Extract figures from the PDF of a published article

Extracts the raster images embedded in a PDF, e.g., the article being checked, so that they can be compared with the reproduced figures using compare_figures. With method = "pdfimages", the images are extracted at their original resolution with the pdfimages tool from poppler-utils, JPEG images as JPEG and all others as PNG, and images smaller than min_size pixels in width or height (e.g., logos, icons) are skipped. With method = "pages", whole pages are rendered with pdftools::pdf_convert() instead, e.g., for vector figures that are not embedded as images; crop them with magick::image_crop() if needed.

Usage

extract_pdf_figures(
  pdf,
  dest_dir = "published",
  pages = NULL,
  min_size = 300,
  method = c("auto", "pdfimages", "pages"),
  dpi = 150,
  prefix = "figure"
)

Arguments

pdf

Path to the PDF file.

dest_dir

Directory to write the images to; created if missing.

pages

Page numbers to extract from; NULL for all pages.

min_size

Minimum width and height in pixels of an embedded image to be extracted (only for method = "pdfimages").

method

"auto" (default) uses "pdfimages" if the pdfimages tool is available and "pages" otherwise.

dpi

Resolution for rendering pages (only for method = "pages").

prefix

File name prefix of the extracted images.

Value

A data frame with one row per extracted image and the columns page, width, height, and file, invisibly.

Details

The order of embedded images in a PDF does not always follow the page layout, so check the extracted images against the figure captions.

Examples

if (FALSE) { # \dontrun{
figures <- extract_pdf_figures("article.pdf", "published", pages = 6:12)
} # }