Unit Tests using pytest and Minor modifications to allow unit testing

aiserver.py

@@ -330,6 +330,7 @@ class vars:
     debug       = False # If set to true, will send debug information to the client for display
     lazy_load   = True  # Whether or not to use torch_lazy_loader.py for transformers models in order to reduce CPU memory usage
     use_colab_tpu = os.environ.get("COLAB_TPU_ADDR", "") != "" or os.environ.get("TPU_NAME", "") != ""  # Whether or not we're in a Colab TPU instance or Kaggle TPU instance and are going to use the TPU rather than the CPU
+    revision    = None
 
 utils.vars = vars
 
@@ -907,7 +908,7 @@ def spRequest(filename):
 #==================================================================#
 # Startup
 #==================================================================#
-def general_startup():
+def general_startup(override_args=None):
     global args
     # Parsing Parameters
     parser = argparse.ArgumentParser(description="KoboldAI Server")
@@ -936,7 +937,14 @@ def general_startup():
     parser.add_argument("--lowmem", action='store_true', help="Extra Low Memory loading for the GPU, slower but memory does not peak to twice the usage")
     parser.add_argument("--savemodel", action='store_true', help="Saves the model to the models folder even if --colab is used (Allows you to save models to Google Drive)")
     #args: argparse.Namespace = None
-    if(os.environ.get("KOBOLDAI_ARGS") is not None):
+    if len(sys.argv) > 1 and override_args is None:
+        if sys.argv[1] == 'unit_tests.py':
+            args = parser.parse_args([])
+            return
+    if override_args is not None:
+        import shlex
+        args = parser.parse_args(shlex.split(override_args))
+    elif(os.environ.get("KOBOLDAI_ARGS") is not None):
         import shlex
         args = parser.parse_args(shlex.split(os.environ["KOBOLDAI_ARGS"]))
     else:

