use strict;
use warnings;
-use CGI qw( -oldstyle_urls );
use POSIX;
use HTML::Entities;
+use DB_File;
+use Lingua::Stem v0.82;
+use Search::Xapian qw(:ops);
use Deb::Versions;
+use Packages::CGI;
use Exporter;
our @ISA = qw( Exporter );
-our @EXPORT_OK = qw( nextlink prevlink indexline
- resperpagelink );
+our @EXPORT_OK = qw( read_entry read_entry_all read_entry_simple
+ read_src_entry read_src_entry_all find_binaries
+ do_names_search do_fulltext_search do_xapian_search
+ );
our %EXPORT_TAGS = ( all => [ @EXPORT_OK ] );
our $VERSION = 0.01;
-our $USE_PAGED_MODE = 1;
-use constant DEFAULT_PAGE => 1;
-use constant DEFAULT_RES_PER_PAGE => 50;
-our %page_params = ( page => { default => DEFAULT_PAGE,
- match => '(\d+)' },
- number => { default => DEFAULT_RES_PER_PAGE,
- match => '(\d+)' } );
-
-our $debug = 0;
-
-sub parse_params {
- my ( $cgi, $params_def ) = @_;
-
- my %params_ret = ( values => {}, errors => {} );
- my %params;
- if ($USE_PAGED_MODE) {
- print "DEBUG: Use PAGED_MODE<br>" if $debug;
- %params = %$params_def;
- foreach (keys %page_params) {
- delete $params{$_};
- }
- %params = ( %params, %page_params );
- } else {
- %params = %$params_def;
- }
-
- foreach my $param ( keys %params ) {
-
- print "<hr><p>DEBUG: Param <strong>$param</strong><br>" if $debug;
-
- my $p_value_orig = $cgi->param($param);
-
- if (!defined($p_value_orig)
- && defined $params_def->{$param}{alias}
- && defined $cgi->param($params_def->{$param}{alias})) {
- $p_value_orig = $cgi->param($params_def->{$param}{alias});
- print "DEBUG: Used alias <strong>$params_def->{$param}{alias}</strong><br>"
- if $debug;
- }
-
- my @p_value = ($p_value_orig);
+our $too_many_hits = 0;
- print "DEBUG: Value (Orig) ".($p_value_orig||"")."<br>" if $debug;
-
- if ($params_def->{$param}{array} && defined $p_value_orig) {
- @p_value = split /$params_def->{$param}{array}/, $p_value_orig;
- print "DEBUG: Value (Array Split) ".
- join('##',@p_value)."<br>" if $debug;
- }
-
- if ($params_def->{$param}{match} && defined $p_value_orig) {
- @p_value = map
- { $_ =~ m/$params_def->{$param}{match}/; $_ = $1 }
- @p_value;
- }
- @p_value = grep { defined $_ } @p_value;
-
- print "DEBUG: Value (Match) ".
- join('##',@p_value)."<br>" if $debug;
-
- unless (@p_value) {
- if (defined $params{$param}{default}) {
- @p_value = ($params{$param}{default});
- } else {
- @p_value = undef;
- $params_ret{errors}{$param} = "undef";
- next;
- }
- }
+sub read_entry_all {
+ my ($hash, $key, $results, $non_results, $opts) = @_;
+ my ($virt, $result) = split /\000/o, $hash->{$key} || "-\01-", 2;
- print "DEBUG: Value (Default) ".
- join('##',@p_value)."<br>" if $debug;
- my @p_value_no_replace = @p_value;
-
- if ($params{$param}{replace} && @p_value) {
- @p_value = ();
- foreach my $pattern (keys %{$params{$param}{replace}}) {
- foreach (@p_value_no_replace) {
- if ($_ eq $pattern) {
- my $replacement = $params{$param}{replace}{$_};
- if (ref $replacement) {
- push @p_value, @$replacement;
- } else {
- push @p_value, $replacement;
- }
- } else {
- push @p_value, $_;
- }
- }
- }
- }
-
- print "DEBUG: Value (Final) ".
