validate.py

# -*- coding: utf-8 -*-
#
import logging
import sys
from functools import wraps
from os import getenv, path
from sys import stderr
from typing import Dict, Iterator, List, Optional, Set, Tuple, Union, cast

import rdflib
from rdflib import BNode, Literal, URIRef
from rdflib.util import from_n3

from .consts import (
    RDF_object,
    RDF_predicate,
    RDF_subject,
    RDF_type,
    RDFS_Resource,
    SH_conforms,
    SH_detail,
    SH_result,
    SH_resultMessage,
    SH_ValidationReport,
    env_truths,
)
from .errors import ReportableRuntimeError, ValidationFailure
from .extras import check_extra_installed
from .functions import apply_functions, gather_functions, unapply_functions
from .monkey import apply_patches, rdflib_bool_patch, rdflib_bool_unpatch
from .pytypes import GraphLike
from .rdfutil import (
    add_baked_in,
    clone_blank_node,
    clone_graph,
    compare_blank_node,
    compare_node,
    inoculate,
    inoculate_dataset,
    load_from_source,
    mix_datasets,
    mix_graphs,
    order_graph_literal,
)
from .rules import apply_rules, gather_rules
from .shapes_graph import ShapesGraph
from .target import apply_target_types, gather_target_types

USE_FULL_MIXIN = getenv("PYSHACL_USE_FULL_MIXIN") in env_truths

log_handler = logging.StreamHandler(stderr)
log = logging.getLogger(__name__)
for h in log.handlers:
    log.removeHandler(h)  # pragma:no cover
log.addHandler(log_handler)
log.setLevel(logging.INFO)
log_handler.setLevel(logging.INFO)


class Validator(object):
    @classmethod
    def _load_default_options(cls, options_dict: dict):
        options_dict.setdefault('debug', False)
        options_dict.setdefault('advanced', False)
        options_dict.setdefault('inference', 'none')
        options_dict.setdefault('inplace', False)
        options_dict.setdefault('use_js', False)
        options_dict.setdefault('iterate_rules', False)
        options_dict.setdefault('abort_on_first', False)
        options_dict.setdefault('allow_infos', False)
        options_dict.setdefault('allow_warnings', False)
        if 'logger' not in options_dict:
            options_dict['logger'] = logging.getLogger(__name__)
            if options_dict['debug']:
                options_dict['logger'].setLevel(logging.DEBUG)

    @classmethod
    def _run_pre_inference(
        cls, target_graph: GraphLike, inference_option: str, logger: Optional[logging.Logger] = None
    ):
        """
        Note, this is the OWL/RDFS pre-inference,
        it is not the Advanced Spec SHACL-Rule inferencing step.
        :param target_graph:
        :type target_graph: rdflib.Graph|rdflib.ConjunctiveGraph|rdflib.Dataset
        :param inference_option:
        :type inference_option: str
        :return:
        :rtype: NoneType
        """
        # Lazy import owlrl
        import owlrl

        from .inference import CustomRDFSOWLRLSemantics, CustomRDFSSemantics

        if logger is None:
            logger = logging.getLogger(__name__)
        try:
            if inference_option == 'rdfs':
                inferencer = owlrl.DeductiveClosure(CustomRDFSSemantics)
            elif inference_option == 'owlrl':
                inferencer = owlrl.DeductiveClosure(owlrl.OWLRL_Semantics)
            elif inference_option == 'both' or inference_option == 'all' or inference_option == 'rdfsowlrl':
                inferencer = owlrl.DeductiveClosure(CustomRDFSOWLRLSemantics)
            else:
                raise ReportableRuntimeError("Don't know how to do '{}' type inferencing.".format(inference_option))
        except Exception as e:  # pragma: no cover
            logger.error("Error during creation of OWL-RL Deductive Closure")
            if isinstance(e, ReportableRuntimeError):
                raise e
            raise ReportableRuntimeError(
                "Error during creation of OWL-RL Deductive Closure\n{}".format(str(e.args[0]))
            )
        if isinstance(target_graph, (rdflib.Dataset, rdflib.ConjunctiveGraph)):
            named_graphs = [
                rdflib.Graph(target_graph.store, i, namespace_manager=target_graph.namespace_manager)
                if not isinstance(i, rdflib.Graph)
                else i
                for i in target_graph.store.contexts(None)
            ]
        else:
            named_graphs = [target_graph]
        try:
            for g in named_graphs:
                inferencer.expand(g)
        except Exception as e:  # pragma: no cover
            logger.error("Error while running OWL-RL Deductive Closure")
            raise ReportableRuntimeError("Error while running OWL-RL Deductive Closure\n{}".format(str(e.args[0])))

