ayechat-dev 0.36.9.20260204171331__tar.gz → 0.36.9.20260205235944__tar.gz

{ayechat_dev-0.36.9.20260204171331/src/ayechat_dev.egg-info → ayechat_dev-0.36.9.20260205235944}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: ayechat-dev
-Version: 0.36.9.20260204171331
+Version: 0.36.9.20260205235944
 Summary: Aye Chat: Terminal-first AI Code Generator
 Author-email: "Acrotron, Inc." <info@acrotron.com>
 License: MIT

{ayechat_dev-0.36.9.20260204171331 → ayechat_dev-0.36.9.20260205235944}/pyproject.toml

@@ -8,7 +8,7 @@ description = "Aye Chat: Terminal-first AI Code Generator"
 readme = "README.md"
 requires-python = ">=3.10, <3.14"
 dependencies = [ "typer>=0.20.0", "httpx>=0.28.1", "keyring>=25.7.0", "prompt-toolkit>=3.0.52", "pathspec>=0.12.1", "chromadb>=1.3.5", "rapidfuzz",]
-version = "0.36.9.20260204171331"
+version = "0.36.9.20260205235944"
 [[project.authors]]
 name = "Acrotron, Inc."
 email = "info@acrotron.com"

{ayechat_dev-0.36.9.20260204171331 → ayechat_dev-0.36.9.20260205235944}/src/aye/controller/llm_invoker.py

@@ -13,12 +13,16 @@ from aye.model.source_collector import collect_sources
 from aye.model.auth import get_user_config
 from aye.model.offline_llm_manager import is_offline_model
 from aye.controller.util import is_truncated_json
-from aye.model.config import SYSTEM_PROMPT, MODELS, DEFAULT_MAX_OUTPUT_TOKENS, DEFAULT_CONTEXT_TARGET_KB
+from aye.model.config import SYSTEM_PROMPT, MODELS, DEFAULT_MAX_OUTPUT_TOKENS, DEFAULT_CONTEXT_TARGET_KB, CONTEXT_HARD_LIMIT_KB
 from aye.model import telemetry
 
 import os
 
 
+def _is_verbose():
+    return get_user_config("verbose", "off").lower() == "on"
+
+
 def _is_debug():
     return get_user_config("debug", "off").lower() == "on"
 
@@ -77,7 +81,7 @@ def _get_context_hard_limit(model_id: str) -> int:
     model_config = _get_model_config(model_id)
     if model_config and "max_prompt_kb" in model_config:
         return model_config["max_prompt_kb"] * 1024
-    return 170 * 1024
+    return CONTEXT_HARD_LIMIT_KB * 1024
 
 
 def _filter_ground_truth(files: Dict[str, str], conf: Any, verbose: bool) -> Dict[str, str]:
@@ -134,6 +138,8 @@ def _get_rag_context_files(prompt: str, conf: Any, verbose: bool) -> Dict[str, s
 
     context_target_size = _get_context_target_size(conf.selected_model)
     context_hard_limit = _get_context_hard_limit(conf.selected_model)
+    #context_target_size = DEFAULT_CONTEXT_TARGET_KB #_get_context_target_size(conf.selected_model)
+    #context_hard_limit = CONTEXT_HARD_LIMIT_KB # _get_context_hard_limit(conf.selected_model)
 
     if _is_debug():
         rprint(f"[yellow]Context target: {context_target_size / 1024:.1f}KB, hard limit: {context_hard_limit / 1024:.1f}KB[/]")
@@ -298,50 +304,54 @@ def invoke_llm(
     model_config = _get_model_config(conf.selected_model)
     max_output_tokens = model_config.get("max_output_tokens", DEFAULT_MAX_OUTPUT_TOKENS) if model_config else DEFAULT_MAX_OUTPUT_TOKENS
 
