R semicolon delimited a column into rows

Question

I am using RStudio 2.15.0 and have created an object from Excel using XLConnect with 3000+ rows and 12 columns I am trying to delimit/split a column into the rows but don\'t

Answer 1

Here is a base R solution. Split the PolId field using strplit and for each such split field cbind it with the corresponding Description. This gives a list of matrices which we rbind together. Finally set the column names.

out <- do.call(rbind, Map(cbind, strsplit(DF$PolId, ";"), DF$Description))
colnames(out) <- colnames(DF)

giving:

> out
      PolId    Description
 [1,] "ABC123" "TEST1"    
 [2,] "ABC456" "TEST1"    
 [3,] "ABC789" "TEST1"    
 [4,] "ABC123" "TEST1"    
 [5,] "ABC456" "TEST1"    
 [6,] "ABC789" "TEST1"    
 [7,] "ABC123" "TEST1"    
 [8,] "ABC456" "TEST1"    
 [9,] "ABC789" "TEST1"    
[10,] "AAA123" "TEST1"    
[11,] "AAA123" "TEST2"    
[12,] "ABB123" "TEST3"    
[13,] "ABC123" "TEST3"    
[14,] "ABB123" "TEST3"    
[15,] "ABC123" "TEST3" 

Note: We used this as the input:

DF <-
structure(list(PolId = c("ABC123;ABC456;ABC789;", "ABC123;ABC456;ABC789;", 
"ABC123;ABC456;ABC789;", "AAA123;", "AAA123;", "ABB123;ABC123;", 
"ABB123;ABC123;"), Description = c("TEST1", "TEST1", "TEST1", 
"TEST1", "TEST2", "TEST3", "TEST3")), .Names = c("PolId", "Description"
), class = "data.frame", row.names = c(NA, -7L))

Answer 2

You could try unnest from tidyr after splitting the "PolId" column and get the unique rows

library(dplyr)
library(tidyr)
 unnest(setNames(strsplit(df$PolId, ';'), df$Description), 
                                  Description) %>% unique()

Or using base R with stack/strsplit/duplicated. Split the "PolId" (strsplit) by the delimiter(;), name the output list elements with "Description" column, stack the list to get a 'data.frame' and use duplicated to remove the duplicate rows.

df1 <- stack(setNames(strsplit(df$PolId, ';'), df$Description))
setNames(df1[!duplicated(df1),], names(df))
#     PolId Description
#1  ABC123       TEST1
#2  ABC456       TEST1
#3  ABC789       TEST1
#10 AAA123       TEST1
#11 AAA123       TEST2
#12 ABB123       TEST3
#13 ABC123       TEST3

Or another option without using strsplit

v1 <- with(df, tapply(PolId, Description, FUN= function(x) {
            x1 <- paste(x, collapse=";")
        gsub('(\\b\\S+\\b)(?=.*\\b\\1\\b.*);', '', x1, perl=TRUE)}))
library(stringr)
Description <- rep(names(v1),  str_count(v1, '\\w+'))
PolId <- scan(text=gsub(';+', ' ', v1), what='', quiet=TRUE)
data.frame(PolId, Description)
#   PolId Description
#1 ABC123       TEST1
#2 ABC456       TEST1
#3 ABC789       TEST1
#4 AAA123       TEST1
#5 AAA123       TEST2
#6 ABB123       TEST3
#7 ABC123       TEST3

Answer 3

Here's a quick data.table possible solution

library(data.table)
unique(setDT(df)[, .(PolId = unlist(strsplit(as.character(PolId), ";"))), by = Description])
#    Description  PolId
# 1:       TEST1 ABC123
# 2:       TEST1 ABC456
# 3:       TEST1 ABC789
# 4:       TEST1 AAA123
# 5:       TEST2 AAA123
# 6:       TEST3 ABB123
# 7:       TEST3 ABC123

Per your edit- Another option (in case you have more than two columns)

library(splitstackshape)
unique(cSplit(df, "PolId", ";", "long"))
#     PolId Description Document.Type
# 1: ABC123       TEST1          Pol1
# 2: ABC456       TEST1          Pol1
# 3: ABC789       TEST1          Pol1
# 4: AAA123       TEST1          End1
# 5: AAA123       TEST2          End2
# 6: ABB123       TEST3          End1
# 7: ABC123       TEST3          End1