Source code for autojob.harvest.harvesters.gaussian

"""Gaussian calculation harvesting utilities.

This module defines the :func:`.harvest_gaussian_results` function which
can be used to harvest Gaussian results from a directory.

Example:
    from pathlib import Path
    from autojob.harvest.harvesters.gaussian import harvest_gaussian_results

    outputs = harvest_gaussian_results(Path.cwd())
"""

import logging
from pathlib import Path
from typing import Any

import ase.io
import cclib.io.ccio
from cclib.parser.data import ccData

logger = logging.getLogger(__name__)

FILES_TO_CARRYOVER = ["Gaussian.chk"]
GAUSSIAN_LOG = "Gaussian.log"




def harvest_gaussian_results(
    src: str | Path,
) -> dict[str, Any]:
    """Harvest Gaussian calculation results from a directory.

    Note that all quantities other than the final energy are reported in atomic
    units (Hartree/Bohr).

    Args:
        src: The directory containing the Gaussian output files.

    Returns:
        A dictionary containing Gaussian calculation outputs.
    """
    logger.debug(
        "Loading calculation outputs for Gaussian calculation in directory: %s",
        src,
    )
    log_file = Path(src, GAUSSIAN_LOG)
    results: dict[str, Any] = {"forces": None, "energy": None}

    if not log_file.exists():
        logger.warning(
            "Gaussian output file %s does not exist in directory: %s",
            GAUSSIAN_LOG,
            src,
        )
    elif cc_data := cclib.io.ccio.ccread(str(log_file)):
        cc_data.listify()
        data = cc_data.getattributes()

        if energies := data.get("scfenergies"):
            results["energy"] = energies[-1] if energies else None

        if forces := data.get("grads"):
            results["forces"] = forces[-1] if forces else None

        results["converged"] = bool(
            # optimization statuses are defined in bit value notation
            getattr(cc_data, "optstatus", False)
            and (cc_data.optstatus[-1] & ccData.OPT_DONE)
        )
        atoms = ase.io.read(Path(src, GAUSSIAN_LOG))
        results["calculator_results"] = {**data, "atoms": atoms}

        logger.debug(
            "Successfully loaded Gaussian calculation outputs "
            "from directory: %s",
            src,
        )
    else:
        logger.warning(
            "Unable to parse Gaussian outputs from directory: %s", src
        )

    return results