Check in diffs for the test db. Try to make the test db identifiers predictable.

initdb.py

@@ -1,26 +1,42 @@
 import logging
 import string
+import shutil
+import os
+import hashlib
 
-from random import SystemRandom
 from datetime import datetime
 
+import storage
+
 from data.database import initialize_db
 from data import model
 from app import app
 
 
 logger = logging.getLogger(__name__)
+store = storage.load()
 logging.basicConfig(**app.config['LOGGING_CONFIG'])
 
 
-def __gen_hex_id(length=64):
-  random = SystemRandom()
-  return ''.join([random.choice('abcdef' + string.digits)
-                  for x in range(length)])
+SAMPLE_DIFFS = [
+  'test/data/sample/diffs/diffs.json',
+  'test/data/sample/diffs/diffs2.json',
+  'test/data/sample/diffs/diffs3.json',
+]
 
 
-def __gen_checksum():
-  return 'tarsum+sha256:' + __gen_hex_id(64)
+def __gen_checksum(image_id):
+  h = hashlib.md5(image_id)
+  return 'tarsum+sha256:' + h.hexdigest()
+
+
+global_image_num = [0]
+def __gen_image_id(repo):
+  str_to_hash = "%s/%s/%s" % (repo.namespace, repo.name, global_image_num[0])
+  global_image_num[0] += 1
+
+  h = hashlib.md5(str_to_hash)
+  return h.hexdigest()
 
 
 def create_subtree(repo, structure, parent):
@@ -28,8 +44,9 @@ def create_subtree(repo, structure, parent):
 
   # create the nodes
   for i in range(num_nodes):
-    docker_image_id = __gen_hex_id()
-    checksum = __gen_checksum()
+    docker_image_id = __gen_image_id(repo)
+    logger.debug('new docker id: %s' % docker_image_id)
+    checksum = __gen_checksum(docker_image_id)
 
     new_image = model.create_image(docker_image_id, repo)
     model.set_image_checksum(docker_image_id, repo, checksum)
@@ -38,6 +55,14 @@ def create_subtree(repo, structure, parent):
                                          repo.name, str(datetime.now()),
                                          'no comment', parent)
 
+    # Populate the diff file
+    diff_path = store.image_file_diffs_path(repo.namespace, repo.name,
+                                          docker_image_id)
+    source_diff = SAMPLE_DIFFS[i % len(SAMPLE_DIFFS)]
+
+    with open(source_diff, 'r') as source_file:
+      store.stream_write(diff_path, source_file)
+
     parent = new_image
 
   if last_node_tags: