How can I tell the resolution of scanned PDF from within a shell script?

Question

I have a large collection of documents scanned into PDF format, and I wish to write a shell script that will convert each document to DjVu format. Some documents were scanned a

Answer 1

Here are the elements to this answer:

• pdfimages will extract images so that the number of dots can be discovered.
• identify will give the size of the image in units of PostScript points (72 to the inch)
• Because some scanners may split a single page into multiple images of varying sizes and shapes, the key is to add up the areas of all the images. Dividing square dots by square inches and taking the square root produces the answer.

Below is a Lua script that solves the problem. I probably could have used a plain shell, but capturing the width and height would have been a greater nuisance.

#!/usr/bin/env lua

require 'osutil'
require 'posixutil'
require 'mathutil'

local function runf(...) return os.execute(string.format(...)) end

assert(arg[1], "no file on command line")

local function dimens(filename)
  local cmd = [[identify -format "return %w, %h\n" $file | sed 1q]]
  cmd = cmd:gsub('$file', os.quote(filename))
  local w, h = assert(loadstring(os.capture(cmd)))()
  assert(w and h)
  return w, h
end

assert(#arg == 1, "dpi of just one file")

for _, pdf in ipairs(arg) do
  local w, h = dimens(pdf)  -- units are points
  local insquared = w * h / (72.00 * 72.00)
  local imagedir = os.capture 'mktemp -d'
  assert(posix.isdir(imagedir))
  runf('pdfimages -f 1 -l 1 %s %s 1>&2', os.quote(pdf),
                                         os.quote(imagedir .. '/img'))
  local dotsquared = 0
  for file in posix.glob(imagedir .. '/img*') do
    local w, h = dimens(file)  -- units are pixels
    dotsquared = dotsquared + w * h
  end
  os.execute('rm -rf ' .. os.quote(imagedir))
  local dpi = math.sqrt(dotsquared / insquared)

  if true then
    io.stderr:write(insquared, " square inches\n")
    io.stderr:write(dotsquared, " square dots\n")
    io.stderr:write(dpi, " exact dpi\n")
    io.stderr:write(math.round(dpi, 10), " rounded dpi\n")
  end
  print(math.round(dpi, 10))
end