Merge pull request #73 from lsst-sqre/tickets/DM-43393

DM-43393: Cast job_try to str when making SlackTextField

.github/workflows/ci.yaml

@@ -23,15 +23,15 @@ jobs:
     timeout-minutes: 5
 
     steps:
-      - uses: actions/checkout@v3
+      - uses: actions/checkout@v4
 
       - name: Set up Python
-        uses: actions/setup-python@v4
+        uses: actions/setup-python@v5
         with:
           python-version: "3.12"
 
       - name: Run pre-commit
-        uses: pre-commit/[email protected].0
+        uses: pre-commit/[email protected].1
 
   test:
     runs-on: ubuntu-latest
@@ -43,7 +43,7 @@ jobs:
           - "3.12"
 
     steps:
-      - uses: actions/checkout@v3
+      - uses: actions/checkout@v4
 
       - name: Run tox
         uses: lsst-sqre/run-tox@v1
@@ -67,7 +67,7 @@ jobs:
           || startsWith(github.head_ref, 'tickets/'))
 
     steps:
-      - uses: actions/checkout@v3
+      - uses: actions/checkout@v4
         with:
           fetch-depth: 0
 

.github/workflows/dependencies.yaml

@@ -11,7 +11,7 @@ jobs:
     timeout-minutes: 10
 
     steps:
-      - uses: actions/checkout@v3
+      - uses: actions/checkout@v4
 
       - name: Run neophile
         uses: lsst-sqre/run-neophile@v1

.github/workflows/docs.yaml

@@ -18,10 +18,10 @@ jobs:
     runs-on: ubuntu-latest
 
     steps:
-      - uses: actions/checkout@v3
+      - uses: actions/checkout@v4
 
       - name: Set up Python
-        uses: actions/setup-python@v4
+        uses: actions/setup-python@v5
         with:
           python-version: "3.12"
 

.github/workflows/periodic-ci.yaml

@@ -21,7 +21,7 @@ jobs:
           - "3.12"
 
     steps:
-      - uses: actions/checkout@v3
+      - uses: actions/checkout@v4
 
       # Use the oldest supported version of Python to update dependencies,
       # not the matrixed Python version, since this accurately reflects

CHANGELOG.md

@@ -2,6 +2,14 @@
 
 <!-- scriv-insert-here -->
 
+<a id='changelog-0.9.1'></a>
+## 0.9.1 (2024-03-21)
+
+### Bug fixes
+
+- Fix Slack error messaging in the `nbexec` worker function.
+- Extract and use the actual XSRF token when communicating with the Hub and Lab.
+
 <a id='changelog-0.9.0'></a>
 
 ## 0.9.0 (2024-03-13)

Dockerfile

@@ -14,7 +14,7 @@
 #   - Runs a non-root user.
 #   - Sets up the entrypoint and port.
 
-FROM python:3.12.1-slim-bullseye as base-image
+FROM python:3.12.2-slim-bullseye as base-image
 
 # Update system packages
 COPY scripts/install-base-packages.sh .

src/noteburst/jupyterclient/jupyterlab.py

@@ -6,16 +6,15 @@
 import contextlib
 import datetime
 import json
-import string
 from collections.abc import AsyncGenerator, AsyncIterator
 from dataclasses import dataclass
-from random import SystemRandom
-from typing import Annotated, Any, Self
+from typing import Annotated, Any
 from urllib.parse import urljoin, urlparse
 from uuid import uuid4
 
 import httpx
 import websockets
+from httpx import Cookies
 from pydantic import BaseModel, Field
 from structlog import BoundLogger
 from websockets.client import WebSocketClientProtocol
@@ -422,83 +421,35 @@ def __init__(
             str(noteburst_config.environment_url), self.config.url_prefix
         )
 
-        self._http_client: httpx.AsyncClient | None = None
-        self._lab_controller_client: LabControllerClient | None = None
-        self._common_headers: dict[str, str]  # set and reset in the context
-
-    @property
-    def http_client(self) -> httpx.AsyncClient:
-        """The HTTPX client instance associated with the Jupyter session.."""
-        if self._http_client is None:
-            self._open_clients()
-        if self._http_client is None:
-            raise RuntimeError("http_client is not set")
-        return self._http_client
-
-    @property
-    def lab_controller(self) -> LabControllerClient:
-        """The Jupyter Lab Controller client, only available in the
-        JupyterClient context.
-        """
-        if self._lab_controller_client is None:
-            self._open_clients()
-        if self._lab_controller_client is None:
-            raise RuntimeError("LabControllerClient is not set set up")
-        return self._lab_controller_client
-
-    async def __aenter__(self) -> Self:
-        self._open_clients()
-        return self
-
-    def _open_clients(self) -> None:
-        if (self._http_client is not None) or (
-            self._lab_controller_client is not None
-        ):
-            raise RuntimeError(
-                "JupyterClient is already open. Call close() before "
-                "re-opening?"
-            )
-
-        alphabet = string.ascii_uppercase + string.digits
-        xsrf_token = "".join(SystemRandom().choices(alphabet, k=16))
-        headers = {
-            "x-xsrftoken": xsrf_token,
+        self._headers = {
             "Authorization": f"Bearer {self.user.token}",
         }
-        self._common_headers = headers
-        cookies = {"_xsrf": xsrf_token}
-
-        self._http_client = httpx.AsyncClient(
-            headers=headers,
-            cookies=cookies,
+        self.http_client = httpx.AsyncClient(
+            headers=self._headers,
             follow_redirects=True,
-            timeout=30.0,  # default is 5, but Hub can be slow
+            timeout=30,
         )
+        self._lab_controller_client: LabControllerClient | None = None
+        self._hub_xsrf: str | None = None
+        self._lab_xsrf: str | None = None
+
+    @property
+    def lab_controller(self) -> LabControllerClient:
+        if self._lab_controller_client:
+            return self._lab_controller_client
 
