test: add more graph variants highlighting bugs

test/data/variants/diverse_quads-asserts.json

@@ -0,0 +1,4 @@
+{
+  "quad_count": 10,
+  "exact_match": true
+}

test/data/variants/diverse_quads.jsonld

@@ -0,0 +1,73 @@
+{
+  "@graph": [
+    {
+      "@graph": [
+        {
+          "@id": "egschema:subject",
+          "egschema:predicate": [
+            12,
+            {
+              "@id": "egschema:object"
+            }
+          ]
+        },
+        {
+          "@id": "eghttp:subject",
+          "predicate": {
+            "@language": "jpx",
+            "@value": "日本語の表記体系"
+          }
+        },
+        {
+          "@id": "egurn:subject",
+          "egschema:predicate": {
+            "@id": "egschema:subject"
+          }
+        }
+      ],
+      "@id": "egschema:graph"
+    },
+    {
+      "@id": "egschema:subject",
+      "egschema:predicate": {
+        "@id": "egschema:object"
+      }
+    },
+    {
+      "@id": "eghttp:subject",
+      "predicate": "typeless"
+    },
+    {
+      "@graph": [
+        {
+          "@id": "egschema:subject",
+          "egschema:predicate": {
+            "@id": "egschema:object"
+          },
+          "predicate": [
+            "XSD string",
+            {
+              "@id": "eghttp:object"
+            }
+          ]
+        }
+      ],
+      "@id": "egurn:graph"
+    },
+    {
+      "@id": "egurn:subject",
+      "egurn:predicate": {
+        "@id": "egurn:object"
+      }
+    }
+  ],
+  "@context": {
+    "predicate": {
+      "@id": "http://example.com/predicate"
+    },
+    "egurn": "urn:example:",
+    "xsd": "http://www.w3.org/2001/XMLSchema#",
+    "egschema": "example:",
+    "eghttp": "http://example.com/"
+  }
+}

test/data/variants/diverse_quads.nq

@@ -0,0 +1,10 @@
+<http://example.com/subject> <http://example.com/predicate> "日本語の表記体系"@jpx <example:graph> .
+<urn:example:subject> <example:predicate> <example:subject> <example:graph> .
+<example:subject> <example:predicate> <example:object> <example:graph> .
+<example:subject> <example:predicate> "12"^^<http://www.w3.org/2001/XMLSchema#integer> <example:graph> .
+<example:subject> <example:predicate> <example:object> <urn:example:graph> .
+<example:subject> <http://example.com/predicate> <http://example.com/object> <urn:example:graph> .
+<example:subject> <http://example.com/predicate> "XSD string" <urn:example:graph> .
+<example:subject> <example:predicate> <example:object> .
+<http://example.com/subject> <http://example.com/predicate> "typeless" .
+<urn:example:subject> <urn:example:predicate> <urn:example:object> .

test/data/variants/diverse_quads.trig

@@ -0,0 +1,27 @@
+@prefix egschema: <example:> .
+@prefix eghttp: <http://example.com/> .
+@prefix egurn: <urn:example:> .
+@prefix xsd: <http://www.w3.org/2001/XMLSchema#> .
+
+egschema:graph {
+    eghttp:subject eghttp:predicate "日本語の表記体系"@jpx .
+
+    egurn:subject egschema:predicate egschema:subject .
+
+    egschema:subject egschema:predicate egschema:object,
+            12 .
+}
+
+egurn:graph {
+    egschema:subject egschema:predicate egschema:object ;
+        eghttp:predicate eghttp:object,
+            "XSD string"^^xsd:string .
+}
+
+{
+    egschema:subject egschema:predicate egschema:object .
+
+    eghttp:subject eghttp:predicate "typeless" .
+
+    egurn:subject egurn:predicate egurn:object .
+}

test/data/variants/diverse_triples-asserts.json

@@ -0,0 +1,4 @@
+{
+  "quad_count": 5,
+  "exact_match": true
+}

test/data/variants/diverse_triples.hext

@@ -0,0 +1,5 @@
+["urn:example:subject", "example:predicate", "example:subject", "globalId", "", ""]
+["example:subject", "example:predicate", "example:object", "globalId", "", ""]
+["example:object", "http://example.com/predicate", "XSD string", "http://www.w3.org/2001/XMLSchema#string", "", ""]
+["http://example.com/subject", "http://example.com/predicate", "日本語の表記体系", "http://www.w3.org/1999/02/22-rdf-syntax-ns#langString", "jpx", ""]
+["example:subject", "example:predicate", "12", "http://www.w3.org/2001/XMLSchema#integer", "", ""]

