llama-deploy-appserver 0.3.27__tar.gz → 0.4.1__tar.gz

{llama_deploy_appserver-0.3.27 → llama_deploy_appserver-0.4.1}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.3
 Name: llama-deploy-appserver
-Version: 0.3.27
+Version: 0.4.1
 Summary: Application server components for LlamaDeploy
 Author: Massimiliano Pippi, Adrian Lyjak
 Author-email: Massimiliano Pippi <mpippi@gmail.com>, Adrian Lyjak <adrianlyjak@gmail.com>
@@ -9,14 +9,13 @@ Requires-Dist: llama-index-workflows[server]>=2.12.0,<2.13.0
 Requires-Dist: pydantic-settings>=2.10.1
 Requires-Dist: fastapi>=0.100.0
 Requires-Dist: websockets>=12.0
-Requires-Dist: llama-deploy-core>=0.3.27,<0.4.0
+Requires-Dist: llama-deploy-core>=0.4.1,<0.5.0
 Requires-Dist: httpx>=0.24.0,<1.0.0
 Requires-Dist: prometheus-fastapi-instrumentator>=7.1.0
 Requires-Dist: packaging>=25.0
 Requires-Dist: structlog>=25.4.0
 Requires-Dist: rich>=14.1.0
 Requires-Dist: pyyaml>=6.0.2
-Requires-Dist: llama-cloud-services>=0.6.60
 Requires-Dist: watchfiles>=1.1.0
 Requires-Dist: uvicorn>=0.35.0
 Requires-Dist: typing-extensions>=4.15.0 ; python_full_version < '3.12'

{llama_deploy_appserver-0.3.27 → llama_deploy_appserver-0.4.1}/pyproject.toml

@@ -14,7 +14,7 @@ dev = [
 
 [project]
 name = "llama-deploy-appserver"
-version = "0.3.27"
+version = "0.4.1"
 description = "Application server components for LlamaDeploy"
 readme = "README.md"
 license = {text = "MIT"}
@@ -28,14 +28,13 @@ dependencies = [
   "pydantic-settings>=2.10.1",
   "fastapi>=0.100.0",
   "websockets>=12.0",
-  "llama-deploy-core>=0.3.27,<0.4.0",
+  "llama-deploy-core>=0.4.1,<0.5.0",
   "httpx>=0.24.0,<1.0.0",
   "prometheus-fastapi-instrumentator>=7.1.0",
   "packaging>=25.0",
   "structlog>=25.4.0",
   "rich>=14.1.0",
   "pyyaml>=6.0.2",
-  "llama-cloud-services>=0.6.60",
   "watchfiles>=1.1.0",
   "uvicorn>=0.35.0",
   "typing-extensions>=4.15.0 ; python_full_version < '3.12'"

{llama_deploy_appserver-0.3.27 → llama_deploy_appserver-0.4.1}/src/llama_deploy/appserver/app.py

@@ -173,13 +173,14 @@ def prepare_server(
     install: bool = False,
     build: bool = False,
     install_ui_deps: bool = True,
+    skip_env_validation: bool = False,
 ) -> None:
     configure_settings(
         deployment_file_path=deployment_file or Path(DEFAULT_DEPLOYMENT_FILE_PATH)
     )
     cfg = get_deployment_config()
     load_environment_variables(cfg, settings.resolved_config_parent)
-    validate_required_env_vars(cfg)
+    validate_required_env_vars(cfg, fill_missing=skip_env_validation)
     if install:
         config = get_deployment_config()
         inject_appserver_into_target(config, settings.resolved_config_parent)
@@ -298,10 +299,16 @@ def start_server_in_target_venv(
 def start_preflight_in_target_venv(
     cwd: Path | None = None,
     deployment_file: Path | None = None,
+    skip_env_validation: bool = False,
 ) -> None:
     """
     Run preflight validation inside the target project's virtual environment using uv.
     Mirrors the venv targeting and invocation strategy used by start_server_in_target_venv.
+
+    Args:
+        cwd: Working directory for the validation.
+        deployment_file: Path to the deployment configuration file.
+        skip_env_validation: If True, skip validation of required environment variables.
     """
     configure_settings(
         app_root=cwd,
@@ -320,6 +327,8 @@ def start_preflight_in_target_venv(
     ]
     if deployment_file:
         args.extend(["--deployment-file", str(deployment_file)])
+    if skip_env_validation:
+        args.append("--skip-env-validation")
 
