app.py

import logging
from contextlib import asynccontextmanager
from typing import Optional

from fastapi import FastAPI, HTTPException, Query
from fastapi.middleware.cors import CORSMiddleware
from fastapi.staticfiles import StaticFiles
from pydantic import BaseModel

from word_vectors import WordVectorAnalyzer

logging.basicConfig(
    level=logging.INFO,
    format="%(asctime)s [%(levelname)s] %(name)s — %(message)s",
)
logger = logging.getLogger(__name__)

analyzer: Optional[WordVectorAnalyzer] = None


@asynccontextmanager
async def lifespan(app: FastAPI):
    global analyzer
    logger.info("Starting up — loading Word2Vec model...")
    try:
        analyzer = WordVectorAnalyzer()
        logger.info("Model loaded and ready.")
    except Exception:
        logger.exception("Failed to load model")
    yield
    logger.info("Shutting down.")


app = FastAPI(title="Word2Vec Galaxy", lifespan=lifespan)

app.add_middleware(
    CORSMiddleware,
    allow_origins=["*"],
    allow_methods=["GET", "POST"],
    allow_headers=["Content-Type"],
)


@app.get("/api/health")
def health():
    return {"status": "ready" if analyzer is not None else "loading",
            "model_loaded": analyzer is not None}


@app.get("/api/similar")
def get_similar(
    word: str = Query(..., min_length=1, max_length=100),
    n: int = Query(20, ge=5, le=50),
):
    if analyzer is None:
        raise HTTPException(503, detail="Model is still loading — please wait.")

    similar = analyzer.find_similar_words(word, n)
    if not similar:
        raise HTTPException(404, detail=f"'{word}' not found in vocabulary.")

    words, vectors = analyzer.reduce_dimensions(similar)

    similarities: list[float] = []
    for w in words:
        if w == word:
            similarities.append(1.0)
        else:
            try:
                similarities.append(float(analyzer.model.similarity(word, w)))
            except Exception:
                similarities.append(0.5)

    return {
        "target": word,
        "words": words,
        "vectors": vectors.tolist(),
        "similarities": similarities,
    }


class AnalogyRequest(BaseModel):
    word1: str   # subtracted  (e.g. "man")
    word2: str   # added       (e.g. "woman")
    word3: str   # base        (e.g. "king")


@app.post("/api/analogy")
def get_analogy(req: AnalogyRequest):
    if analyzer is None:
        raise HTTPException(503, detail="Model is still loading — please wait.")

    result, error = analyzer.word_analogy(req.word1, req.word2, req.word3)
    if error:
        raise HTTPException(400, detail=error)

    words_in = [req.word1, req.word2, req.word3, result]
    valid_words, vectors = analyzer.reduce_dimensions(words_in)

    return {
        "word1": req.word1,
        "word2": req.word2,
        "word3": req.word3,
        "result": result,
        "words": valid_words,
        "vectors": vectors.tolist(),
    }

app.mount("/", StaticFiles(directory="static", html=True), name="static")