    @classmethod
    def create_validation_report(cls, sg, conforms: bool, results: List[Tuple]):
        v_text = "Validation Report\nConforms: {}\n".format(str(conforms))
        result_len = len(results)
        if not conforms and result_len < 1:
            raise RuntimeError("A Non-Conformant Validation Report must have at least one result.")
        if result_len > 0:
            v_text += "Results ({}):\n".format(str(result_len))
        vg = rdflib.Graph(bind_namespaces='core')
        for p, n in sg.graph.namespace_manager.namespaces():
            vg.namespace_manager.bind(p, n)
        vr = BNode()
        vg.add((vr, RDF_type, SH_ValidationReport))
        vg.add((vr, SH_conforms, Literal(conforms)))
        cloned_nodes: Dict[Tuple[GraphLike, str], Union[BNode, URIRef]] = {}
        for result in iter(results):
            _d, _bn, _tr = result
            v_text += _d
            vg.add((vr, SH_result, _bn))
            for tr in iter(_tr):
                s, p, o = tr
                if isinstance(o, tuple):
                    source = o[0]
                    node = o[1]
                    if isinstance(node, Literal):
                        o = node  # No need to clone a literal from the data graph
                    else:
                        _id = str(node)
                        if (source, _id) in cloned_nodes:
                            o = cloned_nodes[(source, _id)]
                        elif isinstance(node, BNode):
                            cloned_nodes[(source, _id)] = o = clone_blank_node(source, node, vg, keepid=True)
                        else:
                            cloned_nodes[(source, _id)] = o = URIRef(_id)
                vg.add((s, p, o))
        return vg, v_text

    def __init__(
        self,
        data_graph: GraphLike,
        *args,
        shacl_graph: Optional[GraphLike] = None,
        ont_graph: Optional[GraphLike] = None,
        options: Optional[dict] = None,
        **kwargs,
    ):
        options = options or {}
        self._load_default_options(options)
        self.options = options  # type: dict
        self.logger = options['logger']  # type: logging.Logger
        self.debug = options['debug']
        self.pre_inferenced = kwargs.pop('pre_inferenced', False)
        self.inplace = options['inplace']
        if not isinstance(data_graph, rdflib.Graph):
            raise RuntimeError("data_graph must be a rdflib Graph object")
        self.data_graph = data_graph  # type: GraphLike
        self._target_graph = None
        self.ont_graph = ont_graph  # type: Optional[GraphLike]
        self.data_graph_is_multigraph = isinstance(self.data_graph, (rdflib.Dataset, rdflib.ConjunctiveGraph))
        if self.ont_graph is not None and isinstance(self.ont_graph, (rdflib.Dataset, rdflib.ConjunctiveGraph)):
            self.ont_graph.default_union = True

        if shacl_graph is None:
            shacl_graph = clone_graph(data_graph, identifier='shacl')
        assert isinstance(shacl_graph, rdflib.Graph), "shacl_graph must be a rdflib Graph object"
        self.shacl_graph = ShapesGraph(shacl_graph, self.debug, self.logger)  # type: ShapesGraph

        if options['use_js']:
            is_js_installed = check_extra_installed('js')
            if is_js_installed:
                self.shacl_graph.enable_js()

