utils/config_loader.py

from pathlib import Path
from typing import List

import yaml

from utils.constants import DEFAULT_COMPRESSION_LEVELS


def load_papers_config(config_path: str) -> List[dict]:
    """
    Load papers configuration from YAML file.

    Args:
        config_path: Path to YAML configuration file

    Returns:
        List of paper dicts with normalized fields: path, name, token_limits

    Raises:
        FileNotFoundError: If config file doesn't exist
        ValueError: If config is invalid or required fields missing
    """
    config_file = Path(config_path)

    if not config_file.exists():
        raise FileNotFoundError(f"Config file not found: {config_path}")

    try:
        with open(config_file, "r") as f:
            config = yaml.safe_load(f)
    except yaml.YAMLError as e:
        raise ValueError(f"Invalid YAML in {config_path}: {e}")

    if not config or "papers" not in config:
        raise ValueError("Config must contain 'papers' key")

    papers = config["papers"]
    if not isinstance(papers, list):
        raise ValueError("'papers' must be a list")

    normalized_papers = []
    for idx, entry in enumerate(papers):
        try:
            normalized = _normalize_paper_entry(entry)
            normalized_papers.append(normalized)
        except ValueError as e:
            raise ValueError(f"Invalid paper entry at index {idx}: {e}")

    return normalized_papers


def _normalize_paper_entry(entry: dict) -> dict:
    """
    Normalize and validate a single paper entry.

    Args:
        entry: Paper entry dict from YAML

    Returns:
        Normalized entry with all required fields and sensible defaults

    Raises:
        ValueError: If required fields missing or invalid
    """
    if not isinstance(entry, dict):
        raise ValueError("Paper entry must be a dictionary")

    # Validate required field: path
    if "path" not in entry:
        raise ValueError("Missing required field: 'path'")

    path = entry["path"]
    if not isinstance(path, str):
        raise ValueError("'path' must be a string")

    if not Path(path).exists():
        raise ValueError(f"Paper path does not exist: {path}")

    # Optional field: name (default to filename)
    name = entry.get("name")
    if name is None:
        name = Path(path).name

    # Optional field: token_limits (default to DEFAULT_COMPRESSION_LEVELS)
    token_limits = entry.get("token_limits")
    if token_limits is None:
        token_limits = DEFAULT_COMPRESSION_LEVELS
    else:
        if not isinstance(token_limits, list):
            raise ValueError("'token_limits' must be a list")

        for limit in token_limits:
            if not isinstance(limit, int) or limit <= 0:
                raise ValueError(
                    f"Token limits must be positive integers, got: {limit}"
                )

    # Optional field: token_tolerances (tolerance values for each token_limit)
    token_tolerances = entry.get("token_tolerances")
    if token_tolerances is not None:
        if not isinstance(token_tolerances, list):
            raise ValueError("'token_tolerances' must be a list")

        for tolerance in token_tolerances:
            if not isinstance(tolerance, int) or tolerance < 0:
                raise ValueError(
                    f"Token tolerances must be non-negative integers, got: {tolerance}"
                )

    # Optional field: context_pdfs (list of reference PDFs)
    context_pdfs = entry.get("context_pdfs")
    if context_pdfs is not None:
        if not isinstance(context_pdfs, list):
            raise ValueError("'context_pdfs' must be a list")

        for pdf_path in context_pdfs:
            if not isinstance(pdf_path, str):
                raise ValueError("context_pdfs must be a list of strings")
            if not Path(pdf_path).exists():
                raise FileNotFoundError(f"Context PDF not found: {pdf_path}")

    return {
        "path": path,
        "name": name,
        "token_limits": token_limits,
        "token_tolerances": token_tolerances,
        "description": entry.get("description", ""),
        "context_pdfs": context_pdfs or [],
    }