From 6e3de0f956f611cc27a592fecd44317650591d15 Mon Sep 17 00:00:00 2001
From: Graham Knop <haarg@haarg.org>
Date: Thu, 25 Apr 2024 12:39:20 +0100
Subject: [PATCH 1/3] move find_download_url from document to query namespace

---
 lib/MetaCPAN/Document/File/Set.pm | 226 +----------------------------
 lib/MetaCPAN/Query/Release.pm     | 233 ++++++++++++++++++++++++++++++
 2 files changed, 234 insertions(+), 225 deletions(-)

diff --git a/lib/MetaCPAN/Document/File/Set.pm b/lib/MetaCPAN/Document/File/Set.pm
index abe2a62b9..c758497df 100644
--- a/lib/MetaCPAN/Document/File/Set.pm
+++ b/lib/MetaCPAN/Document/File/Set.pm
@@ -51,7 +51,7 @@ has query_release => (
     isa     => 'MetaCPAN::Query::Release',
     lazy    => 1,
     builder => '_build_query_release',
-    handles => [qw< get_checksums >],
+    handles => [qw< find_download_url >],
 );
 
 sub _build_query_release {
@@ -203,230 +203,6 @@ sub documented_modules {
         ->source( [qw(name module path documentation distribution)] )->all;
 }
 
