Source code for qcelemental.testing

import copy
import logging
import pprint
import sys
from typing import Callable, Dict, List, Tuple, Union

import numpy as np

try:
    from pydantic.v1 import BaseModel
except ImportError:  # Will also trap ModuleNotFoundError
    from pydantic import BaseModel

from qcelemental.models.basemodels import ProtoModel

pp = pprint.PrettyPrinter(width=120)


def _handle_return(passfail: bool, label: str, message: str, return_message: bool, quiet: bool = False):
    """Function to print a '*label*...PASSED' line to log."""

    if not quiet:
        if passfail:
            logging.info(f"    {label:.<53}PASSED")
        else:
            logging.error(f"    {label:.<53}FAILED")
            logging.error(f"    {message:.<53}")

    if return_message:
        return passfail, message
    else:
        return passfail


def tnm() -> str:
    """Returns the name of the calling function, usually name of test case."""

    return sys._getframe().f_back.f_code.co_name


[docs]def compare_values( expected: Union[float, List, np.ndarray], computed: Union[float, List, np.ndarray], label: str = None, *, atol: float = 1.0e-6, rtol: float = 1.0e-16, equal_nan: bool = False, equal_phase: bool = False, passnone: bool = False, quiet: bool = False, return_message: bool = False, return_handler: Callable = None, ) -> Union[bool, Tuple[bool, str]]: r"""Returns True if two floats or float arrays are element-wise equal within a tolerance. Parameters ---------- expected float or float array-like Reference value against which `computed` is compared. computed float or float array-like Input value to compare against `expected`. atol Absolute tolerance (see formula below). label Label for passed and error messages. Defaults to calling function name. rtol Relative tolerance (see formula below). By default set to zero so `atol` dominates. equal_nan Passed to :func:`numpy.isclose`. Compare NaN's as equal. equal_phase Compare computed *or its opposite* as equal. passnone Return True when both expected and computed are None. quiet Whether to log the return message. return_message Whether to return tuple. See below. Returns ------- allclose : bool Returns True if `expected` and `computed` are equal within tolerance; False otherwise. message : str When return_message=True, also return passed or error message. Other Parameters ---------------- return_handler Function to control printing, logging, raising, and returning. Specialized interception for interfacing testing systems. Notes ----- * Akin to :func:`numpy.allclose`. * For scalar float-comparable types and for arbitrary-dimension, np.ndarray-castable, uniform-type, float-comparable types. For mixed types, use :py:func:`compare_recursive`. * Sets rtol to zero to match expected Psi4 behaviour, otherwise measured as: .. code-block:: python absolute(computed - expected) <= (atol + rtol * absolute(expected)) """ label = label or sys._getframe().f_back.f_code.co_name pass_message = f"\t{label:.<66}PASSED" if return_handler is None: return_handler = _handle_return if passnone: if expected is None and computed is None: return return_handler(True, label, pass_message, return_message, quiet) if np.iscomplexobj(expected): dtype = np.complex else: dtype = float try: xptd, cptd = np.array(expected, dtype=dtype), np.array(computed, dtype=dtype) except Exception: return return_handler( False, label, f"""\t{label}: inputs not cast-able to ndarray of {dtype}.""", return_message, quiet ) if xptd.shape != cptd.shape: return return_handler( False, label, f"""\t{label}: computed shape ({cptd.shape}) does not match ({xptd.shape}).""", return_message, quiet, ) digits1 = abs(int(np.log10(atol))) + 2 digits_str = f"to atol={atol}" if rtol > 1.0e-12: digits_str += f", rtol={rtol}" isclose = np.isclose(cptd, xptd, rtol=rtol, atol=atol, equal_nan=equal_nan) allclose = bool(np.all(isclose)) if not allclose and equal_phase and hasattr(cptd, "__neg__"): n_isclose = np.isclose(-cptd, xptd, rtol=rtol, atol=atol, equal_nan=equal_nan) allclose = bool(np.all(n_isclose)) if allclose: message = pass_message else: if xptd.shape == (): xptd_str = f"{float(xptd):.