linaro-metrics: add logic to find author of PR

Github accounts have no mapping into Linaro's Crowd user database.
However, by looking at the email address of the commit in Git, we can
see if its a linked account.

Change-Id: I9fa151c515cd5ad8bd06bb024aa04d5fad6f473d

1 files changed, 20 insertions, 2 deletions

diff --git a/linaro_metrics/sync_github_changes.py b/linaro_metrics/sync_github_changes.py
index 528f04f..54661ef 100755
--- a/linaro_metrics/sync_github_changes.py
+++ b/linaro_metrics/sync_github_changes.py
@@ -17,6 +17,9 @@ django_setup()  # must be called to get sys.path and django settings in place
 
 from django.conf import settings
 
+from linaro_metrics.crowd import Crowd
+from linaro_metrics.parsemail import get_linaro_person
+
 log = logging.getLogger('sync_github_changes')
 
 GITHUB_REPOS = [
@@ -61,6 +64,16 @@ def get_pull_requests(owner, repo, last_update=None):
                 url = m.group(1)
 
 
+def get_author(crowd, pr):
+    resp = _get(pr['commits_url'])
+    data = json.loads(resp.read())
+    if not len(data):
+        # some PR's have no commits: https://github.com/docker/docker/pull/5894
+        return
+    email = data[0]['commit']['author']['email']
+    return get_linaro_person(crowd, email)
+
+
 @contextlib.contextmanager
 def repo_cache():
     def dt_serialize(obj):
@@ -83,7 +96,9 @@ def repo_cache():
 
 
 def main(args):
-    with repo_cache() as repos:
+    crwd = Crowd(settings.CROWD_USER, settings.CROWD_PASS, settings.CROWD_URL)
+
+    with crwd.cached(settings.CROWD_CACHE), repo_cache() as repos:
         for owner, repo, proj in GITHUB_REPOS:
             repo_path = '%s/%s' % (owner, repo)
             log.info('Looking at: %s', repo_path)
@@ -93,7 +108,10 @@ def main(args):
             try:
                 for pr in get_pull_requests(owner, repo, last_update):
                     x += 1
-                    log.debug('TODO analyze pr: %r', pr)
+                    auth = get_author(crwd, pr)
+                    if auth:
+                        log.debug('checking change: %d', pr['number'])
+                        # TODO check the PR
                 repos[repo_path] = now
             finally:
                 log.info('analayzed %d pull-requests', x)