-=head2 find_download_url
-
-
-cpanm Foo
-=> status: latest, maturity: released
-
-cpanm --dev Foo
-=> status: -backpan, sort_by: version_numified,date
-
-cpanm Foo~1.0
-=> status: latest, maturity: released, module.version_numified: gte: 1.0
-
-cpanm --dev Foo~1.0
--> status: -backpan, module.version_numified: gte: 1.0, sort_by: version_numified,date
-
-cpanm Foo~<2
-=> maturity: released, module.version_numified: lt: 2, sort_by: status,version_numified,date
-
-cpanm --dev Foo~<2
-=> status: -backpan, module.version_numified: lt: 2, sort_by: status,version_numified,date
-
-    $file->find_download_url( 'Foo', { version => $version, dev => 0|1 });
-
-Sorting:
-
-    if it's stable:
-      prefer latest > cpan > backpan
-      then sort by version desc
-      then sort by date descending (rev chron)
-
-    if it's dev:
-      sort by version desc
-      sort by date descending (reverse chronologically)
-
-
-=cut
-
-sub find_download_url {
-    my ( $self, $module, $args ) = @_;
-    $args ||= {};
-
-    my $dev              = $args->{dev};
-    my $version          = $args->{version};
-    my $explicit_version = $version && $version =~ /==/;
-
-    my @filters;
-    if ( !$explicit_version ) {
-        push @filters, { not => { term => { status => 'backpan' } } };
-        if ( !$dev ) {
-            push @filters, { term => { maturity => 'released' } };
-        }
-    }
-
-    my $version_filters = $self->_version_filters($version);
-
-    # filters to be applied to the nested modules
-    my $module_f = {
-        nested => {
-            path       => 'module',
-            inner_hits => { _source => 'version' },
-            filter     => {
-                bool => {
-                    must => [
-                        { term => { 'module.authorized' => 1 } },
-                        { term => { 'module.indexed'    => 1 } },
-                        { term => { 'module.name'       => $module } },
-                        (
-                            exists $version_filters->{must}
-                            ? @{ $version_filters->{must} }
-                            : ()
-                        )
-                    ],
-                    (
-                        exists $version_filters->{must_not}
-                        ? ( must_not => [ $version_filters->{must_not} ] )
-                        : ()
-                    )
-                }
-            }
-        }
-    };
-
-    my $filter
-        = @filters
-        ? { bool => { must => [ @filters, $module_f ] } }
-        : $module_f;
-
-    # sort by score, then version desc, then date desc
-    my @sort = (
-        '_score',
-        {
-            'module.version_numified' => {
-                mode          => 'max',
-                order         => 'desc',
-                nested_path   => 'module',
-                nested_filter => $module_f->{nested}{filter}
-            }
-        },
-        { date => { order => 'desc' } }
-    );
-
-    my $query;
-
-    if ($dev) {
-        $query = { filtered => { filter => $filter } };
-    }
-    else {
-        # if not dev, then prefer latest > cpan > backpan
-        $query = {
-            function_score => {
-                filter     => $filter,
-                score_mode => 'first',
-                boost_mode => 'replace',
-                functions  => [
-                    {
