model-compose 0.3.18__tar.gz → 0.4.1__tar.gz

{model_compose-0.3.18 → model_compose-0.4.1}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: model-compose
-Version: 0.3.18
+Version: 0.4.1
 Summary: model-compose: Declarative AI Model and Workflow Orchestrator
 Author-email: Hanyeol Cho <hanyeol.cho@gmail.com>
 License-Expression: MIT
@@ -153,7 +153,7 @@ By default, `model-compose` automatically looks for a file named `model-compose.
 
 #### 🧪 Minimal Example (with Components and Workflows)
 
-```
+```yaml
 controller:
   type: http-server
   port: 8080
@@ -174,7 +174,7 @@ components:
       model: gpt-4o
       messages:
         - role: user
-          content: "Write an inspiring quote."
+          content: Write an inspiring quote.
     output:
       quote: ${response.choices[0].message.content}
 
@@ -197,7 +197,7 @@ You can easily expand this example by adding more components (e.g., text-to-spee
 
 #### 📡 Listener Example
 
-```
+```yaml
 listener:
   type: http-callback
   port: 8090
@@ -214,7 +214,7 @@ This listener sets up an HTTP callback endpoint at `http://localhost:8090/callba
 
 #### 🌐 Gateway Example
 
-```
+```yaml
 gateway:
   type: http-tunnel
   driver: ngrok
@@ -233,7 +233,7 @@ This gateway configuration exposes the local listener defined above to the publi
 
 #### ✅ Enabling the Web UI
 To enable the Web UI, simply add the `webui` section under your `controller` in the `model-compose.yml` file:
-```
+```yaml
 controller:
   type: http-server
   port: 8080
@@ -253,7 +253,7 @@ By default, `model-compose` uses [Gradio](https://www.gradio.app) as the interac
 
 Here’s how you can do it:
 
-```
+```yaml
 controller:
   type: http-server
   port: 8080
@@ -292,7 +292,7 @@ You can also expose your workflows via the **Model Context Protocol (MCP)** serv
 
 #### ✅ Minimal MCP Server Example
 
-```
+```yaml
 controller:
   type: mcp-server
   port: 8080
@@ -311,7 +311,7 @@ components:
       model: gpt-4o
       messages:
         - role: user
-          content: "Write an inspiring quote."
+          content: Write an inspiring quote.
     output:
       quote: ${response.choices[0].message.content}
 
@@ -326,7 +326,7 @@ This configuration launches the controller as an **MCP server**, which listens o
 
 Once running, you can invoke workflows remotely using a standard MCP request:
 
-```
+```json
 {
   "jsonrpc": "2.0",
   "id": 1,
@@ -349,7 +349,7 @@ You can run the workflow controller inside a Docker container by specifying the
 
 #### ✅ Minimal Docker Runtime Example
 
-```
+```yaml
 controller:
   type: http-server
   port: 8080
@@ -361,7 +361,7 @@ This configuration will launch the controller inside a lightweight Docker contai
 #### ⚙️ Advanced Docker Runtime Options
 You can fully configure the Docker runtime by using an object under the `runtime` key:
 
