swish-e/prog-bin/DirTree.pl

#!/usr/local/bin/perl -w
use strict;

# This is a short example that basically does the same
# thing as the default file system access method by
# recursing directories, but also shows how to process different
# file types -- in this example pdf is converted to xml for indexing.

# for the odd chance of running under Windows
# Now extprog.c expects file in text mode so no need to binmode.
# binmode STDOUT;  

use File::Find;  # for recursing a directory tree
use pdf2xml;     # example module for pdf to xml conversion
                 # Not that you need IndexContents XML .pdf in the
                 # swish-e config file

# See perldoc File::Find for information on following symbolic links
# and other important topics.

use constant DEBUG => 0;


find(
    {
        wanted => \&wanted,
        # no_chdir => 1,  # 5.6 feature
    },
    @ARGV,
);

sub wanted {
    return if -d;

    if ( !-r _ ) {
        warn "$File::Find::name is not readable\n";
        return;
    }

    if ( /\.pdf$/ ) {
        print STDERR "Indexing pdf $File::Find::name\n" if DEBUG;
        print ${ pdf2xml( $File::Find::name ) };

    } elsif ( /\.(txt|log|pl|html|htm)$/ ) {
        print STDERR "Indexing $File::Find::name\n" if DEBUG;
        print ${ get_content( $_ ) };

    } else {
        print STDERR "Skipping $File::Find::name\n" if DEBUG;
    }
}


sub get_content {
    my $path = shift;

    my ( $size, $mtime )  = (stat $path )[7,9];
    open FH, $path or die "Failed to open $path: $!";

    my $content =  <<EOF;
Content-Length: $size
Last-Mtime: $mtime
Path-Name: $File::Find::name

EOF
    local $/ = undef;
    $content .= <FH>;
    return \$content;
}

1	adcroft	1.1	#!/usr/local/bin/perl -w
2			use strict;
3
4			# This is a short example that basically does the same
5			# thing as the default file system access method by
6			# recursing directories, but also shows how to process different
7			# file types -- in this example pdf is converted to xml for indexing.
8
9			# for the odd chance of running under Windows
10			# Now extprog.c expects file in text mode so no need to binmode.
11			# binmode STDOUT;
12
13			use File::Find; # for recursing a directory tree
14			use pdf2xml; # example module for pdf to xml conversion
15			# Not that you need IndexContents XML .pdf in the
16			# swish-e config file
17
18			# See perldoc File::Find for information on following symbolic links
19			# and other important topics.
20
21			use constant DEBUG => 0;
22
23
24			find(
25			{
26			wanted => \&wanted,
27			# no_chdir => 1, # 5.6 feature
28			},
29			@ARGV,
30			);
31
32			sub wanted {
33			return if -d;
34
35			if ( !-r _ ) {
36			warn "$File::Find::name is not readable\n";
37			return;
38			}
39
40			if ( /\.pdf$/ ) {
41			print STDERR "Indexing pdf $File::Find::name\n" if DEBUG;
42			print ${ pdf2xml( $File::Find::name ) };
43
44			} elsif ( /\.(txt\|log\|pl\|html\|htm)$/ ) {
45			print STDERR "Indexing $File::Find::name\n" if DEBUG;
46			print ${ get_content( $_ ) };
47
48			} else {
49			print STDERR "Skipping $File::Find::name\n" if DEBUG;
50			}
51			}
52
53
54			sub get_content {
55			my $path = shift;
56
57			my ( $size, $mtime ) = (stat $path )[7,9];
58			open FH, $path or die "Failed to open $path: $!";
59
60			my $content = <<EOF;
61			Content-Length: $size
62			Last-Mtime: $mtime
63			Path-Name: $File::Find::name
64
65			EOF
66			local $/ = undef;
67			$content .= <FH>;
68			return \$content;
69			}
70