Add 'in_memory' messaging backend and use it during build_module_locally to drop dependency on fedmsg-relay for local builds.

conf/config.py

@@ -125,6 +125,7 @@ class TestConfiguration(BaseConfiguration):
     LOG_LEVEL = 'debug'
     SQLALCHEMY_DATABASE_URI = 'sqlite:///:memory:'
     DEBUG = True
+    MESSAGING = 'in_memory'
 
     # Global network-related values, in seconds
     NET_TIMEOUT = 3

module_build_service/builder.py

@@ -1149,7 +1149,7 @@ $repos
         # what RPMs are output of particular SRPM build yet.
         for artifact in artifacts:
             if artifact.startswith("module-build-macros"):
-                self._execute_cmd(["mock", "-r", self.mock_config, "-i",
+                _execute_cmd(["mock", "-r", self.mock_config, "-i",
                                    "module-build-macros"])
 
     def buildroot_add_repos(self, dependencies):

module_build_service/config.py

@@ -334,7 +334,7 @@ class Config(object):
 
     def _setifok_messaging(self, s):
         s = str(s)
-        if s not in ("fedmsg", "amq"):
+        if s not in ("fedmsg", "amq", "in_memory"):
             raise ValueError("Unsupported messaging system.")
         self.messaging = s
 

module_build_service/manage.py

@@ -117,6 +117,7 @@ def cleardb():
 @manager.command
 def build_module_locally(url):
     conf.set_item("system", "mock")
+    conf.set_item("messaging", "in_memory")
 
     # Use our own local SQLite3 database.
     confdir = os.path.abspath(os.path.dirname(__file__))

module_build_service/messaging.py

@@ -33,7 +33,7 @@ except ImportError:
     from funcsigs import signature
 
 from module_build_service import log
-
+import queue
 
 class BaseMessage(object):
     def __init__(self, msg_id):
@@ -330,6 +330,39 @@ def _amq_publish(topic, msg, conf, service):
     msngr.put(message)
     msngr.send()
 
+# Queue for "in_memory" messaging.
+_in_memory_work_queue = queue.Queue()
+
+# Message id for "in_memory" messaging.
+_in_memory_msg_id = 0
+
+def _in_memory_publish(topic, msg, conf, service):
+    """
+    Puts the message to _in_memory_work_queue".
+    """
+
+    # Increment the message ID.
+    global _in_memory_msg_id
+    _in_memory_msg_id += 1
+
+    # Create fake fedmsg from the message so we can reuse
+    # the BaseMessage.from_fedmsg code to get the particular BaseMessage
+    # class instance.
+    topic = service + "." + topic
+    wrapped_msg = {}
+    wrapped_msg["msg_id"] = str(_in_memory_msg_id)
+    wrapped_msg["msg"] = msg
+    wrapped_msg = BaseMessage.from_fedmsg(topic, wrapped_msg)
+
+    # Put the message to queue.
+    _in_memory_work_queue.put(wrapped_msg)
+
+def _in_memory_listen(conf, **kwargs):
+    """
+    Yields the message from the _in_memory_work_queue when ready.
+    """
+    while True:
+        yield _in_memory_work_queue.get(True)
 
 _messaging_backends = {
     'fedmsg': {
@@ -340,4 +373,8 @@ _messaging_backends = {
         'publish': _amq_publish,
         'listen': _amq_listen,
     },
+    'in_memory': {
+        'publish': _in_memory_publish,
+        'listen': _in_memory_listen,
+    },
 }