    @property
    def target_graph(self):
        return self._target_graph

    def mix_in_ontology(self):
        if USE_FULL_MIXIN:
            if not self.data_graph_is_multigraph:
                return mix_graphs(self.data_graph, self.ont_graph, "inplace" if self.inplace else None)
            return mix_datasets(self.data_graph, self.ont_graph, "inplace" if self.inplace else None)
        if not self.data_graph_is_multigraph:
            if self.inplace:
                to_graph = self.data_graph
            else:
                to_graph = clone_graph(self.data_graph, identifier=self.data_graph.identifier)
            return inoculate(to_graph, self.ont_graph)
        return inoculate_dataset(self.data_graph, self.ont_graph, self.data_graph if self.inplace else None)

    def run(self):
        if self.target_graph is not None:
            the_target_graph = self.target_graph
        else:
            has_cloned = False
            if self.ont_graph is not None:
                self.logger.debug("Cloning DataGraph to temporary memory graph, to add ontology definitions.")
                # creates a copy of self.data_graph, doesn't modify it
                the_target_graph = self.mix_in_ontology()
                has_cloned = True
            else:
                the_target_graph = self.data_graph
            inference_option = self.options.get('inference', 'none')
            if self.inplace and self.debug:
                self.logger.debug("Skipping DataGraph clone because inplace option is passed.")
            if inference_option and not self.pre_inferenced and str(inference_option) != "none":
                if not has_cloned and not self.inplace:
                    self.logger.debug("Cloning DataGraph to temporary memory graph before pre-inferencing.")
                    the_target_graph = clone_graph(the_target_graph)
                    has_cloned = True
                self.logger.debug(f"Running pre-inferencing with option='{inference_option}'.")
                self._run_pre_inference(the_target_graph, inference_option, logger=self.logger)
                self.pre_inferenced = True
            if not has_cloned and not self.inplace and self.options['advanced']:
                # We still need to clone in advanced mode, because of triple rules
                self.logger.debug("Forcing clone of DataGraph because advanced mode is enabled.")
                the_target_graph = clone_graph(the_target_graph)
                has_cloned = True
            if not has_cloned and not self.inplace:
                # No inferencing, no ont_graph, and no advanced mode, now implies inplace mode
                self.logger.debug("Running validation in-place, without modifying the DataGraph.")
                self.inplace = True
            self._target_graph = the_target_graph

        shapes = self.shacl_graph.shapes  # This property getter triggers shapes harvest.
        iterate_rules = self.options.get("iterate_rules", False)
        if self.options['advanced']:
            self.logger.debug("Activating SHACL-AF Features.")
            target_types = gather_target_types(self.shacl_graph)
            advanced = {
                'functions': gather_functions(self.shacl_graph),
                'rules': gather_rules(self.shacl_graph, iterate_rules=iterate_rules),
            }
            for s in shapes:
                s.set_advanced(True)
            apply_target_types(target_types)
        else:
            advanced = {}
        if isinstance(the_target_graph, (rdflib.Dataset, rdflib.ConjunctiveGraph)):
            named_graphs = [
                rdflib.Graph(the_target_graph.store, i, namespace_manager=the_target_graph.namespace_manager)
                if not isinstance(i, rdflib.Graph)
                else i
                for i in the_target_graph.store.contexts(None)
            ]
        else:
            named_graphs = [the_target_graph]
        reports = []
        abort_on_first: bool = bool(self.options.get("abort_on_first", False))
        allow_infos: bool = bool(self.options.get("allow_infos", False))
        allow_warnings: bool = bool(self.options.get("allow_warnings", False))
        non_conformant = False
        aborted = False
        if abort_on_first and self.debug:
            self.logger.debug(
                "Abort on first error is enabled. Will exit at end of first Shape that fails validation."
            )
        if self.debug:
            self.logger.debug(f"Will run validation on {len(named_graphs)} named graph/s.")
        for g in named_graphs:
            if self.debug:
                self.logger.debug(f"Validating DataGraph named {g.identifier}")
            if advanced:
                apply_functions(advanced['functions'], g)
                apply_rules(advanced['rules'], g, iterate=iterate_rules)
            try:
                for s in shapes:
                    _is_conform, _reports = s.validate(
                        g, abort_on_first=abort_on_first, allow_infos=allow_infos, allow_warnings=allow_warnings
                    )
                    non_conformant = non_conformant or (not _is_conform)
                    reports.extend(_reports)
                    if abort_on_first and non_conformant:
                        aborted = True
                        break
                if aborted:
                    break
            finally:
                if advanced:
                    unapply_functions(advanced['functions'], g)
        v_report, v_text = self.create_validation_report(self.shacl_graph, not non_conformant, reports)
        return (not non_conformant), v_report, v_text


