Annotation of /trunk/spider/swishspider

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

use LWP::UserAgent;
use HTTP::Status;
use HTML::LinkExtor;

if (scalar(@ARGV) != 2) {
    print STDERR "Usage: SwishSpider localpath url\n";
    exit(1);
}

my $ua = new LWP::UserAgent;
$ua->agent( "SwishSpider http://swish-e.org" );

my $localpath = shift;
my $url = shift;

my $no_parent_url;
if ($url =~ m/\s/) {
        ($no_parent_url,$url) = split(/\s/,$url,2);
        # old scheme had URL, no parent and new is reverse
        ($url,$no_parent_url) = ($no_parent_url,$url) if ($no_parent_url =~ m/$url/);
}

my $request = new HTTP::Request( "GET", $url );
my $response = $ua->simple_request( $request );

#
# Write out important meta-data.  This includes the HTTP code.  Depending on the
# code, we write out other data.  Redirects have the location printed, everything
# else gets the content-type.
#
open( RESP, ">$localpath.response" ) || die( "Could not open response file $localpath.response" );

print RESP $response->code() . "\n";
if( $response->code() == RC_OK ) {
    print RESP $response->header( "content-type" ) . "\n";
} elsif( $response->is_redirect() ) {
        my $link = $response->header( "location" );
        if ($no_parent_url) {
                if ($link =~ m/$no_parent_url/) {
                        # if this URL is below parent URL o.k....
                        print RESP "$no_parent_url $link\n";
                } else {
                        # if not, crawl just this page!
                        print RESP "$link $link\n";
                }
        } else {
                print RESP "$link\n";
        }
}
close( RESP );

#
# Write out the actual data assuming the retrieval was succesful.  Also, if
# we have actual data and it's of type text/html, write out all the links it
# refers to
#
if( $response->code() == RC_OK ) {
    my $contents = $response->content();

    open( CONTENTS, ">$localpath.contents" ) || die( "Could not open contents file $localpath.contents\n" );
    # fixup just HTML files
    if ($response->header("content-type") =~ "text/html") {
        # if you don't want content to be indexed, include it in
        # <noindex> foobar </noindex> tags or surround it with comments
        # <!-- noindex --> foobar <!-- /noindex -->
        $contents =~ s,<noindex>.+?</noindex>,,isg;
        $contents =~ s,<!--\s*noindex\s*-->.+?<!--\s*/noindex\s*-->,,isg;
        # this will remove all script from indexing content
        $contents =~ s,<script>.+?</script>,,isg;
        # remap Windows charset to ISO-8859-2
        $contents =~ tr/ðèæÐÈÆ/¹ð¾èæ©Ð®ÈÆ/; # 1250 -> iso8859-2
        # this will fix badly formatted html in form:
        # <head><title>some text</title
        # ></head>
        # which will confuse indexer (or libxml2?)
        $contents =~ s/[\n\r]^(>)/$1\n/msg;
        # remove comments between <html> and <head> texi2html inserts them
        # there and swish can't find document title then (libxml or swish bug?)
        while ($contents =~ s/(<html>.*)<!--.*?-->(.*<head>)/$1$2/msi) { };
    }
    print CONTENTS $contents;
    close( CONTENTS );

    if( $response->header("content-type") =~ "text/html" ) {
        open( LINKS, ">$localpath.links" ) || die( "Could not open links file $localpath.links\n" );
        my $p = HTML::LinkExtor->new( \&linkcb, $url );
        $p->parse( $contents );

        close( LINKS );
    }
}


sub linkcb {
    my($tag, %links) = @_;
    if (($tag eq "a" || $tag eq "area") && ($links{"href"}) || ($tag eq "frame" && $links{"src"})) {
        my $link = $links{"href"} || $links{"src"};

        #
        # Remove fragments
        #
        $link =~ s/(.*)#.*/$1/;

        #
        # Remove ../  This is important because the abs() function
        # can leave these in and cause never ending loops.
        #
        $link =~ s/\.\.\///g;

        # hack for apostrophe -- changes URL, but should work for most clients.
        $link =~ s/'/%27/g;

        # hack for Apache directory listings
        $link =~ s,/\?[NMSD]=[AD]$,/,g;

        # speedup, skip pictures
        return if ($link =~ m/\.(gif|jpg|png)/);

        if ($no_parent_url) {
                        if ($link =~ m/$no_parent_url/) {
                                print LINKS "$no_parent_url $link\n";
#                               print STDERR "using $link\n";
#                       } else {
#                               print STDERR "skipping $link\n";
                        }
        } else {
                print LINKS "$link\n";
        }
    }
}