     run_process(
         args,
@@ -386,10 +395,19 @@ def preflight_validate(
     cwd: Path | None = None,
     deployment_file: Path | None = None,
     configure_logging: bool = False,
+    skip_env_validation: bool = False,
 ) -> None:
     """
     Perform the same initialization path as starting the server, without serving.
     This catches import errors and runs workflow validations.
+
+    Args:
+        cwd: Working directory for the validation.
+        deployment_file: Path to the deployment configuration file.
+        configure_logging: Whether to configure logging.
+        skip_env_validation: If True, fill missing required env vars with placeholder
+            values instead of raising an error. Useful when validating workflow structure
+            without actual environment variable values.
     """
     configure_settings(
         app_root=cwd,
@@ -397,7 +415,7 @@ def preflight_validate(
     )
     cfg = get_deployment_config()
     load_environment_variables(cfg, settings.resolved_config_parent)
-    validate_required_env_vars(cfg)
+    validate_required_env_vars(cfg, fill_missing=skip_env_validation)
 
     workflows = load_workflows(cfg)
     # Instantiate Deployment to ensure server wiring doesn't raise
@@ -454,12 +472,17 @@ if __name__ == "__main__":
     parser.add_argument("--deployment-file", type=Path)
     parser.add_argument("--open-browser", action="store_true")
     parser.add_argument("--preflight", action="store_true")
+    parser.add_argument("--skip-env-validation", action="store_true")
     parser.add_argument("--export-json-graph", action="store_true")
     parser.add_argument("--export-output", type=Path)
 
     args = parser.parse_args()
     if args.preflight:
-        preflight_validate(cwd=Path.cwd(), deployment_file=args.deployment_file)
+        preflight_validate(
+            cwd=Path.cwd(),
+            deployment_file=args.deployment_file,
+            skip_env_validation=args.skip_env_validation,
+        )
     elif args.export_json_graph:
         export_json_graph(
             cwd=Path.cwd(),

{llama_deploy_appserver-0.3.27 → llama_deploy_appserver-0.4.1}/src/llama_deploy/appserver/workflow_loader.py

@@ -73,25 +73,37 @@ def load_environment_variables(config: DeploymentConfig, source_root: Path) -> N
             os.environ[key] = value
 
 
-def validate_required_env_vars(config: DeploymentConfig) -> None:
+def validate_required_env_vars(
+    config: DeploymentConfig, *, fill_missing: bool = False
+) -> None:
     """
     Validate that all required environment variables are present and non-empty.
 
+    Args:
+        config: The deployment configuration containing required_env_vars.
+        fill_missing: If True, fill missing env vars with placeholder values instead
+            of raising an error. This is useful for validation where env vars may be
+            read statically during import time.
+
     Raises:
-        RuntimeError: If any required env vars are missing or empty.
+        RuntimeError: If any required env vars are missing or empty and fill_missing is False.
     """
     required = config.required_env_vars
     if not required:
         return
     missing = [name for name in required if not os.environ.get(name)]
     if missing:
-        missing_list = ", ".join(sorted(missing))
-        raise RuntimeError(
-            (
-                "Missing required environment variables defined in required_env_vars: "
-                f"{missing_list}. Provide them via your environment, .env files, or the deployment secrets."
+        if fill_missing:
+            for name in missing:
+                os.environ[name] = f"__PLACEHOLDER_{name}__"
+        else:
+            missing_list = ", ".join(sorted(missing))
+            raise RuntimeError(
+                (
+                    "Missing required environment variables defined in required_env_vars: "
+                    f"{missing_list}. Provide them via your environment, .env files, or the deployment secrets."
+                )
             )
-        )
 
 
 def parse_environment_variables(

llama_deploy_appserver-0.4.1/src/llama_deploy/appserver/workflow_store/agent_data_store.py