def assign_baked_in():
    if getattr(sys, 'frozen', False):
        # runs in a pyinstaller bundle
        HERE = sys._MEIPASS
    else:
        HERE = path.dirname(__file__)
    shacl_file = path.join(HERE, "assets", "shacl.pickle")
    add_baked_in("http://www.w3.org/ns/shacl", shacl_file)
    add_baked_in("https://www.w3.org/ns/shacl", shacl_file)
    add_baked_in("http://www.w3.org/ns/shacl.ttl", shacl_file)
    shacl_shacl_file = path.join(HERE, "assets", "shacl-shacl.pickle")
    add_baked_in("http://www.w3.org/ns/shacl-shacl", shacl_shacl_file)
    add_baked_in("https://www.w3.org/ns/shacl-shacl", shacl_shacl_file)
    add_baked_in("http://www.w3.org/ns/shacl-shacl.ttl", shacl_shacl_file)
    schema_file = path.join(HERE, "assets", "schema.pickle")
    add_baked_in("http://datashapes.org/schema", schema_file)
    add_baked_in("https://datashapes.org/schema", schema_file)
    add_baked_in("http://datashapes.org/schema.ttl", schema_file)


def with_metashacl_shacl_graph_cache(f):
    # noinspection PyPep8Naming
    EMPTY = object()

    @wraps(f)
    def wrapped(*args, **kwargs):
        graph_cache = getattr(wrapped, "graph_cache", None)
        assert graph_cache is not None
        if graph_cache is EMPTY:
            import pickle

            if getattr(sys, 'frozen', False):
                # runs in a pyinstaller bundle
                here_dir = sys._MEIPASS
            else:
                here_dir = path.dirname(__file__)
            pickle_file = path.join(here_dir, "assets", "shacl-shacl.pickle")
            with open(pickle_file, 'rb') as shacl_pickle:
                u = pickle.Unpickler(shacl_pickle, fix_imports=False)
                shacl_shacl_store, identifier = u.load()
            shacl_shacl_graph = rdflib.Graph(store=shacl_shacl_store, identifier=identifier)
            setattr(wrapped, "graph_cache", shacl_shacl_graph)
        return f(*args, **kwargs)

    setattr(wrapped, "graph_cache", EMPTY)
    return wrapped


@with_metashacl_shacl_graph_cache
def meta_validate(shacl_graph: Union[GraphLike, str], inference: Optional[str] = 'rdfs', **kwargs):
    shacl_shacl_graph = meta_validate.graph_cache
    shacl_graph = load_from_source(shacl_graph, rdf_format=kwargs.pop('shacl_graph_format', None), multigraph=True)
    _ = kwargs.pop('meta_shacl', None)
    return validate(shacl_graph, shacl_graph=shacl_shacl_graph, inference=inference, **kwargs)


def validate(
    data_graph: Union[GraphLike, str, bytes],
    *args,
    shacl_graph: Optional[Union[GraphLike, str, bytes]] = None,
    ont_graph: Optional[Union[GraphLike, str, bytes]] = None,
    advanced: Optional[bool] = False,
    inference: Optional[str] = None,
    inplace: Optional[bool] = False,
    abort_on_first: Optional[bool] = False,
    allow_infos: Optional[bool] = False,
    allow_warnings: Optional[bool] = False,
    **kwargs,
):
    """
    :param data_graph: rdflib.Graph or file path or web url of the data to validate
    :type data_graph: rdflib.Graph | str | bytes
    :param args:
    :type args: list
    :param shacl_graph: rdflib.Graph or file path or web url of the SHACL Shapes graph to use to
    validate the data graph
    :type shacl_graph: rdflib.Graph | str | bytes
    :param ont_graph: rdflib.Graph or file path or web url of an extra ontology document to mix into the data graph
    :type ont_graph: rdflib.Graph | str | bytes
    :param advanced: Enable advanced SHACL features, default=False
    :type advanced: bool | None
    :param inference: One of "rdfs", "owlrl", "both", "none", or None
    :type inference: str | None
    :param inplace: If this is enabled, do not clone the datagraph, manipulate it inplace
    :type inplace: bool
    :param abort_on_first: Stop evaluating constraints after first violation is found
    :type abort_on_first: bool | None
    :param allow_infos: Shapes marked with severity of sh:Info will not cause result to be invalid.
    :type allow_infos: bool | None
    :param allow_warnings: Shapes marked with severity of sh:Warning or sh:Info will not cause result to be invalid.
    :type allow_warnings: bool | None
    :param kwargs:
    :return:
    """

