app.py

from typing import List, Union
from fastapi import FastAPI, HTTPException
from pydantic import BaseModel
from transformers import AutoTokenizer
from optimum.onnxruntime import ORTModelForSequenceClassification
import torch

MODEL_DIR = "onnx_model"

CLASS_LABELS = [
    "الطويل", "البسيط", "الكامل", "الوافر", "الهزج",
    "الرجز", "الرمل", "السريع", "المنسرح", "الخفيف",
    "المضارع", "المقتضب", "المجتث", "المتقارب", "المحدث"
]

app = FastAPI(title="Arabic Poetry Meter Predictor (ONNX)")

tokenizer = None
model = None
id2label = None


class PredictRequest(BaseModel):
    sentences: Union[str, List[str]]


@app.on_event("startup")
def startup_event():
    global tokenizer, model, id2label
    tokenizer = AutoTokenizer.from_pretrained(MODEL_DIR)
    model = ORTModelForSequenceClassification.from_pretrained(MODEL_DIR)

    config_labels = getattr(model.config, "id2label", None)
    if isinstance(config_labels, dict) and config_labels:
        fixed = {}
        for k, v in config_labels.items():
            try:
                fixed[int(k)] = v
            except Exception:
                fixed[k] = v
        id2label = fixed
    else:
        id2label = {i: label for i, label in enumerate(CLASS_LABELS)}


def normalize_sentences(value: Union[str, List[str]]) -> List[str]:
    if isinstance(value, str):
        value = value.strip()
        return [value] if value else []

    if isinstance(value, list):
        return [str(v).strip() for v in value if str(v).strip()]

    return []


@app.get("/")
def root():
    return {"message": "Arabic Poetry Meter Predictor API is running"}


@app.get("/health")
def health():
    return {"status": "ok"}



@app.get("/test")
def test():
    return {"arabic": "مرحبا هذا اختبار"}


@app.post("/predict")
def predict(req: PredictRequest):
    if tokenizer is None or model is None:
        raise HTTPException(status_code=500, detail="Model is not loaded")

    sentences = normalize_sentences(req.sentences)
    if not sentences:
        raise HTTPException(status_code=400, detail="No poetry lines provided")

    try:
        inputs = tokenizer(
            sentences,
            return_tensors="pt",
            padding=True,
            truncation=True,
            max_length=256,
        )

        outputs = model(**inputs)
        probs = torch.softmax(outputs.logits, dim=-1)
        top_scores, top_indices = torch.max(probs, dim=-1)

    except Exception as e:
        raise HTTPException(status_code=500, detail=f"Prediction failed: {str(e)}")

    results = []
    for line, score, idx in zip(sentences, top_scores.tolist(), top_indices.tolist()):
        label = id2label.get(int(idx), f"LABEL_{idx}")
        results.append({
            "line": line,
            "predictions": [
                {
                    "rank": 1,
                    "label": label,
                    "score": float(score),
                }
            ],
        })

    return results