iolanta 2.1.18__py3-none-any.whl → 2.1.20__py3-none-any.whl

iolanta/cli/main.py

@@ -54,6 +54,9 @@ def decode_datatype(datatype: str) -> URIRef:
     if datatype.startswith("http"):
         return URIRef(datatype)
 
+    if "/" in datatype:
+        return URIRef(f"https://iolanta.tech/{datatype}")
+
     return URIRef(f"https://iolanta.tech/datatypes/{datatype}")
 
 

iolanta/iolanta.py

@@ -62,6 +62,21 @@ def _create_default_graph():
     return ConjunctiveGraph(identifier=namespaces.LOCAL.term("_inference"))
 
 
+def _loadable_files(path: Path) -> Iterable[Path]:
+    """Paths of loadable files under path, skipping hidden files and dirs."""
+    if path.is_file():
+        if not path.name.startswith("."):
+            yield path
+        return
+    for entry in path.iterdir():
+        if entry.name.startswith("."):
+            continue
+        if entry.is_file():
+            yield entry
+        else:
+            yield from _loadable_files(entry)
+
+
 @dataclass
 class Iolanta:  # noqa: WPS214, WPS338
     """Iolanta is a Semantic web browser."""
@@ -241,10 +256,7 @@ class Iolanta:  # noqa: WPS214, WPS338
         if not isinstance(source, Path):
             source = Path(source)
 
-        for source_file in list(source.rglob("*")) or [source]:
-            if source_file.is_dir():
-                continue
-
+        for source_file in _loadable_files(source):
             try:  # noqa: WPS225
                 ld_rdf = yaml_ld.to_rdf(source_file)
             except ConnectionError as name_resolution_error:

iolanta/kglint/__init__.py

@@ -0,0 +1,5 @@
+"""Knowledge Graph linter facet."""
+
+from iolanta.kglint.facet import KGLint
+
+__all__ = ['KGLint']

