ocr-tools/tools/sortpages.pl

#!/usr/bin/env perl
#
# $Id: sortpages,v 1.8 1997/12/11 19:20:58 mhw Exp $
#

@fileNameFromNumber = ();
@pagesFound = ();
$theProductNumber = 0;

for $fileIndex (0..$#ARGV)
{
	$fileName = $ARGV[$fileIndex];
	open(FILE, "<$fileName") || die;
	while (!eof(FILE))
  	{
  		$filePos = tell(FILE);
  		$_ = <FILE>;
 		if (/^\f?-\S/)
  		{
  			my ($versionHex, $flagsHex, $pageCRCHex, $tabWidthHex,
				$productNumberHex, $fileNumberHex, $pageNumber, $name)
					  = (/^\f?-\S\S{4}\ 		# CRC followed by a space
						 ([0-9a-f])				# Format version
						 ([0-9a-f]{2})			# Flags
						 ([0-9a-f]{8})			# Running CRC32
						 ([0-9a-f])				# Tab width (0 means radix64)
						 ([0-9a-f]{3})			# Product number
						 ([0-9a-f]{4})			# File number
						 \ Page\ (\d+)\ of\ (.*)/x);
			my $version = hex($versionHex);
			my $flags = hex($flagsHex);
			my $productNumber = hex($productNumberHex);
			my $fileNumber = hex($fileNumberHex);

			unless ($version == 0 && $productNumber > 0
						&& $fileNumber > 0 && $pageNumber > 0
						&& $name ne "")
			{
				print STDERR "ERROR: Invalid header info ",
							 "at $fileName line $.\n";
				exit(1);
			}

			if (!defined($fileNameFromNumber[$fileNumber]))
			{
				$fileNameFromNumber[$fileNumber] = $name;
			}
			elsif ($fileNameFromNumber[$fileNumber] ne $name)
			{
				print STDERR "ERROR: Mismatched filename ",
							 "at $fileName line $.\n";
				exit(1);
			}

			if (!$theProductNumber)
			{
				$theProductNumber = $productNumber;
			}
			elsif ($theProductNumber != $productNumber)
			{
				print STDERR "ERROR: Different product number ",
							 "at $fileName line $.\n";
				exit(1);
			}

			push @pagesFound, (sprintf "%5d:%4d:%d:%d:%d",
					 $fileNumber, $pageNumber, $flags, $fileIndex, $filePos);
		}
	}
	close(FILE) || die;
}

@pagesFound = sort @pagesFound;

$result = 0;
$lastFileNumber = 0;
$lastPageNumber = 0;
$nextFileNumber = 1;
$nextPageNumber = 1;
$fileIndexOpen = -1;
foreach (@pagesFound)
{
	my ($fileNumber, $pageNumber, $flags, $fileIndex, $filePos) = split /:/;

	$fileNumber = int($fileNumber);
	$pageNumber = int($pageNumber);

	if ($fileNumber == $lastFileNumber && $pageNumber == $lastPageNumber)
	{
		print STDERR "DUPLICATE: File $fileNumber, page $pageNumber, skipped\n";
		next;
	}

	if ($nextFileNumber < $fileNumber && $nextPageNumber != 1)
	{
		print STDERR "MISSING: File $nextFileNumber, ",
					 "pages $nextPageNumber - END\n";
		$nextPageNumber = 1;
		$nextFileNumber++;
		$result = 1;
	}
	if ($nextFileNumber < $fileNumber)
	{
		print STDERR "MISSING: Files $nextFileNumber - ",
					 $fileNumber-1, "\n";
		$nextFileNumber = $fileNumber;
		$nextPageNumber = 1;
		$result = 1;
	}
	if ($nextFileNumber != $fileNumber)
	{
		print STDERR "ERROR: Internal error, unexpected fileNumber\n";
		exit(1);
	}

	if ($nextPageNumber < $pageNumber)
	{
		print STDERR "MISSING: File $fileNumber, pages $nextPageNumber - ",
					 $pageNumber-1, "\n";
		$nextPageNumber = $pageNumber;
		$result = 1;
	}
	if ($nextPageNumber != $pageNumber)
	{
		print STDERR "ERROR: Internal error, unexpected pageNumber\n";
		exit(1);
	}

	if ($fileIndexOpen != $fileIndex)
	{
		if ($fileIndexOpen >= 0)
		{
			close(FILE) || die;
			$fileIndexOpen = -1;
		}
		$fileName = $ARGV[$fileIndex];
		open(FILE, "<$fileName") || die;
		$fileIndexOpen = $fileIndex;
	}
	seek(FILE, $filePos, 0) || die($!);

	$_ = <FILE>;
	print;
	while (<FILE>)
	{
		last if /^\f?-\S/;
		print;
	}
	$lastFileNumber = $fileNumber;
	$lastPageNumber = $pageNumber;

	if ($flags & 1)		# Bit 0 of flags indicates last page of file
	{
		$nextFileNumber++;
		$nextPageNumber = 1;
	}
	else
	{
		$nextPageNumber++;
	}
}

if ($nextPageNumber != 1)
{
	print STDERR "MISSING: File $nextFileNumber, ",
				 "pages $nextPageNumber - END\n";
	$nextPageNumber = 1;
	$nextFileNumber++;
	$result = 1;
}

print STDERR "Highest file number encountered: ", $nextFileNumber - 1, "\n";

if ($fileIndexOpen >= 0)
{
	close(FILE) || die;
	$fileIndexOpen = -1;
}

exit($result);

#
# vi: ai ts=4
# vim: si
#