fix(craft): fix bad escape in re.sub replacement by using lambda

fix(craft): inject script to rewrite next/font URLs injected client-side by React
fix(craft): remove PROVISIONING from offline HTML auto-refresh condition
2026-03-11 18:52:39 +00:00 · 2026-03-11 08:43:09 -07:00 · 2026-03-10 19:20:43 -07:00 · 2026-03-10 19:01:05 -07:00 · 2026-03-10 18:39:26 -07:00 · 2026-03-10 17:01:17 -07:00
9 changed files with 195 additions and 16 deletions
--- a/backend/onyx/server/auth_check.py
+++ b/backend/onyx/server/auth_check.py
@@ -64,6 +64,13 @@ PUBLIC_ENDPOINT_SPECS = [
    ("/build/sessions/{session_id}/webapp/{path:path}", {"GET"}),
 ]

+# WebSocket routes have no `methods` attribute so they can't go in PUBLIC_ENDPOINT_SPECS.
+# These paths are access-controlled inside their handlers via _check_webapp_access.
+PUBLIC_WEBSOCKET_PATHS = {
+    "/build/sessions/{session_id}/webapp/_next/webpack-hmr",
+    "/build/sessions/{session_id}/webapp/_next/hmr",
+}
+

 def is_route_in_spec_list(
    route: BaseRoute, public_endpoint_specs: list[tuple[str, set[str]]]
@@ -114,6 +121,21 @@ def check_router_auth(
        if is_route_in_spec_list(route, public_endpoint_specs):
            continue

+        # WebSocket routes have no methods; check against the explicit WebSocket allowlist
+        if not hasattr(route, "methods") and hasattr(route, "path"):
+            if route.path in PUBLIC_WEBSOCKET_PATHS:
+                continue
+            # Strip global prefix for comparison
+            processed_global_prefix = (
+                f"/{APP_API_PREFIX.strip('/')}" if APP_API_PREFIX else ""
+            )
+            if processed_global_prefix and route.path.startswith(
+                processed_global_prefix
+            ):
+                stripped = route.path[len(processed_global_prefix) :]
+                if stripped in PUBLIC_WEBSOCKET_PATHS:
+                    continue
+
        # check for auth
        found_auth = False
        route_dependant_obj = cast(
--- a/backend/onyx/server/features/build/api/api.py
+++ b/backend/onyx/server/features/build/api/api.py
@@ -1,3 +1,5 @@
+import asyncio
+import re
 from collections.abc import Iterator
 from pathlib import Path
 from uuid import UUID
@@ -8,6 +10,8 @@ from fastapi import Depends
 from fastapi import HTTPException
 from fastapi import Request
 from fastapi import Response
+from fastapi import WebSocket
+from fastapi import WebSocketDisconnect
 from fastapi.responses import RedirectResponse
 from fastapi.responses import StreamingResponse
 from sqlalchemy.orm import Session
@@ -20,6 +24,7 @@ from onyx.db.engine.sql_engine import get_session
 from onyx.db.enums import ConnectorCredentialPairStatus
 from onyx.db.enums import IndexingStatus
 from onyx.db.enums import ProcessingMode
+from onyx.db.enums import SandboxStatus
 from onyx.db.enums import SharingScope
 from onyx.db.index_attempt import get_latest_index_attempt_for_cc_pair_id
 from onyx.db.models import BuildSession
@@ -239,18 +244,45 @@ def _stream_response(response: httpx.Response) -> Iterator[bytes]:
        yield chunk


