When collecting lead data collect include subproject leads

1 files changed, 25 insertions, 4 deletions

diff --git a/lib/tasks/prepare.rake b/lib/tasks/prepare.rake
index d9a90fc..f45f0dc 100644
--- a/lib/tasks/prepare.rake
+++ b/lib/tasks/prepare.rake
@@ -39,13 +39,34 @@ namespace :prepare do
     # Fetch xml document from uri. Collect all elements with name equal to
     # name parameter and array containing them (and only them).
     def get_all_tags_with_name(uri, name)
-      raw_data      = Net::HTTP.get_response(URI.parse(uri)).body
-      project_data  = REXML::Document.new(raw_data)
-      REXML::XPath.match(project_data, "//#{name}")
+      begin
+        raw_data      = Net::HTTP.get_response(URI.parse(uri)).body
+        project_data  = REXML::Document.new(raw_data)
+        REXML::XPath.match(project_data, "//#{name}")
+      rescue
+        # Warn if there was some error and return empty array.
+        # This way problems with one document won't break whole process.
+        # Down side is that leads may be marked as non-leads (this
+        # wouldn't happen if task crashed when encountering a problem)
+        Rails.logger.error "Error when trying to collect <#{name}> tags from #{uri}:"
+        Rails.logger.error $!
+        []
+      end
     end
 
     devs      = []
-    projects  = get_all_tags_with_name('http://www.gentoo.org/proj/en/metastructure/gentoo.xml?passthru=1', 'subproject')
+
+    projects = []
+    subprojects = get_all_tags_with_name('http://www.gentoo.org/proj/en/metastructure/gentoo.xml?passthru=1', 'subproject')
+    until subprojects.empty?
+      projects += subprojects
+
+      new_subprojects = subprojects.collect do |proj|
+        get_all_tags_with_name("http://www.gentoo.org#{proj.attribute('ref').to_s}?passthru=1", 'subproject')
+      end
+
+      subprojects = new_subprojects.flatten
+    end
 
     for proj in projects
       devs += get_all_tags_with_name("http://www.gentoo.org#{proj.attribute('ref').to_s}?passthru=1", 'dev')