-```
+```yaml
 controller:
   type: http-server
   port: 8080
@@ -373,11 +373,11 @@ controller:
       - ./models:/models
       - ./cache:/cache
     ports:
-      - "5000:8080"
-      - "5001:8081"
+      - 5000:8080
+      - 5001:8081
     env:
       MODEL_STORAGE_PATH: /models
-    command: [ "python", "-m", "mindor.cli.compose", "up" ]
+    command: [python, -m, mindor.cli.compose, up]
     ...
 ```
 

{model_compose-0.3.18 → model_compose-0.4.1}/README.md

@@ -128,7 +128,7 @@ By default, `model-compose` automatically looks for a file named `model-compose.
 
 #### 🧪 Minimal Example (with Components and Workflows)
 
-```
+```yaml
 controller:
   type: http-server
   port: 8080
@@ -149,7 +149,7 @@ components:
       model: gpt-4o
       messages:
         - role: user
-          content: "Write an inspiring quote."
+          content: Write an inspiring quote.
     output:
       quote: ${response.choices[0].message.content}
 
@@ -172,7 +172,7 @@ You can easily expand this example by adding more components (e.g., text-to-spee
 
 #### 📡 Listener Example
 
-```
+```yaml
 listener:
   type: http-callback
   port: 8090
@@ -189,7 +189,7 @@ This listener sets up an HTTP callback endpoint at `http://localhost:8090/callba
 
 #### 🌐 Gateway Example
 
-```
+```yaml
 gateway:
   type: http-tunnel
   driver: ngrok
@@ -208,7 +208,7 @@ This gateway configuration exposes the local listener defined above to the publi
 
 #### ✅ Enabling the Web UI
 To enable the Web UI, simply add the `webui` section under your `controller` in the `model-compose.yml` file:
-```
+```yaml
 controller:
   type: http-server
   port: 8080
@@ -228,7 +228,7 @@ By default, `model-compose` uses [Gradio](https://www.gradio.app) as the interac
 
 Here’s how you can do it:
 
-```
+```yaml
 controller:
   type: http-server
   port: 8080
@@ -267,7 +267,7 @@ You can also expose your workflows via the **Model Context Protocol (MCP)** serv
 
 #### ✅ Minimal MCP Server Example
 
-```
+```yaml
 controller:
   type: mcp-server
   port: 8080
@@ -286,7 +286,7 @@ components:
       model: gpt-4o
       messages:
         - role: user
-          content: "Write an inspiring quote."
+          content: Write an inspiring quote.
     output:
       quote: ${response.choices[0].message.content}
 
@@ -301,7 +301,7 @@ This configuration launches the controller as an **MCP server**, which listens o
 
 Once running, you can invoke workflows remotely using a standard MCP request:
 
-```
+```json
 {
   "jsonrpc": "2.0",
   "id": 1,
@@ -324,7 +324,7 @@ You can run the workflow controller inside a Docker container by specifying the
 
 #### ✅ Minimal Docker Runtime Example
 
-```
+```yaml
 controller:
   type: http-server
   port: 8080
@@ -336,7 +336,7 @@ This configuration will launch the controller inside a lightweight Docker contai
 #### ⚙️ Advanced Docker Runtime Options
 You can fully configure the Docker runtime by using an object under the `runtime` key:
 
-```
+```yaml
 controller:
   type: http-server
   port: 8080
@@ -348,11 +348,11 @@ controller:
       - ./models:/models
       - ./cache:/cache
     ports:
-      - "5000:8080"
-      - "5001:8081"
+      - 5000:8080
+      - 5001:8081
     env:
       MODEL_STORAGE_PATH: /models
-    command: [ "python", "-m", "mindor.cli.compose", "up" ]
+    command: [python, -m, mindor.cli.compose, up]
     ...
 ```
 

{model_compose-0.3.18 → model_compose-0.4.1}/pyproject.toml

@@ -1,6 +1,6 @@
 [project]
 name = "model-compose"
-version = "0.3.18"
+version = "0.4.1"
 description = "model-compose: Declarative AI Model and Workflow Orchestrator"
 authors = [
     { name = "Hanyeol Cho", email = "hanyeol.cho@gmail.com" }

{model_compose-0.3.18 → model_compose-0.4.1}/src/mindor/core/controller/runner/http_client.py

@@ -13,7 +13,6 @@ class HttpControllerClient(ControllerClient):
 
     async def run_workflow(self, workflow_id: Optional[str], input: Any, workflow: WorkflowSchema) -> Any:
         body = {
-            "workflow_id": workflow_id,
             "input": input,
             "wait_for_completion": True,
             "output_only": True
@@ -28,7 +27,7 @@ class HttpControllerClient(ControllerClient):
                 "Content-Type": "multipart/form-data"
             }
 
-        return await self.client.request("/workflows", "POST", None, body, headers)
+        return await self.client.request(f"/workflows/{workflow_id}/runs", "POST", None, body, headers)
 
     async def close(self) -> None:
         await self.client.close()

{model_compose-0.3.18 → model_compose-0.4.1}/src/mindor/core/controller/services/http_server.py

