Use updates.codegate.ai for /versions instead of GitHub (#1264)

Wire up CodeGate to use the new update service to determine the latest
version. This is kept behind a config flag, but I have tested locally.

Author: dmjb

migrations/versions/2025_03_05_2126-e4c05d7591a8_add_installation_table.py

@@ -9,8 +9,6 @@
 from typing import Sequence, Union
 
 from alembic import op
-import sqlalchemy as sa
-
 
 # revision identifiers, used by Alembic.
 revision: str = "e4c05d7591a8"

poetry.lock

@@ -10,7 +10,7 @@
 
 from codegate.config import API_DEFAULT_PAGE_SIZE, API_MAX_PAGE_SIZE
 import codegate.muxing.models as mux_models
-from codegate import __version__
+from codegate import Config, __version__
 from codegate.api import v1_models, v1_processing
 from codegate.db.connection import AlreadyExistsError, DbReader
 from codegate.db.models import AlertSeverity, AlertTriggerType, Persona, WorkspaceWithModel
@@ -20,6 +20,7 @@
     PersonaSimilarDescriptionError,
 )
 from codegate.providers import crud as provendcrud
+from codegate.updates.client import Origin, UpdateClient
 from codegate.workspaces import crud
 
 logger = structlog.get_logger("codegate")
@@ -31,6 +32,7 @@
 
 # This is a singleton object
 dbreader = DbReader()
+update_client = UpdateClient(Config.get_config().update_service_url, __version__, dbreader)
 
 
 def uniq_name(route: APIRoute):
@@ -724,10 +726,12 @@ async def stream_sse():
 
 
 @v1.get("/version", tags=["Dashboard"], generate_unique_id_function=uniq_name)
-def version_check():
+async def version_check():
     try:
-        latest_version = v1_processing.fetch_latest_version()
-
+        if Config.get_config().use_update_service:
+            latest_version = await update_client.get_latest_version(Origin.FrontEnd)
+        else:
+            latest_version = v1_processing.fetch_latest_version()
         # normalize the versions as github will return them with a 'v' prefix
         current_version = __version__.lstrip("v")
         latest_version_stripped = latest_version.lstrip("v")

src/codegate/api/v1.py

@@ -16,8 +16,8 @@
 from codegate.config import Config, ConfigurationError
 from codegate.db.connection import (
     init_db_sync,
-    init_session_if_not_exists,
     init_instance,
+    init_session_if_not_exists,
 )
 from codegate.pipeline.factory import PipelineFactory
 from codegate.pipeline.sensitive_data.manager import SensitiveDataManager

src/codegate/cli.py

@@ -59,6 +59,10 @@ class Config:
     server_key: str = "server.key"
     force_certs: bool = False
 
+    # Update configuration.
+    use_update_service: bool = False
+    update_service_url: str = "https://updates.codegate.ai/api/v1/version"
+
     max_fim_hash_lifetime: int = 60 * 5  # Time in seconds. Default is 5 minutes.
 
     # Min value is 0 (max similarity), max value is 2 (orthogonal)
@@ -165,6 +169,8 @@ def from_file(cls, config_path: Union[str, Path]) -> "Config":
                 force_certs=config_data.get("force_certs", cls.force_certs),
                 prompts=prompts_config,
                 provider_urls=provider_urls,
+                use_update_service=config_data.get("use_update_service", cls.use_update_service),
+                update_service_url=config_data.get("update_service_url", cls.update_service_url),
             )
         except yaml.YAMLError as e:
             raise ConfigurationError(f"Failed to parse config file: {e}")
@@ -209,11 +215,17 @@ def from_env(cls) -> "Config":
             if "CODEGATE_SERVER_KEY" in os.environ:
                 config.server_key = os.environ["CODEGATE_SERVER_KEY"]
             if "CODEGATE_FORCE_CERTS" in os.environ:
-                config.force_certs = os.environ["CODEGATE_FORCE_CERTS"]
+                config.force_certs = cls.__bool_from_string(os.environ["CODEGATE_FORCE_CERTS"])
             if "CODEGATE_DB_PATH" in os.environ:
                 config.db_path = os.environ["CODEGATE_DB_PATH"]
             if "CODEGATE_VEC_DB_PATH" in os.environ:
                 config.vec_db_path = os.environ["CODEGATE_VEC_DB_PATH"]
+            if "CODEGATE_USE_UPDATE_SERVICE" in os.environ:
+                config.use_update_service = cls.__bool_from_string(
+                    os.environ["CODEGATE_USE_UPDATE_SERVICE"]
+                )
+            if "CODEGATE_UPDATE_SERVICE_URL" in os.environ:
+                config.update_service_url = os.environ["CODEGATE_UPDATE_SERVICE_URL"]
 
             # Load provider URLs from environment variables
             for provider in DEFAULT_PROVIDER_URLS.keys():
@@ -246,6 +258,8 @@ def load(
         force_certs: Optional[bool] = None,
         db_path: Optional[str] = None,
         vec_db_path: Optional[str] = None,
+        use_update_service: Optional[bool] = None,
+        update_service_url: Optional[str] = None,
     ) -> "Config":
         """Load configuration with priority resolution.
 
