R/html2pdf.R

In gorkang/html2latex: Converting sjPlot Models To LaTeX

#' html2pdf
#' @description Convert [sjPlot::tab_model()] html tables to tex and pdf
#' @name html2pdf
#' @param filename Name of the sjPlot table object
#' @param build_pdf TRUE/FALSE
#' @param clean TRUE/FALSE clean intermediate files
#' @param silent TRUE/FALSE
#' @param path_w2l Path to the Writer2latex files
#' @param OS String identifying operating system (Linux, macOS or Windows)
#' @param table_width Width of the table
#' @param page_width In inches
#' @param page_height In inches
#' @param style makes table somewhat prettier
#' @return TeX file
#' @export
#' @importFrom tinytex latexmk
#'
#' @examples
#' \dontrun{
#' model = lmer(mpg ~ cyl * disp + (1|vs), mtcars)
#'
#' # We save the sjPlot table to an .html file
#' sjPlot::tab_model(
#'   model,
#'   show.r2 = TRUE,
#'   show.icc = FALSE,
#'   show.re.var = FALSE,
#'   p.style = "scientific",
#'   emph.p = TRUE,
#'   file = "temp.html")
#'
#' # Create tex and pdf
#' html2pdf(filename = "temp.html", table_width = 13, silent = TRUE,
#'           style = TRUE, build_pdf = TRUE, clean = TRUE)
#' }

html2pdf <-
  function(filename,
           table_width = 30,
           page_width = 8.5,
           page_height = 11,
           build_pdf = FALSE,
           clean = FALSE,
           style = TRUE,
           silent = TRUE,
           OS = NULL,
           path_w2l = NULL,
           name_table = "table1") {

    # CHECK
    if (grepl(" ", filename)) stop("filename contains spaces. Things get messy when mixing latex and spaces.\nRename the file to something like '", gsub(" ", "", filename), "' and try again.")

    if (is.null(path_w2l)) {
      path_w2l <- system.file("extdata", package = "html2latex")
    }

    # Find w2l files
    w2l_file <- list.files(path = path_w2l, pattern = "w2l", recursive = TRUE)[1]

    if (length(w2l_file) == 0) {
      stop(
        "Writer2latex not found in folder: '", path_w2l, "'\n",
        "You can use the parameter 'path_w2l' to tell me where Writer2latex is, or: \n\n",
        " 1. Download: https://sourceforge.net/projects/writer2latex/files/writer2latex/Writer2LaTeX%201.6/ \n",
        " 2. Extract 'w2l' and 'writer2latex.jar' to a folder in the project where this function is called \n",
        " 3. chmod +x w2l (see http://writer2latex.sourceforge.net/doc1.6/user-manual10.html)"
      )
    }

    # Operative system detection
    if (is.null(OS)) OS = Sys.info()["sysname"]

    if (OS == "Linux") {

      # Necessary to run soffice command in Ubuntu 20.04
      Sys.setenv(LD_LIBRARY_PATH = "/usr/lib/libreoffice/program/")
      soffice <- "soffice"

    } else if (OS == "Windows") {

      soffice <- "C:\\Program Files\\LibreOffice\\program\\soffice.bin"
      if (!file.exists(soffice)) stop("Did not locate Libreoffice in C:\\Program Files\\LibreOffice\\program\\soffice.bin")
      if (!grepl("\\.bat", w2l_file)) w2l_file = paste0(w2l_file, ".bat")
      if (build_pdf == TRUE) message("Cannot automatically build a pdf on Windows. I set 'build_pdf = FALSE'.\n * You can build a pdf using the tex file with pdflatex.exe. See https://tex.stackexchange.com/questions/49569/where-to-download-pdflatex-exe")
      build_pdf = FALSE

    } else if (OS == "Darwin" | OS == "macOS") {

      soffice <- "/Applications/LibreOffice.app/Contents/MacOS/soffice"

    } else {
      stop("Not sure about your operative system. Use the parameter 'OS' with either 'Linux', 'macOS' or 'Windows'")
    }



    # Location of files
    odt_file <- gsub("html$", "odt", paste0(getwd(), "/", basename(filename))) # Location of generated odt file
    tex_file <- gsub("html$", "tex", paste0(getwd(), "/", basename(filename))) # Location of generated tex file
    pdf_file <- gsub("html$", "pdf", paste0(getwd(), "/", basename(filename))) # Location of generated pdf file

