app.py

"""
app.py
------
FastAPI service exposing the Reddit upvotes prediction pipeline.

Endpoints
---------
POST /predict_single   — single prediction via hours_ago and post_utc
POST /predict_batch    — batch prediction via lists of hours_ago and post_utc
GET  /health           — liveness check
"""

import sys
import os
from pathlib import Path

# Fix sys.path so we can import the refactored code correctly
PROJECT_ROOT = Path(__file__).resolve().parent.parent.parent.parent.parent
sys.path.append(str(PROJECT_ROOT))
sys.path.append(str(PROJECT_ROOT / "models"))

from fastapi import FastAPI, HTTPException
from pydantic import BaseModel, Field, field_validator
import uvicorn
from typing import List, Dict, Any

from models.reddit_predict import predict_single, predict_batch

app = FastAPI(
    title="Reddit Upvotes Predictor",
    description="Predicts the number of upvotes a Reddit post will receive.",
    version="1.0.0",
)

# Reference the artifacts directly in this specific 'test' directory as requested
LOCAL_ARTIFACTS_MODEL_PATH = Path(__file__).resolve().parent / "artifacts" / "model_reddit_final.pth"

# ---------------------------------------------------------------------------
# Request / Response schemas
# ---------------------------------------------------------------------------

class PostContent(BaseModel):
    """Shared post content fields, inherited by all request models."""
    subreddit: str = Field(..., description="Subreddit name (e.g. 'MachineLearning')")
    title: str     = Field(..., min_length=1, description="Post title")
    text: str      = Field(..., min_length=1, description="Post body text")


class PredictSingleRequest(PostContent):
    """Single prediction."""
    hours_ago: float
    post_utc: float


class PredictBatchRequest(PostContent):
    """Batch prediction."""
    hours_ago_list: List[float] = Field(..., min_length=1)
    post_utc_list: List[float] = Field(..., min_length=1)

    @field_validator("post_utc_list")
    @classmethod
    def validate_lengths(cls, v: List[float], info: Any) -> List[float]:
        if 'hours_ago_list' in info.data and len(v) != len(info.data['hours_ago_list']):
            raise ValueError("hours_ago_list and post_utc_list must have the same length.")
        return v


class PredictSingleResponse(BaseModel):
    prediction: float


class BatchPredictionItem(BaseModel):
    hours_ago: float
    post_utc: float
    prediction: float


class PredictBatchResponse(BaseModel):
    predictions: List[BatchPredictionItem]
    count: int


# ---------------------------------------------------------------------------
# Endpoints
# ---------------------------------------------------------------------------

@app.get("/health", tags=["Meta"])
def health():
    """Liveness check."""
    return {"status": "ok"}


@app.post(
    "/predict_single",
    response_model=PredictSingleResponse,
    summary="Single prediction with mutual specific timestamp",
    tags=["Predict"],
)
def api_predict_single(request: PredictSingleRequest):
    """
    Predict upvotes for a single post context given both hours_ago and post_utc.
    """
    try:
        result = predict_single(
            subreddit=request.subreddit,
            title=request.title,
            text=request.text,
            hours_ago=request.hours_ago,
            post_utc=request.post_utc,
            model_path=LOCAL_ARTIFACTS_MODEL_PATH
        )
    except (ValueError, FileNotFoundError) as e:
        raise HTTPException(status_code=422, detail=str(e))

    return PredictSingleResponse(prediction=result)


@app.post(
    "/predict_batch",
    response_model=PredictBatchResponse,
    summary="Batch prediction with matching length lists",
    tags=["Predict"],
)
def api_predict_batch(request: PredictBatchRequest):
    """
    Predict upvotes for an array of potential posting times.
    """
    try:
        results = predict_batch(
            subreddit=request.subreddit,
            title=request.title,
            text=request.text,
            hours_ago_list=request.hours_ago_list,
            post_utc_list=request.post_utc_list,
            model_path=LOCAL_ARTIFACTS_MODEL_PATH
        )
    except (ValueError, FileNotFoundError) as e:
        raise HTTPException(status_code=422, detail=str(e))

    return PredictBatchResponse(predictions=results, count=len(results))


# ---------------------------------------------------------------------------
# Entry point
# ---------------------------------------------------------------------------

if __name__ == "__main__":
    uvicorn.run(app, host="0.0.0.0", port=7860)














# git add <filename>
# git commit -m <message>
# git push

# git remote -v ???
# git config --global credential.helper store

# uvicorn app:app --reload

# git commit conventions?

# https://cyrilfrl-test.hf.space/greet

# git commit conventions

# curl -X POST https://cyrilfrl-test.hf.space/greet -H "Content-Type: application/json" -d '{"name": "Cyril"}'
# (dev) C:\Users\Cyril\Desktop\B3Q2\startup\app\HuggingFace\RedditPredictUpvotes\test>curl -X POST https://cyrilfrl-test.hf.space/greet -H "Content-Type: application/json" -d "{\"name\": \"Cyril\"}"
# {"message":"Hello Cyril!!"}
# (dev) C:\Users\Cyril\Desktop\B3Q2\startup\app\HuggingFace\RedditPredictUpvotes\test>
# curl -X POST "http://127.0.0.1:8000/greet" -H "Content-Type: application/json" -d "{\"name\": \"Cyril\"}"