dsp-tools 9.1.0.post11__py3-none-any.whl → 18.3.0.post13__py3-none-any.whl

dsp_tools/commands/xml_validate/models/data_deserialised.py

@@ -1,108 +0,0 @@
-from __future__ import annotations
-
-from abc import ABC
-from dataclasses import dataclass
-
-
-@dataclass
-class ProjectDeserialised:
-    info: ProjectInformation
-    data: DataDeserialised
-
-
-@dataclass
-class ProjectInformation:
-    shortcode: str
-    default_onto: str
-
-
-@dataclass
-class DataDeserialised:
-    resources: list[AbstractResource]
-
-
-@dataclass
-class AbstractResource(ABC):
-    res_id: str
-    label: str
-
-
-@dataclass
-class ResourceDeserialised(AbstractResource):
-    res_class: str
-    values: list[ValueDeserialised]
-
-
-@dataclass
-class AnnotationDeserialised(AbstractResource): ...
-
-
-@dataclass
-class RegionDeserialised(AbstractResource): ...
-
-
-@dataclass
-class LinkObjDeserialised(AbstractResource): ...
-
-
-@dataclass
-class VideoSegmentDeserialised(AbstractResource): ...
-
-
-@dataclass
-class AudioSegmentDeserialised(AbstractResource): ...
-
-
-@dataclass
-class ValueDeserialised:
-    prop_name: str
-    object_value: str | None
-
-
-@dataclass
-class BooleanValueDeserialised(ValueDeserialised): ...
-
-
-@dataclass
-class ColorValueDeserialised(ValueDeserialised): ...
-
-
-@dataclass
-class DateValueDeserialised(ValueDeserialised): ...
-
-
-@dataclass
-class DecimalValueDeserialised(ValueDeserialised): ...
-
-
-@dataclass
-class GeonameValueDeserialised(ValueDeserialised): ...
-
-
-@dataclass
-class IntValueDeserialised(ValueDeserialised): ...
-
-
-@dataclass
-class LinkValueDeserialised(ValueDeserialised): ...
-
-
-@dataclass
-class ListValueDeserialised(ValueDeserialised):
-    list_name: str
-
-
-@dataclass
-class SimpleTextDeserialised(ValueDeserialised): ...
-
-
-@dataclass
-class RichtextDeserialised(ValueDeserialised): ...
-
-
-@dataclass
-class TimeValueDeserialised(ValueDeserialised): ...
-
-
-@dataclass
-class UriValueDeserialised(ValueDeserialised): ...

