From 27766d44b05b020db39ccee071e9c76fd7380b46 Mon Sep 17 00:00:00 2001
From: Mark Glines <mark@glines.org>
Date: Sun, 28 Feb 2010 07:16:20 -0500
Subject: [PATCH] Github updates.

* Add some sanity checking to make sure the github feed gave us data from the right project.  (They apparently had a temporary bug, and the consequences were quite spammy.)
* Add a warning message and return semigracefully if a feed no longer exists.
* Move the timer registration down a bit, so github parsers can be tested with modules/local/test.pl.
* The feed format changed ever so slightly (they added a rel attribute to the link tag), add that to the testsuite just in case.
---
 modules/local/githubparser.pm | 16 ++++++++++++++--
 t/rakudolog.t                 | 35 +++++++++++++++++++++++++++++++++--
 2 files changed, 47 insertions(+), 4 deletions(-)

diff --git a/modules/local/githubparser.pm b/modules/local/githubparser.pm
index 9994266..86ac14f 100644
--- a/modules/local/githubparser.pm
+++ b/modules/local/githubparser.pm
@@ -108,6 +108,10 @@ sub process_branch {
         my $atom = XML::Atom::Client->new();
         $feed = $atom->getFeed($$self{branches}{$branch}{url});
     }
+    if(!defined($feed)) {
+        warn "could not fetch branch $branch feed " . $$self{branches}{$branch}{url};
+        return;
+    }
 
     my @items = $feed->entries;
     @items = sort { $a->updated cmp $b->updated } @items; # ascending order
@@ -118,12 +122,15 @@ sub process_branch {
     foreach my $item (@items) {
         my $link    = $item->link->href;
         my ($rev)   = $link =~ m|/commit/([a-z0-9]{40})|;
+        my ($proj)  = $link =~ m|http://github.com/[^/]+/([^/]+)/|;
         if(exists($$self{not_first_time})) {
+            return unless $proj = $$self{project};
             # output new entries to channel
             next if exists($$self{seen}{$rev});
             $$self{seen}{$rev} = 1;
             $self->output_item($item, $branch, $link, $rev);
         } else {
+            die "got bad data from github feed" unless $proj = $$self{project};
             # just populate the seen cache
             $$self{seen}{$rev} = 1;
         }
@@ -206,6 +213,7 @@ sub try_link {
 
     # create project, if necessary
     my $self = $objects_by_package{$modulename};
+    my $register_timer = 0;
     if(!defined($self)) {
         $objects_by_package{$modulename} = $self = {
             project    => $project,
@@ -216,8 +224,7 @@ sub try_link {
         # create a dynamic subclass to get the timer callback back to us
         eval "package $modulename; use base 'modules::local::githubparser';";
         $objects_by_package{$modulename} = bless($self, $modulename);
-        main::create_timer($parsername."_process_project_timer", $modulename,
-            "process_project", 300 + $feed_number++);
+        $register_timer = 1;
         main::lprint("github: created project $project ($modulename)");
     }
 
@@ -244,6 +251,11 @@ sub try_link {
             main::lprint("github: $project/$branchname will output to ".join("/",@$target));
         }
     }
+
+    if($register_timer) {
+        main::create_timer($parsername."_process_project_timer", $modulename,
+            "process_project", 300 + $feed_number++);
+    }
 }
 
 
diff --git a/t/rakudolog.t b/t/rakudolog.t
index 05c50b8..fa4d709 100644
--- a/t/rakudolog.t
+++ b/t/rakudolog.t
@@ -82,10 +82,10 @@ $output = [output()];
 is(scalar @$output, 6, "6 lines of output");
 is($$output[0]{net} , 'magnet'  , "line to magnet/#parrot");
 is($$output[0]{chan}, '#parrot' , "line to magnet/#parrot");
-like($$output[0]{text}, qr|rakudo/master: |, "master branch");
+like($$output[0]{text}, qr|rakudo: |, "master branch");
 is($$output[1]{net} , 'freenode', "line to freenode/#perl6");
 is($$output[1]{chan}, '#perl6'  , "line to freenode/#perl6");
-like($$output[1]{text}, qr|rakudo/master: |, "master branch");
+like($$output[1]{text}, qr|rakudo: |, "master branch");
 BEGIN { $tests += 7 };
 
 # update with multiple commits having the same timestamp
@@ -131,3 +131,34 @@ my @message_list = ($$output[2]{text}, $$output[8]{text});
 is(scalar grep(/Small optimizations/ , @message_list), 1, "log message");
 is(scalar grep(/Add some micro-bench/, @message_list), 1, "log message");
 BEGIN { $tests += 7 };
+
+# update with their post-2010-02-26 feed format (added rel attribute to the <link> tag)
+reset_output();
+$xml_footer = << '__XML__' . $xml_footer;
+  <entry>
+    <id>tag:github.com,2008:Grit::Commit/b131f6052a181bdad8f7b9e5abe30c9c2c2e360e</id>
+    <link type="text/html" href="http://github.com/rakudo/rakudo/commit/b131f6052a181bdad8f7b9e5abe30c9c2c2e360e" rel="alternate"/>
+    <title>Split .defined method and defined vtable method for language interop, as per pmichaud++.</title>
+    <updated>2010-02-27T16:46:53-08:00</updated>
+    <content type="html">&lt;pre&gt;m src/builtins/Parcel.pir
+
+Split .defined method and defined vtable method for language interop, as per pmichaud++.&lt;/pre&gt;</content>
+    <author>
+      <name>Jonathan Worthington</name>
+    </author>
+  </entry>
+__XML__
+$xml = $xml_header . '<updated>2010-02-27T16:46:53-08:00</updated>' . $xml_footer;
+$feed = XML::Atom::Feed->new(\$xml);
+call_func('process_branch', 'master', $feed);
+$output = [output()];
+is(scalar @$output, 6, "6 lines of output");
+is($$output[0]{net} , 'magnet'  , "line to magnet/#parrot");
+is($$output[0]{chan}, '#parrot' , "line to magnet/#parrot");
+is($$output[1]{net} , 'freenode', "line to freenode/#perl6");
+is($$output[1]{chan}, '#perl6'  , "line to freenode/#perl6");
+# The module sorts by <updated> time, but the time is the same for these two commits.
+# Do it this way so we don't depend on perl's internal sort algorithm details.
+@message_list = ($$output[2]{text});
+is(scalar grep(/Split .defined method/ , @message_list), 1, "log message");
+BEGIN { $tests += 6 };