surfraw-extra 2.2.8-1 / usr / bin / opensearch-genquery

#!/usr/bin/perl -w
# $Id$
# Ian Beckwith <ianb@erislabs.net>
# 20060830

use Getopt::Long;

use strict;
use vars qw($me);
$me=($0=~/(?:.*\/)?(.*)/)[0];

use vars qw(@wantedtypes $htmlmime $atommime $rssmime);
@wantedtypes=();
$htmlmime="text/html";
$atommime="application/atom+xml";
$rssmime ="application/rss+xml";

my $quiet=0;
my $count=undef;
my $startindex=undef;
my $startpage=undef;
my $language=undef;
my $inputencoding=undef;
my $outputencoding=undef;
my $help=0;

Getopt::Long::Configure("no_ignore_case");

GetOptions("quiet"              => \$quiet,
		   "H|html"             => sub { push(@wantedtypes,$htmlmime); },
		   "A|atom"             => sub { push(@wantedtypes,$atommime); },
		   "R|rss"              => sub { push(@wantedtypes,$rssmime);  },
		   "count=i"            => \$count,
		   "language=s"         => \$language,
		   "i|startindex=i"     => \$startindex,
		   "p|startpage=i"      => \$startpage,
		   "O|outputencoding=s" => \$outputencoding,
		   "I|inputencoding=s"  => \$inputencoding,
		   "h|help"             => \$help);

if ($help || ($#ARGV<1)) { usage(); }

my $url=shift;
my $searchterms=join(' ',@ARGV);

unless($url=~/^http\:\/\//i)
{
	$url="http://".$url;
}

unless (eval { require WWW::OpenSearch; })
{
	unless($quiet)
	{
		print STDERR "$me: Cannot find WWW::OpenSearch, is it installed?\n";
	}
	exit 4;
}

my $engine;
eval { $engine=WWW::OpenSearch->new($url); };
if($@)
{
	my $errstr="";
	my $exitcode=0;
	if($@ =~/fetching/i)
	{
		$errstr="Error fetching $url";
		$exitcode=1;
	}
	elsif(($@ =~/parsing/i) ||
		  ($@ =~/node\s+should/i))
	{
		$errstr="Error parsing OpenSearch Description at $url"; 
		$exitcode=2;
	}
	else
	{
		# Some other error, probably because WWW::OpenSearch has changed
		$errstr=$@;
		$exitcode=5;
	}
	unless($quiet)
	{
		print STDERR "$me: $errstr\n";
	}
	exit $exitcode;
}

my $desc=$engine->description();

# copy options to hash, translating from option name to OpenSearch param
my %options=();
$options{searchTerms} = $searchterms;
if(defined($count))          { $options{count}          = $count;          }
if(defined($startindex))     { $options{startIndex}     = $startindex;     }
if(defined($startpage))      { $options{startPage}      = $startpage;      }
if(defined($language))       { $options{language}       = $language;       }
if(defined($inputencoding))  { $options{inputEncoding}  = $inputencoding;  }
if(defined($outputencoding)) { $options{outputEncoding} = $outputencoding; }


my @urls=();
for (@{$desc->url()})
{
	if($_->method() =~ /^get$/i)
	{
		push(@urls,$_);
	}
}

my $besturl=undef;

# if no preferred type specified use first Url.
if($#wantedtypes<0)
{
	$besturl=$urls[0];
}
else # find best match against @wantedtypes
{
	my $bestindex=undef;
	for my $urlindex (0..$#urls)
	{
		for(my $typeindex=0;$typeindex<=$#wantedtypes;$typeindex++)
		{
			if($urls[$urlindex]->type() eq $wantedtypes[$typeindex])
			{
				if((!defined($bestindex)) || ($typeindex < $bestindex))
				{
					$bestindex=$typeindex;
					$besturl=$urls[$urlindex];
				}
			}
		}
	}
}

if(!defined($besturl))
{
	unless($quiet)
	{
		print STDERR "$me: No matching search type found\n";
	}
	exit 3;
}
	
print $besturl->prepare_query(\%options),"\n";

exit 0;