dsp_tools/commands/xml_validate/models/data_rdf.py

@@ -1,214 +0,0 @@
-from __future__ import annotations
-
-from abc import ABC
-from dataclasses import dataclass
-from uuid import uuid4
-
-from rdflib import RDF
-from rdflib import RDFS
-from rdflib import Graph
-from rdflib import Literal
-from rdflib import Namespace
-from rdflib import URIRef
-
-KNORA_API = Namespace("http://api.knora.org/ontology/knora-api/v2#")
-API_SHAPES = Namespace("http://api.knora.org/ontology/knora-api/shapes/v2#")
-DATA = Namespace("http://data/")
-
-
-@dataclass
-class DataRDF:
-    triples: list[RDFTriples]
-
-    def make_graph(self) -> Graph:
-        g = Graph()
-        for r in self.triples:
-            g += r.make_graph()
-        return g
-
-
-class RDFTriples(ABC):
-    def make_graph(self) -> Graph:
-        raise NotImplementedError
-
-
-@dataclass
-class ResourceRDF(RDFTriples):
-    res_iri: URIRef
-    res_class: URIRef
-    label: Literal
-
-    def make_graph(self) -> Graph:
-        g = Graph()
-        g.add((self.res_iri, RDF.type, self.res_class))
-        g.add((self.res_iri, RDFS.label, self.label))
-        return g
-
-
-@dataclass
-class ValueRDF(RDFTriples):
-    prop_name: URIRef
-    object_value: URIRef | Literal
-    res_iri: URIRef
-
-    def make_graph(self) -> Graph:
-        raise NotImplementedError
-
-
-@dataclass
-class BooleanValueRDF(ValueRDF):
-    object_value: Literal
-
-    def make_graph(self) -> Graph:
-        g = Graph()
-        val_iri = DATA[str(uuid4())]
-        g.add((val_iri, RDF.type, KNORA_API.BooleanValue))
-        g.add((val_iri, KNORA_API.booleanValueAsBoolean, self.object_value))
-        g.add((self.res_iri, self.prop_name, val_iri))
-        return g
-
-
-@dataclass
-class ColorValueRDF(ValueRDF):
-    object_value: Literal
-
-    def make_graph(self) -> Graph:
-        g = Graph()
-        val_iri = DATA[str(uuid4())]
-        g.add((val_iri, RDF.type, KNORA_API.ColorValue))
-        g.add((val_iri, KNORA_API.colorValueAsColor, self.object_value))
-        g.add((self.res_iri, self.prop_name, val_iri))
-        return g
-
-
-@dataclass
-class DateValueRDF(ValueRDF):
-    object_value: Literal
-
-    def make_graph(self) -> Graph:
-        g = Graph()
-        val_iri = DATA[str(uuid4())]
-        g.add((val_iri, RDF.type, KNORA_API.DateValue))
-        g.add((val_iri, KNORA_API.valueAsString, self.object_value))
-        g.add((self.res_iri, self.prop_name, val_iri))
-        return g
-
-
-@dataclass
-class DecimalValueRDF(ValueRDF):
-    object_value: Literal
-
-    def make_graph(self) -> Graph:
-        g = Graph()
-        val_iri = DATA[str(uuid4())]
-        g.add((val_iri, RDF.type, KNORA_API.DecimalValue))
-        g.add((val_iri, KNORA_API.decimalValueAsDecimal, self.object_value))
-        g.add((self.res_iri, self.prop_name, val_iri))
-        return g
-
-
-@dataclass
-class GeonameValueRDF(ValueRDF):
-    object_value: Literal
-
-    def make_graph(self) -> Graph:
-        g = Graph()