@@ -19,8 +19,7 @@ from fastapi.responses import Response, JSONResponse, StreamingResponse
 from starlette.background import BackgroundTask
 import uvicorn
 
-class WorkflowTaskRequestBody(BaseModel):
-    workflow_id: Optional[str] = None
+class WorkflowRunRequestBody(BaseModel):
     input: Optional[Any] = None
     wait_for_completion: bool = True
     output_only: bool = False
@@ -74,6 +73,7 @@ class WorkflowSchemaResult(BaseModel):
     description: Optional[str] = None
     input: List[WorkflowVariableResult]
     output: List[Union[WorkflowVariableResult, WorkflowVariableGroupResult]]
+    default: bool = False
 
     @classmethod
     def from_instance(cls, instance: WorkflowSchema) -> Self:
@@ -83,10 +83,11 @@ class WorkflowSchemaResult(BaseModel):
             description=instance.description,
             input=[ cls._to_variable_result(variable) for variable in instance.input ],
             output=[ cls._to_variable_result(variable) for variable in instance.output ],
+            default=instance.default
         )
-    
+
     @classmethod
-    def _to_variable_result(cls, variable: Union[WorkflowVariableConfig, WorkflowVariableGroupConfig ]) -> Union[WorkflowVariableResult, WorkflowVariableGroupResult]:
+    def _to_variable_result(cls, variable: Union[WorkflowVariableConfig, WorkflowVariableGroupConfig]) -> Union[WorkflowVariableResult, WorkflowVariableGroupResult]:
         if isinstance(variable, WorkflowVariableGroupConfig):
             return WorkflowVariableGroupResult.from_instance(variable)
         return WorkflowVariableResult.from_instance(variable)
@@ -123,23 +124,34 @@ class HttpServerController(ControllerService):
         )
 
     def _configure_routes(self) -> None:
-        @self.router.post("/workflows")
-        async def run_workflow(
+        @self.router.post("/workflows/__default__/runs")
+        async def run_default_workflow(
             request: Request
         ):
-            body = await self._parse_workflow_body(request)
-            state = await self.run_workflow(body.workflow_id, body.input, body.wait_for_completion)
+            return await self._handle_workflow_run_request(request, None)
 
-            if body.output_only and not body.wait_for_completion:
-                raise HTTPException(status_code=400, detail="output_only requires wait_for_completion=true.")
-            
-            if not body.output_only and isinstance(state.output, (StreamResource, AsyncGeneratorType)):
-                raise HTTPException(status_code=400, detail="Streaming output is only allowed when output_only=true.")
+        @self.router.post("/workflows/{workflow_id}/runs")
+        async def run_workflow(
+            request: Request,
+            workflow_id: str
+        ):
+            if workflow_id not in self.workflow_schemas:
+                raise HTTPException(status_code=404, detail="Workflow not found.")
 
-            if body.output_only:
-                return self._render_task_output(state)
-            
-            return self._render_task_state(state)
+            return await self._handle_workflow_run_request(request, workflow_id)
+
+        @self.router.get("/workflows/schemas")
+        async def get_workflow_schemas():
+            return self._render_workflow_schemas(self.workflow_schemas)
+
+        @self.router.get("/workflows/{workflow_id}/schema")
+        async def get_workflow_schema(
+            workflow_id: str
+        ):
+            if workflow_id not in self.workflow_schemas:
+                raise HTTPException(status_code=404, detail="Workflow not found.")
+
+            return self._render_workflow_schema(self.workflow_schemas[workflow_id])
 
         @self.router.get("/tasks/{task_id}")
         async def get_task_state(
@@ -151,34 +163,7 @@ class HttpServerController(ControllerService):
             if not state:
                 raise HTTPException(status_code=404, detail="Task not found.")
             