    do_debug = kwargs.get('debug', False)
    if do_debug:
        log_handler.setLevel(logging.DEBUG)
        log.setLevel(logging.DEBUG)
    apply_patches()
    assign_baked_in()
    do_check_dash_result = kwargs.pop('check_dash_result', False)  # type: bool
    do_check_sht_result = kwargs.pop('check_sht_result', False)  # type: bool
    if kwargs.get('meta_shacl', False):
        to_meta_val = shacl_graph or data_graph
        conforms, v_r, v_t = meta_validate(to_meta_val, inference=inference, **kwargs)
        if not conforms:
            msg = f"SHACL File does not validate against the SHACL Shapes SHACL (MetaSHACL) file.\n{v_t}"
            log.error(msg)
            raise ReportableRuntimeError(msg)
    do_owl_imports = kwargs.pop('do_owl_imports', False)
    data_graph_format = kwargs.pop('data_graph_format', None)
    # force no owl imports on data_graph
    loaded_dg = load_from_source(
        data_graph, rdf_format=data_graph_format, multigraph=True, do_owl_imports=False, logger=log
    )
    ont_graph_format = kwargs.pop('ont_graph_format', None)
    if ont_graph is not None:
        loaded_og = load_from_source(
            ont_graph, rdf_format=ont_graph_format, multigraph=True, do_owl_imports=do_owl_imports, logger=log
        )
    else:
        loaded_og = None
    shacl_graph_format = kwargs.pop('shacl_graph_format', None)
    if shacl_graph is not None:
        rdflib_bool_patch()
        loaded_sg = load_from_source(
            shacl_graph, rdf_format=shacl_graph_format, multigraph=True, do_owl_imports=do_owl_imports, logger=log
        )
        rdflib_bool_unpatch()
    else:
        loaded_sg = None
    use_js = kwargs.pop('js', None)
    iterate_rules = kwargs.pop('iterate_rules', False)
    if "abort_on_error" in kwargs:
        log.warning("Usage of abort_on_error is deprecated. Use abort_on_first instead.")
        ae = kwargs.pop("abort_on_error")
        abort_on_first = bool(abort_on_first) or bool(ae)
    validator = None
    try:
        validator = Validator(
            loaded_dg,
            shacl_graph=loaded_sg,
            ont_graph=loaded_og,
            options={
                'debug': do_debug or False,
                'inference': inference,
                'inplace': inplace,
                'abort_on_first': abort_on_first,
                'allow_infos': allow_infos,
                'allow_warnings': allow_warnings,
                'advanced': advanced,
                'iterate_rules': iterate_rules,
                'use_js': use_js,
                'logger': log,
            },
        )
        conforms, report_graph, report_text = validator.run()
    except ValidationFailure as e:
        conforms = False
        report_graph = e
        report_text = "Validation Failure - {}".format(e.message)
    if do_check_dash_result and validator is not None:
        passes = check_dash_result(validator, report_graph, loaded_sg or loaded_dg)
        return passes, report_graph, report_text
    if do_check_sht_result:
        (sht_graph, sht_result_node) = kwargs.pop('sht_validate', (False, None))
        if not sht_result_node:
            raise RuntimeError("Cannot check SHT result if SHT graph and result node are not given.")
        passes = check_sht_result(report_graph, sht_graph or loaded_sg or loaded_dg, sht_result_node)
        return passes, report_graph, report_text
    do_serialize_report_graph = kwargs.pop('serialize_report_graph', False)
    if do_serialize_report_graph and isinstance(report_graph, rdflib.Graph):
        if not (isinstance(do_serialize_report_graph, str)):
            do_serialize_report_graph = 'turtle'
        report_graph = report_graph.serialize(None, encoding='utf-8', format=do_serialize_report_graph)
    return conforms, report_graph, report_text