-        # Create a LabController client
-        # We also send the XSRF token to Lab Controller because of how we're
-        # sharing the session, but that shouldn't matter.
         self._lab_controller_client = LabControllerClient(
             http_client=self.http_client,
             token=noteburst_config.gafaelfawr_token.get_secret_value(),
             url_prefix=noteburst_config.nublado_controller_path_prefix,
         )
-
-    async def __aexit__(self, *exc_info: object) -> None:
-        await self.close()
+        return self._lab_controller_client
 
     async def close(self) -> None:
-        """Manually close the client.
-
-        Do not use this method for manually closing the Jupyter client when
-        using JupyterClient as an async context manager. The client is
-        closed automatically.
-        """
+        """Close the client."""
         self._lab_controller_client = None
 
         await self.http_client.aclose()
-        self._http_client = None
-        self._common_headers = {}
 
     def url_for(self, path: str) -> str:
         """Create a URL relative to the jupyter_url."""
@@ -515,14 +466,17 @@ def url_for_websocket(self, path: str) -> str:
     async def log_into_hub(self) -> None:
         """Log into JupyterHub or raise a JupyterError."""
         self.logger.debug("Logging into JupyterHub")
-        r = await self.http_client.get(
-            self.url_for("hub/login"), follow_redirects=False
-        )
+        r = await self.http_client.get(self.url_for("hub/home"))
         # JupyterHub returns a 302 redirect to the login page on success,
         # but we don't want to follow that redirect. This request is just
         # to set cookies.
         if r.status_code >= 400:
             raise JupyterError.from_response(self.user.username, r)
+        cookies = Cookies()
+        cookies.extract_cookies(r)
+        xsrf = cookies.get("_xsrf")
+        if xsrf:
+            self._hub_xsrf = xsrf
 
     async def log_into_lab(self) -> None:
         """Log into JupyterLab or raise a JupyterError."""
@@ -532,6 +486,11 @@ async def log_into_lab(self) -> None:
         )
         if r.status_code != 200:
             raise JupyterError.from_response(self.user.username, r)
+        cookies = Cookies()
+        cookies.extract_cookies(r)
+        xsrf = cookies.get("_xsrf")
+        if xsrf:
+            self._lab_xsrf = xsrf
 
     async def spawn_lab(self) -> JupyterImage:
         """Spawn a JupyterLab pod."""
@@ -540,13 +499,16 @@ async def spawn_lab(self) -> JupyterImage:
         # Retrieving the spawn page before POSTing to it appears to trigger
         # some necessary internal state construction (and also more accurately
         # simulates a user interaction).  See DM-23864.
-        _ = await self.http_client.get(spawn_url)
+        headers = dict(self._headers)
+        if self._hub_xsrf:
+            headers["X-XSRFToken"] = self._hub_xsrf
+        _ = await self.http_client.get(spawn_url, headers=headers)
 
         # POST the options form to the spawn page.  This should redirect to
         # the spawn-pending page, which will return a 200.
         image = await self._get_spawn_image()
         data = self._build_jupyter_spawn_form(image)
-        r = await self.http_client.post(spawn_url, data=data)
+        r = await self.http_client.post(spawn_url, data=data, headers=headers)
         if r.status_code != 200:
             raise JupyterError.from_response(self.user.username, r)
 