-                        filter => { term => { status => 'latest' } },
-                        weight => 3
-                    },
-                    {
-                        filter => { term => { status => 'cpan' } },
-                        weight => 2
-                    },
-                    { filter => { match_all => {} }, weight => 1 },
-                ]
-            }
-        };
-    }
-
-    my $res
-        = $self->size(1)->query($query)
-        ->source( [ 'release', 'download_url', 'date', 'status' ] )
-        ->search_type('dfs_query_then_fetch')->sort( \@sort )->raw->all;
-    return unless $res->{hits}{total};
-
-    my @checksums;
-
-    my $hit     = $res->{hits}{hits}[0];
-    my $release = exists $hit->{_source} ? $hit->{_source}{release} : undef;
-
-    if ($release) {
-        my $checksums = $self->get_checksums($release);
-        @checksums = (
-            (
-                $checksums->{checksum_md5}
-                ? ( checksum_md5 => $checksums->{checksum_md5} )
-                : ()
-            ),
-            (
-                $checksums->{checksum_sha256}
-                ? ( checksum_sha256 => $checksums->{checksum_sha256} )
-                : ()
-            ),
-        );
-    }
-
-    return +{
-        %{ $hit->{_source} },
-        %{ $hit->{inner_hits}{module}{hits}{hits}[0]{_source} }, @checksums,
-    };
-}
-
-sub _version_filters {
-    my ( $self, $version ) = @_;
-
-    return () unless $version;
-
-    if ( $version =~ s/^==\s*// ) {
-        return +{
-            must => [ {
-                term => {
-                    'module.version_numified' => $self->_numify($version)
-                }
-            } ]
-        };
-    }
-    elsif ( $version =~ /^[<>!]=?\s*/ ) {
-        my %ops = qw(< lt <= lte > gt >= gte);
-        my ( %filters, %range, @exclusion );
-        my @requirements = split /,\s*/, $version;
-        for my $r (@requirements) {
-            if ( $r =~ s/^([<>]=?)\s*// ) {
-                $range{ $ops{$1} } = $self->_numify($r);
-            }
-            elsif ( $r =~ s/\!=\s*// ) {
-                push @exclusion, $self->_numify($r);
-            }
-        }
-
-        if ( keys %range ) {
-            $filters{must}
-                = [ { range => { 'module.version_numified' => \%range } } ];
-        }
-
-        if (@exclusion) {
-            $filters{must_not} = [];
-            push @{ $filters{must_not} }, map {
-                +{
-                    term => {
-                        'module.version_numified' => $self->_numify($_)
-                    }
-                }
-            } @exclusion;
-        }
-
-        return \%filters;
-    }
-    elsif ( $version !~ /\s/ ) {
-        return +{
-            must => [ {
-                range => {
-                    'module.version_numified' =>
-                        { 'gte' => $self->_numify($version) }
-                },
-            } ]
-        };
-    }
-}
-
-sub _numify {
-    my ( $self, $ver ) = @_;
-    $ver =~ s/_//g;
-    version->new($ver)->numify;
-}
-
 =head2 history
 
 Find the history of a given module/documentation.
diff --git a/lib/MetaCPAN/Query/Release.pm b/lib/MetaCPAN/Query/Release.pm
index ce8d51910..77b189e6a 100644
--- a/lib/MetaCPAN/Query/Release.pm
+++ b/lib/MetaCPAN/Query/Release.pm
@@ -992,5 +992,238 @@ sub modules {
     };
 }
 