@@ -0,0 +1,216 @@
+import asyncio
+import logging
+import os
+import sys
+from typing import Any, List
+
+import httpx
+from llama_deploy.appserver.settings import ApiserverSettings
+from llama_deploy.core.client.ssl_util import get_httpx_verify_param
+from llama_deploy.core.deployment_config import DeploymentConfig
+from workflows.server import AbstractWorkflowStore, HandlerQuery, PersistentHandler
+
+from .keyed_lock import AsyncKeyedLock
+from .lru_cache import LRUCache
+
+if sys.version_info <= (3, 11):
+    from typing_extensions import override
+else:
+    from typing import override
+
+logger = logging.getLogger(__name__)
+
+DEFAULT_BASE_URL = "https://api.cloud.llamaindex.ai"
+
+
+class AgentDataStore(AbstractWorkflowStore):
+    """Workflow store backed by LlamaCloud Agent Data API using raw httpx."""
+
+    def __init__(
+        self, settings: DeploymentConfig, server_settings: ApiserverSettings
+    ) -> None:
+        agent_url_id: str | None = server_settings.cloud_persistence_name
+        collection = "workflow_contexts"
+        if agent_url_id is not None:
+            parts = agent_url_id.split(":")
+            if len(parts) > 1:
+                collection = parts[1]
+            agent_url_id = parts[0]
+        else:
+            agent_url_id = settings.name
+
+        self.settings = settings
+        self.collection = collection
+        self.deployment_name = agent_url_id
+
+        self.base_url = os.getenv("LLAMA_CLOUD_BASE_URL", DEFAULT_BASE_URL).rstrip("/")
+        self.api_key = os.getenv("LLAMA_CLOUD_API_KEY")
+        self.project_id = os.getenv("LLAMA_DEPLOY_PROJECT_ID")
+
+        self.lock = AsyncKeyedLock()
+        # workflow id -> agent data id
+        self.cache = LRUCache[str, str](maxsize=1024)
+
+    def _get_headers(self) -> dict[str, str]:
+        """Build HTTP headers for API requests."""
+        headers = {
+            "Content-Type": "application/json",
+        }
+        if self.api_key:
+            headers["Authorization"] = f"Bearer {self.api_key}"
+        if self.project_id:
+            headers["Project-Id"] = self.project_id
+        return headers
+
+    def _get_client(self) -> httpx.AsyncClient:
+        """Create a new httpx client."""
+        return httpx.AsyncClient(
+            headers=self._get_headers(),
+            verify=get_httpx_verify_param(),
+        )
+
+    @override
+    async def query(self, query: HandlerQuery) -> List[PersistentHandler]:
+        filters = self._build_filters(query)
+        search_request = {
+            "deployment_name": self.deployment_name,
+            "collection": self.collection,
+            "filter": filters,
+            "page_size": 1000,
+        }
+        async with self._get_client() as client:
+            response = await client.post(
+                f"{self.base_url}/api/v1/beta/agent-data/:search",
+                json=search_request,
+            )
+            response.raise_for_status()
+            data = response.json()
+
+        items = data.get("items", [])
+        return [PersistentHandler(**item["data"]) for item in items]
+
+    @override
+    async def update(self, handler: PersistentHandler) -> None:
+        async with self.lock.acquire(handler.handler_id):
+            id = await self._get_item_id(handler)
+            if id is None:
+                item = await self._create_item(handler)
+                item_id = item.get("id")
+                if item_id is None:
+                    raise ValueError(f"Failed to create handler {handler.handler_id}")
+                self.cache.set(handler.handler_id, item_id)
+            else:
+                await self._update_item(id, handler)
+
+    @override
+    async def delete(self, query: HandlerQuery) -> int:
+        filters = self._build_filters(query)
+        search_request = {
+            "deployment_name": self.deployment_name,
+            "collection": self.collection,
+            "filter": filters,