-            if not output_only and isinstance(state.output, (StreamResource, AsyncGeneratorType)):
-                raise HTTPException(status_code=400, detail="Streaming output is only allowed when output_only=true.")
-
-            if output_only:
-                return self._render_task_output(state)
-
-            return self._render_task_state(state)
-
-        @self.router.get("/tasks/{task_id}/jobs")
-        async def get_task_jobs(
-            task_id: str
-        ):
-            return None
-
-        @self.router.get("/workflows")
-        async def get_workflow_schemas():
-            return self._render_workflow_schemas(self.workflow_schemas)
-
-        @self.router.get("/workflows/{workflow_id}")
-        async def get_workflow_schema(
-            workflow_id: str,
-        ):
-            workflow = self.workflow_schemas[workflow_id] if workflow_id in self.workflow_schemas else None
-
-            if not workflow:
-                raise HTTPException(status_code=404, detail="Workflow not found.")
-
-            return self._render_workflow_schema(workflow)
+            return self._render_task_response(state, output_only)
 
     async def _serve(self) -> None:
         self.server = uvicorn.Server(uvicorn.Config(
@@ -196,7 +181,16 @@ class HttpServerController(ControllerService):
         if self.server:
             self.server.should_exit = True
 
-    async def _parse_workflow_body(self, request: Request) -> WorkflowTaskRequestBody:
+    async def _handle_workflow_run_request(self, request: Request, workflow_id: Optional[str]) -> Response:
+        body = await self._parse_workflow_run_body(request)
+        state = await self.run_workflow(workflow_id, body.input, body.wait_for_completion)
+
+        if body.output_only and not body.wait_for_completion:
+            raise HTTPException(status_code=400, detail="output_only requires wait_for_completion=true.")
+
+        return self._render_task_response(state, body.output_only)
+
+    async def _parse_workflow_run_body(self, request: Request) -> WorkflowRunRequestBody:
         content_type, _ = parse_options_header(request.headers, "Content-Type")
 
         if content_type not in [ "application/json", "multipart/form-data", "application/x-www-form-urlencoded" ]:
@@ -206,10 +200,19 @@ class HttpServerController(ControllerService):
                 raise HTTPException(status_code=400, detail=f"Unsupported Content-Type: {content_type}")
 
         try:
-            return WorkflowTaskRequestBody(**await parse_request_body(request, content_type, nested=True))
+            return WorkflowRunRequestBody(**await parse_request_body(request, content_type, nested=True))
         except Exception as e:
             raise HTTPException(status_code=400, detail=f"Invalid request body: {e}")
 
+    def _render_task_response(self, state: TaskState, output_only: bool) -> Response:
+        if not output_only and isinstance(state.output, (StreamResource, AsyncGeneratorType)):
+            raise HTTPException(status_code=400, detail="Streaming output is only allowed when output_only=true.")
+
+        if output_only:
+            return self._render_task_output(state)
+
+        return self._render_task_state(state)
+
     def _render_task_state(self, state: TaskState) -> Response:
         return JSONResponse(content=TaskResult.from_instance(state).model_dump(exclude_none=True))
 

{model_compose-0.3.18 → model_compose-0.4.1}/src/mindor/core/controller/webui/gradio.py

@@ -9,7 +9,9 @@ from mindor.core.utils.image import load_image_from_stream
 from mindor.core.utils.resolvers import FieldResolver
 from PIL import Image as PILImage
 import gradio as gr
-import json
+import json, re
+
+_variable_name_regex = re.compile(r"^([^[]+)(?:\[(\w+)\])?$")
 
 class ComponentGroup:
     def __init__(self, group: gr.Component, components: List[gr.Component]):
@@ -193,20 +195,39 @@ class GradioWebUIBuilder:
         flattened = []
         for variable in variables:
             if isinstance(variable, WorkflowVariableGroupConfig):
-                group = output[variable.name] if isinstance(output, dict) and variable.name in output else None if variable.name else output
+                group = self._resolve_variable_output(output, variable)
                 for value in group or ():
                     flattened.extend(await self._flatten_output_value(value, variable.variables))
             else:
-                value = output[variable.name] if isinstance(output, dict) and variable.name in output else None if variable.name else output
+                value = self._resolve_variable_output(output, variable)
                 flattened.append(await self._convert_output_value(value, variable.type, variable.subtype, variable.format, variable.internal))
         return flattened
 