def clean_validation_reports(actual_graph, actual_report, expected_graph, expected_report):
    # remove rdfs-added stuff
    # remove resultMessage if expected_report does not include result_message
    # remove sh:detail if expected_report does not include details
    # expected_graph.remove((expected_report, RDF_type, RDFS_Resource))
    # actual_graph.remove((actual_report, RDF_type, RDFS_Resource))
    expected_graph.remove((None, RDF_type, RDFS_Resource))
    actual_graph.remove((None, RDF_type, RDFS_Resource))
    expected_results = list(expected_graph.objects(expected_report, SH_result))
    actual_results = list(actual_graph.objects(actual_report, SH_result))
    er_has_messages = None
    er_has_details = False
    for er in expected_results:
        expected_graph.remove((er, RDF_type, RDFS_Resource))
        er_has_messages = list(expected_graph.objects(er, SH_resultMessage))
        er_has_details = er_has_details or expected_graph.value(er, SH_detail) is not None
        # sourceShapes = list(expected_graph.objects(er, SH_sourceShape))
        # for s in sourceShapes:
        #     expected_graph.remove((s, RDF_type, RDFS_Resource))
        # resultPaths = list(expected_graph.objects(er, SH_resultPath))
        # for r in resultPaths:
        #     expected_graph.remove((r, RDF_type, RDFS_Resource))
        # sourceConstraints = list(expected_graph.objects(er, SH_sourceConstraint))
        # for s in sourceConstraints:
        #     expected_graph.remove((s, RDF_type, RDFS_Resource))
    if er_has_messages and len(er_has_messages) > 0:
        # keep messages in actual
        pass
    else:
        for ar in actual_results:
            actual_graph.remove((ar, SH_resultMessage, None))
    if not er_has_details:
        # If no expected result had details, remove all details from actual
        for ar in actual_results:
            for detail in actual_graph.objects(ar, SH_detail):
                actual_graph -= actual_graph.cbd(detail)
                actual_graph.remove((ar, SH_detail, detail))
    return True


def compare_validation_reports(report_graph: GraphLike, expected_graph: GraphLike, expected_result):
    expected_conforms_i = expected_graph.objects(expected_result, SH_conforms)
    expected_conforms = set(cast(Iterator[Literal], expected_conforms_i))
    if len(expected_conforms) < 1:  # pragma: no cover
        raise ReportableRuntimeError(
            "Cannot check the expected result, the given expectedResult does not have an sh:conforms."
        )
    expected_conform = next(iter(expected_conforms))
    expected_result_nodes = expected_graph.objects(expected_result, SH_result)
    expected_result_nodes_set = set(expected_result_nodes)
    expected_result_node_count = len(expected_result_nodes_set)

    validation_reports = report_graph.subjects(RDF_type, SH_ValidationReport)
    validation_reports_set = set(validation_reports)
    if len(validation_reports_set) < 1:  # pragma: no cover
        raise ReportableRuntimeError(
            "Cannot check the validation report, the report graph does not contain a ValidationReport"
        )
    validation_report = next(iter(validation_reports_set))
    clean_validation_reports(report_graph, validation_report, expected_graph, expected_result)
    eq = compare_blank_node(report_graph, validation_report, expected_graph, expected_result)
    if eq != 0:
        return False
    report_conforms_i = report_graph.objects(validation_report, SH_conforms)
    report_conforms = set(cast(Iterator[Literal], report_conforms_i))
    if len(report_conforms) < 1:  # pragma: no cover
        raise ReportableRuntimeError(
            "Cannot check the validation report, the report graph does not have an sh:conforms."
        )
    report_conform = next(iter(report_conforms))

    if bool(expected_conform.value) != bool(report_conform.value):
        # TODO:coverage: write a test for this
        log.error("Expected Result Conforms value is different from Validation Report's Conforms value.")
        return False

    report_result_nodes_i = report_graph.objects(validation_report, SH_result)
    report_result_node_count = len(set(report_result_nodes_i))

    if expected_result_node_count != report_result_node_count:
        # TODO:coverage: write a test for this
        log.error(
            "Number of expected result's sh:result entries is different from Validation Report's sh:result entries.\n"
            "Expected {}, got {}.".format(expected_result_node_count, report_result_node_count)
        )
        return False
    return True