-    # 1. Try local/offline model plugins first (no streaming UI for local models)
-    local_response = plugin_manager.handle_command("local_model_invoke", {
-        "prompt": prompt,
-        "model_id": conf.selected_model,
-        "source_files": source_files,
-        "chat_id": chat_id,
-        "root": conf.root,
-        "system_prompt": system_prompt,
-        "max_output_tokens": max_output_tokens
-    })
-
-    if local_response is not None:
-        return LLMResponse(
-            summary=local_response.get("summary", ""),
-            updated_files=local_response.get("updated_files", []),
-            chat_id=None,
-            source=LLMSource.LOCAL
-        )
-
-    # 2. API call with spinner + streaming display
-    if _is_debug():
-        print(f"[DEBUG] Processing chat message with chat_id={chat_id or -1}, model={conf.selected_model}")
-
-    telemetry_payload = telemetry.build_payload(top_n=20) if telemetry.is_enabled() else None
-
-    # Create spinner - will be stopped when streaming starts
+    # Create spinner - will be shown for ALL model types (local, databricks, API)
     spinner = StoppableSpinner(
         console,
         messages=DEFAULT_THINKING_MESSAGES,
         interval=15.0
     )
     
+    # For API calls, we also have streaming display
+    streaming_display: Optional[StreamingResponseDisplay] = None
+    
     def stop_spinner():
-        """Callback to stop spinner when first content arrives."""
+        """Callback to stop spinner when first content arrives (for streaming API)."""
         spinner.stop()
-    
-    # Create streaming display with callback to stop spinner on first content
-    streaming_display = StreamingResponseDisplay(on_first_content=stop_spinner)
-    stream_callback = create_streaming_callback(streaming_display)
 
     try:
-        # Start the spinner before the API call
+        # Start the spinner before ANY LLM call (local or API)
         spinner.start()
         
+        # 1. Try local/offline model plugins first
+        local_response = plugin_manager.handle_command("local_model_invoke", {
+            "prompt": prompt,
+            "model_id": conf.selected_model,
+            "source_files": source_files,
+            "chat_id": chat_id,
+            "root": conf.root,
+            "system_prompt": system_prompt,
+            "max_output_tokens": max_output_tokens
+        })
+
+        if local_response is not None:
+            # Local model handled the request - spinner will be stopped in finally block
+            return LLMResponse(
+                summary=local_response.get("summary", ""),
+                updated_files=local_response.get("updated_files", []),
+                chat_id=None,
+                source=LLMSource.LOCAL
+            )
+
+        # 2. API call with streaming display
+        if _is_debug():
+            print(f"[DEBUG] Processing chat message with chat_id={chat_id or -1}, model={conf.selected_model}")
+
+        telemetry_payload = telemetry.build_payload(top_n=20) if telemetry.is_enabled() else None
+
+        # Create streaming display with callback to stop spinner on first content
+        streaming_display = StreamingResponseDisplay(on_first_content=stop_spinner)
+        stream_callback = create_streaming_callback(streaming_display)
+
         api_resp = cli_invoke(
             message=prompt,
             chat_id=chat_id or -1,
@@ -352,29 +362,29 @@ def invoke_llm(
             telemetry=telemetry_payload,
             on_stream_update=stream_callback
         )
-    finally:
-        # Ensure spinner is stopped (in case no streaming content was received)
-        spinner.stop()
-        
-        # Always stop the streaming display when done
-        if streaming_display.is_active():
-            streaming_display.stop()
 
-    if telemetry_payload is not None:
-        telemetry.reset()
+        if telemetry_payload is not None:
+            telemetry.reset()
 
-    if _is_debug():
-        print(f"[DEBUG] Chat message processed, response keys: {api_resp.keys() if api_resp else 'None'}")
+        if _is_debug():
+            print(f"[DEBUG] Chat message processed, response keys: {api_resp.keys() if api_resp else 'None'}")
 
-    # Check if we already displayed the response via streaming
-    streamed_summary = bool(api_resp.get("_streamed_summary")) if isinstance(api_resp, dict) else False
+        # Check if we already displayed the response via streaming
+        streamed_summary = bool(api_resp.get("_streamed_summary")) if isinstance(api_resp, dict) else False
 
-    # 3. Parse API response
-    assistant_resp, new_chat_id = _parse_api_response(api_resp)
+        # 3. Parse API response
+        assistant_resp, new_chat_id = _parse_api_response(api_resp)
 