+def _inject_asset_fixer(content: bytes, session_id: str) -> bytes:
+    """Inject a script that rewrites /_next/ in dynamically inserted <style> tags.
+
+    next/font in dev mode injects @font-face declarations via React client-side,
+    bypassing server-side proxy rewriting. Patching appendChild/insertBefore
+    intercepts these before the browser parses the font URL.
+    """
+    base = f"/api/build/sessions/{session_id}/webapp"
+    script = (
+        f"<script>(function(){{var B='{base}';"
+        "function f(n){if(!n||n.nodeType!==1)return;"
+        "if(n.tagName==='STYLE'&&n.textContent)"
+        "n.textContent=n.textContent.replace(/(url\\s*\\(\\s*['\"]?)\\/_next\\//g,'$1'+B+'/_next/');"
+        "else if(n.tagName==='LINK'){var h=n.getAttribute('href')||'';"
+        "if(h.indexOf('/_next/')===0)n.setAttribute('href',B+h);}}"
+        "function w(m){var o=Element.prototype[m];"
+        "Element.prototype[m]=function(n){f(n);return o.apply(this,arguments);}}"
+        "w('appendChild');w('insertBefore');})()</script>"
+    )
+    text = content.decode("utf-8")
+    # Inject immediately after <head> so it runs before React initialises
+    text = re.sub(
+        r"(<head\b[^>]*>)",
+        lambda m: m.group(0) + script,
+        text,
+        count=1,
+        flags=re.IGNORECASE,
+    )
+    return text.encode("utf-8")
+
+
 def _rewrite_asset_paths(content: bytes, session_id: str) -> bytes:
    """Rewrite Next.js asset paths to go through the proxy."""
-    import re
-
-    # Base path includes session_id for routing
    webapp_base_path = f"/api/build/sessions/{session_id}/webapp"

    text = content.decode("utf-8")
-    # Rewrite /_next/ paths to go through our proxy
-    text = text.replace("/_next/", f"{webapp_base_path}/_next/")
-    # Rewrite JSON data file fetch paths (e.g., /data.json, /data/tickets.json)
-    # Matches paths like "/filename.json" or "/path/to/file.json"
+    # Anchor on delimiter so already-prefixed URLs (from assetPrefix) aren't double-rewritten.
+    for delim in ('"', "'", "("):
+        text = text.replace(f"{delim}/_next/", f"{delim}{webapp_base_path}/_next/")
    text = re.sub(
        r'"(/(?:[a-zA-Z0-9_-]+/)*[a-zA-Z0-9_-]+\.json)"',
        f'"{webapp_base_path}\\1"',
@@ -261,7 +293,6 @@ def _rewrite_asset_paths(content: bytes, session_id: str) -> bytes:
        f"'{webapp_base_path}\\1'",
        text,
    )
-    # Rewrite favicon
    text = text.replace('"/favicon.ico', f'"{webapp_base_path}/favicon.ico')
    return text.encode("utf-8")

@@ -348,6 +379,8 @@ def _proxy_request(
            # For HTML/CSS/JS responses, rewrite asset paths
            if any(ct in content_type for ct in REWRITABLE_CONTENT_TYPES):
                content = _rewrite_asset_paths(response.content, str(session_id))
+                if "text/html" in content_type:
+                    content = _inject_asset_fixer(content, str(session_id))
                return Response(
                    content=content,
                    status_code=response.status_code,
@@ -394,13 +427,22 @@ def _check_webapp_access(
 _OFFLINE_HTML_PATH = Path(__file__).parent / "templates" / "webapp_offline.html"


-def _offline_html_response() -> Response:
+def _offline_html_response(auto_refresh: bool) -> Response:
    """Return a branded Craft HTML page when the sandbox is not reachable.