- join('##',@p_value)."<br>" if $debug;
-
- if ($params_def->{$param}{array}) {
- $params_ret{values}{$param} = {
- orig => $p_value_orig,
- no_replace => \@p_value_no_replace,
- final => \@p_value,
- };
+ my %virt = split /\01/o, $virt;
+ while (my ($suite, $provides) = each %virt) {
+ next if $suite eq '-';
+ if ($opts->{h_suites}{$suite}) {
+ push @$results, [ $key, "-", $suite, 'virtual', 'v', 'v', 'v', 'v',
+ $provides];
} else {
- $params_ret{values}{$param} = {
- orig => $p_value_orig,
- no_replace => $p_value_no_replace[0],
- final => $p_value[0],
- };
+ push @$non_results, [ $key, "-", $suite, 'virtual', 'v', 'v', 'v', 'v',
+ $provides];
}
}
- if ($USE_PAGED_MODE) {
- $cgi->delete( "page" );
- $cgi->delete( "number" );
+ foreach (split /\000/o, $result) {
+ my @data = split ( /\s/o, $_, 8 );
+ debug( "Considering entry ".join( ':', @data), 2) if DEBUG;
+ if ($opts->{h_suites}{$data[1]}
+ && ($opts->{h_archs}{$data[2]} || $data[2] eq 'all')
+ && $opts->{h_sections}{$data[3]}) {
+ debug( "Using entry ".join( ':', @data), 2) if DEBUG;
+ push @$results, [ $key, @data ];
+ } else {
+ push @$non_results, [ $key, @data ];
+ }
}
-
- return %params_ret;
}
-
-sub start {
- my $params = shift;
-
- my $page = $params->{values}{page}{final}
- || DEFAULT_PAGE;
- my $res_per_page = $params->{values}{number}{final}
- || DEFAULT_RES_PER_PAGE;
-
- return 1 if $res_per_page =~ /^all$/i;
- return $res_per_page * ($page - 1) + 1;
+sub read_entry {
+ my ($hash, $key, $results, $opts) = @_;
+ my @non_results;
+ read_entry_all( $hash, $key, $results, \@non_results, $opts );
}
-sub end {
- my $params = shift;
-
- my $page = $params->{values}{page}{final}
- || DEFAULT_PAGE;
- my $res_per_page = $params->{values}{number}{final}
- || DEFAULT_RES_PER_PAGE;
-
- return $page * $res_per_page;
+#FIXME: make configurable
+my %fallback_suites = (
+ 'oldstable-backports' => 'oldstable',
+ 'oldstable-volatile' => 'oldstable',
+ 'stable-backports' => 'stable',
+ 'stable-volatile' => 'stable',
+ experimental => 'unstable' );
+
+sub read_entry_simple {
+ my ($hash, $key, $archives, $suite) = @_;
+ # FIXME: drop $archives
+
+ my ($virt, $result) = split /\000/o, $hash->{$key} || "-\01-\0", 2;
+ my %virt = split /\01/o, $virt;
+ debug( "read_entry_simple: key=$key, archives=".
+ join(" ",(keys %$archives)).", suite=$suite", 1) if DEBUG;
+ debug( "read_entry_simple: virt=".join(" ",(%virt)), 2) if DEBUG;
+ # FIXME: not all of the 2^4=16 combinations of empty(results),
+ # empty(virt{suite}), empty(fb_result), empty(virt{fb_suite}) are dealt
+ # with correctly, but it's adequate enough for now
+ return [ $virt{$suite} ] unless defined $result;
+ foreach (split /\000/o, $result) {
+ my @data = split ( /\s/o, $_, 8 );
+ debug( "use entry: @data", 2 ) if DEBUG && $data[1] eq $suite;
+ return [ $virt{$suite}, @data ] if $data[1] eq $suite;
+ }
+ if (my $fb_suite = $fallback_suites{$suite}) {
+ my $fb_result = read_entry_simple( $hash, $key, $archives, $fb_suite );
+ my $fb_virt = shift(@$fb_result);
+ $virt{$suite} .= $virt{$suite} ? " $fb_virt" : $fb_virt if $fb_virt;
+ return [ $virt{$suite}, @$fb_result ] if @$fb_result;
+ }
+ return [ $virt{$suite} ];
}
-sub indexline {
- my ($cgi, $params, $num_res) = @_;
-
- my $index_line = "";
- my $page = $params->{values}{page}{final}
- || DEFAULT_PAGE;
- my $res_per_page = $params->{values}{number}{final}
- || DEFAULT_RES_PER_PAGE;
- my $numpages = ceil($num_res /
- $res_per_page);
- for (my $i = 1; $i <= $numpages; $i++) {
- if ($i == $page) {
- $index_line .= $i;
- } else {
- $index_line .= "<a href=\"".encode_entities($cgi->self_url).
- "&page=$i&number=$res_per_page\">".
