CMFActivity: Remove remaining transcaions.commit and move BTree to ActiveProcess

product/CMFActivity/ActiveProcess.py

@@ -32,6 +32,7 @@ from Products.CMFCore import permissions as CMFCorePermissions
 from Products.ERP5Type.Base import Base
 from Products.ERP5Type import PropertySheet
 from Products.ERP5Type.ConflictFree import ConflictFreeLog
+from BTrees.LOBTree import LOBTree
 from BTrees.Length import Length
 from random import randrange
 from .ActiveResult import ActiveResult
@@ -85,6 +86,15 @@ class ActiveProcess(Base):
   def __init__(self, *args, **kw):
     Base.__init__(self, *args, **kw)
     self.result_list = ConflictFreeLog()
+    self.use_btree = False
+
+  security.declareProtected(CMFCorePermissions.ManagePortal, 'useBTree')
+  def useBTree(self):
+    # Use BTree instead of Linked List
+    # this is used by joblib Backend to store results in a dictionary with
+    # signature as key
+    self.use_btree = True
+    self.result_list = LOBTree()
 
   security.declareProtected(CMFCorePermissions.ManagePortal, 'postResult')
   def postResult(self, result):
@@ -92,8 +102,19 @@ class ActiveProcess(Base):
       result_list = self.result_list
     except AttributeError:
       # BBB: self was created before implementation of __init__
-      self.result_list = result_list = ConflictFreeLog()
+      if self.use_btree:
+        self.result_list = result_list = LOBTree()
+      else:
+        self.result_list = result_list = ConflictFreeLog()
     else:
+      if self.use_btree:
+        if not hasattr(result, 'sig'):
+          result_id = randrange(0, 10000 * (id(result) + 1))
+        else:
+          result_id = result.sig
+        result_list.insert(result_id, result)
+        return
+
       if type(result_list) is not ConflictFreeLog: # BBB: result_list is IOBTree
         # use a random id in order to store result in a way with
         # fewer conflict errors
@@ -103,7 +124,12 @@ class ActiveProcess(Base):
         result_list[random_id] = result
         self.result_len.change(1)
         return
-    result_list.append(result)
+
+    if self.use_btree:
+      signature = int(result.sig, 16)
+      result_list.insert(signature, result)
+    else:
+      result_list.append(result)
 
   security.declareProtected(CMFCorePermissions.ManagePortal, 'postActiveResult')
   def postActiveResult(self, *args, **kw):
@@ -124,6 +150,18 @@ class ActiveProcess(Base):
       return result_list.values()
     return list(result_list)
 
+  security.declareProtected(CMFCorePermissions.ManagePortal, 'getResult')
+  def getResult(self, key, **kw):
+    """
+      Returns the result with requested key else None
+    """
+    try:
+      result_list = self.result_list
+      result = result_list[key]
+    except:
+      return None
+    return result
+
   security.declareProtected(CMFCorePermissions.ManagePortal, 'activateResult')
   def activateResult(self, result):
     if result not in (None, 0, '', (), []):

product/CMFActivity/Activity/SQLBase.py

@@ -139,7 +139,6 @@ class SQLBase(Queue):
       serialization_tag_list = [m.activity_kw.get('serialization_tag', '')
                                 for m in message_list]
       processing_node_list = []
-
       for m in message_list:
         m.order_validation_text = x = self.getOrderValidationText(m)
         processing_node_list.append(0 if x == 'none' else -1)

product/CMFActivity/Activity/SQLJoblib.py

@@ -26,24 +26,11 @@
 #
 ##############################################################################
 
