X-Git-Url: https://git.deb.at/?a=blobdiff_plain;f=bin%2Fparse-packages;h=399a62ca060656a82314f4f80000a7462899b938;hb=fcd2a8e03040589f3048e91737d4b2863c8521fd;hp=ab5fbcef5b12e357fbfcce47cea2093e14356e8c;hpb=5ca21f1ba07a6b559395bec8d6bc1e528eb238d1;p=deb%2Fpackages.git diff --git a/bin/parse-packages b/bin/parse-packages index ab5fbce..399a62c 100755 --- a/bin/parse-packages +++ b/bin/parse-packages @@ -22,11 +22,14 @@ use strict; use lib './lib'; +$| = 1; + my $what = $ARGV[0] ? "non-free" : "*"; # max. distinct results for a given package postfix my $MAX_PACKAGE_POSTFIXES = 100; use DB_File; +use Storable; use Packages::Config qw( $TOPDIR $DBDIR @ARCHIVES @SUITES ); &Packages::Config::init( './' ); my %packages_small = (); @@ -38,11 +41,16 @@ my @descriptions = ("we count lines one-based\000"); my %packages_descriptions = (); my %descriptions_packages = (); +my %sections = (); +my %subsections = (); +my %priorities = (); + $/ = ""; for my $archive (@ARCHIVES) { for my $suite (@SUITES) { + my %virtual_packages = (); print "Reading $archive/$suite...\n"; my %packages_all_db; tie %packages_all_db, "DB_File", "$DBDIR/packages_all_$suite.db.new", @@ -64,16 +72,19 @@ for my $archive (@ARCHIVES) { # Skip double package next if exists($packages_all_db{"$data{'package'} $data{'architecture'} $data{'version'}"}); + if ($data{'provides'}) { + foreach (split /\s*,\s*/, $data{'provides'}) { + $virtual_packages{$_}{$data{'package'}}++; + } + } $package_names{$data{'package'}} = 1; my $src = $data{'package'}; my $src_version = ''; if ($data{'source'}) { $src = $data{'source'}; - $src_version = $1 - if $src =~ s/\s+\(\s*=\s*(.*)\).*//; # strip version info + $src =~ s/\s+.*//; # strip version info } $data{'source'} = $src; - $data{'source-version'} = $src_version; my $descr = $data{'description'}; my $did = undef; if (exists($descriptions{$descr})) { @@ -94,10 +105,15 @@ for my $archive (@ARCHIVES) { my $subsection = $data{section} || '-'; if ($data{section} && ($data{section} =~ m=/=o)) { ($section, $subsection) = split m=/=o, $data{section}, 2; + ($subsection, $section) = split m=/=o, $data{section}, 2 + if $section eq 'non-US'; } $data{'section'} = $section; $data{'subsection'} = $subsection; $data{'priority'} ||= '-'; + $sections{$suite}{$archive}{$section}++; + $subsections{$suite}{$archive}{$subsection}++; + $priorities{$suite}{$archive}{$data{priority}}++; $packages_small{$data{'package'}} .= "$archive $suite $data{'architecture'} ". "$section $subsection $data{'priority'} $data{'version'} $sdescr\000"; $sources_packages{$src} .= @@ -107,12 +123,18 @@ for my $archive (@ARCHIVES) { next if $key eq 'package' or $key eq 'version' or $key eq 'architecture' or $key eq 'maintainer'; print STDERR "WARN: $key ($suite/$archive/$data{package}/$data{architecture}\n" unless defined $value; - $data .= "$key: $value\n"; + $data .= "$key\00$value\00"; } + $data =~ s/.$//s; $packages_all_db{"$data{'package'} $data{'architecture'} $data{'version'}"} = $data; } + foreach (keys %virtual_packages) { + my @provided_by = keys %{$virtual_packages{$_}}; + $packages_small{$_} .= "$archive $suite virtual v v v v @provided_by\000"; + } + untie %packages_all_db; } } @@ -164,8 +186,13 @@ tie %descriptions_db, "DB_File", "$DBDIR/descriptions.db.new", open DESCR, ">", "$DBDIR/descriptions.txt" or die "Error creating descriptions textfile"; for (my $i=1; $i<= $#descriptions; $i++) { my $plain_description = $descriptions[$i]; - $plain_description =~ s/\n .\n/ /og; - $plain_description =~ s/[\n \t]+/ /og; +# WARNING: This needs to correspond with what happens in +# Packages/Search.pm:do_fulltext_search + $plain_description =~ tr [A-Z] [a-z]; + # ensure one space on both ends + $plain_description = " $plain_description "; + $plain_description =~ s/[(),.-]+//og; + $plain_description =~ s#[^a-z0-9_/+]+# #og; print DESCR "$plain_description\n"; $descriptions_db{$i} = $descriptions[$i]; } @@ -197,6 +224,10 @@ while (my ($k, $v) = each(%package_postfixes)) { } untie %package_postfixes_db; +store \%sections, "$DBDIR/sections.info"; +store \%subsections, "$DBDIR/subsections.info"; +store \%priorities, "$DBDIR/priorities.info"; + rename("$DBDIR/packages_small.db.new", "$DBDIR/packages_small.db"); rename("$DBDIR/sources_packages.db.new", "$DBDIR/sources_packages.db"); for my $suite (@SUITES) {