swish-e/conf/example9.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.

# in this example, only .pdf and .config files are indexed.

# the pdf2xml module is in the prog-bin directory of the swish-e distribution
use lib '../prog-bin';

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

use constant DEBUG => 0;

# See if a directory was passed in via the SwishProgParameters swish
# directive

my $dir = shift || '.';

find(
    {
        wanted => \&wanted,
        no_chdir => 1,
    },
    $dir,
);

sub wanted {
    return if -d;

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

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

    } 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, $path or die "$path: $!";

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

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			# in this example, only .pdf and .config files are indexed.
10
11			# the pdf2xml module is in the prog-bin directory of the swish-e distribution
12			use lib '../prog-bin';
13
14			use File::Find; # for recursing a directory tree
15			use pdf2xml; # example module for pdf to xml conversion
16			# Not that you need IndexContents XML .pdf in the
17			# swish-e config file
18
19			# See perldoc File::Find for information on following symbolic links
20
21			use constant DEBUG => 0;
22
23			# See if a directory was passed in via the SwishProgParameters swish
24			# directive
25
26			my $dir = shift \|\| '.';
27
28			find(
29			{
30			wanted => \&wanted,
31			no_chdir => 1,
32			},
33			$dir,
34			);
35
36			sub wanted {
37			return if -d;
38
39			if ( /\.pdf$/ ) {
40			print STDERR "Indexing pdf $File::Find::name\n" if DEBUG;
41			print ${ pdf2xml( $File::Find::name ) };
42
43			} elsif ( /\.config$/ ) {
44			print STDERR "Indexing $File::Find::name\n" if DEBUG;
45			print ${ get_content( $File::Find::name ) };
46
47			} else {
48			print STDERR "Skipping $File::Find::name\n" if DEBUG;
49			}
50			}
51
52
53			sub get_content {
54			my $path = shift;
55
56			my ( $size, $mtime ) = (stat $path )[7,9];
57			open FH, $path or die "$path: $!";
58
59			my $content = <<EOF;
60			Content-Length: $size
61			Last-Mtime: $mtime
62			Path-Name: $path
63
64			EOF
65			local $/ = undef;
66			$content .= <FH>;
67			return \$content;
68			}