-# XXX: Note from Rafael
-# only reimplment the minimal, and only custom the SQL that update this table.
-# Always check if things are there (ie.: If the connection, or the script are present).
-
-import copy
-import hashlib
-import sys
 import transaction
 from functools import total_ordering
 from zLOG import LOG, TRACE, INFO, WARNING, ERROR, PANIC
-from zExceptions import ExceptionFormatter
-from ZODB.POSException import ConflictError
 from SQLBase import SQLBase, sort_message_key
 from Products.CMFActivity.ActivityTool import Message
-from Products.CMFActivity.ActivityTool import (
-  Message, MESSAGE_NOT_EXECUTED, MESSAGE_EXECUTED, SkippedMessage)
-from Products.CMFActivity.ActivityRuntimeEnvironment import (
-  DEFAULT_MAX_RETRY, ActivityRuntimeEnvironment, getTransactionalVariable)
 from Queue import Queue, VALIDATION_ERROR_DELAY, VALID, INVALID_PATH
 
 # Stop validating more messages when this limit is reached
@@ -55,46 +42,6 @@ _DequeueMessageException = Exception()
 
 from SQLDict import SQLDict
 
-# this is improvisation of 
-# http://stackoverflow.com/questions/5884066/hashing-a-python-dictionary/8714242#8714242
-def make_hash(o):
-  """
-  Makes a hash from a dictionary, list, tuple or set to any level, that contains
-  only other hashable types (including any lists, tuples, sets, and
-  dictionaries).
-  """
-
-  if isinstance(o, (set, tuple, list)):
-    return hash(tuple([make_hash(e) for e in o]))
-
-  elif not isinstance(o, dict):
-    try:
-      return hash(o)
-    except TypeError:
-      return hash(int(hashlib.md5(o).hexdigest(), 16))
-  new_o = copy.deepcopy(o)
-  for k, v in new_o.items():
-    new_o[k] = make_hash(v)
-
-  return hash(tuple(frozenset(sorted(new_o.items()))))
-
-@total_ordering
-class MyBatchedSignature(object):
-  """Create hashable signature"""
-  def __init__(self, batch):
-    #LOG('CMFActivity', INFO, batch.items)
-    items = batch.items[0]
-    self.func = items[0].__name__
-    self.args = items[1]
-    self.kwargs = items[2]
-
-  def __eq__(self, other):
-    return (self.func, self.args) == (other.func, other.args)
-
-  def __lt__(self, other):
-    return (self.func, self.args) < (other.func, other.args)
-
-
 class SQLJoblib(SQLDict):
   """
     XXX SQLJoblib
@@ -124,7 +71,7 @@ class SQLJoblib(SQLDict):
 
   def register(self, activity_buffer, activity_tool, message):
     """
-      Send message to mysql directly
+      Send messages to mysql directly
     """
     assert not message.is_registered, message
     message.is_registered = True
@@ -137,7 +84,6 @@ class SQLJoblib(SQLDict):
     if m.is_registered:
       uid = portal.portal_ids.generateNewIdList(self.uid_group,
         id_count=1, id_generator='uid')[0]
-      #import pdb; pdb.set_trace()
       m.order_validation_text = x = self.getOrderValidationText(m)
       processing_node = (0 if x == 'none' else -1)
       portal.SQLJoblib_writeMessage(
@@ -151,82 +97,122 @@ class SQLJoblib(SQLDict):
         group_method_id=m.getGroupId(),
         date=m.activity_kw.get('at_date'),
         tag=m.activity_kw.get('tag', ''),
+        signature=m.activity_kw.get('signature', ''),
         processing_node=processing_node,
         serialization_tag=m.activity_kw.get('serialization_tag', ''))
 