@@ -565,6 +527,8 @@ async def spawn_progress(self) -> AsyncIterator[SpawnProgressMessage]:
         )
         referer_url = self.url_for("hub/home")
         headers = {"Referer": referer_url}
+        if self._hub_xsrf:
+            headers["X-XSRFToken"] = self._hub_xsrf
         while True:
             async with self.http_client.stream(
                 "GET", progress_url, headers=headers
@@ -626,6 +590,8 @@ async def stop_lab(self) -> None:
         server_url = self.url_for(f"hub/api/users/{user}/server")
         referer_url = self.url_for("hub/home")
         headers = {"Referer": referer_url}
+        if self._hub_xsrf:
+            headers["X-XSRFToken"] = self._hub_xsrf
         r = await self.http_client.delete(server_url, headers=headers)
         if r.status_code not in [200, 202, 204]:
             raise JupyterError.from_response(self.user.username, r)
@@ -642,6 +608,8 @@ async def is_lab_stopped(self, *, final: bool = False) -> bool:
         user_url = self.url_for(f"hub/api/users/{self.user.username}")
         referer_url = self.url_for("hub/home")
         headers = {"Referer": referer_url}
+        if self._hub_xsrf:
+            headers["X-XSRFToken"] = self._hub_xsrf
         r = await self.http_client.get(user_url, headers=headers)
         if r.status_code != 200:
             raise JupyterError.from_response(self.user.username, r)
@@ -669,7 +637,12 @@ async def open_lab_session(
             "path": notebook_name if notebook_name else uuid4().hex,
             "type": session_type,
         }
-        r = await self.http_client.post(session_url, json=body)
+        headers = {}
+        if self._lab_xsrf:
+            headers["X-XSRFToken"] = self._lab_xsrf
+        r = await self.http_client.post(
+            session_url, json=body, headers=headers
+        )
         if r.status_code != 201:
             raise JupyterError.from_response(self.user.username, r)
         session_resource = r.json()
@@ -685,10 +658,11 @@ async def open_lab_session(
         # Generate a mock request and copy its headers / cookies over to the
         # websocket connection.
         mock_request = self.http_client.build_request("GET", http_channels_uri)
-        copied_headers = ["x-xsrftoken", "authorization", "cookie"]
-        websocket_headers = {
-            header: mock_request.headers[header] for header in copied_headers
+        headers = {
+            h: mock_request.headers[h] for h in ("authorization", "cookie")
         }
+        if self._lab_xsrf:
+            headers["X-XSRFToken"] = self._lab_xsrf
 
         session_id: str | None = None  # will be set if a session is opened
         self.logger.debug("Trying to create websocket connection")
@@ -699,7 +673,7 @@ async def open_lab_session(
             # long lived clients
             # https://websockets.readthedocs.io/en/stable/reference/client.html#using-a-connection
             async with websockets.connect(
-                wss_channels_uri, extra_headers=websocket_headers
+                wss_channels_uri, extra_headers=headers
             ) as websocket:
                 self.logger.info("Created websocket connection")
                 jupyter_lab_session = JupyterLabSession(
@@ -720,7 +694,9 @@ async def open_lab_session(
                 session_id_url = self.url_for(
                     f"user/{self.user.username}/api/sessions/{session_id}"
                 )
-                r = await self.http_client.delete(session_id_url)
+                r = await self.http_client.delete(
+                    session_id_url, headers=headers
+                )
                 if r.status_code != 204:
                     raise JupyterError.from_response(self.user.username, r)
 
@@ -747,10 +723,14 @@ async def execute_notebook(
             Notebook execution extension.
         """
         exec_url = self.url_for(f"user/{self.user.username}/rubin/execution")
+        headers = {}
+        if self._lab_xsrf:
+            headers["X-XSRFToken"] = self._lab_xsrf
         try:
             r = await self.http_client.post(
                 exec_url,
                 content=json.dumps(notebook).encode("utf-8"),
+                headers=headers,
             )
         except httpx.HTTPError as e:
             # This often occurs from timeouts, so we want to convert the
@@ -777,7 +757,10 @@ async def get_jupyterlab_env(self) -> dict[str, Any]:
         environment_url = self.url_for(
             f"user/{self.user.username}/rubin/environment"
         )
-        r = await self.http_client.get(environment_url)
+        headers = {}
+        if self._lab_xsrf:
+            headers["X-XSRFToken"] = self._lab_xsrf
+        r = await self.http_client.get(environment_url, headers=headers)
         if r.status_code != 200:
             raise JupyterError.from_response(self.user.username, r)
         return r.json()

src/noteburst/jupyterclient/labcontroller.py

@@ -137,7 +137,11 @@ class LabControllerClient:
     """
 
     def __init__(
-        self, *, http_client: httpx.AsyncClient, token: str, url_prefix: str
+        self,
+        *,
+        http_client: httpx.AsyncClient,
+        token: str,
+        url_prefix: str,
     ) -> None:
         self._http_client = http_client
         self._token = token
@@ -204,7 +208,9 @@ async def get_by_reference(self, reference: str) -> JupyterImage:
         return image
 
     async def _get_images(self) -> LabControllerImages:
-        headers = {"Authorization": f"bearer {self._token}"}
+        headers = {
+            "Authorization": f"bearer {self._token}",
+        }
         url = urljoin(
             str(config.environment_url),
             f"{self._url_prefix}/spawner/v1/images",