+=head2 find_download_url
+
+cpanm Foo
+=> status: latest, maturity: released
+
+cpanm --dev Foo
+=> status: -backpan, sort_by: version_numified,date
+
+cpanm Foo~1.0
+=> status: latest, maturity: released, module.version_numified: gte: 1.0
+
+cpanm --dev Foo~1.0
+-> status: -backpan, module.version_numified: gte: 1.0, sort_by: version_numified,date
+
+cpanm Foo~<2
+=> maturity: released, module.version_numified: lt: 2, sort_by: status,version_numified,date
+
+cpanm --dev Foo~<2
+=> status: -backpan, module.version_numified: lt: 2, sort_by: status,version_numified,date
+
+    $release->find_download_url( 'Foo', { version => $version, dev => 0|1 });
+
+Sorting:
+
+    if it's stable:
+      prefer latest > cpan > backpan
+      then sort by version desc
+      then sort by date descending (rev chron)
+
+    if it's dev:
+      sort by version desc
+      sort by date descending (reverse chronologically)
+
+
+=cut
+
+sub find_download_url {
+    my ( $self, $module, $args ) = @_;
+    $args ||= {};
+
+    my $dev              = $args->{dev};
+    my $version          = $args->{version};
+    my $explicit_version = $version && $version =~ /==/;
+
+    my @filters;
+    if ( !$explicit_version ) {
+        push @filters, { not => { term => { status => 'backpan' } } };
+        if ( !$dev ) {
+            push @filters, { term => { maturity => 'released' } };
+        }
+    }
+
+    my $version_filters = $self->_version_filters($version);
+
+    # filters to be applied to the nested modules
+    my $module_f = {
+        nested => {
+            path       => 'module',
+            inner_hits => { _source => 'version' },
+            filter     => {
+                bool => {
+                    must => [
+                        { term => { 'module.authorized' => 1 } },
+                        { term => { 'module.indexed'    => 1 } },
+                        { term => { 'module.name'       => $module } },
+                        (
+                            exists $version_filters->{must}
+                            ? @{ $version_filters->{must} }
+                            : ()
+                        )
+                    ],
+                    (
+                        exists $version_filters->{must_not}
+                        ? ( must_not => [ $version_filters->{must_not} ] )
+                        : ()
+                    )
+                }
+            }
+        }
+    };
+
+    my $filter
+        = @filters
+        ? { bool => { must => [ @filters, $module_f ] } }
+        : $module_f;
+
+    # sort by score, then version desc, then date desc
+    my @sort = (
+        '_score',
+        {
+            'module.version_numified' => {
+                mode          => 'max',
+                order         => 'desc',
+                nested_path   => 'module',
+                nested_filter => $module_f->{nested}{filter}
+            }
+        },
+        { date => { order => 'desc' } }
+    );
+
+    my $query;
+
+    if ($dev) {
+        $query = { filtered => { filter => $filter } };
+    }
+    else {
+        # if not dev, then prefer latest > cpan > backpan
+        $query = {
+            function_score => {
+                filter     => $filter,
+                score_mode => 'first',
+                boost_mode => 'replace',
+                functions  => [
+                    {
+                        filter => { term => { status => 'latest' } },
+                        weight => 3
+                    },
+                    {
+                        filter => { term => { status => 'cpan' } },
+                        weight => 2
+                    },
+                    { filter => { match_all => {} }, weight => 1 },
+                ]
+            }
+        };
+    }
+
+    my $body = {
+        query => $query,
+        size => 1,
+        sort => \@sort,
+        _source => [ 'release', 'download_url', 'date', 'status' ],
+        search_type => 'dfs_query_then_fetch',
+    };
+
+    my $ret = $self->es->search(
+        index => $self->index_name,
+        type  => 'file',
+        body  => $body,
+    );
+
+    return unless $res->{hits}{total};
+
+    my @checksums;
+
+    my $hit     = $res->{hits}{hits}[0];
+    my $release = exists $hit->{_source} ? $hit->{_source}{release} : undef;
+
+    if ($release) {
+        my $checksums = $self->get_checksums($release);
+        @checksums = (
+            (
+                $checksums->{checksum_md5}
+                ? ( checksum_md5 => $checksums->{checksum_md5} )
+                : ()
+            ),
+            (
+                $checksums->{checksum_sha256}
+                ? ( checksum_sha256 => $checksums->{checksum_sha256} )
+                : ()
+            ),
+        );
+    }
+
+    return +{
+        %{ $hit->{_source} },
+        %{ $hit->{inner_hits}{module}{hits}{hits}[0]{_source} }, @checksums,
+    };
+}
+
+sub _version_filters {
+    my ( $self, $version ) = @_;
+
+    return () unless $version;
+
+    if ( $version =~ s/^==\s*// ) {
+        return +{
+            must => [ {
+                term => {
+                    'module.version_numified' => $self->_numify($version)
+                }
+            } ]
+        };
+    }
+    elsif ( $version =~ /^[<>!]=?\s*/ ) {
+        my %ops = qw(< lt <= lte > gt >= gte);
+        my ( %filters, %range, @exclusion );
+        my @requirements = split /,\s*/, $version;
+        for my $r (@requirements) {
+            if ( $r =~ s/^([<>]=?)\s*// ) {
+                $range{ $ops{$1} } = $self->_numify($r);
+            }
+            elsif ( $r =~ s/\!=\s*// ) {
+                push @exclusion, $self->_numify($r);
+            }
+        }
+
+        if ( keys %range ) {
+            $filters{must}
+                = [ { range => { 'module.version_numified' => \%range } } ];
+        }
+
+        if (@exclusion) {
+            $filters{must_not} = [];
+            push @{ $filters{must_not} }, map {
+                +{
+                    term => {
+                        'module.version_numified' => $self->_numify($_)
+                    }
+                }
+            } @exclusion;
+        }
+
+        return \%filters;
+    }
+    elsif ( $version !~ /\s/ ) {
+        return +{
+            must => [ {
+                range => {
+                    'module.version_numified' =>
+                        { 'gte' => $self->_numify($version) }
+                },
+            } ]
+        };
+    }
+}
+
+sub _numify {
+    my ( $self, $ver ) = @_;
+    $ver =~ s/_//g;
+    version->new($ver)->numify;
+}
+
 __PACKAGE__->meta->make_immutable;
 1;