-  # Queue semantic
-  def dequeueMessage(self, activity_tool, processing_node):
-    message_list, group_method_id, uid_to_duplicate_uid_list_dict = \
-      self.getProcessableMessageList(activity_tool, processing_node)
-    if message_list:
-      # Remove group_id parameter from group_method_id
-      if group_method_id is not None:
-        group_method_id = group_method_id.split('\0')[0]
-      if group_method_id not in (None, ""):
-        method  = activity_tool.invokeGroup
-        args = (group_method_id, message_list, self.__class__.__name__,
-                hasattr(self, 'generateMessageUID'))
-        activity_runtime_environment = ActivityRuntimeEnvironment(None)
-      else:
-        method = activity_tool.invoke
-        message = message_list[0]
-        args = (message, )
-        activity_runtime_environment = ActivityRuntimeEnvironment(message)
-      # Commit right before executing messages.
-      # As MySQL transaction does not start exactly at the same time as ZODB
-      # transactions but a bit later, messages available might be called
-      # on objects which are not available - or available in an old
-      # version - to ZODB connector.
-      # So all connectors must be committed now that we have selected
-      # everything needed from MySQL to get a fresh view of ZODB objects.
-      transaction.commit()
-      transaction.begin()
-      tv = getTransactionalVariable()
-      tv['activity_runtime_environment'] = activity_runtime_environment
-      # Try to invoke
-      try:
-        method(*args)
-        # Abort if at least 1 message failed. On next tic, only those that
-        # succeeded will be selected because their at_date won't have been
-        # increased.
-        for m in message_list:
-          if m.getExecutionState() == MESSAGE_NOT_EXECUTED:
-            raise _DequeueMessageException
+  def getProcessableMessageLoader(self, activity_tool, processing_node):
+    path_and_method_id_dict = {}
+    def load(line):
+      # getProcessableMessageList already fetch messages with the same
+      # group_method_id, so what remains to be filtered on are path, method_id
+      # and signature
+      path = line.path
+      method_id = line.method_id
+      key = path, method_id
+      uid = line.uid
+      signature = line.signature
+      original_uid = path_and_method_id_dict.get(key)
+      if original_uid is None:
+        m = Message.load(line.message, uid=uid, line=line, signature=signature)
+        try:
+          result = activity_tool.SQLJoblib_selectDuplicatedLineList(
+            path=path,
+            method_id=method_id,
+            group_method_id=line.group_method_id,
+            signature=signature)
+          reserve_uid_list = uid_list = [x.uid for x in result]
+          if reserve_uid_list:
+            activity_tool.SQLJoblib_reserveDuplicatedLineList(
+              processing_node=processing_node, uid=reserve_uid_list)
+        except:
+          self._log(WARNING, 'getDuplicateMessageUidList got an exception')
+          raise
+        if uid_list:
+          self._log(TRACE, 'Reserved duplicate messages: %r' % uid_list)
+        path_and_method_id_dict[key] = uid
+        return m, uid, uid_list
+      # We know that original_uid != uid because caller skips lines we returned
+      # earlier.
+      return None, original_uid, [uid]
+    return load
+
+  def generateMessageUID(self, m):
+    return (tuple(m.object_path), m.method_id, m.activity_kw.get('signature'),
+                  m.activity_kw.get('tag'), m.activity_kw.get('group_id'))
+
+  def distribute(self, activity_tool, node_count):
+    offset = 0
+    assignMessage = getattr(activity_tool, 'SQLBase_assignMessage', None)
+    if assignMessage is not None:
+      now_date = self.getNow(activity_tool)
+      validated_count = 0
+      while 1:
+        result = self._getMessageList(activity_tool, processing_node=-1,
+                                      to_date=now_date,
+                                      offset=offset, count=READ_MESSAGE_LIMIT)
+        if not result:
+          return
         transaction.commit()
-        
-        for m in message_list:
-          if m.getExecutionState() == MESSAGE_EXECUTED:
-            transaction.begin()
-            # Create a signature and then store result into the dict
-            signature = MyBatchedSignature(m.args[0].batch)
-            # get active process
-            active_process = activity_tool.unrestrictedTraverse(m.active_process)
-            active_process.process_result_map.update({signature: m.result})
