Making gsub only replace entire words?

问题

(I'm using R.) For a list of words that's called "goodwords.corpus", I am looping through the documents in a corpus, and replacing each of the words on the list "goodwords.corpus" with the word + a number.

So for example if the word "good" is on the list, and "goodnight" is NOT on the list, then this document:

I am having a good time goodnight

would turn into:

I am having a good 1234 time goodnight

**I'm using this code (EDIT- made this reproducible):

goodwords.corpus <- c("good")
test <- "I am having a good time goodnight"
for (i in 1:length(goodwords.corpus)){
test <-gsub(goodwords.corpus[[i]], paste(goodwords.corpus[[i]], "1234"), test)
}

However, the problem is I want gsub to only replace ENTIRE words. The issue that arises is that: "good" is on the "goodwords.corpus" list, but then "goodnight", which is NOT on the list, is also affected. So I get this:

I am having a good 1234 time good 1234night

Is there anyway I can tell gsub to only replace ENTIRE words, and not words that might be a part of other words?

I want to use this:

test <-gsub("\\<goodwords.corpus[[i]]\\>", paste(goodwords.corpus[[i]], "1234"), test)
}

I've read that the \< and \> will tell gsub to only look for whole words. But obviously that doesn't work, because goodwords.corpus[[i]] won't work when it's in quotes.

Any suggestions?

回答1:

You are so close to getting this. You're already using paste to form the replacement string, why not use it to form the pattern string?

goodwords.corpus <- c("good")
test <- "I am having a good time goodnight"
for (i in 1:length(goodwords.corpus)){
    test <-gsub(paste0('\\<', goodwords.corpus[[i]], '\\>'), paste(goodwords.corpus[[i]], "1234"), test)
}
test
# [1] "I am having a good 1234 time goodnight"

(paste0 is merely paste(..., sep='').)

(I posted this the same time as @MatthewLundberg, and his is also correct. I'm actually more familiar with using \b vice \<, but I thought I'd continue with using your code.)

回答2:

Use \b to indicate a word boundary:

> text <- "good night goodnight"
> gsub("\\bgood\\b", paste("good", 1234), text)
[1] "good 1234 night goodnight"

In your loop, something like this:

for (word in goodwords.corpus){
  patt <- paste0('\\b', word, '\\b')
  repl <- paste(word, "1234")

  test <-gsub(patt, repl, test)
}

来源：https://stackoverflow.com/questions/22888646/making-gsub-only-replace-entire-words

标签

gsub

topic-modeling