From 804b967250750f910071b73fd57827186f2fedd2 Mon Sep 17 00:00:00 2001
From: Graham Knop <haarg@haarg.org>
Date: Thu, 25 Apr 2024 13:04:10 +0100
Subject: [PATCH 2/3] extend find_download_url to be able to query dists rather
 than modules

---
 lib/MetaCPAN/Query/Release.pm                 | 134 ++++++++++--------
 .../Server/Controller/Search/DownloadURL.pm   |   4 +-
 2 files changed, 76 insertions(+), 62 deletions(-)

diff --git a/lib/MetaCPAN/Query/Release.pm b/lib/MetaCPAN/Query/Release.pm
index 77b189e6a..70e443859 100644
--- a/lib/MetaCPAN/Query/Release.pm
+++ b/lib/MetaCPAN/Query/Release.pm
@@ -1012,7 +1012,7 @@ cpanm Foo~<2
 cpanm --dev Foo~<2
 => status: -backpan, module.version_numified: lt: 2, sort_by: status,version_numified,date
 
-    $release->find_download_url( 'Foo', { version => $version, dev => 0|1 });
+    $release->find_download_url( 'module', 'Foo', { version => $version, dev => 0|1 });
 
 Sorting:
 
@@ -1029,7 +1029,7 @@ Sorting:
 =cut
 
 sub find_download_url {
-    my ( $self, $module, $args ) = @_;
+    my ( $self, $type, $module, $args ) = @_;
     $args ||= {};
 
     my $dev              = $args->{dev};
@@ -1037,6 +1037,11 @@ sub find_download_url {
     my $explicit_version = $version && $version =~ /==/;
 
     my @filters;
+
+    die 'type must be module or dist'
+        unless $type eq 'module' || $type eq 'dist';
+    my $module_filter = $type eq 'module';
+
     if ( !$explicit_version ) {
         push @filters, { not => { term => { status => 'backpan' } } };
         if ( !$dev ) {
@@ -1044,53 +1049,65 @@ sub find_download_url {
         }
     }
 
-    my $version_filters = $self->_version_filters($version);
+    my $prefix = $module_filter ? 'module.' : '';
+
+    my $version_filters
+        = $self->_version_filters( $version, $prefix . 'version_numified' );
+
+    my $entity_filter = {
+        bool => {
+            must => [
+                { term => { $prefix . 'authorized' => 1 } },
+                { term => { $prefix . 'indexed'    => 1 } },
+                { term => { $prefix . 'name'       => $module } },
+                (
+                    exists $version_filters->{must}
+                    ? @{ $version_filters->{must} }
+                    : ()
+                )
+            ],
+            (
+                exists $version_filters->{must_not}
+                ? ( must_not => [ $version_filters->{must_not} ] )
+                : ()
+            )
+        }
+    };
 
     # filters to be applied to the nested modules
-    my $module_f = {
-        nested => {
-            path       => 'module',
-            inner_hits => { _source => 'version' },
-            filter     => {
-                bool => {
-                    must => [
-                        { term => { 'module.authorized' => 1 } },
-                        { term => { 'module.indexed'    => 1 } },
-                        { term => { 'module.name'       => $module } },
-                        (
-                            exists $version_filters->{must}
-                            ? @{ $version_filters->{must} }
-                            : ()
-                        )
-                    ],
-                    (
-                        exists $version_filters->{must_not}
-                        ? ( must_not => [ $version_filters->{must_not} ] )
-                        : ()
-                    )
-                }
+    if ($module_filter) {
+        push @filters,
+            {
+            nested => {
+                path       => 'module',
+                inner_hits => { _source => 'version' },
+                filter     => $entity_filter,
             }
-        }
-    };
+            };
+    }
+    else {
+        push @filters, $entity_filter;
+    }
 
     my $filter
         = @filters
-        ? { bool => { must => [ @filters, $module_f ] } }
-        : $module_f;
+        ? { bool => { must => \@filters } }
+        : $filters[0];
+
+    my $version_sort
+        = $module_filter
+        ? {
+        'module.version_numified' => {
+            mode          => 'max',
+            order         => 'desc',
+            nested_path   => 'module',
+            nested_filter => $entity_filter,
+        }
+        }
+        : { version_numified => { order => 'desc' } };
 
     # sort by score, then version desc, then date desc
-    my @sort = (
-        '_score',
-        {
-            'module.version_numified' => {
-                mode          => 'max',
-                order         => 'desc',
-                nested_path   => 'module',
-                nested_filter => $module_f->{nested}{filter}
-            }
-        },
-        { date => { order => 'desc' } }
-    );
+    my @sort = ( '_score', $version_sort, { date => { order => 'desc' } } );
 
     my $query;
 
@@ -1120,17 +1137,17 @@ sub find_download_url {
     }
 
     my $body = {
-        query => $query,
-        size => 1,
-        sort => \@sort,
+        query   => $query,
+        size    => 1,
+        sort    => \@sort,
         _source => [ 'release', 'download_url', 'date', 'status' ],
-        search_type => 'dfs_query_then_fetch',
     };
 