-    return LLMResponse(
-        summary="" if streamed_summary else assistant_resp.get("answer_summary", ""),
-        updated_files=assistant_resp.get("source_files", []),
-        chat_id=new_chat_id,
-        source=LLMSource.API
-    )
+        return LLMResponse(
+            summary="" if streamed_summary else assistant_resp.get("answer_summary", ""),
+            updated_files=assistant_resp.get("source_files", []),
+            chat_id=new_chat_id,
+            source=LLMSource.API
+        )
+    finally:
+        # Ensure spinner is stopped for ALL code paths (local model, API, or error)
+        spinner.stop()
+        
+        # Stop the streaming display if it was created and is active
+        if streaming_display is not None and streaming_display.is_active():
+            streaming_display.stop()

{ayechat_dev-0.36.9.20260204171331 → ayechat_dev-0.36.9.20260205235944}/src/aye/model/config.py

@@ -17,12 +17,17 @@ SMALL_PROJECT_FILE_LIMIT = 200
 # Projects smaller than this will skip RAG and include all files directly.
 # Set to match default max_prompt_kb (170KB) so all files can fit in context.
 SMALL_PROJECT_TOTAL_SIZE_LIMIT = 170 * 1024  # 170KB
+#SMALL_PROJECT_TOTAL_SIZE_LIMIT = 100 * 1024  # 170KB
 
 # Default maximum output tokens for LLM responses
 DEFAULT_MAX_OUTPUT_TOKENS = 32000
+#DEFAULT_MAX_OUTPUT_TOKENS = 16000
 
 # Default context target size in KB (used when model doesn't specify one)
 DEFAULT_CONTEXT_TARGET_KB = 150
+#DEFAULT_CONTEXT_TARGET_KB = 20
+
+CONTEXT_HARD_LIMIT_KB = 170
 
 # Shared system prompt for all LLM interactions
 SYSTEM_PROMPT = (

{ayechat_dev-0.36.9.20260204171331 → ayechat_dev-0.36.9.20260205235944}/src/aye/plugins/databricks_model.py

@@ -245,7 +245,8 @@ class DatabricksModelPlugin(Plugin):
         try:
             with httpx.Client(timeout=LLM_TIMEOUT) as client:
                 response = client.post(api_url, json=payload, headers=headers)
-                if self.debug:
+                if self.verbose and response.status_code != 200:
+                    print(f"Status code: {response.status_code}")
                     print("-----------------")
                     print(response.text)
                     print("-----------------")

{ayechat_dev-0.36.9.20260204171331 → ayechat_dev-0.36.9.20260205235944}/src/aye/plugins/local_model.py

@@ -136,48 +136,6 @@ class LocalModelPlugin(Plugin):
             "updated_files": []
         }
 
-    def _handle_databricks(self, prompt: str, source_files: Dict[str, str], chat_id: Optional[int] = None, system_prompt: Optional[str] = None, max_output_tokens: int = DEFAULT_MAX_OUTPUT_TOKENS) -> Optional[Dict[str, Any]]:
-        api_url = os.environ.get("AYE_DBX_API_URL")
-        api_key = os.environ.get("AYE_DBX_API_KEY")
-        model_name = os.environ.get("AYE_DBX_MODEL", "gpt-3.5-turbo")
-        
-        if not api_url or not api_key:
-            return None
-        
-        conv_id = self._get_conversation_id(chat_id)
-        if conv_id not in self.chat_history:
-            self.chat_history[conv_id] = []
-        
-        user_message = self._build_user_message(prompt, source_files)
-        effective_system_prompt = system_prompt if system_prompt else SYSTEM_PROMPT
-        
-        messages = [{"role": "system", "content": effective_system_prompt}] + self.chat_history[conv_id] + [{"role": "user", "content": user_message}]
-        
-        headers = {"Content-Type": "application/json", "Authorization": f"Bearer {api_key}"}
-        payload = {"model": model_name, "messages": messages, "temperature": 0.7, "max_tokens": max_output_tokens, "response_format": {"type": "json_object"}}
-        
-        try:
-            with httpx.Client(timeout=LLM_TIMEOUT) as client:
-                response = client.post(api_url, json=payload, headers=headers)
-                response.raise_for_status()
-                result = response.json()
-                if result.get("choices") and result["choices"][0].get("message"):
-                    generated_text = result["choices"][0]["message"]["content"][0]["text"]
-                    self.chat_history[conv_id].append({"role": "user", "content": user_message})
-                    self.chat_history[conv_id].append({"role": "assistant", "content": generated_text})
-                    self._save_history()
-                    return self._parse_llm_response(generated_text)
-                return self._create_error_response("Failed to get a valid response from the Databricks API")
-        except httpx.HTTPStatusError as e:
-            error_msg = f"DBX API error: {e.response.status_code}"
-            try:
-                error_detail = e.response.json()
-                if "error" in error_detail:
-                    error_msg += f" - {error_detail['error'].get('message', str(error_detail['error']))}"
-            except: error_msg += f" - {e.response.text[:200]}"
-            return self._create_error_response(error_msg)
-        except Exception as e:
-            return self._create_error_response(f"Error calling Databricks API: {str(e)}")
 
