File pom2spec of Package obs-service-generator_pom

#!/usr/bin/perl
# vim: ai:ts=4:sw=4:et
#
# Pulls metadata from a Maven2 POM file on Maven Central
# to generate a .spec file (on STDOUT) that merely warps
# the .jar file(s) into noarch RPM files.
#
# Note that it does *NOT* create a .spec file that builds
# the upstream sources, that would just be too much awesome
# magic.
#
# Copyright 2011 Pascal Bleser <pascal.bleser@opensuse.org>
#
# This file is licensed under the
# GNU Lesser General Public License version 2.1 or later:
# http://www.gnu.org/licenses/lgpl-2.1.html
#

use strict;
use warnings;
use LWP::UserAgent;
use XML::LibXML;
use Cwd;
use File::Basename;
use Carp;

# These are the very few configuration options:

my $repourl = "http://repo1.maven.org/maven2";
my $include_requires_as_buildrequires = 1;
my $agent = "pom2spec/1.0";
my $email = 'pascal.bleser@opensuse.org';
my $max_redirect = 2;
my $timeout = 20; # in seconds

# Now some helper functions to reduce verbosity:

# Returns either the content of the file that is specified
# as parameter or undef if the file does not exist.
sub r($) {
    my $filename = shift;
    if (-e $filename) {
        open(my $fh, '<', $filename) or croak "failed to read existing file $filename: $!";
        chomp(my $content = do { local $/; <$fh>; });
        close($fh);
        return $content;
    } else {
        return undef;
    }
}

# "ue" = "unique element": seeks out a unique child element
# under a node ($p) with a given tag name ($n) and, when
# found, returns its chomped text content.
# Croaks out when no matching element could be found, or when
# more than one matching element could be found.
sub ue($$) {
    my $p = shift;
    my $n = shift;
    my @results = $p->getChildrenByTagName($n);
    croak "invalid metadata: has more than one <${n}/> element" if scalar(@results) > 1;
    croak "invalid metadata: has no <${n}/> element" if scalar(@results) < 1;
    chomp(my $t = $results[0]->textContent);
    return $t;
}

# Same as "ue", but when no matching element is found, it returns
# a fallback value ($alt) instead of croaking.
sub uec($$$) {
    my $p = shift;
    my $n = shift;
    my $alt = shift;
    my @results = $p->getChildrenByTagName($n);
    croak "invalid metadata: has more than one <${n}/> element" if scalar(@results) > 1;
    return $alt if scalar(@results) < 1;
    chomp(my $t = $results[0]->textContent);
    return $t;
}

# Conditionally puts a match (see ue() and uec() above) into a hash.
sub ueh($$$$) {
    my $p = shift;
    my $n = shift;
    my $h = shift;
    my $k = shift;
    my @results = $p->getChildrenByTagName($n);
    croak "invalid metadata: has more than one <${n}/> element" if scalar(@results) > 1;
    return undef if scalar(@results) < 1;
    chomp(my $t = $results[0]->textContent);
    $h->{$k} = $t if defined $t;
    return $t;
}

# Simple and naive transliteration (replaces ${...} placeholders by
# their value in a hash).
sub trl($$) {
    my $text = shift;
    return $text unless defined $text;

    my $dict = shift;

    while ($text =~ /^(.*?)\$\{(.+?)\}(.*)$/) {
        croak "unresolvable placeholder \${$2}" unless exists $dict->{$2};
        $text = $1 . $dict->{$2} . $3;
    }
    return $text;
}

my $group;
my $name;
my $version;
my $pgroup;
my $path;
{
    if (scalar(@ARGV) == 0 or scalar(@ARGV) == 1) {
        my $d = basename(cwd());
        ($group, $name, $version) = $d =~ /^(.+)\.([^\.]+)(?:-(\d.?))?$/;
        die "when groupId and artifactId are not specified as parameters, the current directory must reflect groupId.artifactId" unless defined $group and defined $name;
        my $versionOverride = shift;
        if (defined($versionOverride)) {
            die "version is already specified in the name of the current working directory ($d -> $version)" if defined $version;
            $version = $versionOverride;
        }
    } elsif (scalar(@ARGV) == 2 or scalar(@ARGV) == 3) {
        $group = shift;
        $name = shift;
        $version = shift;

        $group =~ s/\//./g;
    }
    # $version might and may be undefined at this point