def compare_inferencing_reports(data_graph: GraphLike, expected_graph: GraphLike, expected_results: Union[List, Set]):
    all_good = True
    for expected_result in expected_results:
        expected_objects = set(expected_graph.objects(expected_result, RDF_object))
        if len(expected_objects) < 1:
            raise ReportableRuntimeError(
                "Cannot check the expected result, the given expectedResult does not have an rdf:object."
            )
        expected_object = next(iter(expected_objects))
        expected_subjects = set(expected_graph.objects(expected_result, RDF_subject))
        if len(expected_subjects) < 1:
            raise ReportableRuntimeError(
                "Cannot check the expected result, the given expectedResult does not have an rdf:subject."
            )
        expected_subject = next(iter(expected_subjects))
        expected_predicates = set(expected_graph.objects(expected_result, RDF_predicate))
        if len(expected_predicates) < 1:
            raise ReportableRuntimeError(
                "Cannot check the expected result, the given expectedResult does not have an rdf:predicate."
            )
        expected_predicate = next(iter(expected_predicates))
        if isinstance(expected_object, Literal):
            found_objs = set(data_graph.objects(expected_subject, expected_predicate))
            if len(found_objs) < 1:
                all_good = False
                print("Found no sub/pred matching {} {}".format(expected_subject, expected_predicate))
                continue
            found = False
            for o in found_objs:
                if isinstance(o, Literal):
                    found = 0 == order_graph_literal(expected_graph, expected_object, data_graph, o)
                    if found:
                        break
            if not found:
                print(
                    "Found no sub/pred/obj matching {} {} {}".format(
                        expected_subject, expected_predicate, expected_object
                    )
                )
            all_good = all_good and found
            continue

        elif isinstance(expected_object, BNode):
            found_objs = set(data_graph.objects(expected_subject, expected_predicate))
            if len(found_objs) < 1:
                all_good = False
                print("Found no sub/pred matching {} {}".format(expected_subject, expected_predicate))
                continue
            found = False
            for o in found_objs:
                if isinstance(o, BNode):
                    found = 0 == compare_blank_node(expected_graph, expected_object, data_graph, o)
                    if found:
                        break
            if not found:
                print(
                    "Found no sub/pred/obj matching {} {} {}".format(
                        expected_subject, expected_predicate, expected_object
                    )
                )
            all_good = all_good and found
            continue
        else:
            found_triples = set(data_graph.triples((expected_subject, expected_predicate, expected_object)))
            if len(found_triples) < 1:
                all_good = False

