From 35132ef3567b569aeb9bc0355983600bbb4e08a6 Mon Sep 17 00:00:00 2001
From: franck cuny <franck@lumberjaph.net>
Date: Sun, 31 Jan 2010 21:41:05 +0100
Subject: don't check follow(ers|ing) more than once

---
 lib/githubexplorer/Profile.pm                | 71 ++++++++++++++++------------
 lib/githubexplorer/Schema/Result/Profiles.pm |  3 +-
 2 files changed, 44 insertions(+), 30 deletions(-)

(limited to 'lib')

diff --git a/lib/githubexplorer/Profile.pm b/lib/githubexplorer/Profile.pm
index 1f38452..94fe040 100644
--- a/lib/githubexplorer/Profile.pm
+++ b/lib/githubexplorer/Profile.pm
@@ -9,13 +9,11 @@ has banned_profiles =>
 sub fetch_profile {
     my ( $self, $login, $depth ) = @_;
 
-    return if grep {$_ =~ /$login/i} @{$self->banned_profiles};
-    return if $depth > 2;
+    return if grep { $_ =~ /$login/i } @{ $self->banned_profiles };
+    return if $depth > 3;
 
     my $profile = $self->_profile_exists($login);
 
-    return $profile if $profile;
-
     say "fetch profile for $login ($depth)...";
     my $github = Net::GitHub::V2::Users->new(
         owner => $login,
@@ -25,46 +23,61 @@ sub fetch_profile {
     sleep(1);
 
     if ( !$profile ) {
+        my $desc = $github->show;
+        if (!$desc || ($desc && exists $desc->{error})) {
+            sleep(60);
+            $self->fetch_profile($login, $depth);
+        }
         $profile = $self->_create_profile( $login, $github->show, $depth );
-        sleep(1);
+        sleep(2);
         if ( $self->with_repo ) {
             $self->fetch_repositories( $profile, $github->list );
         }
     }
 
-    my $followers   = $github->followers();
+    my $followers = $github->followers();
     sleep(1);
     my $following   = $github->following();
+    sleep(1);
     my $local_depth = $depth + 1;
 
-    foreach my $f (@$followers) {
-        my $p = $self->fetch_profile( $f, $local_depth );
-        next unless $p;
-        $self->schema->txn_do(
-            sub {
-                $self->schema->resultset('Follow')
-                    ->find_or_create(
-                    { id_following => $profile->id, id_follower => $p->id } );
-            }
-        );
-    }
+    unless ( $profile->done ) {
+        foreach my $f (@$followers) {
+            my $p = $self->fetch_profile( $f, $local_depth );
+            next unless $p;
+            $self->schema->txn_do(
+                sub {
+                    $self->schema->resultset('Follow')->find_or_create(
+                        {
+                            id_following => $profile->id,
+                            id_follower  => $p->id
+                        }
+                    );
+                }
+            );
+        }
 
-    foreach my $f (@$following) {
-        my $p = $self->fetch_profile( $f, $local_depth );
-        next unless $p;
-        $self->schema->txn_do(
-            sub {
-                $self->schema->resultset('Follow')
-                    ->find_or_create(
-                    { id_following => $p->id, id_follower => $profile->id } );
-            },
-
-        );
+        foreach my $f (@$following) {
+            my $p = $self->fetch_profile( $f, $local_depth );
+            next unless $p;
+            $self->schema->txn_do(
+                sub {
+                    $self->schema->resultset('Follow')->find_or_create(
+                        {
+                            id_following => $p->id,
+                            id_follower  => $profile->id
+                        }
+                    );
+                },
+
+            );
+        }
+        $profile->update( { done => 1 } );
     }
+    sleep(1);
     $profile;
 }
 
-
 sub _profile_exists {
     my ( $self, $login ) = @_;
     my $profile
diff --git a/lib/githubexplorer/Schema/Result/Profiles.pm b/lib/githubexplorer/Schema/Result/Profiles.pm
index 376103d..e0349d7 100644
--- a/lib/githubexplorer/Schema/Result/Profiles.pm
+++ b/lib/githubexplorer/Schema/Result/Profiles.pm
@@ -18,7 +18,8 @@ __PACKAGE__->add_columns(
     name              => { data_type => 'varchar', is_nullable => 1 },
     public_gist_count => { data_type => 'int' },
     public_repo_count => { data_type => 'int' },
-    depth             => { data_type => 'boolean' },
+    depth             => { data_type => 'int' },
+    done              => { data_type => 'boolean', default_value => 0 },
     perl_total_bytes =>
         { data_type => 'int', is_nullable => 1, default_value => 0 },
 );
-- 
cgit 1.4.1