{digits1}f}" else: xptd_str = np.array_str(xptd, max_line_width=120, precision=12, suppress_small=True) xptd_str = "\n".join(" " + ln for ln in xptd_str.splitlines()) if cptd.shape == (): cptd_str = f"{float(cptd):.{digits1}f}" else: cptd_str = np.array_str(cptd, max_line_width=120, precision=12, suppress_small=True) cptd_str = "\n".join(" " + ln for ln in cptd_str.splitlines()) diff = cptd - xptd if xptd.shape == (): diff_str = f"{float(diff):.{digits1}f}" message = """\t{}: computed value ({}) does not match ({}) {} by difference ({}).""".format( label, cptd_str, xptd_str, digits_str, diff_str ) else: diff[isclose] = 0.0 diff_str = np.array_str(diff, max_line_width=120, precision=12, suppress_small=False) diff_str = "\n".join(" " + ln for ln in diff_str.splitlines()) with np.errstate(divide="ignore", invalid="ignore"): diffrel = np.divide(diff, xptd) np.nan_to_num(diffrel, copy=False) diffraw = cptd - xptd digits_str += f" (o-e: RMS {_rms(diffraw):.1e}, MAX {np.amax(np.absolute(diffraw)):.1e}, RMAX {np.amax(np.absolute(diffrel)):.1e})" message = """\t{}: computed value does not match {}.\n Expected:\n{}\n Observed:\n{}\n Difference (passed elements are zeroed):\n{}\n""".format( label, digits_str, xptd_str, cptd_str, diff_str ) return return_handler(allclose, label, message, return_message, quiet)
def _rms(arr: np.ndarray) -> float: return np.sqrt(np.mean(np.square(arr)))
[docs]def compare( expected: Union[int, bool, str, List[int], np.ndarray], computed: Union[int, bool, str, List[int], np.ndarray], label: str = None, *, equal_phase: bool = False, quiet: bool = False, return_message: bool = False, return_handler: Callable = None, ) -> Union[bool, Tuple[bool, str]]: r"""Returns True if two integers, strings, booleans, or integer arrays are element-wise equal. Parameters ---------- expected int, bool, str or array-like of same. Reference value against which `computed` is compared. computed int, bool, str or array-like of same. Input value to compare against `expected`. label Label for passed and error messages. Defaults to calling function name. equal_phase Compare computed *or its opposite* as equal. quiet Whether to log the return message. return_message Whether to return tuple. See below. Returns ------- allclose : bool Returns True if `expected` and `computed` are equal; False otherwise. message : str When return_message=True, also return passed or error message. Other Parameters ---------------- return_handler Function to control printing, logging, raising, and returning. Specialized interception for interfacing testing systems. Notes ----- * Akin to :func:`numpy.array_equal`. * For scalar exactly-comparable types and for arbitrary-dimension, np.ndarray-castable, uniform-type, exactly-comparable types. For mixed types, use :py:func:`compare_recursive`. """ label = label or sys._getframe().f_back.f_code.co_name pass_message = f"\t{label:.<66}PASSED" if return_handler is None: return_handler = _handle_return try: xptd, cptd = np.array(expected), np.array(computed) except Exception: return return_handler(False, label, f"""\t{label}: inputs not cast-able to ndarray.""", return_message, quiet) if xptd.shape != cptd.shape: return return_handler( False, label, f"""\t{label}: computed shape ({cptd.shape}) does not match ({xptd.shape}).""", return_message, quiet, ) isclose = np.asarray(xptd == cptd) allclose = bool(isclose.all()) if not allclose and equal_phase: try: n_isclose = np.asarray(xptd == -cptd) except TypeError: pass else: allclose = bool(n_isclose.all()) if allclose: message = pass_message else: if xptd.shape == (): xptd_str = f"{xptd}" else: xptd_str = np.array_str(xptd, max_line_width=120, precision=12, suppress_small=True) xptd_str = "\n".join(" " + ln for ln in xptd_str.splitlines()) if cptd.shape == (): cptd_str = f"{cptd}" else: cptd_str = np.array_str(cptd, max_line_width=120, precision=12, suppress_small=True) cptd_str = "\n".join(" " + ln for ln in cptd_str.splitlines()) try: diff = cptd - xptd except TypeError: diff_str = "(n/a)" else: if xptd.shape == (): diff_str = f"{diff}" else: diff_str = np.array_str(diff, max_line_width=120, precision=12, suppress_small=False) diff_str = "\n".join(" " + ln for ln in diff_str.splitlines()) if xptd.shape == (): message = """\t{}: computed value ({}) does not match ({}) by difference ({}).""".format( label, cptd_str, xptd_str, diff_str ) else: message = """\t{}: computed value does not match.\n Expected:\n{}\n Observed:\n{}\n Difference:\n{}\n""".format( label, xptd_str, cptd_str, diff_str ) return return_handler(allclose, label, message, return_message, quiet)