-        val_iri = DATA[str(uuid4())]
-        g.add((val_iri, RDF.type, KNORA_API.GeonameValue))
-        g.add((val_iri, KNORA_API.geonameValueAsGeonameCode, self.object_value))
-        g.add((self.res_iri, self.prop_name, val_iri))
-        return g
-
-
-@dataclass
-class IntValueRDF(ValueRDF):
-    object_value: Literal
-
-    def make_graph(self) -> Graph:
-        g = Graph()
-        val_iri = DATA[str(uuid4())]
-        g.add((val_iri, RDF.type, KNORA_API.IntValue))
-        g.add((val_iri, KNORA_API.intValueAsInt, self.object_value))
-        g.add((self.res_iri, self.prop_name, val_iri))
-        return g
-
-
-@dataclass
-class LinkValueRDF(ValueRDF):
-    object_value: URIRef
-
-    def make_graph(self) -> Graph:
-        g = Graph()
-        val_iri = DATA[str(uuid4())]
-        g.add((val_iri, RDF.type, KNORA_API.LinkValue))
-        g.add((val_iri, API_SHAPES.linkValueHasTargetID, self.object_value))
-        g.add((self.res_iri, self.prop_name, val_iri))
-        return g
-
-
-@dataclass
-class ListValueRDF(ValueRDF):
-    prop_name: URIRef
-    object_value: Literal
-    list_name: Literal
-
-    def make_graph(self) -> Graph:
-        g = Graph()
-        val_iri = DATA[str(uuid4())]
-        g.add((val_iri, RDF.type, KNORA_API.ListValue))
-        g.add((val_iri, API_SHAPES.listNodeAsString, self.object_value))
-        g.add((val_iri, API_SHAPES.listNameAsString, self.list_name))
-        g.add((self.res_iri, self.prop_name, val_iri))
-        return g
-
-
-@dataclass
-class SimpleTextRDF(ValueRDF):
-    object_value: Literal
-
-    def make_graph(self) -> Graph:
-        g = Graph()
-        val_iri = DATA[str(uuid4())]
-        g.add((val_iri, RDF.type, KNORA_API.TextValue))
-        g.add((val_iri, KNORA_API.valueAsString, self.object_value))
-        g.add((self.res_iri, self.prop_name, val_iri))
-        return g
-
-
-@dataclass
-class RichtextRDF(ValueRDF):
-    object_value: Literal
-
-    def make_graph(self) -> Graph:
-        g = Graph()
-        val_iri = DATA[str(uuid4())]
-        g.add((val_iri, RDF.type, KNORA_API.TextValue))
-        g.add((val_iri, KNORA_API.textValueAsXml, self.object_value))
-        g.add((self.res_iri, self.prop_name, val_iri))
-        return g
-
-
-@dataclass
-class TimeValueRDF(ValueRDF):
-    object_value: Literal
-
-    def make_graph(self) -> Graph:
-        g = Graph()
-        val_iri = DATA[str(uuid4())]
-        g.add((val_iri, RDF.type, KNORA_API.TimeValue))
-        g.add((val_iri, KNORA_API.timeValueAsTimeStamp, self.object_value))
-        g.add((self.res_iri, self.prop_name, val_iri))
-        return g
-
-
-@dataclass
-class UriValueRDF(ValueRDF):
-    object_value: Literal
-
-    def make_graph(self) -> Graph:
-        g = Graph()
-        val_iri = DATA[str(uuid4())]
-        g.add((val_iri, RDF.type, KNORA_API.UriValue))
-        g.add((val_iri, KNORA_API.uriValueAsUri, self.object_value))
-        g.add((self.res_iri, self.prop_name, val_iri))
-        return g

