import github::collector

Signed-off-by: franck cuny <franck@lumberjaph.net>

6 files changed, 265 insertions, 0 deletions

diff --git a/lib/GitHub/Collector/Role/Graph/Edges.pm b/lib/GitHub/Collector/Role/Graph/Edges.pm
new file mode 100644
index 0000000..bea5914
--- /dev/null
+++ b/lib/GitHub/Collector/Role/Graph/Edges.pm
@@ -0,0 +1,13 @@
+package GitHub::Collector::Role::Graph::Edges;
+
+use Moose::Role;
+
+has edges => (
+    is         => 'rw',
+    isa        => 'HashRef',
+    lazy       => 1,
+    default    => sub { {} },
+    auto_deref => 1,
+);
+
+1;
diff --git a/lib/GitHub/Collector/Role/Graph/Gexf.pm b/lib/GitHub/Collector/Role/Graph/Gexf.pm
new file mode 100644
index 0000000..ab83dd4
--- /dev/null
+++ b/lib/GitHub/Collector/Role/Graph/Gexf.pm
@@ -0,0 +1,47 @@
+package GitHub::Collector::Role::Graph::Gexf;
+
+use Moose::Role;
+use Graph::GEXF;
+
+has output => (
+    is        => 'ro',
+    isa       => 'Int',
+    predicate => 'should_export',
+);
+
+sub export {
+    my ($self, ) = @_;
+
+    my $gexf = Graph::GEXF->new();
+    $gexf->add_node_attribute( name    => 'string' );
+    $gexf->add_node_attribute( lang    => 'string' );
+    $gexf->add_node_attribute( size    => 'int' );
+    $gexf->add_node_attribute( country => 'string' );
+    $gexf->add_node_attribute( indegree => 'int' );
+
+    my $nodes = {};
+    foreach my $node ( keys %{ $self->nodes } ) {
+        my $n = $gexf->add_node( $self->nodes->{$node}->{id} );
+        $n->label($node);
+        $n->attribute( name     => $node );
+        $n->attribute( size     => $self->nodes->{$node}->{size} );
+        $n->attribute( lang     => $self->nodes->{$node}->{language} || '' );
+        $n->attribute( country  => $self->nodes->{$node}->{country} || '' );
+        $n->attribute( indegree => $self->nodes->{$node}->{indegree} );
+        $nodes->{$node} = $n;
+    }
+
+    foreach my $edge (keys %{$self->edges}){
+        $nodes->{ $self->edges->{$edge}->{sourceId} }->link_to(
+            {
+                target => $self->edges->{$edge}->{targetId},
+                weight => $self->edges->{$edge}->{weight}
+            }
+        );
+    }
+
+    my $xml = $gexf->to_xml();
+    print $xml;
+}
+
+1;
diff --git a/lib/GitHub/Collector/Role/Graph/Neighbors.pm b/lib/GitHub/Collector/Role/Graph/Neighbors.pm
new file mode 100644
index 0000000..627de94
--- /dev/null
+++ b/lib/GitHub/Collector/Role/Graph/Neighbors.pm
@@ -0,0 +1,128 @@
+package GitHub::Collector::Role::Graph::Neighbors;
+
+use Moose::Role;
+with qw(GitHub::Collector::Role::Languages);
+
+sub neighbors {
+    my ( $self, $name, $with_connections ) = @_;
+
+    $self->_neighbors($name);
+
+    if ($with_connections) {
+        foreach my $id ( keys %{ $self->edges } ) {
+            $self->_connections( $id, $name );
+        }
+    }
+}
+
+sub _neighbors {
+    my ( $self, $name ) = @_;
+
+    if ( !defined $self->nodes->{$name} ) {
+        $self->_create_node($name);
+    }
+
+    $self->_fetch_edges_from($name);
+    $self->_fetch_edges_to($name);
+}
+
+sub _connections {
+    my ( $self, $id, $name ) = @_;
+
+    if ( $self->edges->{$id}->{sourceId} eq $name ) {
+        $self->_neighbors( $self->edges->{$id}->{targetId} );
+    }
+    else {
+        $self->_neighbors( $self->edges->{$id}->{sourceId} );
+    }
+}
+
+sub remove_leaves {
+    my $self = shift;
+
+    foreach my $id ( keys %{$self->nodes} ) {
+        if ( $self->nodes->{$id}->{size} < 2 ) {
+            delete $self->nodes->{$id};
+        }
+    }
+
+    foreach my $id ( keys %{$self->edges} ) {
+        unless ( $self->nodes->{ $self->edges->{$id}->{sourceId} }
+            && $self->nodes->{ $self->edges->{$id}->{targetId} } )
+        {
+            delete $self->edges->{ $id };
+        }
+    }
+}
+
+sub _fetch_edges_from {
+    my ( $self, $name ) = @_;
+
+    my $connections = $self->db_edges->find( { source => $name } );
+
+    while ( my $edge = $connections->next ) {
