Replace single backslash in R

Question

I have a string that looks like:

str<-"a\f\r"

I'm trying to remove the backslashes but nothing works:

gsub("\","",str, fixed=TRUE)
gsub("\\","",str)
gsub("(\)","",str)
gsub("([\])","",str)

...basically all the variations you can imagine. I have even tried the string_replace_all function. ANY HELP??

I'm using R version 3.1.1; Mac OSX 10.7; the dput for a single string in my vector of strings gives:

dput(line)
"ud83d\ude21\ud83d\udd2b"

I imported the file using readLines from a standard .txt file. The content of the file looks something like: got an engineer booked for this afternoon \ud83d\udc4d all now hopefully sorted\ud83d\ude0a I m going to go insane ud83d\ude21\ud83d\udd2b in utf8towcs …

Thanks.

Solution 1

Since there isn't any direct ways to dealing with single backslashes, here's the closest solution to the problem as provided by David Arenburg in the comments section

gsub("[^A-Za-z0-9]", "", str) #remove all besides the alphabets & numbers

Solution 2

One quite universal solution is

gsub("\\\\", "", str)

Thanks to the comment above.

Solution 3

When inputting backslashes from the keyboard, always escape them.

str <-"this\\is\\my\\string"    # note doubled backslashes -> 'this\is\my\string'
gsub("\\", "", str, fixed=TRUE) # ditto

str2 <- "a\\f\\r"               # ditto -> 'a\f\r'
gsub("\\", "", str2, fixed=TRUE)# ditto

Note that if you do

str <- "a\f\r"

then str contains no backslashes. It consists of the 3 characters a, \f (which is not normally printable, except as \f, and \r (same).

And just to head off a possible question. If your data was read from a file, the file doesn't have to have doubled backslashes. For example, if you have a file test.txt containing

a\b\c\d\e\f

and you do

str <- readLines("test.txt")

then str will contain the string a\b\c\d\e\f as you'd expect: 6 letters separated by 5 single backslashes. But you still have to type doubled backslashes if you want to work with it.

str <- gsub("\\", "", str, fixed=TRUE)  # now contains abcdef

---

From the dput, it looks like what you've got there is UTF-16 encoded text, which probably came from a Windows machine. According to

• https://en.wikipedia.org/wiki/Unicode#Character_General_Category
• https://en.wikipedia.org/wiki/UTF-16

it encodes glyphs in the Supplementary Multilingual Plane, which is pretty obscure. I'll guess that you need to supply the argument encoding="UTF-16" to readLines when you read in the file.

Solution 4

This might be helpful :)

require(stringi)
stri_escape_unicode("ala\\ma\\kota")
## [1] "ala\\\\ma\\\\kota"
stri_unescape_unicode("ala\\ ma\\ kota")
## [1] "ala ma kota"

Solution 5

As of R 4.0.0, you can now use raw strings to avoid confusion with backlashes, just use the following syntax: r"(your_raw_expression)" (parentheses included):

str<-r"(ud83d\ude21\ud83d\udd2b)" #Equivalent of "ud83d\\ude21\\ud83d\\udd2b"
gsub(r"(\\)", "", str)
# [1] "ud83dude21ud83dudd2b"

View more solutions

Comments

[TooTone]

related stackoverflow.com/questions/11806501/backslash-in-r-string

[Tavi]

@TooTone it is not quite the same problem. I've worked out dealing with double slashes, but I couldn't find any solutions to single slashes at all.

[David Arenburg]

@maryam str<-"this\is\my\string" can't be read into R because \i, \m and \s are stored letters in regex, so your example isn't reproducible (and confusing). You could change it to something like str <- "a\f\r"

[Tavi]

@DavidArenburg okay i'll do that

[Hong Ooi]

@DavidArenburg That doesn't contain any backslashes either. You need to escape them if you're inputting them from the keyboard: str <- "a\\f\\r"

[TooTone]

@DavidArenburg I've deleted my answer which addressed the original version of the question.

[David Arenburg]

@TooTone, seems like third exactly the same answer was just posted...

[TooTone]

@HongOoi I'm not sure about your edit to the post. I think we should leave it to the OP to clarify whether it's "this\\is\\my\\string" or "a\f\r" as David suggested.

[Hong Ooi]

I've rolled it back to the original.

[Tavi]

@HongOoi its like David suggested "a\f\r" all I want to do is remove the darn single backslashes

[Hong Ooi]

@maryam You have to be careful. If you want to have a string that contains a backslash, you have to type two backslashes, because \ is a special character to the R parser. For example str <- "a\\f\\r" contains the string 'a\f\r' (two single backslashes). Conversely, str <- "a\f\r" contains no backslashes: it has 3 characters, a, \f and \r.

[Tavi]

@HongOoi but I don't want a string that contain backslashes in the first place. I just got a messed up text dataset that I'm trying to clean up and it has all sords of weird characters in it. I take it there is NO way to deal with single backslashes then?

[Hong Ooi]

@maryam See my edited answer.

[Hong Ooi]

It sounds like your problem is not actually backslashes, but mangled text encoding. If R encounters a character it doesn't know how to print, you'll see lots of backslashes over the place, representing hex codes. Post an actual sample of your code and data, using dput.

[Tavi]

@HongOoi alright, but what is dput? sorry

[Hong Ooi]

See ?dput. It prints something that you can cut and paste into the edit window here on StackOverflow, that will show R's internal representation of your data.

