Added SPARQL unicode numeric codepoint escape tests

Added four (marked-as-xfail) unicode numeric codepoint escape tests to the RDFLib-specific SPARQL test suite `test/data/suites/rdflib/sparql`. Also: - Fix the base URI in the RDFLib SPARQL test manifest.
RDFLib · May 14, 2022 · 03e8cac · 03e8cac
1 parent 7d496bc
commit 03e8cac
Show file tree

Hide file tree

Showing 7 changed files with 67 additions and 1 deletion.
diff --git a/test/data/suites/rdflib/sparql/manifest.ttl b/test/data/suites/rdflib/sparql/manifest.ttl
@@ -5,7 +5,7 @@
 
 @prefix dawgt: <http://www.w3.org/2001/sw/DataAccess/tests/test-dawg#> .
 
-@prefix :      <http://raw.github.com/RDFLib/rdflib/master/test/DAWG/rdflib/manifest.ttl#> .
+@prefix :      <http://raw.github.com/RDFLib/rdflib/master/test/data/rdflib-sparql/manifest.ttl#> .
 
 <> a mf:Manifest ;
 	rdfs:label "RDFLib Extra SPARQL tests" ;
@@ -41,6 +41,12 @@
     :bnode-ppath-mix-g
     :bnode-ppath-mix-h
 
+    :test-codepoint-escape-01
+    :test-codepoint-escape-02
+    :test-codepoint-escape-03
+    :test-codepoint-escape-04
+    :test-codepoint-escape-bad
+
 	) .
 
 
@@ -363,3 +369,35 @@ From https://github.com/RDFLib/rdflib/issues/615, contributed by https://github.
            qt:data <bnode-ppath-mix.ttl> ] ;
     mf:result  <bnode-ppath-mix.srx>
     .
+
+:test-codepoint-escape-01 rdf:type   mf:PositiveSyntaxTest11 ;
+    dawgt:approval dawgt:Proposed ;
+    mf:name    "\\U unicode codepoint escaping in literal" ;
+    rdfs:comment """From https://github.com/RDFLib/rdflib/issues/1884, contributed by https://github.com/gjhiggins""" ;
+    mf:action  <syn-codepoint-escape-01.rq> ;.
+
+:test-codepoint-escape-02 rdf:type   mf:PositiveSyntaxTest11 ;
+    dawgt:approval dawgt:Proposed ;
+    rdfs:comment """From https://github.com/RDFLib/rdflib/issues/1884, contributed by https://github.com/gjhiggins""" ;
+    mf:name    "\\u unicode codepoint escaping in literal" ;
+    mf:action  <syn-codepoint-escape-02.rq> ;.
+
+:test-codepoint-escape-03 rdf:type   mf:PositiveSyntaxTest11 ;
+    dawgt:approval dawgt:Proposed ;
+    rdfs:comment """From https://github.com/RDFLib/rdflib/issues/1884, contributed by https://github.com/gjhiggins""" ;
+    mf:name    "\\u and \\U unicode codepoint escaping in literal" ;
+    mf:action  <syn-codepoint-escape-03.rq> ;.
+
+:test-codepoint-escape-04 rdf:type   mf:PositiveSyntaxTest11 ;
+    dawgt:approval dawgt:Proposed ;
+    rdfs:comment """From https://github.com/RDFLib/rdflib/issues/1884, contributed by https://github.com/gjhiggins""" ;
+    mf:name    "\\U and \\U unicode codepoints escaping in literal" ;
+    mf:action  <syn-codepoint-escape-04.rq> ;.
+
+:test-codepoint-escape-bad rdf:type   mf:NegativeSyntaxTest11 ;
+    dawgt:approval dawgt:Proposed ;
+    rdfs:comment """From https://github.com/RDFLib/rdflib/issues/1884, contributed by https://github.com/gjhiggins""" ;
+    mf:name    "Invalid multi-pass codepoint escaping (\\u then \\U)" ;
+    mf:description "Unescaping one escape sequence must not produce content that is used in another escape sequence" ;
+    mf:action  <syn-codepoint-escape-bad.rq> ;.
+
diff --git a/test/data/suites/rdflib/sparql/syn-codepoint-escape-01.rq b/test/data/suites/rdflib/sparql/syn-codepoint-escape-01.rq
@@ -0,0 +1,3 @@
+SELECT * WHERE {
+    ?s ?p "\U0001f46a" .
+}
diff --git a/test/data/suites/rdflib/sparql/syn-codepoint-escape-02.rq b/test/data/suites/rdflib/sparql/syn-codepoint-escape-02.rq
@@ -0,0 +1,3 @@
+SELECT * WHERE {
+    ?s ?p "\u00a71234" .
+}
diff --git a/test/data/suites/rdflib/sparql/syn-codepoint-escape-03.rq b/test/data/suites/rdflib/sparql/syn-codepoint-escape-03.rq
@@ -0,0 +1,3 @@
+SELECT * WHERE {
+    ?s ?p "\U0001f46a\u00a71234" .
+}
diff --git a/test/data/suites/rdflib/sparql/syn-codepoint-escape-04.rq b/test/data/suites/rdflib/sparql/syn-codepoint-escape-04.rq
@@ -0,0 +1,3 @@
+SELECT * WHERE {
+    ?s ?p "\U0001f46afoo\U0000005c" .
+}
diff --git a/test/data/suites/rdflib/sparql/syn-codepoint-escape-bad.rq b/test/data/suites/rdflib/sparql/syn-codepoint-escape-bad.rq
@@ -0,0 +1,3 @@
+SELECT * WHERE {
+    ?s ?p "\U0001HHHH" .
+}
diff --git a/test/test_w3c_spec/test_sparql_w3c.py b/test/test_w3c_spec/test_sparql_w3c.py
@@ -551,9 +551,22 @@ def test_dawg_data_sparql11(rdf_test_uri: URIRef, type: Node, rdf_test: RDFTest)
     testers[type](rdf_test)
 
 
+EXPECTED_FAILURES: Dict[str, str] = {}
+
+for test in [
+    "test-codepoint-escape-02",
+    "test-codepoint-escape-03",
+    "test-codepoint-escape-04",
+]:
+    EXPECTED_FAILURES[test] = "known codepoint escape issue"
+
+
 @pytest.mark.parametrize(
     "rdf_test_uri, type, rdf_test",
     read_manifest("test/data/suites/rdflib/sparql/manifest.ttl"),
 )
 def test_dawg_rdflib(rdf_test_uri: URIRef, type: Node, rdf_test: RDFTest):
+    suffix = rdf_test_uri.split("#")[1]
+    if suffix in EXPECTED_FAILURES:
+        pytest.xfail(EXPECTED_FAILURES[suffix])
     testers[type](rdf_test)