+        my $edge_id = $edge->{source} . $edge->{target};
+
+        if ( !defined $self->edges->{$edge_id} ) {
+            $self->edges->{$edge_id} = {
+                id       => $edge_id,
+                targetId => $edge->{target},
+                sourceId => $name,
+                weight   => $edge->{weight},
+            };
+
+            $self->nodes->{$name}->{size}++;
+
+            if ( defined $self->nodes->{ $edge->{target} } ) {
+                $self->nodes->{ $edge->{target} }->{size}++;
+            }
+            else {
+                $self->_create_node( $edge->{target}, 1 );
+            }
+        }
+    }
+}
+
+sub _fetch_edges_to {
+    my ($self, $name) = @_;
+    my $connections = $self->db_edges->find( { target => $name } );
+    while ( my $edge = $connections->next ) {
+        my $edge_id = $edge->{source} . $edge->{target};
+
+        if ( !defined $self->edges->{$edge_id} ) {
+            $self->edges->{$edge_id} = {
+                id       => $edge_id,
+                targetId => $name,
+                sourceId => $edge->{source},
+                weight   => $edge->{weight},
+            };
+
+            $self->nodes->{$name}->{size}++;
+
+            if ( defined $self->nodes->{ $edge->{source} } ) {
+                $self->nodes->{ $edge->{source} }->{size}++;
+            }
+            else {
+                $self->_create_node( $edge->{source}, 1 );
+            }
+        }
+    }
+}
+
+sub _create_node {
+    my ( $self, $login, $size ) = @_;
+
+    $size ||= 0;
+
+    my $info = $self->db_profiles->find_one( { login => $login } );
+    $self->nodes->{$login} = {
+        id       => $login,
+        label    => $login,
+        size     => $size,
+        language => $self->map_languages($info->{language}),
+        country  => $info->{country} || '',
+        indegree => $info->{indegree},
+    };
+}
+
+1;
diff --git a/lib/GitHub/Collector/Role/Graph/Nodes.pm b/lib/GitHub/Collector/Role/Graph/Nodes.pm
new file mode 100644
index 0000000..5109e9a
--- /dev/null
+++ b/lib/GitHub/Collector/Role/Graph/Nodes.pm
@@ -0,0 +1,13 @@
+package GitHub::Collector::Role::Graph::Nodes;
+
+use Moose::Role;
+
+has nodes => (
+    is         => 'rw',
+    isa        => 'HashRef',
+    lazy       => 1,
+    default    => sub { {} },
+    auto_deref => 1,
+);
+
+1;
diff --git a/lib/GitHub/Collector/Role/Graph/Query.pm b/lib/GitHub/Collector/Role/Graph/Query.pm
new file mode 100644
index 0000000..d3ab7f0
--- /dev/null
+++ b/lib/GitHub/Collector/Role/Graph/Query.pm
@@ -0,0 +1,45 @@
+package GitHub::Collector::Role::Graph::Query;
+
+use Moose::Role;
+
+has language => (
+    is        => 'rw',
+    isa       => 'Str',
+    predicate => 'has_language',
+);
+
+has location => (
+    is        => 'rw',
+    isa       => 'Str',
+    predicate => 'has_location',
+);
+
+has company => (
+    is        => 'ro',
+    isa       => 'Str',
+    predicate => 'has_company',
+);
+
+has country => (
+    is        => 'ro',
+    isa       => 'Str',
+    predicate => 'has_country',
+);
+
+
+sub build_query {
+    my $self = shift;
+
+    my $search = {};
+
+    foreach my $attr (qw/language location company country/) {
+        my $predicate = "has_$attr";
+        if ( $self->$predicate ) {
+            $search->{$attr} = $self->$attr;
+        }
+    }
+
+    return $search;
+}
+
+1;
diff --git a/lib/GitHub/Collector/Role/Graph/Search.pm b/lib/GitHub/Collector/Role/Graph/Search.pm
new file mode 100644
index 0000000..fbd0d07
--- /dev/null
+++ b/lib/GitHub/Collector/Role/Graph/Search.pm
@@ -0,0 +1,19 @@
+package GitHub::Collector::Role::Graph::Search;
+
+use Moose::Role;
+
+sub build_from_query {
+    my ($self, $search) = @_;
+
+    $search ||= $self->build_query();
+
+    my $profiles = $self->db_profiles->find($search);
+
+    while ( my $profile = $profiles->next ) {
+       $self->neighbors( $profile->{login}, 0 );
+    }
+
+    $self->remove_leaves();
+}
+
+1;