gci: Fetch old GCI data from last deploy

The cleansed data can be fetched from the production
website if the origin remote is set to be production
repository, as it is on pull requests for Travis CI.
This does not yet allow branch builds on forks,
as their origin will not be the production
repository.

Related to coala#3

.ci/build.sh

@@ -4,8 +4,12 @@ set -e -x
 
 mkdir private _site public
 
-python manage.py fetch_gci_task_data private
-python manage.py cleanse_gci_task_data private _site
+if [[ -n "$GCI_TOKEN" ]]; then
+  python manage.py fetch_gci_task_data private
+  python manage.py cleanse_gci_task_data private _site
+else
+  python manage.py fetch_old_gci_task_data _site || true
+fi
 
 python manage.py collectstatic --noinput
 python manage.py distill-local public --force

community/git.py

@@ -66,3 +66,19 @@ def get_owner():
     """
     url = get_remote_url()
     return url.owner
+
+
+def get_deploy_url():
+    """Obtain the http where deploys appear.
+    """
+    # Use environment variable URL when Netlify detected
+    if os.environ.get('REPOSITORY_URL'):
+        return os.environ.get('URL')
+
+    url = get_remote_url()
+    if url.resource != 'github.com':
+        raise Exception('remotes %s is not supported' % url)
+
+    deploy_url = 'https://{url.owner}.github.io/{url.name}'.format(url=url)
+
+    return deploy_url

gci/management/commands/fetch_old_gci_task_data.py

@@ -0,0 +1,29 @@
+import requests
+
+from django.core.management.base import BaseCommand
+
+from community.git import get_deploy_url
+
+
+class Command(BaseCommand):
+    help = 'Fetch old GCI data'
+
+    def add_arguments(self, parser):
+        parser.add_argument('output_dir', nargs='?', type=str)
+
+    def handle(self, *args, **options):
+        output_dir = options.get('output_dir')
+
+        deploy_url = get_deploy_url()
+
+        r = requests.get(deploy_url + '/static/tasks.yaml')
+        r.raise_for_status()
+
+        with open(os.path.join(output_dir, 'tasks.yaml'), 'w') as f:
+            yaml.dump(r.content, f)
+
+        r = requests.get(deploy_url + '/static/instances.yaml')
+        r.raise_for_status()
+
+        with open(os.path.join(output_dir, 'instances.yaml'), 'w') as f:
+            yaml.dump(r.content, f)