From bce1cc2d4f287e925c81dc4d42009b7f8ced0d71 Mon Sep 17 00:00:00 2001
From: Ilya Shakhat <ishakhat@mirantis.com>
Date: Thu, 31 Oct 2013 20:29:51 +0400
Subject: [PATCH] Project 'openstack/openstack' is excluded from analysis

That project has all core projects as sub-modules. As result commits
are assigned incorrectly and all emails are assigned to 'openstack/openstack'.

The change introduces parameter 'exclude' for project source.

Resolves bug 1246785

Change-Id: I35a004382b187e4322e65f898177f91deced1dac
---
 etc/default_data.json                            | 3 ++-
 etc/default_data.schema.json                     | 6 ++++++
 stackalytics/processor/default_data_processor.py | 4 +++-
 3 files changed, 11 insertions(+), 2 deletions(-)

diff --git a/etc/default_data.json b/etc/default_data.json
index ef3ddba58..9ddb61e9b 100644
--- a/etc/default_data.json
+++ b/etc/default_data.json
@@ -5339,7 +5339,8 @@
         {
             "organization": "openstack",
             "project_type": "openstack",
-            "project_group": "other"
+            "project_group": "other",
+            "exclude": ["openstack"]
         },
         {
             "organization": "openstack-dev",
diff --git a/etc/default_data.schema.json b/etc/default_data.schema.json
index 3b52e54cf..ad1e17065 100644
--- a/etc/default_data.schema.json
+++ b/etc/default_data.schema.json
@@ -143,6 +143,12 @@
                     },
                     "project_group": {
                         "type": ["string", "null"]
+                    },
+                    "exclude": {
+                        "type": "array",
+                        "items": {
+                            "type": "string"
+                        }
                     }
                 },
                 "required": ["organization", "project_type"],
diff --git a/stackalytics/processor/default_data_processor.py b/stackalytics/processor/default_data_processor.py
index a0297fc3c..0dd430cdf 100644
--- a/stackalytics/processor/default_data_processor.py
+++ b/stackalytics/processor/default_data_processor.py
@@ -62,11 +62,13 @@ def _retrieve_project_list(default_data):
             LOG.warn('Fail to retrieve list of projects. Keep it unmodified')
             return False
 
+        exclude = set(project_source.get('exclude', []))
+
         for repo in repos:
             repo_uri = repo.git_url
             repo_name = repo.name
 
-            if repo_uri not in repo_index:
+            if (repo_uri not in repo_index) and (repo_name not in exclude):
                 r = {
                     'branches': ['master'],
                     'module': repo_name,