@@ -274,6 +288,8 @@ def load(
             force_certs: Optional flag to force certificate generation
             db_path: Optional path to the main SQLite database file
             vec_db_path: Optional path to the vector SQLite database file
+            use_update_service: Optional flag to enable the update service
+            update_service_url: Optional URL for the update service
 
         Returns:
             Config: Resolved configuration
@@ -326,6 +342,10 @@ def load(
             config.db_path = env_config.db_path
         if "CODEGATE_VEC_DB_PATH" in os.environ:
             config.vec_db_path = env_config.vec_db_path
+        if "CODEGATE_USE_UPDATE_SERVICE" in os.environ:
+            config.use_update_service = env_config.use_update_service
+        if "CODEGATE_UPDATE_SERVICE_URL" in os.environ:
+            config.update_service_url = env_config.update_service_url
 
         # Override provider URLs from environment
         for provider, url in env_config.provider_urls.items():
@@ -366,6 +386,10 @@ def load(
             config.vec_db_path = vec_db_path
         if force_certs is not None:
             config.force_certs = force_certs
+        if use_update_service is not None:
+            config.use_update_service = use_update_service
+        if update_service_url is not None:
+            config.update_service_url = update_service_url
 
         # Set the __config class attribute
         Config.__config = config
@@ -375,3 +399,7 @@ def load(
     @classmethod
     def get_config(cls) -> "Config":
         return cls.__config
+
+    @staticmethod
+    def __bool_from_string(raw_value) -> bool:
+        return raw_value.lower() == "true"

src/codegate/config.py

@@ -619,7 +619,7 @@ async def init_instance(self) -> None:
             await self._execute_with_no_return(sql, instance.model_dump())
         except IntegrityError as e:
             logger.debug(f"Exception type: {type(e)}")
-            raise AlreadyExistsError(f"Instance already initialized.")
+            raise AlreadyExistsError("Instance already initialized.")
 
 
 class DbReader(DbCodeGate):

src/codegate/db/connection.py

@@ -0,0 +1,54 @@
+from enum import Enum
+
+import cachetools.func
+import requests
+import structlog
+
+from codegate.db.connection import DbReader
+
+logger = structlog.get_logger("codegate")
+
+
+# Enum representing whether the request is coming from the front-end or the back-end.
+class Origin(Enum):
+    FrontEnd = "FE"
+    BackEnd = "BE"
+
+
+class UpdateClient:
+    def __init__(self, update_url: str, current_version: str, db_reader: DbReader):
+        self.__update_url = update_url
+        self.__current_version = current_version
+        self.__db_reader = db_reader
+        self.__instance_id = None
+
+    async def get_latest_version(self, origin: Origin) -> str:
+        """
+        Retrieves the latest version of CodeGate from updates.codegate.ai
+        """
+        logger.info(f"Fetching latest version from {self.__update_url}")
+        instance_id = await self.__get_instance_id()
+        return self.__fetch_latest_version(instance_id, origin)
+
+    @cachetools.func.ttl_cache(maxsize=128, ttl=20 * 60)
+    def __fetch_latest_version(self, instance_id: str, origin: Origin) -> str:
+        headers = {
+            "X-Instance-ID": instance_id,
+            "User-Agent": f"codegate/{self.__current_version} {origin.value}",
+        }
+
+        try:
+            response = requests.get(self.__update_url, headers=headers, timeout=10)
+            # Throw if the request was not successful.
+            response.raise_for_status()
+            return response.json()["version"]
+        except Exception as e:
+            logger.error(f"Error fetching latest version from f{self.__update_url}: {e}")
+            return "unknown"
+
+    # Lazy load the instance ID from the DB.
+    async def __get_instance_id(self):
+        if self.__instance_id is None:
+            instance_data = await self.__db_reader.get_instance()
+            self.__instance_id = instance_data[0].id
+        return self.__instance_id