    return all_good


def check_dash_result(validator: Validator, report_graph: GraphLike, expected_result_graph: GraphLike):
    DASH = rdflib.namespace.Namespace('http://datashapes.org/dash#')
    DASH_GraphValidationTestCase = DASH.GraphValidationTestCase
    DASH_InferencingTestCase = DASH.InferencingTestCase
    DASH_FunctionTestCase = DASH.FunctionTestCase
    DASH_expectedResult = DASH.expectedResult
    DASH_expression = DASH.expression
    was_default_union = None
    if isinstance(expected_result_graph, (rdflib.ConjunctiveGraph, rdflib.Dataset)):
        was_default_union = expected_result_graph.default_union
        expected_result_graph.default_union = True  # Force default-union to make all of this a bit easier
    gv_test_cases = expected_result_graph.subjects(RDF_type, DASH_GraphValidationTestCase)
    gv_test_cases_set = set(gv_test_cases)
    inf_test_cases = expected_result_graph.subjects(RDF_type, DASH_InferencingTestCase)
    inf_test_cases_set = set(inf_test_cases)
    fn_test_cases = expected_result_graph.subjects(RDF_type, DASH_FunctionTestCase)
    fn_test_cases_set = set(fn_test_cases)
    if len(gv_test_cases_set) > 0:
        test_case = next(iter(gv_test_cases_set))
        expected_results = expected_result_graph.objects(test_case, DASH_expectedResult)
        expected_results_set = set(expected_results)
        if len(expected_results_set) < 1:  # pragma: no cover
            raise ReportableRuntimeError(
                "Cannot check the expected result, the given GraphValidationTestCase does not have an expectedResult."
            )
        expected_result = next(iter(expected_results_set))
        gv_res: Union[bool, None] = compare_validation_reports(report_graph, expected_result_graph, expected_result)
    else:
        gv_res = None
    if len(inf_test_cases_set) > 0:
        data_graph = validator.target_graph
        if isinstance(data_graph, (rdflib.ConjunctiveGraph, rdflib.Dataset)):
            named_graphs = list(data_graph.contexts())
        else:
            named_graphs = [data_graph]
        inf_res: Union[bool, None] = True
        for test_case in inf_test_cases_set:
            expected_results = expected_result_graph.objects(test_case, DASH_expectedResult)
            expected_results_set = set(expected_results)
            if len(expected_results_set) < 1:  # pragma: no cover
                raise ReportableRuntimeError(
                    "Cannot check the expected result, the given InferencingTestCase does not have an expectedResult."
                )
            found = False
            for g in named_graphs:
                found = found or compare_inferencing_reports(g, expected_result_graph, expected_results_set)
            inf_res = inf_res and found
    else:
        inf_res = None
    if len(fn_test_cases_set) > 0:
        data_graph = validator.target_graph
        fns = gather_functions(validator.shacl_graph)
        apply_functions(fns, data_graph)
        fn_res: Union[bool, None] = True
        for test_case in fn_test_cases_set:
            expected_results_set = set(expected_result_graph.objects(test_case, DASH_expectedResult))
            if len(expected_results_set) < 1:  # pragma: no cover
                raise ReportableRuntimeError(
                    "Cannot check the expected result, the given FunctionTestCase does not have an expectedResult."
                )
            expected_result = next(iter(expected_results_set))
            expressions = set(expected_result_graph.objects(test_case, DASH_expression))
            if len(expressions) < 1:
                raise ReportableRuntimeError(
                    "Cannot check the expected result, the given FunctionTestCase does not have an expression."
                )
            expression_node = next(iter(expressions))
            expression = str(expression_node).strip()
            parts = [e.strip() for e in expression.split("(", 1)]
            if len(parts) < 1:
                expression = parts[0]
                eargs: List[Union[str, URIRef]] = []
            else:
                expression, sargs = parts
                sargs = sargs.rstrip(")")
                if len(sargs) < 1:
                    eargs = []
                else:
                    eargs = [a.strip() for a in sargs.split(',')]
                eargs = [
                    from_n3(e, None, expected_result_graph.store, expected_result_graph.namespace_manager)
                    for e in eargs
                ]
            find_uri = from_n3(expression, None, expected_result_graph.store, expected_result_graph.namespace_manager)
            try:
                fn, options = validator.shacl_graph.get_shacl_function(find_uri)
            except KeyError:
                raise ReportableRuntimeError(
                    "Cannot execute function {}.\nCannot find it in the ShapesGraph object.".format(find_uri)
                )
            result = fn(data_graph, *eargs)
            fn_res = fn_res and 0 == compare_node(expected_result_graph, expected_result, data_graph, result)

    else:
        fn_res = None
    if was_default_union is not None:
        expected_result_graph.default_union = was_default_union
    if gv_res is None and inf_res is None and fn_res is None:  # pragma: no cover
        raise ReportableRuntimeError(
            "Cannot check the expected result, the given expected result graph does not have a GraphValidationTestCase or InferencingTestCase."
        )
    return (gv_res or gv_res is None) and (inf_res or inf_res is None) and (fn_res or fn_res is None)


def check_sht_result(report_graph: GraphLike, sht_graph: GraphLike, sht_result_node: Union[URIRef, BNode]):
    SHT = rdflib.namespace.Namespace('http://www.w3.org/ns/shacl-test#')
    types = set(sht_graph.objects(sht_result_node, RDF_type))
    expected_failure = sht_result_node == SHT.Failure
    if expected_failure and isinstance(report_graph, ValidationFailure):
        return True
    elif isinstance(report_graph, ValidationFailure):
        # TODO:coverage: write a test for this
        log.error("Validation Report indicates a Validation Failure, but the SHT entry does not expect a failure.")
        return False
    elif expected_failure:
        # TODO:coverage: write a test for this
        log.error(
            "SHT expects a Validation Failure, but the Validation Report does not indicate a Validation Failure."
        )
        return False
    if SH_ValidationReport not in types:
        raise ReportableRuntimeError("SHT expected result must have type sh:ValidationReport")
    return compare_validation_reports(report_graph, sht_graph, sht_result_node)