dsp_tools/commands/xml_validate/models/input_problems.py

@@ -1,191 +0,0 @@
-from __future__ import annotations
-
-import warnings
-from abc import ABC
-from collections import defaultdict
-from dataclasses import dataclass
-from datetime import datetime
-from pathlib import Path
-
-import pandas as pd
-from rdflib import Graph
-
-from dsp_tools.commands.xml_validate.models.validation import UnexpectedComponent
-from dsp_tools.models.custom_warnings import DspToolsUserWarning
-
-LIST_SEPARATOR = "\n    - "
-GRAND_SEPARATOR = "\n\n----------------------------\n"
-
-
-@dataclass
-class UnexpectedResults:
-    components: list[UnexpectedComponent]
-
-    def save_inform_user(self, results_graph: Graph, shacl: Graph, data: Graph) -> None:
-        cwdr = Path.cwd()
-        prefix = f"{datetime.now()!s}_"
-        components = sorted(x.component_type for x in self.components)
-        msg = (
-            f"Unexpected violations were found in the validation results:"
-            f"{LIST_SEPARATOR}{LIST_SEPARATOR.join(components)}\n"
-            f"Please contact the development team with the files starting with the timestamp '{prefix}' "
-            f"in the directory '{cwdr}'."
-        )
-        save_path = cwdr / f"{prefix}validation_result.ttl"
-        results_graph.serialize(save_path)
-        shacl_p = cwdr / f"{prefix}shacl.ttl"
-        shacl.serialize(shacl_p)
-        data_p = cwdr / f"{prefix}data.ttl"
-        data.serialize(data_p)
-        warnings.warn(DspToolsUserWarning(msg))
-
-
-@dataclass
-class AllProblems:
-    problems: list[InputProblem]
-    unexpected_results: UnexpectedResults | None
-
-    def get_msg(self, file_path: Path) -> str:
-        coll = self._make_collection()
-        title_msg = f"\nDuring the validation of the data {len(self.problems)} errors were found:\n\n"
-        if len(self.problems) > 50:
-            out_path = file_path.parent / f"{file_path.stem}_validation_errors.csv"
-            self._save_as_csv(out_path)
-            out_message = (
-                title_msg + f"Due to the large number or errors, the validation errors were saved at:\n{out_path}"
-            )
-        else:
-            msg = [x.get_msg() for x in coll]
-            out_message = title_msg + GRAND_SEPARATOR.join(msg)
-        return out_message
-
-    def _make_collection(self) -> list[ResourceProblemCollection]:
-        d = defaultdict(list)
-        for e in self.problems:
-            d[e.res_id].append(e)
-        collection_list = []
-        for k, v in d.items():
-            collection_list.append(ResourceProblemCollection(k, v))
-        return sorted(collection_list, key=lambda x: x.res_id)
-
-    def _save_as_csv(self, out_path: Path) -> None:
-        all_problems = [x.to_dict() for x in self.problems]
-        df = pd.DataFrame.from_records(all_problems)
-        df = df.sort_values(by=["Resource Type", "Resource ID", "Property"])
-        df.to_csv(out_path, index=False)
-
-
-@dataclass
-class ResourceProblemCollection:
-    res_id: str
-    problems: list[InputProblem]
-
-    def get_msg(self) -> str:
-        prop_msg = self._msg_for_properties()
-        return f"Resource ID: {self.res_id} | Resource Type: {self.problems[0].res_type}\n{prop_msg}"
-
-    def _msg_for_properties(self) -> str:
-        grouped = self._make_collection()
-        out_list = []
-        for prop_name, problem in grouped.items():
-            problem_list = [x.get_msg() for x in problem]
-            out_list.append((prop_name, f"{prop_name}{LIST_SEPARATOR}{LIST_SEPARATOR.join(problem_list)}"))
-        sorted_list = sorted(out_list, key=lambda x: x[0])
-        return "\n".join([x[1] for x in sorted_list])
-
-    def _make_collection(self) -> dict[str, list[InputProblem]]:
-        grouped_dict = defaultdict(list)
-        for problem in self.problems:
-            grouped_dict[problem.sort_value()].append(problem)
-        return grouped_dict
-
-    def sort_value(self) -> str:
-        return self.res_id
-
-
-@dataclass
-class InputProblem(ABC):
-    res_id: str
-    res_type: str
-    prop_name: str
-
-    @property
-    def problem(self) -> str:
-        raise NotImplementedError
-
-    def get_msg(self) -> str:
-        raise NotImplementedError
-
-    def to_dict(self) -> dict[str, str]:
-        raise NotImplementedError
-
-    def _base_dict(self) -> dict[str, str]:
-        return {
-            "Resource Type": self.res_type,
-            "Resource ID": self.res_id,
-            "Property": self.prop_name,
-            "Problem": self.problem,
-        }
-
-    def sort_value(self) -> str:
-        raise NotImplementedError
-
-
-#######################
-# Cardinality Violation
-
-
-@dataclass
-class MaxCardinalityViolation(InputProblem):
-    expected_cardinality: str
-
-    @property
-    def problem(self) -> str:
-        return "Maximum Cardinality Violation"
-
-    def get_msg(self) -> str:
-        return f"{self.problem} | Expected Cardinality: {self.expected_cardinality}"
-
-    def to_dict(self) -> dict[str, str]:
-        problm_dict = self._base_dict()
-        problm_dict["Expected"] = f"Cardinality: {self.expected_cardinality}"
-        return problm_dict
-
-    def sort_value(self) -> str:
-        return self.prop_name
-
-
-@dataclass
-class MinCardinalityViolation(InputProblem):
-    expected_cardinality: str
-
-    @property
-    def problem(self) -> str:
-        return "Minimum Cardinality Violation"
-
-    def get_msg(self) -> str:
-        return f"{self.problem} | Expected Cardinality: {self.expected_cardinality}"
-
-    def to_dict(self) -> dict[str, str]:
-        problm_dict = self._base_dict()
-        problm_dict["Expected"] = f"Cardinality: {self.expected_cardinality}"
-        return problm_dict
-
-    def sort_value(self) -> str:
-        return self.prop_name
-
-
-@dataclass
-class NonExistentCardinalityViolation(InputProblem):
-    @property
-    def problem(self) -> str:
-        return "The resource class does not have a cardinality for this property."
-
-    def get_msg(self) -> str:
-        return self.problem
-
-    def to_dict(self) -> dict[str, str]:
-        return self._base_dict()
-
-    def sort_value(self) -> str:
-        return self.prop_name

dsp_tools/commands/xml_validate/models/validation.py

@@ -1,29 +0,0 @@
-from __future__ import annotations
-
-from dataclasses import dataclass
-
-from rdflib import Graph
-from rdflib.term import Node
-
-
-@dataclass
-class ValidationReport:
-    conforms: bool
-    validation_graph: Graph
-    shacl_graph: Graph
-    data_graph: Graph
-
-
-@dataclass
-class UnexpectedComponent:
-    component_type: str
-
-
-@dataclass
-class ValidationResult:
-    source_constraint_component: Node
-    res_iri: Node
-    res_class: Node
-    property: Node
-    results_message: str
-    value: str | None = None