unit_tests.py

@@ -0,0 +1,172 @@
+import pytest, time
+import aiserver
+
+#Test Model List:
+test_models = [
+                ('EleutherAI/gpt-neo-1.3B', {'key': False, 'gpu': False, 'layer_count': 24, 'breakmodel': True, 'url': False}), 
+                ('gpt2', {'key': False, 'gpu': False, 'layer_count': 12, 'breakmodel': True, 'url': False}), 
+                ('facebook/opt-350m', {'key': False, 'gpu': False, 'layer_count': 24, 'breakmodel': True, 'url': False})
+              ]
+
+@pytest.fixture
+def client_data():
+    app = aiserver.app
+    #app.test_client_class = FlaskLoginClient
+    client_conn = app.test_client()
+    socketio_client = aiserver.socketio.test_client(app, flask_test_client=client_conn)
+    #Clear out the connection message
+    response = socketio_client.get_received()
+    return (client_conn, app, socketio_client)
+
+
+def get_model_menu(model):
+    for menu in aiserver.model_menu:
+        for item in aiserver.model_menu[menu]:
+            if item[1] == model:
+                for main_menu_line in aiserver.model_menu['mainmenu']:
+                    if main_menu_line[1] == menu:
+                        return (menu, main_menu_line, item)
+    return None
+    
+def generate_story_data(client_data):
+    (client, app, socketio_client) = client_data
+    socketio_client.emit('message',{'cmd': 'submit', 'allowabort': False, 'actionmode': 0, 'chatname': None, 'data': ''})
+    
+    #wait until the game state turns back to start
+    state = 'wait'
+    new_text = None
+    start_time = time.time()
+    timeout = time.time() + 60*1
+    while state == 'wait':
+        if time.time() > timeout:
+            break
+        responses = socketio_client.get_received()
+        for response in responses:
+            response = response['args'][0]
+            print(response)
+            if response['cmd'] == 'setgamestate':
+                state = response['data']
+            elif response['cmd'] == 'updatechunk' or response['cmd'] == 'genseqs':
+                new_text = response['data']
+        time.sleep(0.1)
+    
+    assert new_text is not None
+
+def test_basic_connection(client_data):
+    (client, app, socketio_client) = client_data
+    response = client.get("/")
+    assert response.status_code == 200
+
+def test_load_story_from_web_ui(client_data):
+    (client, app, socketio_client) = client_data
+    
+    #List out the stories and make sure we have the sample story
+    socketio_client.emit('message',{'cmd': 'loadlistrequest', 'data': ''})
+    response = socketio_client.get_received()[0]['args'][0]['data']
+    found_sample_story = False
+    for story in response:
+        if story['name'] == 'sample_story':
+            found_sample_story = True
+    assert found_sample_story
+    
+    #Click on the sample story, then click load
+    socketio_client.emit('message',{'cmd': 'loadselect', 'data': 'sample_story'})
+    socketio_client.emit('message',{'cmd': 'loadrequest', 'data': ''})
+    
+    #Wait until we get the data back from the load
+    loaded_story = False
+    timeout = time.time() + 60*2
+    while not loaded_story:
+        if time.time() > timeout:
+            break
+        responses = socketio_client.get_received()
+        for response in responses:
+            response = response['args'][0]
+            if 'cmd' not in response:
+                print(response)
+                assert False
+            if response['cmd'] == 'updatescreen':
+                loaded_story = True
+                story_text = response['data']
+                break
+    assert loaded_story
+    
+    #Verify that it's the right story data
+    assert story_text == '<chunk n="0" id="n0" tabindex="-1">Niko the kobold stalked carefully down the alley, his small scaly figure obscured by a dusky cloak that fluttered lightly in the cold winter breeze. Holding up his tail to keep it from dragging in the dirty snow that covered the cobblestone, he waited patiently for the butcher to turn his attention from his stall so that he could pilfer his next meal: a tender-looking</chunk><chunk n="1" id="n1" tabindex="-1"> chicken. He crouched just slightly as he neared the stall to ensure that no one was watching, not that anyone would be dumb enough to hassle a small kobold. What else was there for a lowly kobold to</chunk><chunk n="2" id="n2" tabindex="-1"> do in a city? All that Niko needed to know was</chunk><chunk n="3" id="n3" tabindex="-1"> where to find the chicken and then how to make off with it.<br/><br/>A soft thud caused Niko to quickly lift his head. Standing behind the stall where the butcher had been cutting his chicken,</chunk>'  
+
+@pytest.mark.parametrize("model, expected_load_options", test_models)
+def test_load_model_from_web_ui(client_data, model, expected_load_options):
+    (client, app, socketio_client) = client_data
+    
+    #Clear out any old messages
+    response = socketio_client.get_received()
+    
+    (menu, menu_line, model_line) = get_model_menu(model)
+    
+    #Send the ai load model menu option
+    socketio_client.emit('message',{'cmd': 'list_model', 'data': 'mainmenu'})
+    response = socketio_client.get_received()[0]['args'][0]['data']
+    assert menu_line in response
+    
+    #Send the click model menu option
+    socketio_client.emit('message',{'cmd': 'list_model', 'data': menu, 'pretty_name': ""})
+    response = socketio_client.get_received()[0]['args'][0]['data']
+    assert model_line in response
+    
+    #Click the model
+    socketio_client.emit('message',{'cmd': 'selectmodel', 'data': model})
+    response = socketio_client.get_received()[0]['args'][0]
+    #Check that we're getting the right load options
+    print(response)
+    assert response['key'] == expected_load_options['key']
+    assert response['gpu'] == expected_load_options['gpu']
+    assert response['layer_count'] == expected_load_options['layer_count']
+    assert response['breakmodel'] == expected_load_options['breakmodel']
+    assert response['url'] == expected_load_options['url']
+    
+    #Now send the load 
+    socketio_client.emit('message',{'cmd': 'load_model', 'use_gpu': True, 'key': '', 'gpu_layers': '', 'url': '', 'online_model': ''})
+    #wait until the game state turns back to start
+    state = 'wait'
+    start_time = time.time()
+    timeout = time.time() + 60*2
+    while state == 'wait':
+        if time.time() > timeout:
+            break
+        responses = socketio_client.get_received()
+        for response in responses:
+            response = response['args'][0]
+            if response['cmd'] == 'setgamestate':
+                state = response['data']
+        time.sleep(0.1)
+    
+    #Give it a second to get all of the settings, etc and clear out the messages
+    responses = socketio_client.get_received()
+    
+    #check the model info to see if it's loaded
+    socketio_client.emit('message',{'cmd': 'show_model', 'data': ''})
+    response = socketio_client.get_received()[0]['args'][0]
+    assert response == {'cmd': 'show_model_name', 'data': model}
+    
+    generate_story_data(client_data)
+  
+@pytest.mark.parametrize("model, expected_load_options", test_models)
+def test_load_model_from_command_line(client_data, model, expected_load_options):
+    (client, app, socketio_client) = client_data
+    
+    #Clear out any old messages
+    response = socketio_client.get_received()
+    
+    (menu, menu_line, model_line) = get_model_menu(model)
+    
+    aiserver.general_startup("--model {}".format(model))
+    
+    aiserver.load_model(initial_load=True)
+    
+    #check the model info to see if it's loaded
+    socketio_client.emit('message',{'cmd': 'show_model', 'data': ''})
+    response = socketio_client.get_received()[0]['args'][0]
+    assert response == {'cmd': 'show_model_name', 'data': model}
+    
+    generate_story_data(client_data)
+