def _compare_recursive(expected, computed, atol, rtol, _prefix=False, equal_phase=False): errors = [] name = _prefix or "root" prefix = name + "." # Initial conversions if required if isinstance(expected, BaseModel): expected = expected.dict() if isinstance(computed, BaseModel): computed = computed.dict() if isinstance(expected, (str, int, bool, complex)): if expected != computed: errors.append((name, "Value {} did not match {}.".format(expected, computed))) elif isinstance(expected, (list, tuple)): try: if len(expected) != len(computed): errors.append((name, "Iterable lengths did not match")) else: for i, item1, item2 in zip(range(len(expected)), expected, computed): errors.extend( _compare_recursive( item1, item2, _prefix=prefix + str(i), atol=atol, rtol=rtol, equal_phase=equal_phase ) ) except TypeError: errors.append((name, "Expected computed to have a __len__()")) elif isinstance(expected, dict): expected_extra = computed.keys() - expected.keys() computed_extra = expected.keys() - computed.keys() if len(expected_extra): errors.append((name, "Found extra keys {}".format(expected_extra))) if len(computed_extra): errors.append((name, "Missing keys {}".format(computed_extra))) for k in expected.keys() & computed.keys(): name = prefix + str(k) errors.extend( _compare_recursive( expected[k], computed[k], _prefix=name, atol=atol, rtol=rtol, equal_phase=equal_phase ) ) elif isinstance(expected, (float, np.number)): passfail, msg = compare_values( expected, computed, atol=atol, rtol=rtol, equal_phase=equal_phase, return_message=True, quiet=True ) if not passfail: errors.append((name, "Arrays differ." + msg)) elif isinstance(expected, np.ndarray): if np.issubdtype(expected.dtype, np.floating): passfail, msg = compare_values( expected, computed, atol=atol, rtol=rtol, equal_phase=equal_phase, return_message=True, quiet=True ) else: passfail, msg = compare(expected, computed, equal_phase=equal_phase, return_message=True, quiet=True) if not passfail: errors.append((name, "Arrays differ." + msg)) elif isinstance(expected, type(None)): if expected is not computed: errors.append((name, "'None' does not match.")) else: errors.append((name, f"Type {type(expected)} not understood -- stopping recursive compare.")) return errors
[docs]def compare_recursive( expected: Union[Dict, BaseModel, "ProtoModel"], # type: ignore computed: Union[Dict, BaseModel, "ProtoModel"], # type: ignore label: str = None, *, atol: float = 1.0e-6, rtol: float = 1.0e-16, forgive: List[str] = None, equal_phase: Union[bool, List] = False, quiet: bool = False, return_message: bool = False, return_handler: Callable = None, ) -> Union[bool, Tuple[bool, str]]: r""" Recursively compares nested structures such as dictionaries and lists. Parameters ---------- expected Reference value against which `computed` is compared. Dict may be of any depth but should contain Plain Old Data. computed Input value to compare against `expected`. Dict may be of any depth but should contain Plain Old Data. atol Absolute tolerance (see formula below). label Label for passed and error messages. Defaults to calling function name. rtol Relative tolerance (see formula below). By default set to zero so `atol` dominates. forgive Keys in top level which may change between `expected` and `computed` without triggering failure. equal_phase Compare computed *or its opposite* as equal. quiet Whether to log the return message. return_message Whether to return tuple. See below. Returns ------- allclose : bool Returns True if `expected` and `computed` are equal within tolerance; False otherwise. message : str When return_message=True, also return passed or error message. Notes ----- .. code-block:: python absolute(computed - expected) <= (atol + rtol * absolute(expected)) """ label = label or sys._getframe().f_back.f_code.co_name if atol >= 1: raise ValueError( "Prior