iolanta/kglint/facet.py

@@ -0,0 +1,183 @@
+"""KGLint facet: deterministic KG linter with assertions and labels."""
+
+import re
+from pathlib import Path
+
+import funcy
+from rdflib import BNode, Literal, Node, URIRef
+
+from iolanta import Facet
+from iolanta.kglint.models import (
+    Assertion,
+    AssertionCode,
+    LabelEntry,
+    NodeBlank,
+    NodeLiteral,
+    NodeObject,
+    NodeURI,
+    Report,
+    TripleRef,
+)
+from iolanta.namespaces import DATATYPES
+
+QNAME_RE = re.compile(r'^\S+:\S+$')
+
+
+def _serialize(node: Node) -> str:
+    """Serialize node to stable string (N-Triples style)."""
+    return node.n3()
+
+
+def _looks_like_qname(label: str) -> bool:
+    """True if label has the form prefix:localName (QName-shaped)."""
+    return bool(QNAME_RE.match(label))
+
+
+class KGLint(Facet[str]):
+    """Deterministic Knowledge Graph linter: assertions + URI/blank labels."""
+
+    META = Path(__file__).parent / 'kglint.yamlld'
+
+    def _render_label(self, node: Node) -> str:
+        """Human-readable label for a node (title facet)."""
+        return str(self.render(node, as_datatype=DATATYPES.title))
+
+    @funcy.cached_property
+    @funcy.post_processing(list)
+    def _triples(self):
+        """All triples in the graph (cached)."""
+        for row in self.stored_query('triples.sparql', graph=self.this):
+            yield (row['subject'], row['predicate'], row['object'])
+
+    @funcy.cached_property
+    @funcy.post_processing(set)
+    def _nodes(self):
+        """URIRef and BNode terms that appear in triples (cached)."""
+        for s, p, o in self._triples:
+            for term in (s, p, o):
+                if isinstance(term, (URIRef, BNode)):
+                    yield term
+
+    @funcy.cached_property
+    @funcy.post_processing(dict)
+    def label_by_node(self):
+        """Rendered label per node (cached)."""
+        for node in self._nodes:
+            yield node, self._render_label(node)
+
+    @funcy.cached_property
+    @funcy.post_processing(funcy.group_values)
+    def _triples_by_node(self):
+        """Map each node to list of triples it participates in (cached)."""
+        nodes = self._nodes
+        for s, p, o in self._triples:
+            for term in (s, p, o):
+                if term in nodes:
+                    yield term, (s, p, o)
+
+    def _node_to_object(self, node: Node) -> NodeObject:
+        """Convert RDF node to JSON node object (literal, uri, or blank)."""
+        match node:
+            case Literal():
+                return NodeLiteral(
+                    value=str(node),
+                    datatype=str(node.datatype) if node.datatype else None,
+                    language=node.language or None,
+                )
+            case URIRef():
+                return NodeURI(value=str(node), label=self.label_by_node[node])
+            case BNode():
+                return NodeBlank(value=node.n3(), label=self.label_by_node[node])
+
+    def _label_entries(self):
+        """Yield label entries (node + triples) in deterministic order."""
+        label_by_node = self.label_by_node
+        triples_by_node = self._triples_by_node
+        for node in sorted(self._nodes, key=_serialize):
+            triples_refs = [
+                TripleRef(
+                    s=self._node_to_object(s),
+                    p=self._node_to_object(p),
+                    o=self._node_to_object(o),
+                )
+                for s, p, o in sorted(
+                    triples_by_node[node],
+                    key=lambda t: (_serialize(t[0]), _serialize(t[1]), _serialize(t[2])),
+                )
+            ]
+            if isinstance(node, URIRef):
+                entry_node = NodeURI(value=str(node), label=label_by_node[node])
+            else:
+                entry_node = NodeBlank(value=node.n3(), label=label_by_node[node])
+            yield LabelEntry(node=entry_node, triples=triples_refs)
+
+    def _assertions(self):
+        """Yield all assertions (literal/uri/blank checks)."""
+        label_by_node = self.label_by_node
+        for s, p, o in self._triples:
+            if isinstance(o, Literal):
+                label = self._render_label(o)
+                if label.startswith('http'):
+                    yield Assertion(
+                        severity='warning',
+                        code=AssertionCode.LITERAL_LOOKS_LIKE_URI,
+                        target=TripleRef(
+                            s=self._node_to_object(s),
+                            p=self._node_to_object(p),
+                            o=self._node_to_object(o),
+                        ),
+                        message=(
+                            'This RDF literal seems to be actually a URL. '
+                            'Good chance is that it should not be a literal.'
+                        ),
+                    )
+                elif _looks_like_qname(label):
+                    yield Assertion(
+                        severity='warning',
+                        code=AssertionCode.LITERAL_LOOKS_LIKE_QNAME,
+                        target=TripleRef(
+                            s=self._node_to_object(s),
+                            p=self._node_to_object(p),
+                            o=self._node_to_object(o),
+                        ),
+                        message=(
+                            'This RDF literal seems to be actually a QName '
+                            '(prefixed URI). Good chance is that it should '
+                            'not be a literal.'
+                        ),
+                    )
+        for node in self._nodes:
+            label = label_by_node[node]
+            if isinstance(node, URIRef):
+                if str(node) == label:
+                    yield Assertion(
+                        severity='error',
+                        code=AssertionCode.URI_LABEL_IDENTICAL,
+                        target=NodeURI(value=str(node), label=label),
+                        message=(
+                            'For this URI, the label is the same as the URI. '
+                            'We were unable to render that URI.'
+                        ),
+                    )
+            elif isinstance(node, BNode):
+                if str(node) == label:
+                    yield Assertion(
+                        severity='warning',
+                        code=AssertionCode.BLANK_LABEL_IDENTICAL,
+                        target=NodeBlank(value=node.n3(), label=label),
+                        message=(
+                            'For this blank node, the label is the same as '
+                            'the blank node. We were unable to render that '
+                            'blank node.'
+                        ),
+                    )
+
+    def show(self) -> str:
+        """Build report (assertions + labels) and return JSON string."""
+        labels_list = list(self._label_entries())
+        assertions_list = sorted(
+            self._assertions(),
+            key=lambda a: (a.severity, a.code, a.target.model_dump_json()),
+        )
+        report = Report(assertions=assertions_list, labels=labels_list)
+        return report.model_dump_json(indent=2, exclude_none=True)

iolanta/kglint/kglint.yamlld

@@ -0,0 +1,23 @@
+"@context":
+  "@import": https://json-ld.org/contexts/dollar-convenience.jsonld
+  iolanta: https://iolanta.tech/
+  rdfs: "http://www.w3.org/2000/01/rdf-schema#"
+
+  iolanta:outputs:
+    "@type": "@id"
+
+  $: rdfs:label
+  →:
+    "@type": "@id"
+    "@id": iolanta:outputs
+
+  ↦: iolanta:matches
+
+$id: pkg:pypi/iolanta#kglint
+$: KGLint
+→:
+  $id: https://iolanta.tech/kglint/json
+  $: KGLint JSON
+↦:
+  - ASK WHERE { GRAPH $this { ?s ?p ?o } }
+  - ASK WHERE { $this <https://iolanta.tech/has-sub-graph> ?subgraph }