    Design mirrors the default Craft web template (outputs/web/app/page.tsx):
    terminal window aesthetic with Minecraft-themed typing animation.
+
+    Args:
+        auto_refresh: When True, inject a meta refresh tag so the page polls
+            for the sandbox to wake up. Should only be True when the sandbox is
+            genuinely asleep (SLEEPING/TERMINATED), not when it is RUNNING but
+            Next.js is momentarily restarting — that case causes jarring
+            periodic iframe reloads and the frontend's own polling handles it.
    """
    html = _OFFLINE_HTML_PATH.read_text()
+    meta_tag = '<meta http-equiv="refresh" content="15" />' if auto_refresh else ""
+    html = html.replace("{auto_refresh_meta}", meta_tag)
    return Response(content=html, status_code=503, media_type="text/html")


@@ -435,10 +477,61 @@ def get_webapp(
        return _proxy_request(path, request, session_id, db_session)
    except HTTPException as e:
        if e.status_code in (502, 503, 504):
-            return _offline_html_response()
+            session = db_session.get(BuildSession, session_id)
+            sandbox = (
+                get_sandbox_by_user_id(db_session, session.user_id)
+                if session and session.user_id
+                else None
+            )
+            sandbox_is_asleep = sandbox is None or sandbox.status in (
+                SandboxStatus.SLEEPING,
+                SandboxStatus.TERMINATED,
+            )
+            return _offline_html_response(auto_refresh=sandbox_is_asleep)
        raise


+async def _hmr_websocket_sink(
+    websocket: WebSocket,
+    session_id: UUID,
+    user: User | None,
+    db_session: Session,
+) -> None:
+    """Accept the HMR WebSocket silently to prevent the retry/reload cycle."""
+    try:
+        _check_webapp_access(session_id, user, db_session)
+    except Exception:
+        await websocket.close(code=4003)
+        return
+
+    await websocket.accept()
+    try:
+        while True:
+            await asyncio.sleep(30)
+    except WebSocketDisconnect:
+        pass
+
+
+@public_build_router.websocket("/sessions/{session_id}/webapp/_next/webpack-hmr")
+async def webapp_hmr_sink_webpack(
+    websocket: WebSocket,
+    session_id: UUID,
+    user: User | None = Depends(optional_user),
+    db_session: Session = Depends(get_session),
+) -> None:
+    await _hmr_websocket_sink(websocket, session_id, user, db_session)
+
+
+@public_build_router.websocket("/sessions/{session_id}/webapp/_next/hmr")
+async def webapp_hmr_sink_turbopack(
+    websocket: WebSocket,
+    session_id: UUID,
+    user: User | None = Depends(optional_user),
+    db_session: Session = Depends(get_session),
+) -> None:
+    await _hmr_websocket_sink(websocket, session_id, user, db_session)
+
+
 # =============================================================================
 # Sandbox Management Endpoints
 # =============================================================================
--- a/backend/onyx/server/features/build/api/templates/webapp_offline.html
+++ b/backend/onyx/server/features/build/api/templates/webapp_offline.html
@@ -3,7 +3,7 @@
  <head>
    <meta charset="UTF-8" />
    <meta name="viewport" content="width=device-width, initial-scale=1.0" />
-    <meta http-equiv="refresh" content="15" />
+    {auto_refresh_meta}
    <title>Craft — Starting up</title>
    <style>
      *,
--- a/backend/onyx/server/features/build/sandbox/kubernetes/docker/templates/outputs/web/next.config.ts
+++ b/backend/onyx/server/features/build/sandbox/kubernetes/docker/templates/outputs/web/next.config.ts
@@ -1,7 +1,8 @@
 import type { NextConfig } from "next";

 const nextConfig: NextConfig = {
-  /* config options here */
+  // Routes all asset URLs (including HMR WebSocket) through the Craft proxy path.
+  assetPrefix: process.env.CRAFT_ASSET_PREFIX || undefined,
 };