    ($pgroup = $group) =~ s/\./\//g;
    $path = $pgroup.'/'.$name;
}

my $ua;
{
    $ua = LWP::UserAgent->new;
    $ua->agent($agent);
    $ua->timeout($timeout);
    $ua->max_redirect($max_redirect);
    $ua->from($email);
}

my $groupId;
my $artifactId;
my @versions;
my $defaultVersion;
{
    my $url = "${repourl}/${path}/maven-metadata.xml";
    my $req = HTTP::Request->new(GET => $url);
    $req->content_type('text/xml');
    my $res = $ua->request($req);
    unless ($res->is_success) {
        die "failed to get $url (".$res->status_line.")";
    }

    my $parser = XML::LibXML->new;
    $parser->validation(0);
    $parser->no_network(1);
    $parser->load_ext_dtd(0);

    my $doc = $parser->parse_string($res->content);
    my $metadata = $doc->getDocumentElement();

    $groupId = ue($metadata, 'groupId');
    warn "WARNING: groupId in maven-metadata.xml file is \"$groupId\"\n" if $groupId ne $group;
    $artifactId = ue($metadata, 'artifactId');
    $defaultVersion = uec($metadata, 'version', undef);

    @versions = ();
    foreach my $versioning ($metadata->getChildrenByTagName('versioning')) {
        foreach my $versions ($versioning->getChildrenByTagName('versions')) {
            foreach my $ve ($versions->getChildrenByTagName('version')) {
                chomp(my $v = $ve->textContent);
                push(@versions, $v);
            }
        }
    }

    if (defined $version) {
        die "requested version ${version} is not in metadata" unless @versions // $version;
    } else {
        die "call again and specify the exact version, one of:\n".join("\n", map { " - ".$_ } @versions)."\n";
    }
}


$name = "${groupId}.${artifactId}-${version}";
my $out = "${name}.spec";

sub resolve($$$$);

# global because we load once and cache
my %ignoredeps;
{
    if (-e "_ignoredeps") {
        open(my $fh, "<", "_ignoredeps") or die "failed to open _ignoredeps: $!";
        while (<$fh>) {
            chomp;
            s/#.+//;
            next if /^$/;
            croak "invalid line \"$_\" in _ignoredeps" unless /^(.+?):(.+)$/;
            $ignoredeps{$_} = 1;
        }
        close($fh);
    } else {
        %ignoredeps = ();
    }
}

sub resolve($$$$) {
    my $g = shift;
    my $a = shift;
    my $v = shift;
    my $is_parent = shift;

    my $r = {};

    (my $p = $g) =~ s/\./\//g;

    my $url = "${repourl}/${p}/${a}/${v}/${a}-${v}.pom";
    my $req = HTTP::Request->new(GET => $url);
    $req->content_type('text/xml');
    my $res = $ua->request($req);
    unless ($res->is_success) {
        croak "failed to get $url (".$res->status_line.")";
    }

    $r->{pomurl} = $url;

    my $parser = XML::LibXML->new;
    $parser->validation(0);
    $parser->no_network(1);
    $parser->load_ext_dtd(0);

    my $doc = $parser->parse_string($res->content);
    my $project = $doc->getDocumentElement();

    my $parent;
    {
        my @parentElems = $project->getChildrenByTagName('parent');
        croak "invalid metadata: has more than one <parent/> element" if scalar(@parentElems) > 1;
        if (scalar(@parentElems) > 0) {
            my $parentElem = $parentElems[0];
            $parent = resolve(ue($parentElem, 'groupId'), ue($parentElem, 'artifactId'), uec($parentElem, 'version', $v), 1);
        } else {
            $parent = {};
        }
    }

