feat: upadte logging in hf spaces

Dockerfile

@@ -40,12 +40,13 @@ COPY --chown=appuser *.py ./
 COPY --chown=appuser --from=frontend-build /app/frontend/dist ./frontend/dist
 
 # Data directories (HF cache, engine state, trained models)
-RUN mkdir -p /data/huggingface /data/engine_state /data/trained_model \
+RUN mkdir -p /data/huggingface /data/engine_state /data/w2v_state /data/trained_model \
     && chown -R appuser:appuser /app /data
 
 ENV HF_HOME=/data/huggingface
 ENV TRANSFORMERS_CACHE=/data/huggingface
 ENV ENGINE_STATE_DIR=/data/engine_state
+ENV W2V_STATE_DIR=/data/w2v_state
 
 # Switch to non-root user
 USER appuser

frontend/src/api.ts

@@ -107,6 +107,9 @@ export const api = {
   getStats: () =>
     client.get<CorpusStats>("/stats").then(r => r.data),
 
+  pollLogs: (cursor: number = 0) =>
+    client.get<{ lines: string[]; cursor: number }>(`/logs/poll?cursor=${cursor}`).then(r => r.data),
+
   getCorpusTexts: (maxDocs: number = 500) =>
     client.get<{ documents: { doc_id: string; text: string }[]; count: number }>(`/corpus/texts?max_docs=${maxDocs}`).then(r => r.data),
 

frontend/src/components/LogViewer.tsx

@@ -1,39 +1,41 @@
 import { useState, useEffect, useRef } from "react";
+import { api } from "../api";
 
 interface Props {
-  /** Whether to actively stream logs */
+  /** Whether to actively poll for logs */
   active: boolean;
 }
 
 export default function LogViewer({ active }: Props) {
   const [lines, setLines] = useState<string[]>([]);
   const containerRef = useRef<HTMLDivElement>(null);
+  const cursorRef = useRef(0);
 
   useEffect(() => {
     if (!active) return;
 
     setLines([]);
-    const evtSource = new EventSource("/api/logs/stream");
+    cursorRef.current = 0;
 
-    evtSource.onmessage = (event) => {
-      setLines((prev) => {
-        const next = [...prev, event.data];
-        // Keep last 200 lines
-        return next.length > 200 ? next.slice(-200) : next;
-      });
-    };
+    const interval = setInterval(async () => {
+      try {
+        const res = await api.pollLogs(cursorRef.current);
+        if (res.lines.length > 0) {
+          setLines((prev) => {
+            const next = [...prev, ...res.lines];
+            return next.length > 200 ? next.slice(-200) : next;
+          });
+        }
+        cursorRef.current = res.cursor;
+      } catch {
+        // ignore polling errors
+      }
+    }, 800);
 
-    evtSource.onerror = () => {
-      // SSE will auto-reconnect, no action needed
-    };
-
-    return () => {
-      evtSource.close();
-    };
+    return () => clearInterval(interval);
   }, [active]);
 
   useEffect(() => {
-    // Auto-scroll to bottom
     if (containerRef.current) {
       containerRef.current.scrollTop = containerRef.current.scrollHeight;
     }

server.py

@@ -133,7 +133,9 @@ app = FastAPI(
 
 app.add_middleware(
     CORSMiddleware,
-    allow_origins=["http://localhost:5173", "http://localhost:3000"],
+    allow_origins=["http://localhost:5173", "http://localhost:3000",
+                    "https://huggingface.co", "https://*.hf.space"],
+    allow_origin_regex=r"https://.*\.hf\.space",
     allow_credentials=True,
     allow_methods=["GET", "POST"],
     allow_headers=["Content-Type", "Authorization"],
@@ -194,6 +196,13 @@ async def stream_logs():
     )
 
 
+@app.get("/api/logs/poll")
+def poll_logs(cursor: int = Query(default=0, ge=0)):
+    """Polling fallback for log streaming (works through HF Spaces proxy)."""
+    lines, new_cursor = log_buffer.get_new_lines(cursor)
+    return {"lines": lines, "cursor": new_cursor}
+
+
 # ------------------------------------------------------------------ #
 #  Request models (with input validation)
 # ------------------------------------------------------------------ #