交叉引用列表列表，如果匹配则附加子列表

Question

Sorry for insufficient clarity in the title – I'm having a hard time explaining what I need.

I have a data.frame containing text, eg:

text <- c("a",
          "bb",
          "c ccc",
          "fff")

text_df <- data.frame(line = 1:length(text), text = text, stringsAsFactors = FALSE)

Additionally, I have a list with text that I want to cross-reference:

lol <- list(c('a', 'aa', 'aaa'),
            c('d', 'dd', 'ddd'),
            c('e', 'ee', 'eee'),
            c('c', 'cc', 'ccc', 'cccc'),
            c('b', 'bb', 'bbb'),
            c('f', 'ff', 'fff'))

What I want to do is this: for every string in every row in the text_df I want to see whether there is a corresponding string in any of the sublists in lol , if match is TRUE , I want to append this sublist to the row in the text_df .

So that the end result of this operation is this:

>text_df_new

line          text
   1      a aa aaa
   2      b bb bbb
   3 c cc ccc cccc
   4      f ff fff

I can't really understand how to go about it. The pseudocode, I imagine, would look something like this:

for text in texts:
    for l in lol:
        if strsplit(text[text]) %in% lol[l]:
            text <- c(text, lol[l])

Or maybe there is a way to vectorize this?

Answer 1

I think this accomplishes what you want given your data above:

check_text <- function(df, list){
  tdf <- df
  for(i in 1:length(df$text)){
    x <- unlist(strsplit(df$text[i], split = " "))
    for(j in x){
      for(k in lol){
        for(l in k){
          if(j == l){
            tdf$text[i] <- paste(k, collapse =  " ")
          }
        }
      }
    }
  }
  return(tdf)
}


text_df_new <- check_text(text_df, lol)

> text_df_new

  line          text
1    1      a aa aaa
2    2      b bb bbb
3    3 c cc ccc cccc
4    4      f ff fff

I know that this is not a very "R" approach, so I'm guessing that a real R user will have an idea of how to do the same thing in about 2 lines using apply or one of the other functions that I still haven't really wrapped my head around yet. However, if your data set is small this is probably ok.