1	dpavlin	1	#!/usr/local/bin/perl -w
2			use strict;
3
4			use LWP::UserAgent;
5			use HTTP::Status;
6			use HTML::LinkExtor;
7
8			if (scalar(@ARGV) != 2) {
9			print STDERR "Usage: SwishSpider localpath url\n";
10			exit(1);
11			}
12
13			my $ua = new LWP::UserAgent;
14			$ua->agent( "SwishSpider http://swish-e.org" );
15
16			my $localpath = shift;
17			my $url = shift;
18
19			my $no_parent_url;
20			if ($url =~ m/\s/) {
21	dpavlin	40	($no_parent_url,$url) = split(/\s/,$url,2);
22			# old scheme had URL, no parent and new is reverse
23			($url,$no_parent_url) = ($no_parent_url,$url) if ($no_parent_url =~ m/$url/);
24	dpavlin	1	}
25
26			my $request = new HTTP::Request( "GET", $url );
27			my $response = $ua->simple_request( $request );
28
29			#
30			# Write out important meta-data. This includes the HTTP code. Depending on the
31			# code, we write out other data. Redirects have the location printed, everything
32			# else gets the content-type.
33			#
34			open( RESP, ">$localpath.response" ) \|\| die( "Could not open response file $localpath.response" );
35
36			print RESP $response->code() . "\n";
37			if( $response->code() == RC_OK ) {
38			print RESP $response->header( "content-type" ) . "\n";
39			} elsif( $response->is_redirect() ) {
40			my $link = $response->header( "location" );
41			if ($no_parent_url) {
42			if ($link =~ m/$no_parent_url/) {
43			# if this URL is below parent URL o.k....
44	dpavlin	40	print RESP "$no_parent_url $link\n";
45	dpavlin	1	} else {
46			# if not, crawl just this page!
47			print RESP "$link $link\n";
48			}
49			} else {
50			print RESP "$link\n";
51			}
52			}
53			close( RESP );
54
55			#
56			# Write out the actual data assuming the retrieval was succesful. Also, if
57			# we have actual data and it's of type text/html, write out all the links it
58			# refers to
59			#
60			if( $response->code() == RC_OK ) {
61			my $contents = $response->content();
62
63			open( CONTENTS, ">$localpath.contents" ) \|\| die( "Could not open contents file $localpath.contents\n" );
64	dpavlin	40	# fixup just HTML files
65			if ($response->header("content-type") =~ "text/html") {
66			# if you don't want content to be indexed, include it in
67			# <noindex> foobar </noindex> tags or surround it with comments
68			# <!-- noindex --> foobar <!-- /noindex -->
69			$contents =~ s,<noindex>.+?</noindex>,,isg;
70			$contents =~ s,<!--\snoindex\s-->.+?<!--\s/noindex\s-->,,isg;
71			# this will remove all script from indexing content
72			$contents =~ s,<script>.+?</script>,,isg;
73			# remap Windows charset to ISO-8859-2
74			$contents =~ tr/ðèæÐÈÆ/¹ð¾èæ©Ð®ÈÆ/; # 1250 -> iso8859-2
75			# this will fix badly formatted html in form:
76			# <head><title>some text</title
77			# ></head>
78			# which will confuse indexer (or libxml2?)
79			$contents =~ s/[\n\r]^(>)/$1\n/msg;
80			# remove comments between <html> and <head> texi2html inserts them
81			# there and swish can't find document title then (libxml or swish bug?)
82			while ($contents =~ s/(<html>.)<!--.?-->(.*<head>)/$1$2/msi) { };
83			}
84	dpavlin	1	print CONTENTS $contents;
85			close( CONTENTS );
86
87			if( $response->header("content-type") =~ "text/html" ) {
88			open( LINKS, ">$localpath.links" ) \|\| die( "Could not open links file $localpath.links\n" );
89			my $p = HTML::LinkExtor->new( \&linkcb, $url );
90			$p->parse( $contents );
91
92			close( LINKS );
93			}
94			}
95
96
97			sub linkcb {
98			my($tag, %links) = @_;
99	dpavlin	40	if (($tag eq "a" \|\| $tag eq "area") && ($links{"href"}) \|\| ($tag eq "frame" && $links{"src"})) {
100			my $link = $links{"href"} \|\| $links{"src"};
101	dpavlin	1
102			#
103			# Remove fragments
104			#
105			$link =~ s/(.)#./$1/;
106
107			#
108			# Remove ../ This is important because the abs() function
109			# can leave these in and cause never ending loops.
110			#
111			$link =~ s/\.\.\///g;
112
113			# hack for apostrophe -- changes URL, but should work for most clients.
114			$link =~ s/'/%27/g;
115
116			# hack for Apache directory listings
117			$link =~ s,/\?[NMSD]=[AD]$,/,g;
118
119	dpavlin	15	# speedup, skip pictures
120			return if ($link =~ m/\.(gif\|jpg\|png)/);
121
122	dpavlin	1	if ($no_parent_url) {
123			if ($link =~ m/$no_parent_url/) {
124	dpavlin	40	print LINKS "$no_parent_url $link\n";
125	dpavlin	1	# print STDERR "using $link\n";
126			# } else {
127			# print STDERR "skipping $link\n";
128			}
129			} else {
130			print LINKS "$link\n";
131			}
132			}
133			}
134