    $r->{artifactId} = uec($project, 'artifactId', $name);
    {
        my $nameOverride = $is_parent ? undef : r("_name.override");
        if (defined $nameOverride) {
            $r->{name} = $nameOverride;
        } else {
            $r->{name} = $r->{artifactId};
        }
    }
    {
        my $groupIdOverride = $is_parent ? undef : r("_groupId.override");
        if (defined $groupIdOverride) {
            $r->{groupId} = $groupIdOverride;
        } else {
            $r->{groupId} = uec($project, 'groupId', $r->{artifactId});
            warn "WARNING: groupId in pom file is \"$groupId\", was expecting \"$g\": $url\n" if $r->{groupId} ne $g;
        }
    }
    $r->{version} = uec($project, 'version', exists $parent->{version} ? $parent->{version} : $v);
    $r->{packaging} = uec($project, 'packaging', 'jar');
    $r->{summary} = uec($project, 'name', r('_summary'));
    ueh($project, 'description', $r, 'description');
    ueh($project, 'url', $r, 'url');
    {
        my $license;
        {
            my @licenses;
            {
                my @l = ();
                foreach my $licenses ($project->getChildrenByTagName('licenses')) {
                    foreach my $licenseElem ($licenses->getChildrenByTagName('license')) {
                        chomp(my $name = ue($licenseElem, 'name'));
                        push(@l, $name);
                    }
                }
                if (scalar(@l) > 0) {
                    @licenses = @l;
                } elsif (exists $parent->{license}) {
                    @licenses = ($parent->{license});
                } else {
                    @licenses = ();
                }
            }
            if (scalar(@licenses) > 1) {
                $license = join(", ", @licenses);
            } elsif (scalar(@licenses) > 0) {
                $license = $licenses[0];
            } else {
                $license = undef;
            }
        }
        if (defined $license) {
            $r->{license} = $license;
            warn "WARNING: unnecessary (and unused) file _license\n" if -e "_license" and $is_parent;
        } else {
            if (-e "_license") {
                $r->{license} = '';
                open(my $lf, '<', "_license") or croak "failed to open _license: $!";
                while (<$lf>) {
                    chomp;
                    $r->{license} .= $_;
                }
                close($lf);
            } else {
                warn "WARNING: no license specified for ".join(":", map { $r->{$_} } qw(groupId artifactId version))."\n";
            }
        }
    }

    {
        my @dependencies;
        {
            my $props;
            {
                $props = {};
                while (my ($k, $v) = each(%{$r})) {
                    $props->{"project.${k}"} = $v;
                }
                
                foreach my $properties ($project->getChildrenByTagName('properties')) {
                    foreach my $prop (grep { $_->nodeType == XML_ELEMENT_NODE } $properties->childNodes()) {
                        chomp(my $k = $prop->nodeName);
                        chomp(my $v = $prop->textContent);
                        $props->{$k} = $v;
                    }
                }
            }

            if (exists $parent->{dependencies}) {
                @dependencies = @{$parent->{dependencies}};
            } else {
                @dependencies = ();
            }

            foreach my $dependency ($project->getChildrenByTagName('dependencies')) {
                foreach my $dependency ($dependency->getChildrenByTagName('dependency')) {
                    my $scope = uec($dependency, 'scope', 'runtime');
                    next if $scope ne 'runtime';

                    my $dg = trl(ue($dependency, 'groupId'), $props);
                    my $da = trl(ue($dependency, 'artifactId'), $props);
                    my $dv = trl(uec($dependency, 'version', undef), $props);

                    if (exists $ignoredeps{"$dg:$da"}) {
                        warn "ignoring dependency $dg:$da\n";
                        next;
                    }

                    my $d = {
                        groupId => $dg,
                        artifactId => $da,
                    };
                    $d->{version} = $dv if defined $dv;
                    push(@dependencies, $d);
                }
            }
        }

        $r->{dependencies} = \@dependencies;
    }

    {
        my $sources_url = "${repourl}/${p}/${a}/${v}/${a}-${v}-sources.jar";
        my $head = HTTP::Request->new(HEAD => $sources_url);
        $head->content_type('application/java-archive');
        my $res = $ua->request($head);
        $r->{sources} = $sources_url if $res->is_success and $res->code == 200;
    }
    if (0) {
        # not sufficiently implemented, would actually need to explode the javadoc jar
        # and strip off the package directory from it to properly install it
        # under %{_javadocdir} in the spec file
        my $javadoc_url = "${repourl}/${p}/${a}/${v}${a}-${v}-javadoc.jar";
        my $head = HTTP::Request->new(HEAD => $javadoc_url);
        $head->content_type('application/java-archive');
        my $res = $ua->request($head);
        $r->{javadoc} = $javadoc_url if $res->is_success;
    }

    return $r;
}

my %x = %{resolve($groupId, $artifactId, $version, 0)};

{
    my $tty;
    {
        open($tty, '>>', '/dev/tty') or $tty = undef;
    }
    if (defined $tty) {
        say $tty "name:         $x{name}";
        say $tty "groupId:      $x{groupId}";
        say $tty "artifactId:   $x{artifactId}";
        say $tty "version:      $x{version}";
        say $tty "license:      $x{license}" if exists $x{license};
        say $tty "packaging:    $x{packaging}";
        say $tty "summary:      $x{summary}" if exists $x{summary};
        say $tty "url:          $x{url}" if exists $x{url};
        say $tty "description:  $x{description}" if exists $x{description};
        if (scalar(@{$x{dependencies}}) > 0) {
            say $tty "dependencies:";
            foreach my $d (@{$x{dependencies}}) {
                say $tty "    $d->{groupId}:$d->{artifactId}" . (exists $d->{version} ? ":".$d->{version} : "");
            }
        }
    }
}