test/data/variants/diverse_triples.jsonld

@@ -0,0 +1,40 @@
+{
+  "@graph": [
+    {
+      "@id": "egschema:object",
+      "eghttp:predicate": "XSD string"
+    },
+    {
+      "@id": "egschema:subject",
+      "egschema:predicate": [
+        12,
+        {
+          "@id": "egschema:object"
+        }
+      ]
+    },
+    {
+      "@id": "eghttp:subject",
+      "eghttp:predicate": {
+        "@language": "jpx",
+        "@value": "日本語の表記体系"
+      }
+    },
+    {
+      "@id": "egurn:subject",
+      "egschema:predicate": {
+        "@id": "egschema:subject"
+      }
+    }
+  ],
+  "@context": {
+    "predicate": {
+      "@id": "example:predicate",
+      "@type": "http://www.w3.org/2001/XMLSchema#integer"
+    },
+    "egurn": "urn:example:",
+    "xsd": "http://www.w3.org/2001/XMLSchema#",
+    "egschema": "example:",
+    "eghttp": "http://example.com/"
+  }
+}

test/data/variants/diverse_triples.nt

@@ -0,0 +1,5 @@
+<http://example.com/subject> <http://example.com/predicate> "日本語の表記体系"@jpx .
+<urn:example:subject> <example:predicate> <example:subject> .
+<example:object> <http://example.com/predicate> "XSD string" .
+<example:subject> <example:predicate> <example:object> .
+<example:subject> <example:predicate> "12"^^<http://www.w3.org/2001/XMLSchema#integer> .

test/data/variants/diverse_triples.ttl

@@ -0,0 +1,13 @@
+@prefix eghttp: <http://example.com/> .
+@prefix egurn: <urn:example:> .
+@prefix egschema: <example:> .
+@prefix xsd: <http://www.w3.org/2001/XMLSchema#> .
+
+eghttp:subject eghttp:predicate "日本語の表記体系"@jpx .
+
+egurn:subject egschema:predicate egschema:subject .
+
+egschema:object eghttp:predicate "XSD string"^^xsd:string .
+
+egschema:subject egschema:predicate egschema:object,
+        12 .

test/data/variants/more_quads-asserts.json

@@ -0,0 +1,4 @@
+{
+  "quad_count": 6,
+  "exact_match": true
+}

test/data/variants/more_quads.jsonld

@@ -0,0 +1,56 @@
+{
+  "@graph": [
+    {
+      "@graph": [
+        {
+          "@id": "example:s20",
+          "example:p20": {
+            "@id": "example:o20"
+          }
+        },
+        {
+          "@id": "example:s21",
+          "example:p21": {
+            "@id": "example:o21"
+          }
+        }
+      ],
+      "@id": "example:g2"
+    },
+    {
+      "@id": "example:s00",
+      "p00": "example:o02"
+    },
+    {
+      "@id": "example:s01",
+      "p01": "example:o01"
+    },
+    {
+      "@id": "example:s10",
+      "p10": "example:o10"
+    },
+    {
+      "@id": "example:s11",
+      "p11": "example:o11"
+    }
+  ],
+  "@context": {
+    "p10": {
+      "@id": "http://example.org/p10",
+      "@type": "@id"
+    },
+    "p01": {
+      "@id": "http://example.org/p01",
+      "@type": "@id"
+    },
+    "p00": {
+      "@id": "http://example.org/p00",
+      "@type": "@id"
+    },
+    "p11": {
+      "@id": "http://example.org/p11",
+      "@type": "@id"
+    },
+    "example": "http://example.org/"
+  }
+}

test/data/variants/more_quads.nq

@@ -0,0 +1,6 @@
+<http://example.org/s00> <http://example.org/p00> <http://example.org/o02> .
+<http://example.org/s01> <http://example.org/p01> <http://example.org/o01> .
+<http://example.org/s10> <http://example.org/p10> <http://example.org/o10> .
+<http://example.org/s11> <http://example.org/p11> <http://example.org/o11> .
+<http://example.org/s20> <http://example.org/p20> <http://example.org/o20> <http://example.org/g2> .
+<http://example.org/s21> <http://example.org/p21> <http://example.org/o21> <http://example.org/g2> .