[Tavi]

@HongOoi > dput(line) gives "ud83d\ude21\ud83d\udd2b" where line is the name of my string

[Hong Ooi]

Put that into your question, along with the code you used to read the file and the OS you're using. Each \uXXXX represents a Unicode character which R doesn't know how to handle in your locale/encoding.

[Tavi]

@HongOoi thanks, done!

[Hong Ooi]

Can you also post a fragment of the file you're reading, and the code you used to read it with?

[David Arenburg]

It seems like you have lots of gebberish there, try gsub("[^A-Za-z0-9]", "", "ud83d\ude21\ud83d\udd2b") maybe?

[Tavi]

@HongOoi there you go. thanks

[Tavi]

@DavidArenburg it works david but only the first part of the string is returned "ud83d"

[Hong Ooi]

utf8towcs... you're not working with funky foreign languages are you?

[Dason]

Yeah that looks like an encoding issue to me.

[Hong Ooi]

Can you upload the file somewhere? It looks like whatever editor you're pasting it from has already turned the weird characters into \xxxx form.

[Tavi]

@HongOoi haha noooo its all english :) actually david's idea to try to substitute everything besides the standard alphabets has the potential to work but for now its not doing exactly what its supposed to do

[Hong Ooi]

Right, that's because each \uXXXX is actually one character.

[Tavi]

@HongOoi one character? oh yeah, now I see… I guess the problem is solved then, I'll use gsub("[^A-Za-z0-9]", "", "ud83d\ude21\ud83d\udd2b") thanks a lot Hong do you know how to mark a comment as answer? so perhaps others shall find it useful?

[Dason]

Like I said - it's an encoding issue ;)

[Tavi]

@Dason yes it is, glad its over now been on this for the past 6 hours. thanks to everyone that contributed

[David Arenburg]

Did you just adopt the question to your answer or am I missing something? She has str<-"a\f\r" string, not str <-"this\\is\\my\\string"

[Hong Ooi]

The original, before your edits, was str <- "this\is\my\string". This isn't something you can enter into R from the keyboard. If it was meant to represent the string "this\is\my\string", you have to escape the backslashes.

[David Arenburg]

Actually I didn't edit it to str <- "a\f\r", the OP did. The error was cause not because of the backslashes but because of the \i, \m and \s

[Hong Ooi]

True, but it was on your suggestion.

[David Arenburg]

Because the error wasn't caused due to backslashes, but due to stored letters, so her example wasn't reproducible. You have 10K rep, you can see deleted posts, don't you find any similarity between your answer and the other two deleted?

[Tavi]

please clues anyone? I've wasted so much time trying to do this

[Hong Ooi]

@DavidArenburg The deleted answers would have worked fine, there was no reason to delete them.

[David Arenburg]

If you adjust the question to the answer (like you did), they would work fine, yes.

[Hong Ooi]

I didn't adjust the question, which is "how to replace a single backslash". All the answers do exactly this; or they did, until people tried to replace a nonexistent backslash.... I can't undelete an answer, but martin and tootone are free to undelete if they wish.

[Adrian]

This is not a good solution, because it removes spaces as well. Better would be gsub("\\\\", "", str)

[jon]

Has this ever been resolved? @HongOoi here is a reproducible example that your solution does not fix: tst <- "[179,\"3 min\"]" gsub("\\", "", tst, fixed=TRUE)

[Hong Ooi]

@jon once you understand how many backslashes there are in your string, all will become clear

[Peter Prevos]

Add a space to the excluded characters: gsub("[^A-Za-z0-9 ]", "", str)

[Carl Witthoft]

Except that this doesn't in fact work: > str <- '\a\f\r' > str [1] "\a\f\r" > gsub("\\\\", "", str) [1] "\a\f\r"

[Carl Witthoft]

This does not work in R 4.X . There's still a mess with character strings getting interpreted, and "raw" data maniplulations such as the new r"('this\string')"

[Carl Witthoft]

The only problem here is: what if I have a string mystring <- 'ud83d\ud21' . I can't insert that inside the r"( construct because nothing gets interpreted inside that.

[Hong Ooi]

@CarlWitthoft once you understand how many backslashes there are in your string, all will become clear

[Carl Witthoft]

@HongOoi I do understand, e.g. if I have this \f has none I can search for \\\\f . But I cannot search purely for "\\\\"because R lumps \f as a single special character when interpreting a character string. What the R-lang needs is, maybe, a new class "literally_characters" so that nothing gets handled by the console interpreter.

[Hong Ooi]

@CarlWitthoft once you understand how many backslashes there are in your string, all will become clear

[Carl Witthoft]

@HongOoi , Yes, I know that the \f is a single character and that the console does not "believe" there are any backslash characters in my character string. The point is that we shouldn't have to search for "\\\\f" and "\\\\g" and "\\\\h" ad infinitum. So, backing up: how do we handle a character string that we've been given from elsewhere which has single backslashes that we have to modify? There's no guarantee it was read from a text file (thus bypassing the console parser)

[Hong Ooi]

@CarlWitthoft "read from a text file"? What does that have to do with the price of tea in China?

[Carl Witthoft]

@HongOoi if you try it, you'll see that R's text reader preserves the backslash as a separate character

[Hong Ooi]

@CarlWitthoft and?