(my $uversion = $x{version}) =~ s/\./_/g;
my $summary = exists $x{summary} ? $x{summary} : $x{groupId};
my $description = exists $x{description} ? $x{description} : $summary;

open(my $outfd, '>', $out) or die "failed to create $out: $!";
my $oldselectfd = select $outfd;

print<<EOF;
# vim: set sw=4 ts=4 et:

# POM:        $x{pomurl}
# groupId:    $x{groupId}
# artifactId: $x{artifactId}

Name:               ${name}
%define uname       %{name}-${uversion}
Version:            $x{version}
Release:            0
Summary:            ${summary}
Source0:            ${repourl}/${path}/${version}/$x{artifactId}-$x{version}.jar
EOF
if (exists $x{sources}) {
    print<<EOF;
Source1:            $x{sources}
EOF
}
if (exists $x{url}) {
    print<<EOF;
URL:                $x{url}
EOF
}
print<<EOF;
Group:              Development/Libraries/Java
EOF
if (exists $x{license}) {
    print<<EOF;
License:            $x{license}
EOF
}
print<<EOF;
BuildRoot:          %{_tmppath}/build-%{name}-%{version}
BuildArch:          noarch
Requires:           %{uname} = %{version}
EOF
if ($include_requires_as_buildrequires) {
    foreach my $d (@{$x{dependencies}}) {
        my $version_text = exists $d->{version} ? " = ".$d->{version} : "";
        print<<EOF;
BuildRequires:      java($d->{groupId}:$d->{artifactId})${version_text}
EOF
    }
}
print<<EOF;

%description
${description}

%package -n %{uname}
Summary:            ${summary}
Group:              Development/Libraries/Java
Provides:           java($x{groupId}:$x{artifactId}) = $x{version}
EOF
foreach my $d (@{$x{dependencies}}) {
    my $version_text = exists $d->{version} ? " = ".$d->{version} : "";
    print<<EOF;
Requires:           java($d->{groupId}:$d->{artifactId})${version_text}
EOF
}
print<<EOF;
%description -n %{uname}
${description}

EOF
if (exists $x{sources}) {
    print<<EOF;
%package -n %{name}-sources
Summary:            Sources for %{name}
Group:              Development/Libraries/Java
Requires:           %{uname}-sources = %{version}
%description -n %{name}-sources
This package contains the source bundle for %{name}.

%package -n %{uname}-sources
Summary:            Sources for %{uname}
Group:              Development/Libraries/Java
Requires:           %{uname} = %{version}
%description -n %{uname}-sources
This package contains the source bundle for %{uname}.
EOF
}
print<<EOF;

%prep
%setup -q -T -c "%{name}-%{version}"

%build

%install
%__install -D -m0644 "%{SOURCE0}" "%{buildroot}%{_javadir}/%{name}-%{version}.jar"
%__ln_s "%{name}-%{version}.jar" "%{buildroot}%{_javadir}/%{name}.jar"

cat<<EOCP > "%{buildroot}%{_javadir}/%{name}-%{version}.classpath"
EOF
print join(":", map { "%{_javadir}/$_->{artifactId}".(exists $_->{version} ? "-$_->{version}" : "").".jar" } @{$x{dependencies}}), "\n";
print<<EOF;
EOCP
%__ln_s "%{name}-%{version}.classpath" "%{buildroot}%{_javadir}/%{name}.classpath"

EOF
if (exists $x{sources}) {
    print<<EOF;
%__install -m0644 "%{SOURCE1}" "%{buildroot}%{_javadir}/%{name}-%{version}-sources.jar"
%__ln_s "%{name}-%{version}-sources.jar" "%{buildroot}%{_javadir}/%{name}-sources.jar"
EOF
}
print<<EOF;

%clean
%{?buildroot:%__rm -rf "%{buildroot}"}

%files
%defattr(-,root,root)
%{_javadir}/%{name}.jar
%{_javadir}/%{name}.classpath

%files -n %{uname}
%defattr(-,root,root)
%{_javadir}/%{name}-%{version}.jar
%{_javadir}/%{name}-%{version}.classpath

EOF
if (exists $x{sources}) {
    print<<EOF;
%files -n %{name}-sources
%defattr(-,root,root)
%{_javadir}/%{name}-sources.jar

%files -n %{uname}-sources
%defattr(-,root,root)
%{_javadir}/%{name}-%{version}-sources.jar
EOF
}
print<<EOF;

EOF

select $oldselectfd;