Insert Text from One Markdown Document into Another

While assembling a reply letter for a revise-and-resubmit document, I was documenting for reviewers how we addressed their concerns. The reply letter quoted numerous sentences and paragraphs that had been changed. Unfortunately, each time my co-authors and I edited the paper, I lost track of which sentences and paragraphs I had previously copied into the reply letter. I did not want to misquote my own paper.

Copying-and-pasting with each edit was getting tedious, and it was an error-prone process. I decided to write a function that would retrieve a named div or span from the paper and print it as a quote in my reply letter. Here it is:

get_quote <- function(id, file, blockquote = TRUE) {
  blocktext <- ifelse(blockquote, "\n> ", "")

  refreplace <- function(x) {
    # List of sequence types
    crossrefs <- c(`@fig-` = "Figure",
                   `@fig-` = "Table",
                   `@eq-` = "Equation",
                   `@thm-` = "Lemma",
                   `@lem-` = "Corollary",
                   `@cor-` = "Proposition",
                   `@prp-` = "Conjecture",
                   `@cnj-` = "Definition",
                   `@def-` = "Example",
                   `@exm-` = "Example",
                   `@exr-` = "Exercise",
                   `apafg-` = "Figure",
                   `apatb-` = "Figure"
    )

    # Find all crossreference types and number them
    make_replacements <- function(x, reftype, prefix) {
      regstring <- ifelse(stringr::str_starts(reftype, "\\@"),
                          paste0("\\", reftype, "(.*?)(?=[\\.\\?\\!\\]\\}\\s,])"),
                          paste0("\\{", reftype, "(.*?)\\}"))
      patterns <- stringr::str_extract_all(string = x, pattern = regstring) |>
        unlist() |>
        unique() |>
        stringr::str_replace("\\{", "\\\\{") |>
        stringr::str_replace("\\}", "\\\\}")

      if (all(is.na(patterns))) return(NULL)
      replacements <- paste(prefix, seq_along(patterns))
      names(replacements) <- patterns
      replacements
    }
    allreplacements <- purrr::map2(
      names(crossrefs),
      crossrefs,
      \(rt, pf) make_replacements(
        x = x,
        reftype = rt,
        prefix = pf)) |>
      unlist()

    stringr::str_replace_all(x, allreplacements)

  }

  filetext <- readLines(file) |>
    refreplace()

  idcount <- sum(stringr::str_count(filetext, paste0("#", id)))
  if (idcount > 1)
    stop(paste0(
      "The id (",
      id ,
      ") is not unique. There are ",
      idcount,
      " instances of id = ",
      id,
      "."
    ))

  s <- filetext |>
    paste0(collapse = "\n") |>
    stringr::str_match(pattern = paste0("(?<=\\[).+(?=\\]\\{\\#",
                               id,
                               "\\})"))  |>
    getElement(1)

  if (is.na(s)) {
    s <- filetext |>
      paste0(collapse = "|||") |>
      stringr::str_match(pattern = paste0(":::\\{\\#",
                                 id,
                                 "\\}(.*?):::"))  |>
      getElement(2) |>
      stringr::str_replace_all("\\|\\|\\|", blocktext)
  } else {
    s <- paste0(blocktext, s)
  }

  if (is.na(s)) stop("Could not find a div or span with id = ", id)

  s
}

For your convenience and mine, I have added this function into the WJSmisc package.

Using the get_quote function

If the file has a span or div with a specified id, the get_quote function can find it. Suppose the file has a span with id of id1 and a div with id of id2.

[Text in a span]{#id1}

:::{#id2}
Here are two pagagraphs in a fenced div.

This is the second paragraph.
:::

cat(get_quote("id2", "index.qmd"))

You can pull text from the current file or another file. This file happened to be named index.qmd. The text inside the span with id id1 can be extracted with inline chunks like so:

`r get_quote("id1", "index.qmd")`

Text in a span

Here we extract the two paragraphs in div id2

`r get_quote("id2", "index.qmd")`

Here are two pagagraphs in a fenced div.

This is the second paragraph.

Remove blockquote formatting

By default, the function adds a > before each line in the quoted markdown so that it appears as a block quote. You can turn off the block quote formatting like so:

`r get_quote("id1", "index.qmd", blockquote = FALSE)`

Text in a span

Curly braces with additional information

What if you need to put additional information in the curly braces of the span or div (e.g., a css class)?

[Here is more text with extra stuff in the curly braces.]{#id3 .myclass}

You can trick the function into thinking that the extra stuff is part of the id like so:

`r get_quote("id3 .myclass", "index.qmd")`

Here is more text with extra stuff in the curly braces.