+            "page_size": 1000,
+        }
+        async with self._get_client() as client:
+            response = await client.post(
+                f"{self.base_url}/api/v1/beta/agent-data/:search",
+                json=search_request,
+            )
+            response.raise_for_status()
+            data = response.json()
+
+        items = data.get("items", [])
+        await asyncio.gather(
+            *[self._delete_item(item["id"]) for item in items if item.get("id")]
+        )
+        return len(items)
+
+    async def _create_item(self, handler: PersistentHandler) -> dict[str, Any]:
+        """Create a new agent data item."""
+        create_request = {
+            "deployment_name": self.deployment_name,
+            "collection": self.collection,
+            "data": handler.model_dump(mode="json"),
+        }
+        async with self._get_client() as client:
+            response = await client.post(
+                f"{self.base_url}/api/v1/beta/agent-data",
+                json=create_request,
+            )
+            response.raise_for_status()
+            return response.json()
+
+    async def _update_item(self, item_id: str, handler: PersistentHandler) -> None:
+        """Update an existing agent data item."""
+        update_request = {
+            "data": handler.model_dump(mode="json"),
+        }
+        async with self._get_client() as client:
+            response = await client.put(
+                f"{self.base_url}/api/v1/beta/agent-data/{item_id}",
+                json=update_request,
+            )
+            response.raise_for_status()
+
+    async def _delete_item(self, item_id: str) -> None:
+        """Delete an agent data item."""
+        async with self._get_client() as client:
+            response = await client.delete(
+                f"{self.base_url}/api/v1/beta/agent-data/{item_id}",
+            )
+            response.raise_for_status()
+
+    async def _get_item_id(self, handler: PersistentHandler) -> str | None:
+        cached_id = self.cache.get(handler.handler_id)
+        if cached_id is not None:
+            return cached_id
+        search_filter = {"handler_id": {"eq": handler.handler_id}}
+        search_request = {
+            "deployment_name": self.deployment_name,
+            "collection": self.collection,
+            "filter": search_filter,
+            "page_size": 1,
+        }
+        async with self._get_client() as client:
+            response = await client.post(
+                f"{self.base_url}/api/v1/beta/agent-data/:search",
+                json=search_request,
+            )
+            response.raise_for_status()
+            data = response.json()
+
+        items = data.get("items", [])
+        if not items:
+            return None
+        id = items[0].get("id")
+        if id is None:
+            return None
+        self.cache.set(handler.handler_id, id)
+        return id
+
+    def _build_filters(self, query: HandlerQuery) -> dict[str, Any]:
+        filters: dict[str, Any] = {}
+        if query.handler_id_in is not None:
+            filters["handler_id"] = {
+                "includes": query.handler_id_in,
+            }
+        if query.workflow_name_in is not None:
+            filters["workflow_name"] = {
+                "includes": query.workflow_name_in,
+            }
+        if query.status_in is not None:
+            filters["status"] = {
+                "includes": query.status_in,
+            }
+        if query.is_idle is not None:
+            if query.is_idle:
+                # Filter for handlers where idle_since is set (any valid datetime)
+                filters["idle_since"] = {
+                    "gte": "1970-01-01T00:00:00Z",
+                }
+            else:
+                # Filter for handlers where idle_since is not set (null)
+                filters["idle_since"] = {
+                    "eq": None,
+                }
+        return filters

