tools/read_file.py

import os
import requests
import tempfile
import pandas as pd
from io import StringIO, BytesIO

def read_file(file_source: str) -> str:
    """
    Read a file from a URL or local path and return its content as text.
    Supports CSV, Excel (.xlsx), and plain text files.
    """
    try:
        is_url = file_source.startswith(("http://", "https://"))
        if is_url:
            headers = {"Authorization": f"Bearer {os.getenv('HF_TOKEN')}"} if "huggingface.co" in file_source else {}
            response = requests.get(file_source, headers=headers, timeout=15)
            response.raise_for_status()
            content_type = response.headers.get('Content-Type', '').lower()
            if 'excel' in content_type or file_source.endswith('.xlsx'):
                df = pd.read_excel(BytesIO(response.content))
                return df.to_csv(index=False)[:8000]
            elif 'csv' in content_type or file_source.endswith('.csv'):
                df = pd.read_csv(StringIO(response.text))
                return df.to_csv(index=False)[:8000]
            else:
                return response.text[:8000]
        else:
            if not os.path.exists(file_source):
                return f"File not found: {file_source}"
            ext = os.path.splitext(file_source)[1].lower()
            if ext == ".xlsx":
                df = pd.read_excel(file_source)
                return df.to_csv(index=False)[:8000]
            elif ext == ".csv":
                df = pd.read_csv(file_source)
                return df.to_csv(index=False)[:8000]
            else:
                with open(file_source, "r", encoding="utf-8", errors="replace") as f:
                    return f.read(8000)
    except Exception as e:
        return f"Error reading file: {str(e)}"