+    def _resolve_variable_output(self, output: Any, variable: Union[WorkflowVariableConfig, WorkflowVariableGroupConfig]) -> Any:
+        if isinstance(output, dict) and variable.name:
+            m = re.match(_variable_name_regex, variable.name)
+            if not m:
+                return None
+
+            name, index = m.group(1, 2)
+            if name not in output:
+                return None
+
+            if isinstance(output[name], list) and index:
+                if int(index) < len(output[name]):
+                    return output[name][int(index)]
+                return None
+
+            return output[name]
+
+        return None if variable.name else output
+
     async def _convert_output_value(self, value: Any, type: WorkflowVariableType, subtype: Optional[str], format: Optional[WorkflowVariableFormat], internal: bool) -> Any:
         if format == WorkflowVariableFormat.SSE_JSON:
-            return await self._resolve_json_field_from_bytes(value, subtype, format)
+            return self._resolve_json_field_from_bytes(value, subtype, format)
 
         if type in [ WorkflowVariableType.STRING, WorkflowVariableType.TEXT ]:
-            return await self._convert_value_to_string(value, subtype, format)
+            return self._convert_value_to_string(value, subtype, format)
 
         if type == WorkflowVariableType.IMAGE:
             return await self._load_image_from_value(value, subtype, format)
@@ -216,7 +237,7 @@ class GradioWebUIBuilder:
 
         return value
 
-    async def _convert_value_to_string(self, value: Any, subtype: Optional[str], format: Optional[WorkflowVariableFormat]) -> Optional[str]:
+    def _convert_value_to_string(self, value: Any, subtype: Optional[str], format: Optional[WorkflowVariableFormat]) -> Optional[str]:
         if isinstance(value, (dict, list)):
             return json.dumps(value)
         
@@ -228,7 +249,7 @@ class GradioWebUIBuilder:
 
         return None
     
-    async def _resolve_json_field_from_bytes(self, value: Any, subtype: Optional[str], format: Optional[WorkflowVariableFormat]) -> Optional[Any]:
+    def _resolve_json_field_from_bytes(self, value: Any, subtype: Optional[str], format: Optional[WorkflowVariableFormat]) -> Optional[Any]:
         try:
             return self.field_resolver.resolve(json.loads(value), subtype)
         except Exception:

{model_compose-0.3.18 → model_compose-0.4.1}/src/mindor/core/workflow/schema.py

@@ -292,7 +292,8 @@ class WorkflowSchema:
         title: Optional[str], 
         description: Optional[str], 
         input: List[WorkflowVariableConfig], 
-        output: List[Union[WorkflowVariableConfig, WorkflowVariableGroupConfig]]
+        output: List[Union[WorkflowVariableConfig, WorkflowVariableGroupConfig]],
+        default: bool
     ):
         self.workflow_id: str = workflow_id
         self.name: Optional[str] = name
@@ -300,6 +301,7 @@ class WorkflowSchema:
         self.description: Optional[str] = description
         self.input: List[WorkflowVariableConfig] = input
         self.output: List[Union[WorkflowVariableConfig, WorkflowVariableGroupConfig]] = output
+        self.default: bool = default
 
 def create_workflow_schemas(workflows: List[WorkflowConfig], components: List[ComponentConfig]) -> Dict[str, WorkflowSchema]:
     schema: Dict[str, WorkflowSchema] = {}
@@ -311,7 +313,8 @@ def create_workflow_schemas(workflows: List[WorkflowConfig], components: List[Co
             title=workflow.title, 
             description=workflow.description,
             input=WorkflowInputVariableResolver().resolve(workflow, workflows, components),
-            output=WorkflowOutputVariableResolver().resolve(workflow, workflows, components)
+            output=WorkflowOutputVariableResolver().resolve(workflow, workflows, components),
+            default=workflow.default
         )
 
     return schema

{model_compose-0.3.18 → model_compose-0.4.1}/src/mindor/dsl/schema/compose.py

@@ -24,7 +24,7 @@ class ComposeConfig(BaseModel):
             if component_values:
                 values["components"] = [ component_values ]
         return values
-    
+
     @model_validator(mode="before")
     def inflate_single_listener(cls, values: Dict[str, Any]):
         if "listeners" not in values:

{model_compose-0.3.18 → model_compose-0.4.1}/src/mindor/dsl/schema/job/impl/if_.py

@@ -16,7 +16,7 @@ class IfConditionOperator(str, Enum):
     STARTS_WITH = "starts-with"
     ENDS_WITH   = "ends-with"
     MATCH       = "match"
-
+ 
 class IfConditionConfig(BaseModel):
     operator: IfConditionOperator = Field(default=IfConditionOperator.EQ, description="Condition operator.")
     input: Optional[Any] = Field(default=None, description="Input to evaluate.")

