#!/bin/bash
set -e

# If HF_DATASET_ID is set, download PDFs from HuggingFace Dataset at startup
if [ -n "$HF_DATASET_ID" ]; then
    echo "Downloading PDFs from HF Dataset: $HF_DATASET_ID ..."
    python3 - <<EOF
from huggingface_hub import snapshot_download
import shutil, os

local = snapshot_download(
    repo_id="$HF_DATASET_ID",
    repo_type="dataset",
    local_dir="/tmp/hf_pdfs",
    token=os.environ.get("HF_TOKEN"),
)
pdf_src = os.path.join(local, "downloads")
if os.path.isdir(pdf_src):
    for f in os.listdir(pdf_src):
        if f.endswith(".pdf"):
            shutil.copy2(os.path.join(pdf_src, f), "/app/downloads/" + f)
    print(f"Downloaded {len(os.listdir('/app/downloads'))} PDFs.")
else:
    print("No 'downloads' folder found in dataset — read_doc will be unavailable.")
EOF
fi

exec uvicorn rag.api:app --host 0.0.0.0 --port "${PORT:-7860}"