     def _handle_openai_compatible(self, prompt: str, source_files: Dict[str, str], chat_id: Optional[int] = None, system_prompt: Optional[str] = None, max_output_tokens: int = DEFAULT_MAX_OUTPUT_TOKENS) -> Optional[Dict[str, Any]]:
         """Handle OpenAI-compatible API endpoints.
@@ -288,9 +246,6 @@ class LocalModelPlugin(Plugin):
             result = self._handle_openai_compatible(prompt, source_files, chat_id, system_prompt, max_output_tokens)
             if result is not None: return result
 
-            result = self._handle_databricks(prompt, source_files, chat_id, system_prompt, max_output_tokens)
-            if result is not None: return result
-
             if model_id == "google/gemini-2.5-pro":
                 return self._handle_gemini_pro_25(prompt, source_files, chat_id, system_prompt, max_output_tokens)
             

{ayechat_dev-0.36.9.20260204171331 → ayechat_dev-0.36.9.20260205235944/src/ayechat_dev.egg-info}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: ayechat-dev
-Version: 0.36.9.20260204171331
+Version: 0.36.9.20260205235944
 Summary: Aye Chat: Terminal-first AI Code Generator
 Author-email: "Acrotron, Inc." <info@acrotron.com>
 License: MIT

{ayechat_dev-0.36.9.20260204171331 → ayechat_dev-0.36.9.20260205235944}/tests/test_llm_handler.py

@@ -123,6 +123,7 @@ class TestLlmHandler(TestCase):
 
         mock_print_error.assert_called_once_with(exc)
 
+    @patch('aye.controller.llm_handler.is_autodiff_enabled', return_value=False)
     @patch('aye.controller.llm_handler.get_user_config', return_value='off')
     @patch('aye.controller.llm_handler.filter_unchanged_files')
     @patch('aye.controller.llm_handler.make_paths_relative')
@@ -135,6 +136,7 @@ class TestLlmHandler(TestCase):
         mock_relative,
         mock_filter,
         _mock_get_user_config,
+        _mock_autodiff,
     ):
         updated_files = [{"file_name": "file1.py", "file_content": "content"}]
         llm_resp = LLMResponse(
@@ -175,6 +177,7 @@ class TestLlmHandler(TestCase):
         # Per-session gate is stored on conf
         self.assertTrue(getattr(self.conf, "_restore_tip_shown", False))
 
+    @patch('aye.controller.llm_handler.is_autodiff_enabled', return_value=False)
     @patch('aye.controller.llm_handler.get_user_config', return_value='on')
     @patch('aye.controller.llm_handler.filter_unchanged_files')
     @patch('aye.controller.llm_handler.make_paths_relative')
@@ -187,6 +190,7 @@ class TestLlmHandler(TestCase):
         mock_relative,
         mock_filter,
         _mock_get_user_config,
+        _mock_autodiff,
     ):
         updated_files = [{"file_name": "file1.py", "file_content": "content"}]
         llm_resp = LLMResponse(

{ayechat_dev-0.36.9.20260204171331 → ayechat_dev-0.36.9.20260205235944}/tests/test_local_model_plugin.py

@@ -137,47 +137,6 @@ class TestLocalModelPlugin(TestCase):
     def test_handle_openai_no_key(self):
         self.assertIsNone(self.plugin._handle_openai_compatible("p", {}))
 
