add mutool as PDF image conversion backend

NAMESPACE

@@ -44,6 +44,7 @@ export(logit)
 export(lst2message)
 export(min_max)
 export(month)
+export(mutool_convert)
 export(n_samples)
 export(pandoc_html)
 export(pandoc_pdf)

R/image_format_conversion.R

@@ -2,24 +2,20 @@
 #'
 #' Convert a PDF/SVG file to images of other formats with Inkscape.
 #'
-#' To use this function, `inkscape` (v1.2.2) has to be available from the
-#' command line. Alternatively, users can specify the absolute path to inkscape
-#' by the option `stom.inkscape`
+#' To use this function, `mutool` (v1.24.0) or `inkscape` (v1.2.2) has to be
+#' available from the command line. Alternatively, users can specify the
+#' absolute path to inkscape by the option `stom.inkscape`
 #' (e.g., `options(stom.inkscape = "/home/stom/bin/inkscape")`).
 #'
 #' @param infile String. Path to input PDF file.
 #' @param dpi Integer. "Dots Per Inch" for PNG outputs with `pdf2png()`.
-#' @param outfile Character vector. Output file paths without file extensions.
-#'        File extensions will be set to `.png` for `*2png()`; `.svg` for
-#'        `pdf2svg()`; and `.pdf` for `svg2pdf()`. The length of `outfile`
-#'        should be `1` or corresponds to the length of `page`. If there is a
-#'        mismatch, only the first element in `outfile` will be taken as the
-#'        file stem, and the page numbers will be appended to the file stem as
-#'        suffixes. By default, `outfile` is set to `NULL`, in which the
+#' @param outfile String. By default, `outfile` is set to `NULL`, in which the
 #'        input PDF file stem is used.
 #' @param page Integer vector. The pages (starting with 1) to extract from the
 #'        PDF file, defaults to `NULL`, in which every page is converted to
 #'        an image.
+#' @param backend String. The tool for converting images from PDFs. Defaults to `"mutool"`.
+#'        Currently supported options: `"mutool"` and `"inkscape"`
 #'
 #' @examples
 #' \dontrun{
@@ -33,8 +29,116 @@
 pdf2png = function(infile,
                    outfile = NULL,
                    page = NULL,
-                   dpi = 1200,
-                   white = FALSE) {
+                   dpi = 400,
+                   white = FALSE,
+                   backend = c("mutool", "inkscape") ) {
+    if (backend[1] == "inkscape")
+        pdf2png_inkscape(infile, outfile, page, dpi, white)
+
+    if (backend[1] == "mutool") {
+        if (!is.null(outfile)) outfile = xfun::with_ext(outfile, "png")
+        mutool_convert(infile, outfile, format = "png", page = page, dpi = dpi, white = white)
+    }
+}
+
+
+#' @export
+#' @rdname pdf2png
+pdf2svg = function(infile,
+                   outfile = NULL,
+                   page = NULL,
+                   white = FALSE,
+                   backend = c("mutool", "inkscape")) {
+    if (backend[1] == "inkscape")
+        pdf2svg_inkscape(infile, outfile, page, white)
+
+    if (backend[1] == "mutool") {
+        if (!is.null(outfile)) outfile = xfun::with_ext(outfile, "svg")
+        mutool_convert(infile, outfile, format = "svg", page = page, white = white)
+    }
+
+}
+
+
+#' Convert PDF to images by mutool
+#'
+#' @param infile String. Path to input PDF file.
+#' @param outfile String Output file path. Defaults to NULL, and is inferred from `infile`.
+#' @param dpi Integer. "Dots Per Inch" for PNG outputs.
+#' @param format String. Output format. Ignored when `outfile` is given.
+#' @param page Integer vector. The pages (starting with 1) to extract from the
+#'        PDF file, defaults to `NULL`, in which every page is converted to
+#'        an image.
+#' @export
+mutool_convert = function(infile,
+                          outfile = NULL,
+                          format = c("png", "svg"),
+                          page = NULL,
+                          dpi = 300,
+                          white = FALSE) {
+    # Set output format
+    if (!is.null(outfile)) {
+        fmt = xfun::file_ext(outfile)[1]
+        if (!fmt %in% format) {
+            format = format[1]
+        } else {
+            format = fmt
+        }
+    } else {
+        format = format[1]
+    }
+
+    # Set output directory & file name
+    if (is.null(outfile)) {
+        outdir = dirname(infile)
+        outfn = basename(xfun::with_ext(infile, format))
+    } else {
+        outdir = dirname(outfile)
+        outfn = basename(xfun::with_ext(outfile, format))[1]
+    }
+
+    # Set page
+    if (is.null(page)) {
+        page = "1-N"
+    } else {
+        page = paste(page, collapse=",")
+    }
+
+    # Convert pages to temp dir
+    tmpdir = tempdir()
+    opts = c(
+        paste0("resolution=",dpi),
+        ifelse(white, "alpha", "")
+    )
+    args = c(
+        "convert", "-F", format, "-o", file.path(tmpdir,outfn),
+        "-O", paste(opts, collapse=","),
+        infile, page
+    )
+    # mutool convert -F png -o dir/.png -O resolution=300,alpha dag-dawid.pdf 1-2
+    cat("\n  mutools", args, "\n")
+    system2("mutool", args)
+
+    # Copy files to output directory
+    # Rename according to the number of files generated
+    fps = msg = list.files(tmpdir, full.names=T, pattern=paste0(format,"$"))
+    if (length(fps) == 1) {
+        outfp = msg = file.path(outdir, outfn)
+        file.copy(fps, outfp, overwrite=T)
+
+    } else {
+        file.copy(fps, outdir, overwrite=T)
+    }
+    file.remove(fps)
+    message("\nOutput file(s):\n  ", paste(basename(msg), collapse=", ") )
+}
+
+
+pdf2png_inkscape = function(infile,
+                       outfile = NULL,
+                       page = NULL,
+                       dpi = 1200,
+                       white = FALSE) {
     base_cmd = c(
         "--export-background-opacity=0"[!white],
         "--export-background=white"[white],
@@ -51,9 +155,7 @@ pdf2png = function(infile,
 }
 
 
-#' @export
-#' @rdname pdf2png
-pdf2svg = function(infile,
+pdf2svg_inkscape = function(infile,
                    outfile = NULL,
                    page = NULL,
                    white = FALSE) {