 export default nextConfig;
--- a/backend/onyx/server/features/build/sandbox/kubernetes/kubernetes_sandbox_manager.py
+++ b/backend/onyx/server/features/build/sandbox/kubernetes/kubernetes_sandbox_manager.py
@@ -118,6 +118,7 @@ RESOURCE_DELETION_POLL_INTERVAL_SECONDS = 0.5
 def _build_nextjs_start_script(
    session_path: str,
    nextjs_port: int,
+    session_id: UUID,
    check_node_modules: bool = False,
 ) -> str:
    """Build shell script to start the NextJS dev server.
@@ -146,6 +147,7 @@ cd {session_path}/outputs/web
 {npm_install_check}
 # Start npm run dev in background
 echo "Starting Next.js dev server on port {nextjs_port}..."
+export CRAFT_ASSET_PREFIX=/api/build/sessions/{session_id}/webapp
 nohup npm run dev -- -p {nextjs_port} > {session_path}/nextjs.log 2>&1 &
 NEXTJS_PID=$!
 echo "Next.js server started with PID $NEXTJS_PID"
@@ -1329,7 +1331,7 @@ fi

        # Build NextJS startup script (npm install already done in outputs_setup)
        nextjs_start_script = _build_nextjs_start_script(
-            session_path, nextjs_port, check_node_modules=False
+            session_path, nextjs_port, session_id=session_id, check_node_modules=False
        )

        setup_script = f"""
@@ -1703,7 +1705,10 @@ echo "SNAPSHOT_RESTORED"

            # Start NextJS dev server (check node_modules since restoring from snapshot)
            start_script = _build_nextjs_start_script(
-                safe_session_path, nextjs_port, check_node_modules=True
+                safe_session_path,
+                nextjs_port,
+                session_id=session_id,
+                check_node_modules=True,
            )
            k8s_stream(
                self._stream_core_api.connect_get_namespaced_pod_exec,
--- a/backend/onyx/server/features/build/sandbox/local/local_sandbox_manager.py
+++ b/backend/onyx/server/features/build/sandbox/local/local_sandbox_manager.py
@@ -526,7 +526,7 @@ class LocalSandboxManager(SandboxManager):
            logger.info(f"Starting Next.js server at {web_dir} on port {nextjs_port}")

            nextjs_process = self._process_manager.start_nextjs_server(
-                web_dir, nextjs_port
+                web_dir, nextjs_port, session_id=session_id
            )
            # Store process for clean shutdown on session delete
            with self._nextjs_lock:
@@ -843,7 +843,7 @@ class LocalSandboxManager(SandboxManager):
                    )
                    return
                process = self._process_manager.start_nextjs_server(
-                    web_dir, nextjs_port
+                    web_dir, nextjs_port, session_id=session_id
                )
                with self._nextjs_lock:
                    self._nextjs_processes[process_key] = process
--- a/backend/onyx/server/features/build/sandbox/local/process_manager.py
+++ b/backend/onyx/server/features/build/sandbox/local/process_manager.py
@@ -8,6 +8,7 @@ import time
 import urllib.error
 import urllib.request
 from pathlib import Path
+from uuid import UUID

 from onyx.utils.logger import setup_logger

@@ -27,6 +28,7 @@ class ProcessManager:
        self,
        web_dir: Path,
        port: int,
+        session_id: UUID,
        timeout: float = 180.0,
    ) -> subprocess.Popen[bytes]:
        """Start Next.js dev server.
@@ -76,6 +78,10 @@ class ProcessManager:
            cwd=web_dir,
            stdout=None,
            stderr=None,
+            env={
+                **os.environ,
+                "CRAFT_ASSET_PREFIX": f"/api/build/sessions/{session_id}/webapp",
+            },
        )
        logger.info(f"Next.js process started with PID {process.pid}")