iolanta/kglint/models.py

@@ -0,0 +1,80 @@
+"""Pydantic models for kglint report output."""
+
+from enum import StrEnum
+from typing import Annotated, Literal as LiteralType
+
+from pydantic import BaseModel, Field
+
+
+class AssertionCode(StrEnum):
+    """Deterministic assertion codes emitted by the linter."""
+
+    LITERAL_LOOKS_LIKE_URI = 'literal-looks-like-uri'
+    LITERAL_LOOKS_LIKE_QNAME = 'literal-looks-like-qname'
+    URI_LABEL_IDENTICAL = 'uri-label-identical'
+    BLANK_LABEL_IDENTICAL = 'blank-label-identical'
+
+
+class NodeLiteral(BaseModel):
+    """Literal term as object: value, datatype, language."""
+
+    type: LiteralType['literal'] = 'literal'
+    value: str
+    datatype: str | None = None
+    language: str | None = None
+
+
+class NodeURI(BaseModel):
+    """URI term as object: value and rendered label."""
+
+    type: LiteralType['uri'] = 'uri'
+    value: str
+    label: str
+
+
+class NodeBlank(BaseModel):
+    """Blank node term as object: value and rendered label."""
+
+    type: LiteralType['blank'] = 'blank'
+    value: str
+    label: str
+
+
+NodeObject = Annotated[
+    NodeLiteral | NodeURI | NodeBlank,
+    Field(discriminator='type'),
+]
+
+
+class TripleRef(BaseModel):
+    """A triple (s, p, o as node objects). Used in assertions and label entries."""
+
+    type: LiteralType['triple'] = 'triple'
+    s: NodeObject
+    p: NodeObject
+    o: NodeObject
+
+
+class Assertion(BaseModel):
+    """Single validation assertion (error or warning)."""
+
+    severity: LiteralType['error', 'warning']
+    code: AssertionCode
+    target: NodeURI | NodeBlank | TripleRef
+    message: str
+
+
+class LabelEntry(BaseModel):
+    """URI/blank node (as object with type, value, label) and participating triples."""
+
+    node: NodeURI | NodeBlank
+    triples: list[TripleRef] = Field(default_factory=list)
+
+
+
+
+class Report(BaseModel):
+    """KGLint report: assertions and labels."""
+
+    assertions: list[Assertion] = Field(default_factory=list)
+    labels: list[LabelEntry] = Field(default_factory=list)

iolanta/kglint/sparql/triples.sparql

@@ -0,0 +1,5 @@
+SELECT * WHERE {
+    GRAPH $graph {
+        ?subject ?predicate ?object .
+    }
+}

{iolanta-2.1.18.dist-info → iolanta-2.1.20.dist-info}/METADATA

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: iolanta
-Version: 2.1.18
+Version: 2.1.20
 Summary: Semantic Web browser
 License: MIT
 Author: Anatoly Scherbakov

