just says short or long .

Thanks in advance for the help, SO community! ^_^

Answer 1

It's difficult to extract pieces of a string in base R. Using stringr instead:

library(stringr)
str_extract(vars, 'long|short')
# [1] "long"  "short" "long"  "short" "short" "long" 

(You can use it in mutate or however else.)

---

So your complete example (I would parametrize the pattern for good measure)

pattern = "long|short"
mutate(testdf,
   notes = gsub(pattern, "", vars),
   notes2 = str_replace(vars, pattern, ""), # stringr alternative for consistent syntax
   ls = str_extract(vars, pattern))

Answer 2

testdf in the question is a matrix so convert it to a data frame with one of these two alternatives:

1) sub a mutate with two sub invocations with the same pattern pat but with different replacements.

pat <- "(long|short)(.*)"
testdf %>% 
       as.data.frame %>%
       mutate(notes = sub(pat, "\\2", vars), 
              vars = sub(pat, "\\1", vars))

giving:

   vars  species notes
1  long      sp1      
2 short      sp2      
3  long      sp3     G
4 short      sp4     ?
5 short      sp5    ?F
6  long      sp6    ?G

2) separate Insert a semicolon (or other character) after long or short and then use separate from tidyr. Note that this works even if the notes contain a semicolon since it only splits at the first semicolon.

library(tidyr)

testdf %>% 
       as.data.frame %>%
       mutate(vars = sub("(long|short)", "\\1;", vars)) %>%
       separate(vars, c("vars", "notes"), sep = ";", extra = "merge")

giving:

   vars notes  species
1  long            sp1
2 short            sp2
3  long     G      sp3
4 short     ?      sp4
5 short    ?F      sp5
6  long    ?G      sp6

Note that if there is always a ? separating the notes then it could be reduced to:

testdf %>% 
       as.data.frame %>%
       separate(vars, c("vars", "notes"), sep = "\\?", extra = "merge")