    # Output files
    tex_file_output <- gsub("html$", "tex", paste0(getwd(), "/", filename))
    pdf_file_output <- gsub("html$", "pdf", paste0(getwd(), "/", filename))

    if (silent) {
      invisible(system2(soffice, paste0("--convert-to odt ", filename))) # HTML to ODT
      invisible(system2(paste0(file.path(path_w2l, w2l_file)), odt_file))  # ODT to TEX
    } else {
      cli::cli_h1("html -> odt")
      system2(soffice, paste0("--convert-to odt ", filename))  # HTML to ODT
      cli::cli_alert_success("odt created")

      cli::cli_h1("odt -> tex")
      system2(paste0(file.path(path_w2l, w2l_file)), odt_file) # ODT to TEX
      cli::cli_alert_success("tex created")
    }

    if (style) {
      if (!silent) cli::cli_h1("Clean tex")
      clean_tex(tex_file_output, table_width = table_width,
                page_width = page_width, page_height = page_height, name_table = name_table)
      if (!silent) cli::cli_alert_success("tex cleaned up")

    }
    if (build_pdf) {
      if (!silent) cli::cli_h1("tex -> pdf")
      # build_PDF(file = tex_file_output, silent = silent, filename = filename,
      #           pdf_file = pdf_file, pdf_file_output = pdf_file_output)

      tinytex::latexmk(file = tex_file_output, pdf_file = pdf_file_output)
      if (!silent) cli::cli_alert_success("pdf built in {pdf_file_output}")

    }
    if (clean) {
      if (!silent) cli::cli_h1("Cleaning temp files")
      clean_debris(filename)
      if (!silent) cli::cli_alert_success("All cleaned")

    }
    if (filename != basename(filename)) file.copy(tex_file, tex_file_output, overwrite = TRUE)
    if (filename != basename(filename)) file.remove(gsub("html", "tex", basename(filename)))
    # message("\ntex file created in: ", tex_file_output)
  }


#' Minimal improvements to format
#' @name clean_tex
#' @param file TeX file
#' @param table_width With of the table
#' @param page_width In inches
#' @param page_height In inches
#' @keywords internal
#' @importFrom stringr str_count
#' @return TeX file
clean_tex <- function(file, table_width, page_width, page_height, name_table = "table1") {
  # Get rid of linejumps
  raw_file <- readLines(file)
  clean_file <- gsub("\\\\newline", "", raw_file)
  # Replace supertabular by longtable (so we can control the width of the table and the columns will adjust automagically)
  supertabular_line_n1 <- grep("begin\\{supertabular\\}", clean_file)
  supertabular_line_n2 <- grep("\\usepackage\\{supertabular\\}", clean_file)
  document_class_line <- grep("documentclass", clean_file)
  supertabular_line <- clean_file[supertabular_line_n1]

  number_columns <- stringr::str_count(supertabular_line, "m\\{") - 1
  longtable_line <- paste0("\\begin{longtable}{p{", table_width, "em}l", paste(rep("c", number_columns), collapse = ""), "}") # Table width
  clean_file[supertabular_line_n1] <- longtable_line
  clean_file <- append(clean_file, "\\usepackage{longtable}", after = supertabular_line_n2)
  clean_file <- append(clean_file, paste0("\\usepackage[paperheight=", page_height, "in,paperwidth=", page_width, "in]{geometry}"), after = document_class_line) # Page width

  clean_file <- gsub("\\\\end\\{supertabular\\}", paste0("\\\\label\\{", name_table, "\\}\n\\\\end\\{longtable\\}"), clean_file)
  # clean_file <- gsub("\\\\end\\{supertabular\\}", paste0("\\\\end\\{longtable\\}"), clean_file)

  # Write re-formatted version
  writeLines(clean_file, file)
}



#' Clean up latex debris
#' @name clean_debris
#' @param filename file path to clean up
#' @keywords internal
#' @return pdf file
clean_debris <- function(filename) {
  # file.remove(gsub("html", "out", basename(filename)))
  # file.remove(gsub("html", "aux", basename(filename)))
  # file.remove(gsub("html", "log", basename(filename)))
  if (filename != basename(filename)) file.remove(gsub("html", "pdf", basename(filename)))
  # Clean up odt
  file.remove(gsub("html", "odt", basename(filename)))
}