Upload folder using huggingface_hub

Dockerfile

@@ -1,16 +1,12 @@
-# Read the doc: https://huggingface.co/docs/hub/spaces-sdks-docker
-# you will also find guides on how best to write your Dockerfile
+FROM python:3.11-slim
 
-FROM python:3.9
+WORKDIR /app
 
-RUN useradd -m -u 1000 user
-USER user
-ENV PATH="/home/user/.local/bin:$PATH"
+COPY requirements.txt .
+RUN pip install --no-cache-dir -r requirements.txt
 
-WORKDIR /app
+COPY app.py .
 
-COPY --chown=user ./requirements.txt requirements.txt
-RUN pip install --no-cache-dir --upgrade -r requirements.txt
+EXPOSE 7860
 
-COPY --chown=user . /app
 CMD ["uvicorn", "app:app", "--host", "0.0.0.0", "--port", "7860"]

README.md

@@ -1,11 +1,40 @@
----
-title: Urtox Api
-emoji: 🐢
-colorFrom: pink
-colorTo: purple
-sdk: docker
-pinned: false
-license: mit
----
-
-Check out the configuration reference at https://huggingface.co/docs/hub/spaces-config-reference
+# URTOX Hugging Face Space API
+
+This folder is a small FastAPI backend scaffold for the open-house deployment.
+
+## Local run
+
+```bash
+pip install -r requirements.txt
+uvicorn app:app --host 0.0.0.0 --port 7860
+```
+
+Then set the React app environment variable:
+
+```bash
+REACT_APP_API_URL=http://localhost:7860
+```
+
+## Hugging Face Spaces
+
+Create a new Space with:
+
+- SDK: Docker
+- Root files from this folder: `Dockerfile`, `app.py`, `requirements.txt`
+- Port: `7860`
+
+This scaffold currently returns demo predictions. Replace `demo_prediction()` in `app.py` with the real model inference from your notebooks.
+
+## Test endpoint
+
+After the Space starts, call:
+
+```bash
+POST https://your-space-name.hf.space/detect
+Content-Type: application/json
+
+{
+  "mode": "text",
+  "text": "yeh bad aur toxic jumla hai"
+}
+```

app.py

@@ -1,7 +1,86 @@
+from typing import Optional
+
 from fastapi import FastAPI
+from fastapi.middleware.cors import CORSMiddleware
+from pydantic import BaseModel
+
+
+app = FastAPI(title="URTOX Toxic Span Detection API")
+
+app.add_middleware(
+    CORSMiddleware,
+    allow_origins=["*"],
+    allow_credentials=True,
+    allow_methods=["*"],
+    allow_headers=["*"],
+)
+
+
+class DetectRequest(BaseModel):
+    mode: str
+    text: Optional[str] = None
+    audio: Optional[str] = None
+
+
+def demo_prediction(text: str, mode: str):
+    tokens = [token for token in text.split() if token]
+    if not tokens:
+        tokens = ["demo", "toxic", "span", "result"]
+
+    toxic_hints = {"bad", "hate", "idiot", "stupid", "gali", "toxic"}
+    toxic_indexes = [
+        index
+        for index, token in enumerate(tokens)
+        if any(hint in token.lower() for hint in toxic_hints)
+    ]
+    if not toxic_indexes:
+        toxic_indexes = [1 if len(tokens) > 1 else 0]
+
+    words = []
+    for index, token in enumerate(tokens):
+        is_toxic = index in toxic_indexes
+        words.append(
+            {
+                "text": token,
+                "toxic": is_toxic,
+                "bioTag": "B-Toxic" if is_toxic and index == toxic_indexes[0] else ("I-Toxic" if is_toxic else "O"),
+                "confidence": 0.89 if is_toxic else 0.08,
+            }
+        )
+
+    return {
+        "isToxic": True,
+        "confidence": 0.91,
+        "subLabel": "offensive",
+        "subLabelConfidence": 0.87,
+        "toxicSpanCount": len(toxic_indexes),
+        "transcript": text if mode == "audio" else None,
+        "words": words,
+        "xai": {
+            "modelExplanation": "Demo API response. Replace demo_prediction with the real XLM-R/Wav2Vec2 inference pipeline.",
+            "topToxicTokens": [
+                {
+                    "token": tokens[index],
+                    "attribution": 0.74,
+                    "confidence": 0.89,
+                }
+                for index in toxic_indexes
+            ],
+            "integratedGradients": 0.78,
+        },
+    }
 
-app = FastAPI()
 
 @app.get("/")
-def greet_json():
-    return {"Hello": "World!"}
+def health():
+    return {"status": "ok", "service": "urtox-api"}
+
+
+@app.post("/detect")
+def detect(payload: DetectRequest):
+    if payload.mode == "audio":
+        text = "Demo transcript from uploaded or recorded Urdu audio"
+    else:
+        text = payload.text or "yeh demo toxic span detection result hai"
+
+    return demo_prediction(text, payload.mode)

requirements.txt

@@ -1,2 +1,3 @@
-fastapi
-uvicorn[standard]
+fastapi==0.115.6
+uvicorn[standard]==0.34.0
+python-multipart==0.0.20