-    @patch('httpx.Client')
-    def test_handle_databricks_success(self, mock_client):
-        os.environ["AYE_DBX_API_URL"] = "http://fake.dbx.api"
-        os.environ["AYE_DBX_API_KEY"] = "fake_key"
-        
-        mock_response = MagicMock()
-        mock_response.status_code = 200
-        mock_response.json.return_value = {
-            "choices": [{"message": {"content": [{"type": "text", "text": json.dumps({"answer_summary": "dbx response"})}]}}]
-        }
-        mock_response.raise_for_status.return_value = None
-        mock_client.return_value.__enter__.return_value.post.return_value = mock_response
-        
-        result = self.plugin._handle_databricks("prompt", {})
-        
-        self.assertIsNotNone(result)
-        self.assertEqual(result["summary"], "dbx response")
-
-    @patch('httpx.Client')
-    def test_handle_databricks_http_error(self, mock_client):
-        os.environ["AYE_DBX_API_URL"] = "http://fake.dbx.api"
-        os.environ["AYE_DBX_API_KEY"] = "fake_key"
-        mock_response = MagicMock(status_code=500, text="Server Error")
-        mock_response.json.side_effect = json.JSONDecodeError("", "", 0)
-        mock_client.return_value.__enter__.return_value.post.side_effect = httpx.HTTPStatusError(
-            "Error", request=MagicMock(), response=mock_response
-        )
-        result = self.plugin._handle_databricks("prompt", {})
-        self.assertIn("DBX API error: 500 - Server Error", result["summary"])
-
-    @patch('httpx.Client')
-    def test_handle_databricks_generic_error(self, mock_client):
-        os.environ["AYE_DBX_API_URL"] = "http://fake.dbx.api"
-        os.environ["AYE_DBX_API_KEY"] = "fake_key"
-        mock_client.return_value.__enter__.return_value.post.side_effect = Exception("Network down")
-        result = self.plugin._handle_databricks("prompt", {})
-        self.assertIn("Error calling Databricks API: Network down", result["summary"])
-
-    def test_handle_databricks_no_key(self):
-        self.assertIsNone(self.plugin._handle_databricks("p", {}))
-
     @patch('httpx.Client')
     def test_handle_gemini_success(self, mock_client):
         os.environ["GEMINI_API_KEY"] = "fake_key"
@@ -248,25 +207,8 @@ class TestLocalModelPlugin(TestCase):
         self.assertEqual(result, {"summary": "openai handled"})
 
     @patch.object(LocalModelPlugin, '_handle_openai_compatible', return_value=None)
-    @patch.object(LocalModelPlugin, '_handle_databricks')
-    def test_on_command_invoke_falls_back_to_dbx(self, mock_handle_dbx, mock_handle_openai):
-        mock_handle_dbx.return_value = {"summary": "dbx handled"}
-        
-        params = {
-            "prompt": "test",
-            "source_files": {},
-            "root": self.root
-        }
-        result = self.plugin.on_command("local_model_invoke", params)
-        
-        mock_handle_openai.assert_called_once()
-        mock_handle_dbx.assert_called_once()
-        self.assertEqual(result, {"summary": "dbx handled"})
-
-    @patch.object(LocalModelPlugin, '_handle_openai_compatible', return_value=None)
-    @patch.object(LocalModelPlugin, '_handle_databricks', return_value=None)
     @patch.object(LocalModelPlugin, '_handle_gemini_pro_25')
-    def test_on_command_invoke_routes_to_gemini_by_id(self, mock_handle_gemini, mock_handle_dbx, mock_handle_openai):
+    def test_on_command_invoke_routes_to_gemini_by_id(self, mock_handle_gemini, mock_handle_openai):
         mock_handle_gemini.return_value = {"summary": "gemini handled"}
         
         params = {
@@ -278,7 +220,6 @@ class TestLocalModelPlugin(TestCase):
         result = self.plugin.on_command("local_model_invoke", params)
         
         mock_handle_openai.assert_called_once()
-        mock_handle_dbx.assert_called_once()
         mock_handle_gemini.assert_called_once()
         self.assertEqual(result, {"summary": "gemini handled"})