refactor(xmlvalidate): improve error message (DEV-4201) (#1196)

src/dsp_tools/commands/xml_validate/models/input_problems.py

@@ -8,45 +8,35 @@
 from pathlib import Path
 
 from rdflib import Graph
-from rdflib.term import Node
 
+from dsp_tools.commands.xml_validate.models.validation import UnexpectedComponent
 from dsp_tools.models.custom_warnings import DspToolsUserWarning
 
 LIST_SEPARATOR = "\n    - "
 INDENT = "\n    "
 GRAND_SEPARATOR = "\n\n----------------------------\n"
 
 
-@dataclass
-class ValidationResult:
-    source_constraint_component: Node
-    res_iri: Node
-    res_class: Node
-    property: Node
-    results_message: str
-    value: str | None = None
-
-
-@dataclass
-class UnexpectedComponent:
-    component_type: str
-
-
 @dataclass
 class UnexpectedResults:
     components: list[UnexpectedComponent]
-    validation_result: Graph
 
-    def save_inform_user(self, cwdr: Path) -> None:
-        save_path = cwdr / f"validation_result_{datetime.now()!s}.ttl"
+    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"The validation report was saved here: {save_path}\n"
-            f"Please contact the dsp-tools development team with this information."
+            f"Please contact the development team with the files starting with the timestamp '{prefix}' "
+            f"in the directory '{cwdr}'."
         )
-        self.validation_result.serialize(save_path)
+        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))
 
 
@@ -55,11 +45,6 @@ class AllProblems:
     problems: list[InputProblem]
     unexpected_results: UnexpectedResults | None
 
-    def communicate_with_the_user(self, cwdr: Path) -> str:
-        if self.unexpected_results:
-            self.unexpected_results.save_inform_user(cwdr)
-        return self.get_msg()
-
     def get_msg(self) -> str:
         coll = self._make_collection()
         msg = [x.get_msg() for x in coll]
@@ -82,10 +67,23 @@ class ResourceProblemCollection:
     problems: list[InputProblem]
 
     def get_msg(self) -> str:
-        msg = [f"Resource ID: {self.res_id} | Resource Type: {self.problems[0].res_type}"]
-        sorted_problems = sorted(self.problems, key=lambda x: x.sort_value())
-        msg.extend([x.get_msg() for x in sorted_problems])
-        return "\n".join(msg)
+        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
@@ -113,10 +111,7 @@ class MaxCardinalityViolation(InputProblem):
     expected_cardinality: str
 
     def get_msg(self) -> str:
-        return (
-            f"Maximum Cardinality Violation:"
-            f"{INDENT}Property: {self.prop_name} | Expected Cardinality: {self.expected_cardinality}"
-        )
+        return f"Maximum Cardinality Violation | Expected Cardinality: {self.expected_cardinality}"
 
     def sort_value(self) -> str:
         return self.prop_name
@@ -127,10 +122,7 @@ class MinCardinalityViolation(InputProblem):
     expected_cardinality: str
 
     def get_msg(self) -> str:
-        return (
-            f"Minimum Cardinality Violation:"
-            f"{INDENT}Property: {self.prop_name} | Expected Cardinality: {self.expected_cardinality}"
-        )
+        return f"Minimum Cardinality Violation | Expected Cardinality: {self.expected_cardinality}"
 
     def sort_value(self) -> str:
         return self.prop_name
@@ -139,7 +131,7 @@ def sort_value(self) -> str:
 @dataclass
 class NonExistentCardinalityViolation(InputProblem):
     def get_msg(self) -> str:
-        return f"The resource class does not have a cardinality for{INDENT}Property: {self.prop_name}"
+        return "The resource class does not have a cardinality for this property."
 
     def sort_value(self) -> str:
         return self.prop_name

src/dsp_tools/commands/xml_validate/models/validation.py

@@ -0,0 +1,29 @@
+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

src/dsp_tools/commands/xml_validate/reformat_validaton_result.py

@@ -8,9 +8,9 @@
 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 UnexpectedComponent
 from dsp_tools.commands.xml_validate.models.input_problems import UnexpectedResults
-from dsp_tools.commands.xml_validate.models.input_problems import ValidationResult
+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:
@@ -27,7 +27,7 @@ def reformat_validation_graph(results_graph: Graph, data_graph: Graph) -> AllPro
     """
     reformatted_results = _reformat_result_graph(results_graph, data_graph)
     input_problems, unexpected_components = _transform_violations_into_input_problems(reformatted_results)
-    unexpected = UnexpectedResults(unexpected_components, results_graph) if unexpected_components else None
+    unexpected = UnexpectedResults(unexpected_components) if unexpected_components else None
     return AllProblems(input_problems, unexpected)
 
