Extract all text between two delimiters when both delimiters appear multiple times in a string

Question

I have several rows of chat data that contain transcripts which look like this:

"Participant 1 (Me): I don't know the answer to this. Participant 2: What do you think? Maybe 20%? Participant 2: I don't know either. Participant 1 (Me): that was my guess Participant 2: ok, let's go for it! ...what do you think? Participant 1 (Me): sure! Participant 2: ok! Participant 2: aww! sorry!"

I would like to extract only the text that follows the prefix Participant 1 (Me): up until it says either Participant 1 or Participant 2. All the text that follows immediately after Participant 1 up until the aforementioned delimiters should be stored in variable called participant_1_text. I'd like to store all the remaining text in a separate variable called participant_2_text, like so:

participant_1_text = "I don't know the answer to this. that was my guess. sure!
participant_2_text = "What do you think?   Maybe 20%? I don't know either. ok, let's go for
it! ...what do you think? ok! aww! sorry!"

So all of Participant 1's text and all of Participant 2's texts are now separated.

I tried something like the following regex:

(?<=Participant 1)(.*)(?=Participant 2)

But that will match all text between the first and last occurrence of those two delimiters, instead of every match.

---

Edit: I'm trying to now take below versions of code and apply them to a dataframe containing lots of chat transcripts:

So, taking @akrun's code I made a function that separates out a given chat log to my_chat or partner_chat and returns a named list:

extract_chat <- function(chat_text){
  final_output = chat_text %>% 
    tibble(col1 = chat_text) %>% 
    mutate(col1 = str_replace_all(col1, "Participant", "\nParticipant")) %>% 
    separate_rows(col1, sep="\n") %>% 
    filter(nzchar(col1)) %>% #filter the non-empty strings
    separate(col1, into = c('Participant', "text"), sep=":") %>% 
    group_by(Participant) %>% 
    summarise(text = str_c(text, collapse= ' ')) %>% 
    mutate(Participant = ifelse(str_detect(Participant, "(Me)"), "my_chat_extracted", "partner_chat_extracted")) %>%
    spread(Participant, text)   

  return(list(my_chat_extracted = final_output$my_chat_extracted,
              partner_chat_extracted = final_output$partner_chat_extracted))
}

This seems to work fine, but I'm not sure how to mutate the actual columns in my data-frame to use this function.

Here's an example of a data.frame to use:

str1 <- "Participant 1 (Me): I don't know the answer to this. Participant 2: What do you think? Maybe 20%? Participant 2: I don't know either. Participant 1 (Me): that was my guess Participant 2: ok, let's go for it! ...what do you think? Participant 1 (Me): sure! Participant 2: ok! Participant 2: aww! sorry!"
str2 <- "Participant 1 (Me): Hey, how are you? Participant 2: I'm good, how about you? Participant 2: I'm excited.  Participant 1 (Me): I'm also good."
test = data.frame(chat = c(str1, str2))

I want to do something like:

   tester = test %>% 
      rowwise() %>% 
      mutate(my_chat_extracted = extract_chat(chat)$my_chat_extracted)

But this seems to be pretty slow on my actual dataset, and feels sloppy.

Answer 1

Here's another method using stringr:

library(stringr)  

txt <- "Participant 1 (Me): I don't know the answer to this. Participant 2: What do you think? Maybe 20%? Participant 2: I don't know either. Participant 1 (Me): that was my guess Participant 2: ok, let's go for it! ...what do you think? Participant 1 (Me): sure! Participant 2: ok! Participant 2: aww! sorry!"

txt %>%
  str_split("(?=Participant.+:)", simplify = TRUE)  %>%
  str_split(": ", simplify = TRUE) %>%
  .[-1, ]

#>      [,1]                 [,2]                                         
#> [1,] "Participant 1 (Me)" "I don't know the answer to this. "          
#> [2,] "Participant 2"      "What do you think? Maybe 20%? "             
#> [3,] "Participant 2"      "I don't know either. "                      
#> [4,] "Participant 1 (Me)" "that was my guess "                         
#> [5,] "Participant 2"      "ok, let's go for it! ...what do you think? "
#> [6,] "Participant 1 (Me)" "sure! "                                     
#> [7,] "Participant 2"      "ok! "                                       
#> [8,] "Participant 2"      "aww! sorry!"

^{Created on 2020-06-17 by the reprex package (v0.3.0)}

Answer 2

Another way to do this using stringr, where s is the given string:

r <- "Participant \\d( \\(Me\\))?: "
cbind(unlist(stringr::str_extract_all(s, r)), strsplit(s, r)[[1]][-1])
#>      [,1]                   [,2]                                         
#> [1,] "Participant 1 (Me): " "I don't know the answer to this. "          
#> [2,] "Participant 2: "      "What do you think? Maybe 20%? "             
#> [3,] "Participant 2: "      "I don't know either. "                      
#> [4,] "Participant 1 (Me): " "that was my guess "                         
#> [5,] "Participant 2: "      "ok, let's go for it! ...what do you think? "
#> [6,] "Participant 1 (Me): " "sure! "                                     
#> [7,] "Participant 2: "      "ok! "                                       
#> [8,] "Participant 2: "      "aww! sorry!"  

Answer 3

We can insert a next line character before the Participant (with str_replace_all), then split at the \n with separate_rows, filter out any blanks (nzchar), separate the column into two at :, grouped by 'Participant', paste the 'text' strings into a single string

library(dplyr)
library(stringr)
library(tidyr)
out <- tibble(col1 = str1) %>% 
  mutate(col1 = str_replace_all(col1, "Participant", "\nParticipant")) %>% 
  separate_rows(col1, sep="\n") %>% 
  filter(nzchar(col1)) %>%
  separate(col1, into = c('Participant', "text"), sep=":") %>% 
  group_by(Participant = str_remove(Participant, "\\s*\\(.*")) %>% 
  summarise(text = str_c(text, collapse= ' '))

out
# A tibble: 2 x 2
#  Participant   text                                                                                                                    
#  <chr>         <chr>                                                                                                                   
#1 Participant 1 " I don't know the answer to this.   that was my guess   sure! "                                                        
#2 Participant 2 " What do you think? Maybe 20%?   I don't know either.   ok, let's go for it! ...what do you think?   ok!   aww! sorry!"

It may be better to keep it in a data.frame, but if we need separate objects use list2env after deframeing

library(tibble)
list2env(as.list(deframe(out)), .GlobalEnv)
`Participant 1`
#[1] " I don't know the answer to this.   that was my guess   sure! "

data

str1 <- "Participant 1 (Me): I don't know the answer to this. Participant 2: What do you think? Maybe 20%? Participant 2: I don't know either. Participant 1 (Me): that was my guess Participant 2: ok, let's go for it! ...what do you think? Participant 1 (Me): sure! Participant 2: ok! Participant 2: aww! sorry!"