llama_deploy_appserver-0.3.27/src/llama_deploy/appserver/workflow_store/agent_data_store.py

@@ -1,124 +0,0 @@
-import asyncio
-import logging
-import os
-import sys
-from typing import Any, List, cast
-
-from llama_cloud.client import AsyncLlamaCloud, httpx
-from llama_cloud_services.beta.agent_data import AsyncAgentDataClient
-from llama_deploy.appserver.settings import ApiserverSettings
-from llama_deploy.core.client.ssl_util import get_httpx_verify_param
-from llama_deploy.core.deployment_config import DeploymentConfig
-from workflows.server import AbstractWorkflowStore, HandlerQuery, PersistentHandler
-
-from .keyed_lock import AsyncKeyedLock
-from .lru_cache import LRUCache
-
-if sys.version_info <= (3, 11):
-    from typing_extensions import override
-else:
-    from typing import override
-
-logger = logging.getLogger(__name__)
-
-
-class AgentDataStore(AbstractWorkflowStore):
-    def __init__(
-        self, settings: DeploymentConfig, server_settings: ApiserverSettings
-    ) -> None:
-        agent_url_id: str | None = server_settings.cloud_persistence_name
-        collection = "workflow_contexts"
-        if agent_url_id is not None:
-            parts = agent_url_id.split(":")
-            if len(parts) > 1:
-                collection = parts[1]
-            agent_url_id = parts[0]
-        else:
-            agent_url_id = settings.name
-
-        self.settings = settings
-        project_id = os.getenv("LLAMA_DEPLOY_PROJECT_ID")
-        self.client = AsyncAgentDataClient(
-            type=PersistentHandler,
-            collection=collection,
-            agent_url_id=agent_url_id,
-            client=AsyncLlamaCloud(
-                base_url=os.getenv("LLAMA_CLOUD_BASE_URL"),
-                token=os.getenv("LLAMA_CLOUD_API_KEY"),
-                httpx_client=httpx.AsyncClient(
-                    headers={"Project-Id": project_id} if project_id else None,
-                    verify=get_httpx_verify_param(),
-                ),
-            ),
-        )
-        self.lock = AsyncKeyedLock()
-        # workflow id -> agent data id
-        self.cache = LRUCache[str, str](maxsize=1024)
-
-    @override
-    async def query(self, query: HandlerQuery) -> List[PersistentHandler]:
-        filters = self._build_filters(query)
-        results = await self.client.search(
-            filter=filters,
-            page_size=1000,
-        )
-        return [x.data for x in results.items]
-
-    @override
-    async def update(self, handler: PersistentHandler) -> None:
-        async with self.lock.acquire(handler.handler_id):
-            id = await self._get_item_id(handler)
-            if id is None:
-                item = await self.client.create_item(
-                    data=handler,
-                )
-                if item.id is None:
-                    raise ValueError(f"Failed to create handler {handler.handler_id}")
-                self.cache.set(handler.handler_id, item.id)
-            else:
-                await self.client.update_item(
-                    item_id=id,
-                    data=handler,
-                )
-
-    @override
-    async def delete(self, query: HandlerQuery) -> int:
-        filters = self._build_filters(query)
-        results = await self.client.search(filter=filters, page_size=1000)
-        await asyncio.gather(
-            *[self.client.delete_item(item_id=x.id) for x in results.items if x.id]
-        )
-        return len(results.items)
-
-    async def _get_item_id(self, handler: PersistentHandler) -> str | None:
-        cached_id = self.cache.get(handler.handler_id)
-        if cached_id is not None:
-            return cached_id
-        search_filter = {"handler_id": {"eq": handler.handler_id}}
-        results = await self.client.search(
-            filter=cast(Any, search_filter),
-            page_size=1,
-        )
-        if not results.items:
-            return None
-        id = results.items[0].id
-        if id is None:
-            return None
-        self.cache.set(handler.handler_id, id)
-        return id
-
-    def _build_filters(self, query: HandlerQuery) -> dict[str, Any]:
-        filters: dict[str, Any] = {}
-        if query.handler_id_in is not None:
-            filters["handler_id"] = {
-                "includes": query.handler_id_in,
-            }
-        if query.workflow_name_in is not None:
-            filters["workflow_name"] = {
-                "includes": query.workflow_name_in,
-            }
-        if query.status_in is not None:
-            filters["status"] = {
-                "includes": query.status_in,
-            }
-        return filters