-            transaction.commit()
-      except:
-        exc_info = sys.exc_info()
-        if exc_info[1] is not _DequeueMessageException:
-          self._log(WARNING,
-            'Exception raised when invoking messages (uid, path, method_id) %r'
-            % [(m.uid, m.object_path, m.method_id) for m in message_list])
-          for m in message_list:
-            m.setExecutionState(MESSAGE_NOT_EXECUTED, exc_info, log=False)
-        self._abort()
-        exc_info = message_list[0].exc_info
-        if exc_info:
-          try:
-            # Register it again.
-            tv['activity_runtime_environment'] = activity_runtime_environment
-            cancel = message.on_error_callback(*exc_info)
-            del exc_info, message.exc_info
-            transaction.commit()
-            if cancel:
-              message.setExecutionState(MESSAGE_EXECUTED)
-          except:
-            self._log(WARNING, 'Exception raised when processing error callbacks')
-            message.setExecutionState(MESSAGE_NOT_EXECUTED)
-            self._abort()
-      self.finalizeMessageExecution(activity_tool, message_list,
-                                    uid_to_duplicate_uid_list_dict)
-    transaction.commit()
-    return not message_list
\ No newline at end of file
+
+        validation_text_dict = {'none': 1}
+        message_dict = {}
+        for line in result:
+          message = Message.load(line.message, uid=line.uid, line=line)
+          if not hasattr(message, 'order_validation_text'): # BBB
+            message.order_validation_text = self.getOrderValidationText(message)
+          self.getExecutableMessageList(activity_tool, message, message_dict,
+                                        validation_text_dict, now_date=now_date)
+
+        if message_dict:
+          message_unique_dict = {}
+          serialization_tag_dict = {}
+          distributable_uid_set = set()
+          deletable_uid_list = []
+
+          # remove duplicates
+          # SQLDict considers object_path, method_id, tag to unify activities,
+          # but ignores method arguments. They are outside of semantics.
+          for message in message_dict.itervalues():
+            message_unique_dict.setdefault(self.generateMessageUID(message),
+                                           []).append(message)
+
+          for message_list in message_unique_dict.itervalues():
+            if len(message_list) > 1:
+              # Sort list of duplicates to keep the message with highest score
+              message_list.sort(key=sort_message_key)
+              deletable_uid_list += [m.uid for m in message_list[1:]]
+            message = message_list[0]
+            serialization_tag = message.activity_kw.get('serialization_tag')
+            if serialization_tag is None:
+              distributable_uid_set.add(message.uid)
+            else:
+              serialization_tag_dict.setdefault(serialization_tag,
+                                                []).append(message)
+          # Don't let through if there is the same serialization tag in the
+          # message dict. If there is the same serialization tag, only one can
+          # be validated and others must wait.
+          # But messages with group_method_id are exceptions. serialization_tag
+          # does not stop validating together. Because those messages should
+          # be processed together at once.
+          for message_list in serialization_tag_dict.itervalues():
+            # Sort list of messages to validate the message with highest score
+            message_list.sort(key=sort_message_key)
+            distributable_uid_set.add(message_list[0].uid)
+            group_method_id = message_list[0].line.group_method_id
+            if group_method_id == '\0':
+              continue
+            for message in message_list[1:]:
+              if group_method_id == message.line.group_method_id:
+                distributable_uid_set.add(message.uid)
+          if deletable_uid_list:
+            activity_tool.SQLBase_delMessage(table=self.sql_table,
+                                             uid=deletable_uid_list)
+          distributable_count = len(distributable_uid_set)
+          if distributable_count:
+            assignMessage(table=self.sql_table,
+              processing_node=0, uid=tuple(distributable_uid_set))
+            validated_count += distributable_count
+            if validated_count >= MAX_VALIDATED_LIMIT:
+              return
+        offset += READ_MESSAGE_LIMIT