{model_compose-0.3.18 → model_compose-0.4.1}/src/model_compose.egg-info/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: model-compose
-Version: 0.3.18
+Version: 0.4.1
 Summary: model-compose: Declarative AI Model and Workflow Orchestrator
 Author-email: Hanyeol Cho <hanyeol.cho@gmail.com>
 License-Expression: MIT
@@ -153,7 +153,7 @@ By default, `model-compose` automatically looks for a file named `model-compose.
 
 #### 🧪 Minimal Example (with Components and Workflows)
 
-```
+```yaml
 controller:
   type: http-server
   port: 8080
@@ -174,7 +174,7 @@ components:
       model: gpt-4o
       messages:
         - role: user
-          content: "Write an inspiring quote."
+          content: Write an inspiring quote.
     output:
       quote: ${response.choices[0].message.content}
 
@@ -197,7 +197,7 @@ You can easily expand this example by adding more components (e.g., text-to-spee
 
 #### 📡 Listener Example
 
-```
+```yaml
 listener:
   type: http-callback
   port: 8090
@@ -214,7 +214,7 @@ This listener sets up an HTTP callback endpoint at `http://localhost:8090/callba
 
 #### 🌐 Gateway Example
 
-```
+```yaml
 gateway:
   type: http-tunnel
   driver: ngrok
@@ -233,7 +233,7 @@ This gateway configuration exposes the local listener defined above to the publi
 
 #### ✅ Enabling the Web UI
 To enable the Web UI, simply add the `webui` section under your `controller` in the `model-compose.yml` file:
-```
+```yaml
 controller:
   type: http-server
   port: 8080
@@ -253,7 +253,7 @@ By default, `model-compose` uses [Gradio](https://www.gradio.app) as the interac
 
 Here’s how you can do it:
 
-```
+```yaml
 controller:
   type: http-server
   port: 8080
@@ -292,7 +292,7 @@ You can also expose your workflows via the **Model Context Protocol (MCP)** serv
 
 #### ✅ Minimal MCP Server Example
 
-```
+```yaml
 controller:
   type: mcp-server
   port: 8080
@@ -311,7 +311,7 @@ components:
       model: gpt-4o
       messages:
         - role: user
-          content: "Write an inspiring quote."
+          content: Write an inspiring quote.
     output:
       quote: ${response.choices[0].message.content}
 
@@ -326,7 +326,7 @@ This configuration launches the controller as an **MCP server**, which listens o
 
 Once running, you can invoke workflows remotely using a standard MCP request:
 
-```
+```json
 {
   "jsonrpc": "2.0",
   "id": 1,
@@ -349,7 +349,7 @@ You can run the workflow controller inside a Docker container by specifying the
 
 #### ✅ Minimal Docker Runtime Example
 
-```
+```yaml
 controller:
   type: http-server
   port: 8080
@@ -361,7 +361,7 @@ This configuration will launch the controller inside a lightweight Docker contai
 #### ⚙️ Advanced Docker Runtime Options
 You can fully configure the Docker runtime by using an object under the `runtime` key:
 
-```
+```yaml
 controller:
   type: http-server
   port: 8080
@@ -373,11 +373,11 @@ controller:
       - ./models:/models
       - ./cache:/cache
     ports:
-      - "5000:8080"
-      - "5001:8081"
+      - 5000:8080
+      - 5001:8081
     env:
       MODEL_STORAGE_PATH: /models
-    command: [ "python", "-m", "mindor.cli.compose", "up" ]
+    command: [python, -m, mindor.cli.compose, up]
     ...
 ```