pdfdir-verify

#!/bin/bash
#
# pdfdir-verify   http://github.com/bronson/pdfdir
# Scott Bronson
# 26 Mar 2009
#
# This script tries to determine if your pdf files are good.
# It prints warnings if it suspects any of them are flawed or corrupt,
#   -q --quick: only use the quick-n-dirty alorithm.  It's much faster
#       but some of your PDFs will probably appear to have 0 pages.
#   -v --verbose: print the result for each file.  Normally this utility
#       only produces output when its tests fail.
#
# Thanks to fpmurphy's fast PDF page counting snippet:
# http://www.unix.com/shell-programming-scripting/55661-how-get-number-pages-pdf-file.html


quick=false
verbose=false

while (( "$#" )); do
	[ "x$1" = "x-v" ] || [ "x$1" = "x--verbose" ] && verbose=true
	[ "x$1" = "x-q" ] || [ "x$1" = "x--quick" ] && quick=true
	shift
done


root="$1"
oldifs="$IFS"
if [ "x$root" = "x" ]; then root="."; fi


# Given the name of a pdf file, outputs the number of pages in the file.
# This is just a quick-n-dirty algorithm; it's known to fail.
count_pdf_pages() {
	page_count=0

	while read num; do
		# Turns "BLA BLA/Count 21314BLA BLA" into 21314
		num="${num#*/Count }"
		num="${num%%[!0-9]*}"
		(($num > $page_count)) && page_count=$num
	done < <(strings "$1" | grep "/Count")
	echo -n "$page_count"
}


filepat=': PDF document, version [0-9.]*$'
numpat='[0-9][0-9]*'

find "$root" -name "*.pdf" -print | sort | while read path; do
	[ -z "$path" ] && exit

	type=$(file "$path")
	# This RE is probably way too restrictive. I'll accept patches.
	if [[ ! "$type" =~ $filepat ]]; then
		echo "ERROR: $type <-- bad or unrecognized file type!"
		continue
	fi

	quick_count=$(count_pdf_pages "$path")
	if [[ ! "$quick_count" =~ $numpat ]]; then
		echo "ERROR: $path: bad quick count '$quick_count'"
		continue
	fi

	if ! $quick; then
		slow_count=$(pdf2ps -sOutputFile=- "$path" | grep -c "%%Page: ")
		if [[ ! "$slow_count" =~ $numpat ]]; then
			echo "ERROR: $path: bad slow count '$slow_count'"
			continue
		fi

		if [ "$quick_count" -lt 1 ]; then
			# the quick count algorithm sometimes fails
			# we just have to assume the slow count is correct
			echo "warning: $path: quick count failed ($quick_count).  No problem."
			quick_count="$slow_count"
		fi

		if [ "$quick_count" -ne "$slow_count" ]; then
			echo "ERROR: $path: quick count ($quick_count) and slow count ($slow_count) differ!"
			continue
		fi
	fi

	$verbose && echo "$path: $quick_count"
done