product/CMFActivity/ActivityJoblibBackend.py

@@ -26,16 +26,17 @@
 ##############################################################################
 ENABLE_JOBLIB = True
 
+import copy
+import hashlib
 import sys
 import time
 
-import transaction
-from BTrees.OOBTree import OOBTree
 from zLOG import LOG, INFO, WARNING
 from ZODB.POSException import ConflictError
 
 try:
   from sklearn.externals.joblib import register_parallel_backend
+  from sklearn.externals.joblib.hashing import hash as joblib_hash
   from sklearn.externals.joblib.parallel import ParallelBackendBase, parallel_backend
   from sklearn.externals.joblib.parallel import FallbackToBackend, SequentialBackend
   from sklearn.externals.joblib._parallel_backends import SafeFunction
@@ -45,13 +46,11 @@ except ImportError:
   LOG("CMFActivityBackend", WARNING, "CLASS NOT LOADED!!!")
   ENABLE_JOBLIB = False
 
-from Activity.SQLJoblib import MyBatchedSignature
-
 if ENABLE_JOBLIB:
+
   class MySafeFunction(SafeFunction):
     """Wrapper around a SafeFunction that catches any exception
-  
-    The exception can be handled in CMFActivityResult.get
+       The exception can be handled in CMFActivityResult.get
     """
     def __init__(self, *args, **kwargs):
       super(MySafeFunction, self).__init__(*args, **kwargs)
@@ -67,23 +66,12 @@ if ENABLE_JOBLIB:
       self.active_process = active_process
       self.active_process_sig = active_process_sig
       self.callback = callback
+ 
     def get(self, timeout=None):
-      
-      '''
-      while not self.active_process.getResultList():
-        time.sleep(1)
-        if timeout is not None:
-          timeout -= 1
-          if timeout < 0:
-            raise RuntimeError('Timeout reached')
-        transaction.commit()
-      '''
-
-      if self.active_process.process_result_map[self.active_process_sig] is None:
+      if self.active_process.getResult(self.active_process_sig) is None:
         raise ConflictError
-      result = self.active_process.process_result_map[self.active_process_sig]
+      result = self.active_process.getResult(self.active_process_sig).result
 
-      # TODO raise before or after the callback?
       if isinstance(result, Exception):
         raise result
       if self.callback is not None:
@@ -94,10 +82,7 @@ if ENABLE_JOBLIB:
     def __init__(self, *args, **kwargs):
       self.count = 1
       self.active_process = kwargs['active_process']
-      if not hasattr(self.active_process, 'process_result_map'):
-        self.active_process.process_result_map = OOBTree()
-      transaction.commit()
- 
+
     def effective_n_jobs(self, n_jobs):
       """Dummy implementation to prevent n_jobs <=0
 
@@ -113,14 +98,16 @@ if ENABLE_JOBLIB:
       active_process_id = self.active_process.getId()
       joblib_result = None
 
-      sig = MyBatchedSignature(batch)
-      if not self.active_process.process_result_map.has_key(sig):
-        self.active_process.process_result_map.insert(sig, None)
+      # create a signature and convert it to integer
+      sig = joblib_hash(batch.items[0])
+      sigint = int(sig, 16) % (10 ** 16)
+      if not self.active_process.getResult(sigint):
         joblib_result = portal_activities.activate(activity='SQLJoblib',
           tag="joblib_%s" % active_process_id,
-          active_process=self.active_process).Base_callSafeFunction(MySafeFunction(batch))
+          signature=sig,
+          active_process=self.active_process).Base_callSafeFunction(sigint, MySafeFunction(batch))
       if joblib_result is None:
-        joblib_result = CMFActivityResult(self.active_process, sig, callback)
+        joblib_result = CMFActivityResult(self.active_process, sigint, callback)
       return joblib_result
 
     def configure(self, n_jobs=1, parallel=None, **backend_args):
@@ -138,11 +125,6 @@ if ENABLE_JOBLIB:
 
     def abort_everything(self, ensure_ready=True):
       # All jobs will be aborted here while they are still processing our backend
-
-      # remove job with no results
-      #self.active_process.process_result_map = dict((k, v) 
-      #  for k, v in self.active_process.process_result_map.iteritems() if v)
-      transaction.commit()
       if ensure_ready:
         self.configure(n_jobs=self.parallel.n_jobs, parallel=self.parallel,
                         **self.parallel._backend_args)

