Split

Below is a script in bash the uses the awk tokenizer to do the work.

This assumes that your input and output delimiter is space. The number of consecutive delimiters in
the input is not important. This also assumes that the input file does not have a header row. That
is easy to modify if you want. I always keep header rows in my data files as I think that removing
them is asking for trouble down the road.

I added a NULL for cases where there is no value for the last field. You could use "." if you want.

You should be able to find how to run this from inside R if you want. You will, of course, need a
bash environment to run this, so if you are not in linux you will need cygwin or something similar.

This should be very fast, but let me know if needs to be faster. If the X1_X2 variant occurs less
frequently than not then we should switch the order in which the logic evaluates the options.

LMH


#! /bin/bash

# input filename
input_file=$1

# output filename
output_file=$2

# make sure the input file exists
if [ ! -f $input_file ]; then
   echo $input_file "  cannot be found"
   exit 0
fi

# create the output file
touch $output_file

# make sure the output was created
if [ ! -f $output_file ]; then
   echo $output_file "  was not created"
   exit 0
fi

# write the header row
echo "ID1 ID2 Y1 X1 X2" >> $output_file

# character to find in the third token
look_for='_'

# process with awk
# if the 3rd token contains '_'
#   split the third token on '_' into F[1] and F[2]
#   print the first two tokens, the indicator value of 1, and the split fields F[1] and F[2]
# otherwise,
#   print the first two tokens, the indicator value of 0, the 3rd token, and NULL

cat $input_file | \
awk -v find_char=$look_for '{ if($3 ~ find_char) { { split ($3, F, "_") }
                                                   { print $1, $2, "1", F[1], F[2] }
                                                 }
                              else { print $1, $2, "0", $3, "NULL" }
                            }' >> $output_file







Val wrote:

Thank you all for the help!

LMH, Yes I would like to see the alternative.  I am using this for a
large data set and if the  alternative is more efficient than this
then I would be happy.

On Tue, Sep 22, 2020 at 6:25 PM Bert Gunter <bgunter.4567 at gmail.com> wrote:

To be clear, I think Rui's solution is perfectly fine and probably better than what I offer below. But just for fun, I wanted to do it without the lapply().  Here is one way. I think my comments suffice to explain.

## which are the  non "_" indices?
wh <- grep("_",F1$text, fixed = TRUE, invert = TRUE)
## paste "_." to these
F1[wh,"text"] <- paste(F1[wh,"text"],".",sep = "_")
## Now strsplit() and unlist() them to get a vector
z <- unlist(strsplit(F1$text, "_"))
## now cbind() to the data frame
F1 <- cbind(F1, matrix(z, ncol = 2, byrow = TRUE))
F1

  ID1 ID2   text    1  2
1  A1  B1 NONE_. NONE  .
2  A1  B1  cf_12   cf 12
3  A1  B1 NONE_. NONE  .
4  A2  B2  X2_25   X2 25
5  A2  B3  fd_15   fd 15

## You can change the names of the 2 columns yourself

Cheers,
Bert

Bert Gunter

"The trouble with having an open mind is that people keep coming along and sticking things into it."
-- Opus (aka Berkeley Breathed in his "Bloom County" comic strip )


On Tue, Sep 22, 2020 at 12:19 PM Rui Barradas <ruipbarradas at sapo.pt> wrote:

Hello,

A base R solution with strsplit, like in your code.

F1$Y1 <- +grepl("_", F1$text)

tmp <- strsplit(as.character(F1$text), "_")
tmp <- lapply(tmp, function(x) if(length(x) == 1) c(x, ".") else x)
tmp <- do.call(rbind, tmp)
colnames(tmp) <- c("X1", "X2")
F1 <- cbind(F1[-3], tmp)    # remove the original column
rm(tmp)

F1
#  ID1 ID2 Y1   X1 X2
#1  A1  B1  0 NONE  .
#2  A1  B1  1   cf 12
#3  A1  B1  0 NONE  .
#4  A2  B2  1   X2 25
#5  A2  B3  1   fd 15


Note that cbind dispatches on F1, an object of class "data.frame".
Therefore it's the method cbind.data.frame that is called and the result
is also a df, though tmp is a "matrix".


Hope this helps,

Rui Barradas


?s 20:07 de 22/09/20, Rui Barradas escreveu:

Hello,

Something like this?


F1$Y1 <- +grepl("_", F1$text)
F1 <- F1[c(1, 2, 4, 3)]
F1 <- tidyr::separate(F1, text, into = c("X1", "X2"), sep = "_", fill =
"right")
F1


Hope this helps,

Rui Barradas

?s 19:55 de 22/09/20, Val escreveu:

HI All,

I am trying to create   new columns based on another column string
content. First I want to identify rows that contain a particular
string.  If it contains, I want to split the string and create two
variables.

Here is my sample of data.
F1<-read.table(text="ID1  ID2  text
A1 B1   NONE
A1 B1   cf_12
A1 B1   NONE
A2 B2   X2_25
A2 B3   fd_15  ",header=TRUE,stringsAsFactors=F)
If the variable "text" contains this "_" I want to create an indicator
variable as shown below

F1$Y1 <- ifelse(grepl("_", F1$text),1,0)


Then I want to split that string in to two, before "_" and after "_"
and create two variables as shown below
x1= strsplit(as.character(F1$text),'_',2)

My problem is how to combine this with the original data frame. The
desired  output is shown   below,


ID1 ID2  Y1   X1    X2
A1  B1    0   NONE   .
A1  B1   1    cf        12
A1  B1   0  NONE   .
A2  B2   1    X2    25
A2  B3   1    fd    15

Any help?
Thank you.