Add methods for viewing stored builds

2 files changed, 40 insertions, 2 deletions

diff --git a/morph-cache-server b/morph-cache-server
index 61c1640..32c57a5 100755
--- a/morph-cache-server
+++ b/morph-cache-server
@@ -399,9 +399,12 @@ class MorphCacheServer(cliapp.Application):
 
         @app.get('/builds')
         def get_builds():
-            '''Return info on all known builds of a given artifact.'''
+            '''Return info on all known builds.'''
+            return {'builds': db.view_artifact_statistics()}
 
-            cache_name = self._unescape_parameter(request.query.cache_name)
+        @app.get('/builds/<cache_name>')
+        def get_builds_for_artifact(cache_name):
+            '''Return info on builds of a given artifact.'''
             results = sorted(db.iter_builds_for_artifact_file(cache_name))
 
             if len(results) == 0:
diff --git a/morphcacheserver/artifact_database.py b/morphcacheserver/artifact_database.py
index f893b49..3745a61 100644
--- a/morphcacheserver/artifact_database.py
+++ b/morphcacheserver/artifact_database.py
@@ -140,3 +140,38 @@ class ArtifactDatabase(object):
                 'build_datetime': build_datetime,
                 'hash_sha1': hash_sha1
             }
+
+    def view_artifact_statistics(self, start=0, page_size=50):
+        '''Return information on number of builds for each artifact.
+
+        Returns a dict.
+
+        '''
+        n_builds_per_artifact_sql = '''
+        SELECT builds.cache_name,
+            COUNT(builds.hash_sha1) as n_builds,
+            COUNT(DISTINCT builds.hash_sha1) as n_different_builds
+        FROM artifact_files INNER JOIN builds
+            ON builds.cache_name = artifact_files.cache_name
+        GROUP BY builds.cache_name
+        '''
+
+        # Return the artifacts with the highest number of mismatched builds
+        # first, they are the most important. Sort order should be configurable,
+        # really.
+        sql = n_builds_per_artifact_sql + ' ORDER BY n_different_builds DESC'
+
+        sql += ' LIMIT %i OFFSET %i' % (page_size, start)
+
+        cursor = self.db.cursor()
+        logging.debug('Running: %s', sql)
+        cursor.execute(sql)
+
+        result = []
+        for row in cursor:
+            result.append({
+                'cache_name': row[0],
+                'n_builds': row[1],
+                'n_different_builds': row[2],
+            })
+        return result