sub usage
{
	die("Usage: $me [OPTIONS] URL SEARCH TERMS...\n".
		" Outputs the url of a query generated from an OpenSearch Description\n".
		" at URL with SEARCH TERMS filled in.\n".
		"Options:\n".
		" -q\t\tQuiet: Don't display errors.\n".
		"   -H, -A, -R can be ordered to express response type priority.\n".
		" -H\t\tRequest HTML response.\n".
		" -A\t\tRequest Atom response.\n".
		" -R\t\tRequest RSS response.\n".
		"   Not all sites implement all (or any) of the following options.\n".
		"   Unimplemented options will be silently ignored.\n".
		" -c COUNT\tRequest COUNT results (OpenSearch param: count).\n".
		" -i INDEX\tRequest results start at NUM (OpenSearch param: startIndex).\n".
		" -p PAGENUM\tRequest results page PAGENUM (OpenSearch param: pageIndex).\n".
		" -l LANG\tRequest results in LANG (2-letter ISO country code) (OpenSearch param: language).\n".
		" -I ENC\t\tSpecify search terms are encoded in ENC (OpenSearch param: inputEncoding).\n".
		" -O ENC\t\tRequest results encoded in ENC (OpenSearch param: outputEncoding).\n".
		"Return codes:\n".
		" 0 - Success.\n".
		" 1 - Error fetching OpenSearch Description URL.\n".
		" 2 - Error parsing OpenSearch Description.\n".
		" 3 - No matching search type found.\n".
		" 4 - Cannot find perl module WWW::OpenSearch.\n".
		" 5 - Unhandled error from WWW::OpenSearch.\n");
}

__END__


=head1 NAME

opensearch-genquery - Output the URL of a query generated from an OpenSearch Description

=head1 SYNOPSIS

B<opensearch-genquery> [B<-q>] [B<-H>] [B<-A>] [B<-R>]
S<[B<-c> I<COUNT>]> S<[B<-i> I<INDEX>]>
S<[B<-p> I<PAGENUM>]> S<[B<-l> I<LANG>]>
S<[B<-O> I<ENC>]> S<[B<-I> I<ENC>]>
I<URL> S<I<SEARCH TERMS...>>

=head1 DESCRIPTION

Output the URL of a query generated from an OpenSearch Description
file at I<URL>, with the specified S<I<SEARCH TERMS>> and other
options filled in. 

If multiple search types are available, B<-H>,
B<-A> and B<-R> can be used to specify which type is wanted. For
example, S<B<-A -R>> can be used to request an Atom response, or
failing that fall back to RSS. By default the first URL found in the
description file is used.

Not all web sites implement all (or any) of the optional OpenSearch
parameters. When used with those sites, the corresponding options will
be silently ignored.

=head1 OPTIONS

=over 4

=item B<-q>

Give no output on errors. In this case, consult the exit code to
determine errors (See L</DIAGNOSTICS>).

=item B<-H>

Request HTML response.

=item B<-A>

Request Atom response.

=item B<-R>

Request RSS response.

B<-H>, B<-A>, and B<-R> can be combined to express response type
priority and specify fallback types.

=item B<-c COUNT>, B<--count=COUNT>

Request COUNT results.

OpenSearch parameter: I<count>.

=item B<-i INDEX>, B<--startindex=INDEX>

Request results start at offset NUM.

OpenSearch parameter: I<startIndex>.

=item B<-p PAGENUM>, B<--startpage=PAGENUM>

Request results page PAGENUM.

OpenSearch parameter: I<pageIndex>.

=item B<-l LANG>, B<--language=LANG>

Request results in LANG. LANG should be a 2-letter ISO country code
(e.g. B<en> or B<de>).

OpenSearch parameter: I<language>.

=item B<-O ENC>, B<--outputencoding=ENC>

Request results encoded in ENC (default: UTF-8).

OpenSearch parameter: I<outputEncoding>.

=item B<-I ENC>, B<--inputencoding=ENC>

Specify search terms are encoded in ENC (default: UTF-8).

OpenSearch parameter: I<inputEncoding>.

=item B<-h>, B<--help>

Show a short help message.

=item B<-->

End of options.

=back

=head1 DIAGNOSTICS

The exit code can be consulted to determine errors, as follows:

=over 4

=item B<0> - Success.

=item B<1> - Error fetching OpenSearch Description URL.

=item B<2> - Error parsing OpenSearch Description.

=item B<3> - No matching search type found.

=item B<4> - Cannot find perl module WWW::OpenSearch.

=item B<5> - Unhandled error from WWW::OpenSearch.

=back

=head1 DEPENDENCIES

Requires the module WWW::OpenSearch (Debian package
libwww-opensearch-perl).

=head1 BUGS

Only supports GET queries.

Please report any bugs found to ianb@erislabs.net

=head1 SEE ALSO

L<opensearch(1)>, L<opensearch-discover(1)>, L<surfraw(1)>,
L<WWW::OpenSearch(3pm)>, L<http://www.opensearch.org/>

=head1 AUTHOR

Ian Beckwith <ianb@erislabs.net>

=head1 COPYRIGHT AND LICENSE

Copyright 2006 Ian Beckwith <ianb@erislabs.net>

Licensed under the same terms as surfraw.

=cut