 

src/dsp_tools/commands/xml_validate/xml_validate.py

@@ -4,13 +4,15 @@
 from lxml import etree
 from rdflib import SH
 from rdflib import Graph
+from termcolor import cprint
 
 from dsp_tools.commands.xml_validate.api_connection import OntologyConnection
 from dsp_tools.commands.xml_validate.api_connection import ShaclValidator
 from dsp_tools.commands.xml_validate.deserialise_input import deserialise_xml
 from dsp_tools.commands.xml_validate.make_data_rdf import make_data_rdf
 from dsp_tools.commands.xml_validate.models.data_deserialised import ProjectDeserialised
 from dsp_tools.commands.xml_validate.models.data_rdf import DataRDF
+from dsp_tools.commands.xml_validate.models.validation import ValidationReport
 from dsp_tools.commands.xml_validate.reformat_validaton_result import reformat_validation_graph
 from dsp_tools.commands.xml_validate.sparql.construct_shapes import construct_shapes_graph
 from dsp_tools.models.custom_warnings import DspToolsUserWarning
@@ -40,13 +42,20 @@ def xml_validate(filepath: Path, api_url: str, dev_route: bool) -> bool:  # noqa
     ontologies = _get_project_ontos(onto_con)
     data_graph = data_rdf.make_graph() + ontologies
     val = ShaclValidator(api_url)
-    conforms, result = _validate(val, ontologies, data_graph)
-    if conforms:
-        print("\n\nValidation passed!")
+    report = _validate(val, ontologies, data_graph)
+    if report.conforms:
+        cprint("\n   Validation passed!   ", color="green", attrs=["bold", "reverse"])
     else:
-        reformatted = reformat_validation_graph(result, data_graph)
-        msg = reformatted.communicate_with_the_user(Path.cwd())
-        print(msg)
+        reformatted = reformat_validation_graph(report.validation_graph, data_graph)
+        problem_msg = reformatted.get_msg()
+        cprint("\n   Validation errors found!   ", color="light_red", attrs=["bold", "reverse"])
+        print(problem_msg)
+        if reformatted.unexpected_results:
+            reformatted.unexpected_results.save_inform_user(
+                results_graph=report.validation_graph,
+                shacl=report.shacl_graph,
+                data=data_graph,
+            )
     return True
 
 
@@ -59,13 +68,18 @@ def _inform_about_experimental_feature() -> None:
     warnings.warn(DspToolsUserWarning(LIST_SEPARATOR.join(what_is_validated)))
 
 
-def _validate(validator: ShaclValidator, onto_graph: Graph, data_graph: Graph) -> tuple[bool, Graph]:
+def _validate(validator: ShaclValidator, onto_graph: Graph, data_graph: Graph) -> ValidationReport:
     shapes = construct_shapes_graph(onto_graph)
     shape_str = shapes.serialize(format="ttl")
     data_str = data_graph.serialize(format="ttl")
     results = validator.validate(data_str, shape_str)
     conforms = bool(next(results.objects(None, SH.conforms)))
-    return conforms, results
+    return ValidationReport(
+        conforms=conforms,
+        validation_graph=results,
+        shacl_graph=shapes,
+        data_graph=data_graph,
+    )
 
 
 def _get_project_ontos(onto_con: OntologyConnection) -> Graph:

test/unittests/commands/xml_validate/test_reformat_validaton_result.py

@@ -7,8 +7,8 @@
 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 UnexpectedComponent
-from dsp_tools.commands.xml_validate.models.input_problems import ValidationResult
+from dsp_tools.commands.xml_validate.models.validation import UnexpectedComponent
+from dsp_tools.commands.xml_validate.models.validation import ValidationResult
 from dsp_tools.commands.xml_validate.reformat_validaton_result import _extract_one_violation
 from dsp_tools.commands.xml_validate.reformat_validaton_result import _reformat_one_violation