A script to summarize PostgreSQL documentation changes in SGML files between two PostgreSQL versions

doc_sgml_cmp.pl

#!/usr/bin/perl

# -------------------------------------------------------------
# doc_sgml_cmp.pl
#
# A script to summarize PostgreSQL documentation changes
# in SGML files between two PostgreSQL versions.
#
# Written by Satoshi Nagayasu <[email protected]>
# -------------------------------------------------------------

use strict vars;

my $TARGET1=shift;
my $TARGET2=shift;

sub usage {
    print "Usage: $0 [path1] [path2]\n";
    print "\n";
    print "Both path1 and path2 must point a SGML source directory.\n";
    print "\n";
}

# those directories exist?
if ( ! -f "$TARGET1/postgres.sgml" || ! -f "$TARGET2/postgres.sgml" )
{
    &usage;
    exit;
}

open(FILE, "find $TARGET2 -type f -name '*.sgml' | sort |" ) || die($!);

print "file name,status,total lines,modified lines\n";

while(<FILE>)
{
    chomp;

    my $file = $_;
    $file =~ s,^./,,;

    my $basename = $file;
    $basename =~ s,.*sgml/,,;

    my $lc = `wc -l $_ | sed 's/ .*//'`;
    chomp($lc);

    if ( ! -f "$TARGET1/$file" )
    {
	print "$basename,added,$lc,\n";
    }
    else
    {
	my $mode = 0;
	my $change = 0;

	open(DIFF, "diff -rc $TARGET1/$file $file |") || die($!);
	while(<DIFF>)
	{
	    if ( /^\*\*\* \d+,\d+ \*\*\*\*/ )
	    {
		$mode = 0;
	    }
	    if ( /^--- \d+,\d+ ----/ )
	    {
		$mode = 1;
	    }

	    if ( $mode == 1 && /^\!/ )
	    {
		$change++;
	    }
	    if ( $mode == 1 && /^\+/ )
	    {
		$change++;
	    }
#	    print;
	}
	close(DIFF);

	if ( $change == 0 )
	{
	    print "$basename,not changed,$lc,$change\n";
	}
	else
	{
	    print "$basename,changed,$lc,$change\n";
	}
    }
}

close(FILE);