Merge pull request #100 from sashafrey/master

Introduce ThetaMatrix.topics_count field

docs/ref/c_interface.txt

@@ -355,6 +355,9 @@ ArtmAddBatch
    :return: Returns :c:macro:`ARTM_SUCCESS` value if operation succeeded,
             otherwise returns one of the :ref:`error codes <error-codes>`.
 
+   This operation is only allowed when
+   :attr:`MasterComponentConfig.modus_operandi` is set to *Local*.
+
 ArtmInvokeIteration
 -------------------
 

docs/ref/messages.txt

@@ -1429,7 +1429,7 @@ Represents a topic model.
 .. attribute:: TopicModel.topics_count
 
    A value that describes the number of topics in the topic model.
-   This value will match the value, defined in the model config.
+   This value will match :attr:`ModelConfig.topics_count` value, defined in the model config.
 
 .. attribute:: TopicModel.token
 
@@ -1467,6 +1467,7 @@ Represents a theta matrix.
     repeated int32 item_id = 2;
     repeated FloatArray item_weights = 3;
     repeated string topic_name = 4;
+    optional int32 topics_count = 5;
   }
 
 .. attribute:: ThetaMatrix.model_name
@@ -1489,6 +1490,11 @@ Represents a theta matrix.
    A set of values that represent the names of the topics, included in this theta matrix.
    The names correspond to :attr:`ModelConfig.topic_name`.
 
+.. attribute:: TopicModel.topics_count
+
+   A value that describes the number of topics in the topic model.
+   This value will match :attr:`ModelConfig.topics_count` value, defined in the model config.
+
 .. _CollectionParserConfig:
 
 CollectionParserConfig

src/artm/core/helpers.cc

@@ -273,10 +273,12 @@ bool BatchHelpers::PopulateThetaMatrixFromCacheEntry(
     }
   }
 
+  theta_matrix->set_topics_count(topic_indices.size());
   for (int item_index = 0; item_index < cache.item_id_size(); ++item_index) {
     const artm::FloatArray& item_theta = cache.theta(item_index);
     if (all_topics) {
       theta_matrix->add_item_weights()->CopyFrom(item_theta);
+      theta_matrix->set_topics_count(item_theta.value_size());
     } else {
       if (max_topic_index >= item_theta.value_size())
         continue;  // skip the item to avoid crash.

src/artm/core/master_component.cc

@@ -314,8 +314,8 @@ void MasterComponent::AddBatch(const Batch& batch) {
   }
 
   if (isInNetworkModusOperandi()) {
-    BatchHelpers::SaveBatch(batch, config_.get()->disk_path());
-    return;
+    BOOST_THROW_EXCEPTION(InvalidOperation(
+      "AddBatch() is not allowed together with MasterComponentConfig.modus_operandi=Network"));
   }
 
   BOOST_THROW_EXCEPTION(ArgumentOutOfRangeException(

src/artm/messages.proto

@@ -349,6 +349,7 @@ message ThetaMatrix {
   repeated int32 item_id = 2;
   repeated FloatArray item_weights = 3;
   repeated string topic_name = 4;
+  optional int32 topics_count = 5;
 }
 
 // Represents a configuration of a collection parser.

src/artm_tests/cpp_interface_test.cc

@@ -138,7 +138,8 @@ void BasicTest(bool is_network_mode, bool is_proxy_mode, bool online_processing)
   }
 
   // Index doc-token matrix
-  if (!online_processing) master_component->AddBatch(batch);
+  if (!online_processing && !is_network_mode) master_component->AddBatch(batch);
+  else if (is_network_mode) artm::SaveBatch(batch, "00b6d631-46a6-4edf-8ef6-016c7b27d9f0.batch");
 
   std::shared_ptr<artm::TopicModel> topic_model;
   double expected_normalizer = 0;
@@ -224,6 +225,7 @@ void BasicTest(bool is_network_mode, bool is_proxy_mode, bool online_processing)
       std::shared_ptr< ::artm::ThetaMatrix> theta_matrix = master_component->GetThetaMatrix(args);
 
       EXPECT_EQ(theta_matrix->item_id_size(), nDocs);
+      EXPECT_EQ(theta_matrix->topics_count(), nTopics);
       for (int item_index = 0; item_index < theta_matrix->item_id_size(); ++item_index) {
         const ::artm::FloatArray& weights = theta_matrix->item_weights(item_index);
         ASSERT_EQ(weights.value_size(), nTopics);
@@ -240,6 +242,7 @@ void BasicTest(bool is_network_mode, bool is_proxy_mode, bool online_processing)
       args.add_topic_index(2); args.add_topic_index(3);  // retrieve 2nd and 3rd topic
       std::shared_ptr< ::artm::ThetaMatrix> theta_matrix23 = master_component->GetThetaMatrix(args);
       EXPECT_EQ(theta_matrix23->item_id_size(), nDocs);
+      EXPECT_EQ(theta_matrix23->topics_count(), 2);
       for (int item_index = 0; item_index < theta_matrix23->item_id_size(); ++item_index) {
         const ::artm::FloatArray& weights23 = theta_matrix23->item_weights(item_index);
         const ::artm::FloatArray& weights = theta_matrix->item_weights(item_index);
@@ -270,6 +273,7 @@ void BasicTest(bool is_network_mode, bool is_proxy_mode, bool online_processing)
     if (!is_network_mode) {
       std::shared_ptr< ::artm::ThetaMatrix> theta_matrix2 = master_component->GetThetaMatrix(args);
       EXPECT_EQ(theta_matrix2->item_id_size(), nDocs);
+      EXPECT_EQ(theta_matrix2->topics_count(), nTopics);
       for (int item_index = 0; item_index < theta_matrix2->item_id_size(); ++item_index) {
         const ::artm::FloatArray& weights2 = theta_matrix2->item_weights(item_index);
         EXPECT_EQ(weights2.value_size(), nTopics);