-    my $ret = $self->es->search(
-        index => $self->index_name,
-        type  => 'file',
-        body  => $body,
+    my $res = $self->es->search(
+        index       => $self->index_name,
+        type        => $module_filter ? 'file' : 'release',
+        body        => $body,
+        search_type => 'dfs_query_then_fetch',
     );
 
     return unless $res->{hits}{total};
@@ -1163,7 +1180,7 @@ sub find_download_url {
 }
 
 sub _version_filters {
-    my ( $self, $version ) = @_;
+    my ( $self, $version, $field ) = @_;
 
     return () unless $version;
 
@@ -1171,7 +1188,7 @@ sub _version_filters {
         return +{
             must => [ {
                 term => {
-                    'module.version_numified' => $self->_numify($version)
+                    $field => $self->_numify($version)
                 }
             } ]
         };
@@ -1191,18 +1208,14 @@ sub _version_filters {
 
         if ( keys %range ) {
             $filters{must}
-                = [ { range => { 'module.version_numified' => \%range } } ];
+                = [ { range => { $field => \%range } } ];
         }
 
         if (@exclusion) {
             $filters{must_not} = [];
-            push @{ $filters{must_not} }, map {
-                +{
-                    term => {
-                        'module.version_numified' => $self->_numify($_)
-                    }
-                }
-            } @exclusion;
+            push @{ $filters{must_not} },
+                map { +{ term => { $field => $self->_numify($_) } } }
+                @exclusion;
         }
 
         return \%filters;
@@ -1211,8 +1224,7 @@ sub _version_filters {
         return +{
             must => [ {
                 range => {
-                    'module.version_numified' =>
-                        { 'gte' => $self->_numify($version) }
+                    $field => { 'gte' => $self->_numify($version) }
                 },
             } ]
         };
diff --git a/lib/MetaCPAN/Server/Controller/Search/DownloadURL.pm b/lib/MetaCPAN/Server/Controller/Search/DownloadURL.pm
index 9ce7c8f2d..f52e3b92d 100644
--- a/lib/MetaCPAN/Server/Controller/Search/DownloadURL.pm
+++ b/lib/MetaCPAN/Server/Controller/Search/DownloadURL.pm
@@ -13,8 +13,10 @@ has '+type' => ( default => 'file' );
 
 sub get : Local : Path('/download_url') : Args(1) {
     my ( $self, $c, $module ) = @_;
+    my $type = 'module';
     my $data
-        = $self->model($c)->find_download_url( $module, $c->req->params );
+        = $self->model($c)
+        ->find_download_url( $type, $module, $c->req->params );
     return $c->detach( '/not_found', [] ) unless $data;
     $c->stash($data);
 }

From 0d850ff7abaead7dd4e7ac3785c849a4fd364401 Mon Sep 17 00:00:00 2001
From: Graham Knop <haarg@haarg.org>
Date: Thu, 25 Apr 2024 13:06:25 +0100
Subject: [PATCH 3/3] download_url end point should query releases for "perl"

---
 lib/MetaCPAN/Server/Controller/Search/DownloadURL.pm | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/lib/MetaCPAN/Server/Controller/Search/DownloadURL.pm b/lib/MetaCPAN/Server/Controller/Search/DownloadURL.pm
index f52e3b92d..1b16ab609 100644
--- a/lib/MetaCPAN/Server/Controller/Search/DownloadURL.pm
+++ b/lib/MetaCPAN/Server/Controller/Search/DownloadURL.pm
@@ -13,7 +13,7 @@ has '+type' => ( default => 'file' );
 
 sub get : Local : Path('/download_url') : Args(1) {
     my ( $self, $c, $module ) = @_;
-    my $type = 'module';
+    my $type = $module eq 'perl' ? 'dist' : 'module';
     my $data
         = $self->model($c)
         ->find_download_url( $type, $module, $c->req->params );