Remove only some line breaks in R - r

Question

I'm reading a text file into R:
text <- read_delim("textfile.txt", "\n", escape_double = F, col_names = F, trim_ws = T)
The relevant part is that it is delimited by line breaks.
Then I separate it into a speaker column and a comments column:
text2 <- text %>%
separate(X1, into = c("speaker", "comment"), sep = ":")
The result is a data frame with a column of speakers and another column of their comments.
The issue is that some of the long comments have line breaks embedded in them. This messes up the data structure putting the comment after the line break in the speaker column and then an NA in the comments section.
How can I tell R to ignore these embedded line breaks? If it helps, the columns are separated by a colon (i.e. Interviewer: How are you?), so there should be only one colon before the "true" line break.
Thank you!

Answer 1

I'm going to work under the assumption your input file looks like this:
textfile.txt
Interviewer: How are you?
Respondant: I'm fine.
Interviewer: The issue is that some of the long comments have line breaks
embedded in them. This messes up the data structure putting the comment after
the line break in the speaker column and then an NA in the comments section.
Respondant: How can I tell R to ignore these embedded line breaks? If it helps,
the columns are separated by a colon (i.e. Interviewer: How are you?), so there
should be only one colon before the "true" line break.
If so, this process should work:
Read the lines into a vector.
Find out which lines start with a speaker's name.
Categorize all lines by where they fall between those "starting" lines.
Combine the comments into blocks.
Pull out the speaker names for each comment block.
data_frame it.
library(stringi)
library(dplyr)
text <- readLines("textfile.txt")
speaker_pattern <- "^\\w+(?=:)"
comment_starts <- which(stri_detect_regex(text, speaker_pattern))
comment_groups <- findInterval(seq_along(text), comment_starts)
comments <- text %>%
split(comment_groups) %>%
vapply(FUN = paste0, FUN.VALUE = character(1), collapse = "\n")
speakers <- stri_extract_first_regex(comments, speaker_pattern)
comments <- stri_replace_first_regex(comments, "^\\w+: ", "")
text2 <- data_frame(speaker = speakers, comment = comments)
text2
# # A tibble: 4 x 2
# speaker comment
# <chr> <chr>
# 1 Interviewer How are you?
# 2 Respondant I'm fine.
# 3 Interviewer "The issue is that some of the long comments have ~
# 4 Respondant "How can I tell R to ignore these embedded line br~