--- a/backend/tests/unit/build/init.py
+++ b/backend/tests/unit/build/init.py
--- a/backend/tests/unit/build/test_rewrite_asset_paths.py
+++ b/backend/tests/unit/build/test_rewrite_asset_paths.py
@@ -0,0 +1,52 @@
+"""Unit tests for _rewrite_asset_paths in the webapp proxy."""
+
+from onyx.server.features.build.api.api import _rewrite_asset_paths
+
+SESSION_ID = "aaaaaaaa-bbbb-cccc-dddd-eeeeeeeeeeee"
+BASE = f"/api/build/sessions/{SESSION_ID}/webapp"
+
+
+def rewrite(html: str) -> str:
+    return _rewrite_asset_paths(html.encode(), SESSION_ID).decode()
+
+
+class TestNextjsPathRewriting:
+    def test_rewrites_bare_next_script_src(self):
+        html = '<script src="/_next/static/chunks/main.js">'
+        result = rewrite(html)
+        assert f'src="{BASE}/_next/static/chunks/main.js"' in result
+        assert '"/_next/' not in result
+
+    def test_rewrites_bare_next_in_single_quotes(self):
+        html = "<link href='/_next/static/css/app.css'>"
+        result = rewrite(html)
+        assert f"'{BASE}/_next/static/css/app.css'" in result
+
+    def test_rewrites_bare_next_in_url_parens(self):
+        html = "background: url(/_next/static/media/font.woff2)"
+        result = rewrite(html)
+        assert f"url({BASE}/_next/static/media/font.woff2)" in result
+
+    def test_no_double_prefix_when_already_proxied(self):
+        """assetPrefix makes Next.js emit already-prefixed URLs — must not double-rewrite."""
+        already_prefixed = f'<script src="{BASE}/_next/static/chunks/main.js">'
+        result = rewrite(already_prefixed)
+        # Should be unchanged
+        assert result == already_prefixed
+        # Specifically, no double path
+        assert f"{BASE}/{BASE}" not in result
+
+    def test_rewrites_favicon(self):
+        html = '<link rel="icon" href="/favicon.ico">'
+        result = rewrite(html)
+        assert f'"{BASE}/favicon.ico"' in result
+
+    def test_rewrites_json_data_path_double_quoted(self):
+        html = 'fetch("/data/tickets.json")'
+        result = rewrite(html)
+        assert f'"{BASE}/data/tickets.json"' in result
+
+    def test_rewrites_json_data_path_single_quoted(self):
+        html = "fetch('/data/items.json')"
+        result = rewrite(html)
+        assert f"'{BASE}/data/items.json'" in result
Author	SHA1	Message	Date
rohoswagger	4b6edc091f	fix(craft): fix bad escape in re.sub replacement by using lambda	2026-03-11 08:43:09 -07:00
rohoswagger	11f3d41a67	fix(craft): inject script to rewrite next/font URLs injected client-side by React	2026-03-10 19:20:43 -07:00
rohoswagger	40cb943354	fix(craft): remove PROVISIONING from offline HTML auto-refresh condition	2026-03-10 19:01:05 -07:00
rohoswagger	1a0850d5d0	fix(craft): register HMR WebSocket paths in auth allowlist	2026-03-10 18:39:26 -07:00
rohoswagger	f8294c84d6	chore(craft): trim verbose comments added during HMR fix	2026-03-10 17:01:17 -07:00
rohoswagger	84eef6d9c1	fix(craft): add WebSocket sink for Next.js HMR to prevent periodic location.reload()	2026-03-10 16:48:07 -07:00
rohoswagger	6da3e1bae2	fix(craft): avoid double-prefixing _next/ paths when assetPrefix already applied	2026-03-10 16:46:17 -07:00
rohoswagger	474eab8fb9	fix(craft): pass CRAFT_ASSET_PREFIX env var to local Next.js dev server	2026-03-10 16:44:18 -07:00
rohoswagger	8b0ca9a66d	fix(craft): pass CRAFT_ASSET_PREFIX env var to K8s Next.js dev server	2026-03-10 16:42:51 -07:00
rohoswagger	bc1f81c342	fix(craft): configure assetPrefix via env var so HMR WebSocket routes through proxy	2026-03-10 16:41:15 -07:00
rohoswagger	2c45123de5	fix(craft): only auto-refresh offline page when sandbox is genuinely asleep	2026-03-10 16:06:47 -07:00