dsp_tools/commands/xml_validate/reformat_validaton_result.py

@@ -1,89 +0,0 @@
-from rdflib import RDF
-from rdflib import SH
-from rdflib import Graph
-from rdflib.term import Node
-
-from dsp_tools.commands.xml_validate.models.input_problems import AllProblems
-from dsp_tools.commands.xml_validate.models.input_problems import InputProblem
-from dsp_tools.commands.xml_validate.models.input_problems import MaxCardinalityViolation
-from dsp_tools.commands.xml_validate.models.input_problems import MinCardinalityViolation
-from dsp_tools.commands.xml_validate.models.input_problems import NonExistentCardinalityViolation
-from dsp_tools.commands.xml_validate.models.input_problems import UnexpectedResults
-from dsp_tools.commands.xml_validate.models.validation import UnexpectedComponent
-from dsp_tools.commands.xml_validate.models.validation import ValidationResult
-
-
-def reformat_validation_graph(results_graph: Graph, data_graph: Graph) -> AllProblems:
-    """
-    Reformats the validation result from an RDF graph into class instances
-    that are used to communicate the problems with the user.
-
-    Args:
-        results_graph: Contains the possible validation errors
-        data_graph: Graph with the data
-
-    Returns:
-        All Problems
-    """
-    reformatted_results = _reformat_result_graph(results_graph, data_graph)
-    input_problems, unexpected_components = _transform_violations_into_input_problems(reformatted_results)
-    unexpected = UnexpectedResults(unexpected_components) if unexpected_components else None
-    return AllProblems(input_problems, unexpected)
-
-
-def _reformat_result_graph(results_graph: Graph, data_graph: Graph) -> list[ValidationResult]:
-    violations = results_graph.subjects(RDF.type, SH.ValidationResult)
-    return [_extract_one_violation(x, results_graph, data_graph) for x in violations]
-
-
-def _extract_one_violation(bn: Node, results_graph: Graph, data_graph: Graph) -> ValidationResult:
-    focus_nd = next(results_graph.objects(bn, SH.focusNode))
-    res_type = next(data_graph.objects(focus_nd, RDF.type))
-    path = next(results_graph.objects(bn, SH.resultPath))
-    component = next(results_graph.objects(bn, SH.sourceConstraintComponent))
-    msg = str(next(results_graph.objects(bn, SH.resultMessage)))
-    return ValidationResult(
-        source_constraint_component=component,
-        res_iri=focus_nd,
-        res_class=res_type,
-        property=path,
-        results_message=msg,
-    )
-
-
-def _transform_violations_into_input_problems(
-    violations: list[ValidationResult],
-) -> tuple[list[InputProblem], list[UnexpectedComponent]]:
-    input_problems: list[InputProblem] = []
-    unexpected_components: list[UnexpectedComponent] = []
-    for violation in violations:
-        problem = _reformat_one_violation(violation)
-        if isinstance(problem, UnexpectedComponent):
-            unexpected_components.append(problem)
-        else:
-            input_problems.append(problem)
-    return input_problems, unexpected_components
-
-
-def _reformat_one_violation(violation: ValidationResult) -> InputProblem | UnexpectedComponent:
-    subject_id = _reformat_data_iri(str(violation.res_iri))
-    prop_name = _reformat_onto_iri(str(violation.property))
-    res_type = _reformat_onto_iri(str(violation.res_class))
-    match violation.source_constraint_component:
-        case SH.MaxCountConstraintComponent:
-            return MaxCardinalityViolation(subject_id, res_type, prop_name, violation.results_message)
-        case SH.MinCountConstraintComponent:
-            return MinCardinalityViolation(subject_id, res_type, prop_name, violation.results_message)
-        case SH.ClosedConstraintComponent:
-            return NonExistentCardinalityViolation(subject_id, res_type, prop_name)
-        case _:
-            return UnexpectedComponent(str(violation.source_constraint_component))
-
-
-def _reformat_onto_iri(prop: str) -> str:
-    onto = prop.split("/")[-2]
-    return f'{onto}:{prop.split("#")[-1]}'
-
-
-def _reformat_data_iri(iri: str) -> str:
-    return iri.replace("http://data/", "")

dsp_tools/commands/xml_validate/sparql/construct_shapes.py

@@ -1,16 +0,0 @@
-from rdflib import Graph
-
-from dsp_tools.commands.xml_validate.sparql.resource_shacl import construct_resource_class_node_shape
-
-
-def construct_shapes_graph(onto: Graph) -> Graph:
-    """
-    Constructs a shapes graph from a project ontology
-
-    Args:
-        onto: ontology as graph
-
-    Returns:
-        shapes graph
-    """
-    return construct_resource_class_node_shape(onto)