二人如何在句号分隔的句子中以任意顺序找到两个单词

Question

I'm trying to extract any sentence (defined as being between two periods) which have the two words column and Barr in any order in them. This is tricky as at the moment I have created a regex that only finds the two words in any order before a period but if the two words are present in two sentences then all the text between the two sentences are selected. How can I make the regex sentence specific?

Input

try<-c("I am a sentence.I am a sentence and I contain Barr. I contain other things. I contain column as well.","Here we go. I am a sentence and I contain column but also Barr. I only contain Barr. I am too.")

Desired output

[1] NA
[2] "I am a sentence and I contain column but also Barr.

Attempt

str_extract_all(try,"\\..*column. Barr. ?\\.|.*Barr. column. ?\\.")

Current output

[[1]]
[1] "I am a sentence.I am a sentence and I contain Barr. I contain other things. I contain column as well."

[[2]]
[1] ". I am a sentence and I contain column but also Barr. I only contain Barr."

Answer 1

In order to find two words present in any order, you can use two positive lookaheads: For example grepl((?=.*Barr)(?=.*column),x,perl=T) will return TRUE every time both words are present, regardless of their order and FALSE otherwise, but this doesn't take into account the sentence structure. As you want to extract text, and you want to find the two words in between dots, we can change it to:

library(stringr)
## Example data
x <- c("I am a sentence.I am a sentence and I contain Barr. I contain other things. I contain column as well.","Here we go. I am a sentence and I contain column but also Barr. I only contain Barr. I am too.","Barr and column and also column. But just Barr. And just column. Now again column and Barr")
> x
[1] "I am a sentence.I am a sentence and I contain Barr. I contain other things. I contain column as well."
[2] "Here we go. I am a sentence and I contain column but also Barr. I only contain Barr. I am too."       
[3] "Barr and column and also column. But just Barr. And just column. Now again column and Barr"           

str_extract_all(x,"(\\.|^)(?=[^\\.]*Barr)(?=[^\\.]*column)[^\\.]*(\\.|$)")

This looks for a start of sentence or a dot (\\\\.|^) , followed by characters that are not dots and that contain Barr and column (?=[^\\\\.]*Barr)(?=[^\\\\.]*column)[^\\\\.]* , followed by a dot or an end of sentence (\\\\.|$) . This returns a list:

[[1]]
character(0)

[[2]]
[1] ". I am a sentence and I contain column but also Barr."

[[3]]
[1] "Barr and column and also column." ". Now again column and Barr"

Answer 2

This regex seems to do what you need:

(\\.[^.]*column[^.]*Barr[^.]*)|(\\.[^.]*Barr[^.]*column[^.]*)

It will start with a point ( . ) and grabs anything that is not a point but also has column and Barr . Or the same with both words in a different order.

Example:

try = c("I am a sentence.I am a sentence and I contain Barr. I contain other things. I contain column as well.",
        "Here we go. I am a sentence and I contain column but also Barr. I only contain Barr. I am too.",
        "I am a sentence and I contain column but also Barr. I only contain Barr. I am too.",
        "I contain column and Barr. I have Barr and column. I don't.",
        "Hello. I contain Barr and column but also Barr. I only contain Barr. I am too.") 

k = sapply(try, function(x){
  str_extract(paste0(".",x), "(\\.[^.]*column[^.]*Barr[^.]*)|(\\.[^.]*Barr[^.]*column[^.]*)")
})
names(k) = NULL

Result:

[1] NA                                                    
[2] ". I am a sentence and I contain column but also Barr"
[3] ".I am a sentence and I contain column but also Barr" 
[4] ".I contain column and Barr"                          
[5] ". I contain Barr and column but also Barr"

If you use str_extract_all keep in mind that it returns a list of matches.

[[1]]
character(0)

[[2]]
[1] ". I am a sentence and I contain column but also Barr"

[[3]]
[1] ".I am a sentence and I contain column but also Barr"

[[4]]
[1] ".I contain column and Barr" ". I have Barr and column"  

[[5]]
[1] ". I contain Barr and column but also Barr"

I've added a paste0(".",x) in order to detect sentences that contain both words and are first (they don't start with a period).

Answer 3

Here is a more general attempt which does not require creating every permutation of the desired words, helpful when more than two works are required.

The strategy is find the sentences with each individual word and then find the intersection of for the results.

#split the long text into individual sentences.
sentences<-strsplit(try, "\\.")

#create list of matches for each desired word
columnlist<-lapply(sentences, function(x) {grep("(column)", x)})
barrlist<-lapply(sentences, function(x) {grep("(Barr)", x)})

#find intersection between lists
intersection<-lapply(seq_along(columnlist), function(i){intersect(columnlist[[i]], barrlist[[i]])} )

#extract the sentences out
answer<-sapply(seq_along(intersection), function(i) { 
  if(length(intersection[[i]])) 
    {trimws(sentences[[i]][intersection[[i]] ])}  
  else {NA}
})

Result

#[[1]]
#[1] NA
#
#[[2]]
#[1] "I am a sentence and I contain column but also Barr"