to v0.4.0, ``compare_recursive`` used to 10**-atol any atol >=1. That has ceased, so please express your atol literally." ) if return_handler is None: return_handler = _handle_return errors = _compare_recursive(expected, computed, atol=atol, rtol=rtol) if errors and equal_phase: n_errors = _compare_recursive(expected, computed, atol=atol, rtol=rtol, equal_phase=True) n_errors = dict(n_errors) if equal_phase is False: equal_phase = [] elif equal_phase is True: equal_phase = list(dict(errors).keys()) else: equal_phase = [(ep if ep.startswith("root.") else "root." + ep) for ep in equal_phase] phased = [] for nomatch in sorted(errors): for ep in equal_phase or []: if nomatch[0].startswith(ep): if nomatch[0] not in n_errors: phased.append(nomatch) errors.remove(nomatch) if forgive is None: forgive = [] else: forgive = [(fg if fg.startswith("root.") else "root." + fg) for fg in forgive] forgiven = [] for nomatch in sorted(errors): for fg in forgive or []: if nomatch[0].startswith(fg): forgiven.append(nomatch) errors.remove(nomatch) ## print if verbose >= 2 if these functions had that knob # forgiven_message = [] # for e in sorted(forgiven): # forgiven_message.append(e[0]) # forgiven_message.append("forgiven " + e[1]) # pprint.pprint(forgiven) message = [] for e in sorted(errors): message.append(e[0]) message.append(" " + e[1]) ret_msg_str = "\n".join(message) return return_handler(len(ret_msg_str) == 0, label, ret_msg_str, return_message, quiet)
[docs]def compare_molrecs( expected, computed, label: str = None, *, atol: float = 1.0e-6, rtol: float = 1.0e-16, forgive=None, verbose: int = 1, relative_geoms="exact", return_message: bool = False, return_handler: Callable = None, ) -> bool: """Function to compare Molecule dictionaries.""" # Need to manipulate the dictionaries a bit, so hold values xptd = copy.deepcopy(expected) cptd = copy.deepcopy(computed) def massage_dicts(dicary): # if 'fix_symmetry' in dicary: # dicary['fix_symmetry'] = str(dicary['fix_symmetry']) # if 'units' in dicary: # dicary['units'] = str(dicary['units']) if "fragment_files" in dicary: dicary["fragment_files"] = [str(f) for f in dicary["fragment_files"]] # and about int vs long errors # if 'molecular_multiplicity' in dicary: # dicary['molecular_multiplicity'] = int(dicary['molecular_multiplicity']) # if 'fragment_multiplicities' in dicary: # dicary['fragment_multiplicities'] = [(m if m is None else int(m)) # for m in dicary['fragment_multiplicities']] if "fragment_separators" in dicary: dicary["fragment_separators"] = [(s if s is None else int(s)) for s in dicary["fragment_separators"]] # forgive generator version changes if "provenance" in dicary: dicary["provenance"].pop("version") # regularize connectivity ordering if "connectivity" in dicary: conn = [(min(at1, at2), max(at1, at2), bo) for (at1, at2, bo) in dicary["connectivity"]] conn.sort(key=lambda tup: tup[0]) dicary["connectivity"] = conn return dicary xptd = massage_dicts(xptd) cptd = massage_dicts(cptd) if relative_geoms == "exact": pass elif relative_geoms == "align": # can't just expect geometries to match, so we'll align them, check that # they overlap and that the translation/rotation arrays jibe with # fix_com/orientation, then attach the oriented geom to computed before the # recursive dict comparison. from .molutil.align import B787 cgeom = np.array(cptd["geom"]).reshape((-1, 3)) rgeom = np.array(xptd["geom"]).reshape((-1, 3)) rmsd, mill = B787( rgeom=rgeom, cgeom=cgeom, runiq=None, cuniq=None, atoms_map=True, mols_align=True, run_mirror=False, verbose=0, ) if cptd["fix_com"]: return compare( True, np.allclose(np.zeros((3)), mill.shift, atol=atol), "null shift", quiet=(verbose == 0), return_message=return_message, return_handler=return_handler, ) if cptd["fix_orientation"]: return compare( True, np.allclose(np.identity(3), mill.rotation, atol=atol), "null rotation", quiet=(verbose == 0), return_message=return_message, return_handler=return_handler, ) ageom = mill.align_coordinates(cgeom) cptd["geom"] = ageom.reshape((-1)) return compare_recursive( xptd, cptd, atol=atol, rtol=rtol, label=label, forgive=forgive, quiet=(verbose == 0), return_message=return_message, return_handler=return_handler, )