test/data/variants/more_quads.trig

@@ -0,0 +1,15 @@
+@prefix example: <http://example.org/>.
+
+example:s00 example:p00 example:o02 .
+example:s01 example:p01 example:o01 .
+
+
+{
+	example:s10 example:p10 example:o10 .
+	example:s11 example:p11 example:o11 .
+}
+
+example:g2 {
+	example:s20 example:p20 example:o20 .
+	example:s21 example:p21 example:o21 .
+}

test/test_graph/test_variants.py

@@ -2,6 +2,7 @@
 import logging
 import os
 import re
+import sys
 from dataclasses import dataclass, field
 from pathlib import Path, PurePath
 from test.data import TEST_DATA_DIR
@@ -165,8 +166,30 @@ def for_directory(
         """,
         raises=AssertionError,
     ),
+    ("variants/diverse_quads"): pytest.mark.xfail(
+        reason="""
+    TriG parsing gets confused about what graph 'XSD string' appears in:
+        (rdflib.term.URIRef('example:subject'),
+        rdflib.term.URIRef('http://example.com/predicate'),
+        rdflib.term.Literal('XSD string'),
+    -   rdflib.term.URIRef('example:graph')),
+    +   rdflib.term.URIRef('urn:example:graph')),
+    ?                       ++++
+        """,
+        raises=AssertionError,
+    ),
 }
 
+if sys.platform == "win32":
+    EXPECTED_FAILURES["variants/diverse_triples"] = pytest.mark.xfail(
+        reason="""
+    Some encoding issue when parsing hext on windows:
+        >       return codecs.charmap_decode(input,self.errors,decoding_table)[0]
+        E       UnicodeDecodeError: 'charmap' codec can't decode byte 0x81 in position 356: character maps to <undefined>
+        """,
+        raises=UnicodeDecodeError,
+    )
+
 
 def tests_found() -> None:
     logging.debug("VARIANTS_DIR = %s", VARIANTS_DIR)

test/test_roundtrip.py

@@ -164,6 +164,26 @@
         """,
         raises=AssertionError,
     ),
+    ("json-ld", "diverse_quads.trig"): pytest.mark.xfail(
+        reason="""
+    jsonld serializer is dropping datatype:
+        only in first:
+            (rdflib.term.URIRef('example:subject'), rdflib.term.URIRef('http://example.com/predicate'), rdflib.term.Literal('XSD string', datatype=rdflib.term.URIRef('http://www.w3.org/2001/XMLSchema#string')))
+        only in second:
+            (rdflib.term.URIRef('example:subject'), rdflib.term.URIRef('http://example.com/predicate'), rdflib.term.Literal('XSD string'))
+    """,
+        raises=AssertionError,
+    ),
+    ("hext", "diverse_quads.trig"): pytest.mark.xfail(
+        reason="""
+    hext serializer is dropping datatype:
+        only in first:
+            (rdflib.term.URIRef('example:subject'), rdflib.term.URIRef('http://example.com/predicate'), rdflib.term.Literal('XSD string', datatype=rdflib.term.URIRef('http://www.w3.org/2001/XMLSchema#string')))
+        only in second:
+            (rdflib.term.URIRef('example:subject'), rdflib.term.URIRef('http://example.com/predicate'), rdflib.term.Literal('XSD string'))
+    """,
+        raises=AssertionError,
+    ),
 }
 
 # This is for files which can only be represented properly in one format
@@ -291,6 +311,9 @@ def test_n3(checker: Callable[[str, str, Path], None], args: Tuple[str, str, Pat
     (TEST_DATA_DIR / "variants" / "rdf_prefix.jsonld", "json-ld"),
     (TEST_DATA_DIR / "variants" / "simple_quad.trig", "trig"),
     (TEST_DATA_DIR / "variants" / "rdf11trig_eg2.trig", "trig"),
+    (TEST_DATA_DIR / "variants" / "diverse_triples.nt", "ntriples"),
+    (TEST_DATA_DIR / "variants" / "diverse_quads.nq", "nquads"),
+    (TEST_DATA_DIR / "variants" / "diverse_quads.trig", "trig"),
     (TEST_DATA_DIR / "example-lots_of_graphs.n3", "n3"),
     (TEST_DATA_DIR / "issue156.n3", "n3"),
 ]