- "$i</a>";
- }
- if ($i < $numpages) {
- $index_line .= " | ";
+sub read_src_entry_all {
+ my ($hash, $key, $results, $non_results, $opts) = @_;
+ my $result = $hash->{$key} || '';
+ debug( "read_src_entry_all: key=$key", 1) if DEBUG;
+ foreach (split /\000/o, $result) {
+ my @data = split ( /\s/o, $_, 6 );
+ debug( "Considering entry ".join( ':', @data), 2) if DEBUG;
+ if ($opts->{h_archives}{$data[0]}
+ && $opts->{h_suites}{$data[1]}
+ && $opts->{h_sections}{$data[2]}) {
+ debug( "Using entry ".join( ':', @data), 2) if DEBUG;
+ push @$results, [ $key, @data ];
+ } else {
+ push @$non_results, [ $key, @data ];
}
}
- return $index_line;
}
-
-sub nextlink {
- my ($cgi, $params, $no_results ) = @_;
-
- my $page = $params->{values}{page}{final}
- || DEFAULT_PAGE;
- $page++;
- my $res_per_page = $params->{values}{number}{final}
- || DEFAULT_RES_PER_PAGE;
-
- if ((($page-1)*$res_per_page + 1) > $no_results) {
- return ">>";
+sub read_src_entry {
+ my ($hash, $key, $results, $opts) = @_;
+ my @non_results;
+ read_src_entry_all( $hash, $key, $results, \@non_results, $opts );
+}
+sub do_names_search {
+ my ($keywords, $packages, $postfixes, $read_entry, $opts,
+ $results, $non_results) = @_;
+
+ my $first_keyword = lc shift @$keywords;
+ @$keywords = map { lc $_ } @$keywords;
+
+ my ($key, $prefixes) = ($first_keyword, '');
+ my %pkgs;
+ $postfixes->seq( $key, $prefixes, R_CURSOR );
+ while (index($key, $first_keyword) >= 0) {
+ if ($prefixes =~ /^\001(\d+)/o) {
+ debug( "$key has too many hits", 2 ) if DEBUG;
+ $too_many_hits += $1;
+ } else {
+ PREFIX:
+ foreach (split /\000/o, $prefixes) {
+ $_ = '' if $_ eq '^';
+ my $word = "$_$key";
+ foreach my $k (@$keywords) {
+ next PREFIX unless $word =~ /\Q$k\E/;
+ }
+ debug( "add word $word", 2) if DEBUG;
+ $pkgs{$word}++;
+ }
+ }
+ last if $postfixes->seq( $key, $prefixes, R_NEXT ) != 0;
+ last if $too_many_hits or keys %pkgs >= 100;
+ }
+
+ my $no_results = keys %pkgs;
+ if ($too_many_hits || ($no_results >= 100)) {
+ $too_many_hits += $no_results;
+ %pkgs = ( $first_keyword => 1 ) unless @$keywords;
+ }
+ foreach my $pkg (sort keys %pkgs) {
+ &$read_entry( $packages, $pkg, $results, $non_results, $opts );
}
-
- return "<a href=\"".encode_entities($cgi->self_url).
- "&page=$page&number=$res_per_page\">>></a>";
}
-sub prevlink {
- my ($cgi, $params ) = @_;
+sub do_xapian_search {
+ my ($keywords, $db, $did2pkg, $packages, $read_entry, $opts,
+ $results, $non_results) = @_;
- my $page = $params->{values}{page}{final}
- || DEFAULT_PAGE;
- $page--;
- if (!$page) {
- return "<<";
+# NOTE: this needs to correspond with parse-packages!
+ my @tmp;
+ foreach my $keyword (@$keywords) {
+ $keyword =~ tr [A-Z] [a-z];
+ if ($opts->{exact}) {
+ $keyword = " $keyword ";
+ }
+ $keyword =~ s/[(),.-]+//og;
+ $keyword =~ s;[^a-z0-9_/+]+; ;og;
+ push @tmp, $keyword;
}
+ my $stemmer = Lingua::Stem->new();
+ $keywords = $stemmer->stem( @tmp );
+
+ my $db = Search::Xapian::Database->new( $db );
+ my $enq = $db->enquire( OP_AND, @$keywords );
+ debug( "Xapian Query was: ".$enq->get_query()->get_description(), 1) if DEBUG;
+ my @matches = $enq->matches(0, 100);
+
+ my $numres = 0;
+ my %tmp_results;
+ foreach my $match ( @matches ) {
+ my $id = $match->get_docid();
+ my $result = $did2pkg->{$id};
+
+ foreach (split /\000/o, $result) {
+ my @data = split /\s/, $_, 3;
+# debug ("Considering $data[0], arch = $data[2]", 3) if DEBUG;
+# next unless $data[2] eq 'all' || $opts->{h_archs}{$data[2]};
+# debug ("Ok", 3) if DEBUG;
+ $numres++ unless $tmp_results{$data[0]}++;
+ }
+ last if $numres > 100;
+ }
+ undef $db;
+ $too_many_hits++ if $numres > 100;
- my $res_per_page = $params->{values}{number}{final}
- || DEFAULT_RES_PER_PAGE;
-
- return "<a href=\"".encode_entities($cgi->self_url).
- "&page=$page&number=$res_per_page\"><<</a>";
+ foreach my $pkg (keys %tmp_results) {
+ &$read_entry( $packages, $pkg, $results, $non_results, $opts );
+ }
}
-sub resperpagelink {
- my ($cgi, $params, $res_per_page ) = @_;
+sub find_binaries {
+ my ($pkg, $archive, $suite, $src2bin) = @_;
- my $page;
- if ($res_per_page =~ /^all$/i) {
- $page = 1;
- } else {
- $page = ceil(start( $params ) / $res_per_page);
+ my $bins = $src2bin->{$pkg} || '';
+ my %bins;
+ foreach (split /\000/o, $bins) {
+ my @data = split /\s/, $_, 5;
+
+ debug( "find_binaries: considering @data", 3 ) if DEBUG;
+ if (($data[0] eq $archive)
+ && ($data[1] eq $suite)) {
+ $bins{$data[2]}++;
+ debug( "find_binaries: using @data", 3 ) if DEBUG;
+ }
}
- return "<a href=\"".encode_entities($cgi->self_url).
- "&page=$page&number=$res_per_page\">$res_per_page</a>";
+ return [ keys %bins ];
}