{iolanta-2.1.18.dist-info → iolanta-2.1.20.dist-info}/RECORD

@@ -6,7 +6,7 @@ iolanta/cli/formatters/choose.py,sha256=LWzsO_9IBSSgYNIyLlItkp8TNvpW3v6YCQ8-6kbI
 iolanta/cli/formatters/csv.py,sha256=ceJ_DTz0beqeK-d6FPBQqqjXrziEfF0FRSLoGZCt_fs,760
 iolanta/cli/formatters/json.py,sha256=Og5B9UrSM_0NWqW5Afpsy6WH8ZfYgPMVXYvT3i-43Jc,748
 iolanta/cli/formatters/pretty.py,sha256=IypZRAr2vNqcXFY6NOIc75mpyfpFWh56aCBlOPDDieQ,2901
-iolanta/cli/main.py,sha256=RNNrj8Rnh6F9ISbxCPc_LHQrgwPcwvQ2vL6NU2rxNus,8333
+iolanta/cli/main.py,sha256=9TgrPQ2ytKhK--Z4jikY1ngW7HcVh5nc8q8z4KcBwcs,8416
 iolanta/cli/models.py,sha256=cjbpowdzI4wAP0DUk3qoVHyimk6AZwlXi9CGmusZTuM,159
 iolanta/cli/pretty_print.py,sha256=M6E3TmhzA6JY5GeUVmDZLmOh5u70-393PVit4voFKDI,977
 iolanta/context.py,sha256=bZR-tbZIrDQ-Vby01PMDZ6ifxM-0YMK68RJvAsyqCTs,507
@@ -129,7 +129,12 @@ iolanta/facets/title/title.yamlld,sha256=FjfqNPl3EAzZNLYAZNQ6mHfNmFkfxMm1A43gygf
 iolanta/facets/wikibase_statement_title/__init__.py,sha256=_yk1akxgSJOiUBJIc8QGrD2vovvmx_iw_vJDuv1rD7M,91
 iolanta/facets/wikibase_statement_title/facets.py,sha256=gNRqiwOTMxyyHYvb_vIrfd-4ipb8wfyJ4GgPcQdyy9E,726
 iolanta/facets/wikibase_statement_title/sparql/statement-title.sparql,sha256=n07DQWxKqB5c3CA4kacq2HSN0R0dLgnMnLP1AxMo5YA,320
-iolanta/iolanta.py,sha256=qHc4S1u86m8GLM4omdgzGMoB9QXNnsFH8K5U2rq0r7Y,14086
+iolanta/iolanta.py,sha256=f_BlYPxXs8b2E3BYB_3wsxDcycIt8daqQ-WPecfTEbY,14450
+iolanta/kglint/__init__.py,sha256=m16W7nQvAGylFgc40x_QrJJGTH2fNi8vjKVf0OurYQU,99
+iolanta/kglint/facet.py,sha256=84zuYAcEXcJNvyBNLFhGhFYm6s7SwQ98mAm1794dC3w,7057
+iolanta/kglint/kglint.yamlld,sha256=gKiNhn-IHBCTzfHBVSfGoppprxUQEIoc4GGwsZrycVI,515
+iolanta/kglint/models.py,sha256=Lox7XRttiwKyY5_QdE8k--y3T0oqNEBUNlQzUZUG06k,1949
+iolanta/kglint/sparql/triples.sparql,sha256=VsCmYN5AX7jSIiFm-SqLcRcOvUVj8yyZI4PSzKROtQw,82
 iolanta/labeled_triple_set/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
 iolanta/labeled_triple_set/data/labeled_triple_set.yamlld,sha256=P3oAPSPsirpbcRXej-VekuYFTpWqrkysYsxghZc3bTk,1008
 iolanta/labeled_triple_set/labeled_triple_set.py,sha256=o4IgvTvPd0mzBtpgHYd4n1xpujYdAvWBr6gIYwp5vnA,4061
@@ -165,7 +170,7 @@ iolanta/sparqlspace/sparqlspace.py,sha256=Y8_ZPXwuGEXbEes6XQjaQWA2Zv9y8SWxMPDFdq
 iolanta/widgets/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
 iolanta/widgets/description.py,sha256=98Qd3FwT9r8sYqKjl9ZEptaVX9jJ2ULWf0uy3j52p5o,800
 iolanta/widgets/mixin.py,sha256=nDRCOc-gizCf1a5DAcYs4hW8eZEd6pHBPFsfm0ncv7E,251
-iolanta-2.1.18.dist-info/METADATA,sha256=XZU3ors0xfTtQLxAWmGTGaUtUHjJB5AmMJ9JyM7SEsA,2349
-iolanta-2.1.18.dist-info/WHEEL,sha256=3ny-bZhpXrU6vSQ1UPG34FoxZBp3lVcvK0LkgUz6VLk,88
-iolanta-2.1.18.dist-info/entry_points.txt,sha256=Z1f3OaNruE2a6eprkiLgcPw-lZCah_cRT-PTlDD-Y1s,2569
-iolanta-2.1.18.dist-info/RECORD,,
+iolanta-2.1.20.dist-info/METADATA,sha256=NNHUo9jZBtoUWehayFzugGdCamVM1jlh9OYeU9L-Dwk,2349
+iolanta-2.1.20.dist-info/WHEEL,sha256=3ny-bZhpXrU6vSQ1UPG34FoxZBp3lVcvK0LkgUz6VLk,88
+iolanta-2.1.20.dist-info/entry_points.txt,sha256=ArQ-tRMRs-X_w78E1-4f4oaE7beHskz9knYOPgkE-ms,2604
+iolanta-2.1.20.dist-info/RECORD,,

{iolanta-2.1.18.dist-info → iolanta-2.1.20.dist-info}/entry_points.txt

@@ -12,6 +12,7 @@ construct-result-csv-facet=iolanta.facets.query.construct_result_csv:ConstructRe
 construct-result-json-facet=iolanta.facets.query.construct_result_json:ConstructResultJsonFacet
 construct-result-table-facet=iolanta.facets.query.construct_result_table:ConstructResultTableFacet
 icon=iolanta.facets.icon:IconFacet
+kglint=iolanta.kglint.facet:KGLint
 labeled-triple-set=iolanta.labeled_triple_set.labeled_triple_set:LabeledTripleSet
 mermaid-graph=iolanta.mermaid.facet:Mermaid
 mermaid-roadmap=iolanta.facets.mermaid_roadmap.facet:MermaidRoadmap