product/CMFActivity/ActivityTool.py

@@ -177,7 +177,6 @@ class Message(BaseMessage):
     self.method_id = method_id
     self.args = args
     self.kw = kw
-    self.result = None
     if getattr(portal_activities, 'activity_creation_trace', False):
       # Save current traceback, to make it possible to tell where a message
       # was generated.
@@ -316,12 +315,12 @@ class Message(BaseMessage):
             result = method(*self.args, **self.kw)
         finally:
           setSecurityManager(old_security_manager)
+
         if method is not None:
           if self.active_process and result is not None:
             self.activateResult(
               activity_tool.unrestrictedTraverse(self.active_process),
               result, obj)
-          self.result = result
           self.setExecutionState(MESSAGE_EXECUTED)
     except:
       self.setExecutionState(MESSAGE_NOT_EXECUTED, context=activity_tool)
@@ -504,7 +503,6 @@ class Method(object):
       request=self._request,
       portal_activities=portal_activities,
     )
-
     if portal_activities.activity_tracking:
       activity_tracking_logger.info('queuing message: activity=%s, object_path=%s, method_id=%s, args=%s, kw=%s, activity_kw=%s, user_name=%s' % (self._activity, '/'.join(m.object_path), m.method_id, m.args, m.kw, m.activity_kw, m.user_name))
     portal_activities.getActivityBuffer().deferredQueueMessage(
@@ -1065,6 +1063,7 @@ class ActivityTool (Folder, UniqueObject):
         processing_node starts from 1 (there is not node 0)
       """
       global active_threads
+
       # return if the number of threads is too high
       # else, increase the number of active_threads and continue
       tic_lock.acquire()

product/CMFActivity/skins/activity/SQLJoblib_createMessageTable.zsql

@@ -20,6 +20,7 @@ CREATE TABLE <dtml-var table> (
   `priority` TINYINT NOT NULL DEFAULT 0,
   `group_method_id` VARCHAR(255) NOT NULL DEFAULT '',
   `tag` VARCHAR(255) NOT NULL,
+  `signature` VARCHAR(255) NOT NULL,
   `serialization_tag` VARCHAR(255) NOT NULL,
   `retry` TINYINT UNSIGNED NOT NULL DEFAULT 0,
   `message` LONGBLOB NOT NULL,

product/CMFActivity/skins/activity/SQLJoblib_reserveDuplicatedLineList.zsql

+<dtml-comment>
+title:
+connection_id:cmf_activity_sql_connection
+max_rows:0
+max_cache:0
+cache_time:0
+class_name:
+class_file:
+</dtml-comment>
+<params>
+processing_node
+uid
+</params>
+UPDATE
+  message_job
+SET
+  processing_node=<dtml-sqlvar processing_node type="int">
+WHERE
+  <dtml-sqltest uid type="int" multiple>
+<dtml-var sql_delimiter>
+COMMIT

product/CMFActivity/skins/activity/SQLJoblib_selectDuplicatedLineList.zsql

+<dtml-comment>
+title:
+connection_id:cmf_activity_sql_connection
+max_rows:0
+max_cache:0
+cache_time:0
+class_name:
+class_file:
+</dtml-comment>
+<params>
+path
+method_id
+group_method_id
+signature
+</params>
+SELECT uid FROM
+  message_job
+WHERE
+  processing_node = 0
+  AND path = <dtml-sqlvar path type="string">
+  AND method_id = <dtml-sqlvar method_id type="string">
+  AND group_method_id = <dtml-sqlvar group_method_id type="string">
+  AND signature = <dtml-sqlvar signature type="string">
+FOR UPDATE

product/CMFActivity/skins/activity/SQLJoblib_writeMessage.zsql

@@ -18,10 +18,11 @@ processing_node
 date
 group_method_id
 tag
+signature
 serialization_tag
 </params>
 INSERT INTO <dtml-var table>
-(uid, path, active_process_uid, date, method_id, processing_node, processing, priority, group_method_id, tag, serialization_tag, message)
+(uid, path, active_process_uid, date, method_id, processing_node, processing, priority, group_method_id, tag, signature, serialization_tag, message)
 VALUES
 (
 <dtml-sqlvar expr="uid" type="int">,
@@ -34,6 +35,7 @@ VALUES
 <dtml-sqlvar expr="priority" type="int">,
 <dtml-sqlvar expr="group_method_id" type="string">,
 <dtml-sqlvar expr="tag" type="string">,
+<dtml-sqlvar expr="signature" type="string">,
 <dtml-sqlvar expr="serialization_tag" type="string">,
 <dtml-sqlvar expr="message" type="string">
 )