RDF 1.2 and SPARQL 1.2; SEMIJOIN and ANTIJOIN

apache · Dec 11, 2024 · 2f1818c · 2f1818c
1 parent 3d34ac9
commit 2f1818c
Show file tree

Hide file tree

Showing 595 changed files with 15,748 additions and 13,978 deletions.
diff --git a/jena-arq/Grammar/.gitignore b/jena-arq/Grammar/.gitignore
@@ -9,5 +9,7 @@ Y12.html
 sparql-grammar.html
 # Grammar in BNF text
 sparql.bnf
-# Grammar in HTML as a standalone web page.
+## BNF
+sparql.bnf
+## Grammar in HTML as a standalone web page.
 sparql-html.html
diff --git a/jena-arq/Grammar/RDF-Protobuf/binary-rdf.proto b/jena-arq/Grammar/RDF-Protobuf/binary-rdf.proto
@@ -55,7 +55,8 @@ message RDF_Literal {
   string lex = 1 ;
   oneof literalKind {
     bool simple = 9 ;
-    string langtag = 2 ;
+    string langtag = 2 ;    // Language tag only literal
+    string langdir = 5 ;    // Language tag and base direction separated by "--"
     string datatype = 3 ;
     RDF_PrefixName dtPrefix = 4 ;
   }

diff --git a/jena-arq/Grammar/RDF-Thrift/BinaryRDF.thrift b/jena-arq/Grammar/RDF-Thrift/BinaryRDF.thrift
@@ -44,8 +44,10 @@ struct RDF_BNode {
 // union with additional values. 
 
 struct RDF_Literal {
+// Keep this numbering.
 1: required string  lex ;
 2: optional string  langtag ;
+5: optional string  baseDirection ;
 3: optional string  datatype ;          // Either 3 or 4 but UNION is heavy.
 4: optional RDF_PrefixName dtPrefix ;   // datatype as prefix name
 }

diff --git a/jena-arq/Grammar/RDF-Thrift/gen-thrift b/jena-arq/Grammar/RDF-Thrift/gen-thrift
@@ -1,11 +1,11 @@
 #!/bin/bash
 ## Licensed under the terms of http://www.apache.org/licenses/LICENSE-2.0
 
-if [ "$#" != 1 ]
-then
-    echo "Usage: $(basename $0) FILE" 2>&1
-    exit 1
-fi
+## if [ "$#" != 1 ]
+## then
+##     echo "Usage: $(basename $0) FILE" 2>&1
+##     exit 1
+## fi
 
 # Find the namespace
 PKG=../../src/main/java/org/apache/jena/riot/thrift/wire
@@ -16,7 +16,9 @@ rm -f "$PKG"/*.java
 ##                      undated: suppress the date at @Generated annotations
 ##                      suppress: suppress @Generated annotations entirely
 
-thrift -r -out ../../src/main/java -gen 'java:generated_annotations=suppress' "$@"
+THRIFT="${1:-BinaryRDF.thrift}"
+
+thrift -r -out ../../src/main/java -gen 'java:generated_annotations=suppress' "$THRIFT"
 
 for f in "$PKG"/*.java
 do

diff --git a/jena-arq/Grammar/README b/jena-arq/Grammar/README
@@ -0,0 +1,31 @@
+==== SPARQL Parser
+
+Run
+
+    grammar
+
+This produces the code for the current version SPARQL
+and the ARQ (SPARQL with extensions) parsers.
+
+==
+
+To produce the HTML for the SPARQL Grammar for the W3C spec:
+produce the SPARQL 1.2 specific file
+
+    grammar
+
+which also generates Java code and can be used for checking.
+
+Then run
+
+    sparql2html
+
+The output is in sparql-grammar.html.
+
+This replaces the table between "<!-- GRAMMAR -->" in the specific HTML.
+
+Running
+
+    sparql2bnf
+
+produces a text BNF form.
diff --git a/jena-arq/Grammar/Turtle/turtle.jj b/jena-arq/Grammar/Turtle/turtle.jj
@@ -80,9 +80,9 @@ void Statement() : {}
 {
     Directive()
   | DirectiveOld()
-//  | (TriplesSameSubject() (<DOT> | <EOF> ) )
+//  | (Triples() (<DOT> | <EOF> ) )
   // Strict
-  | (TriplesSameSubject() <DOT> )
+  | (Triples() <DOT> )
 }
 
 // Turtle [3] directive
@@ -106,25 +106,22 @@ void DirectiveOld() : { Token t ; String iri ; }
     { setBase(iri, t.beginLine, t.beginColumn) ; }
 }
 
-// Turtle [6] triples
-void TriplesSameSubject() : { Node s ; } 
+void Triples() : { Node n; }
 {
-  s = Subject()
-  PredicateObjectList(s) 
+  n = Subject() PredicateObjectList(n)
 |
-  s = BlankNodePropertyList()
-  ( PredicateObjectList(s) )?
+  n = BlankNodePropertyList() ( PredicateObjectList(n) )?
+|
+  n = ReifiedTriple() ( PredicateObjectList(n) )?
 }
 
-// Turtle [7] predicateObjectList
 void PredicateObjectList(Node s) : { Node p = null ; }
 {
     p = Verb()
     ObjectList(s, p)
     (<SEMICOLON> (p = Verb() ObjectList(s, p))? )*
 }
 
-// Turtle [8] objectList
 void ObjectList(Node s, Node p): { Node o ; } 
 {
   o = Object()
@@ -137,17 +134,6 @@ void ObjectList(Node s, Node p): { Node o ; }
   )*
 }
 
-// RDF-star Annotation Syntax
-void Annotation(Node s, Node p, Node o) : {}
-{
-  (
-    <L_ANN>
-      { Node x = createQuotedTriple(s, p, o, token.beginLine, token.beginColumn); }
-      PredicateObjectList(x)
-    <R_ANN>
-  )?
-}
-
 // Turtle [9] verb
 Node Verb() : { Node p ; }
 {
@@ -163,13 +149,10 @@ Node Subject() : { Node s; String iri ; }
     s = BlankNode()
   |
     s = Collection()
-  |
-    s = QuotedTriple()
   )
   { return s ; }
 }
 
-
 // Turtle [11] predicate
 Node Predicate() : { String iri;}
 {
@@ -185,43 +168,12 @@ Node Object(): { Node o ; String iri; }
   | o = Collection()
   | o = BlankNodePropertyList()
   | o = Literal()
-  | o = QuotedTriple()
+  | o = TripleTerm()
+  | o = ReifiedTriple()
   )
   { return o; }
 }
 
-Node QuotedTripleSubject(): { Node o ; String iri; }
-{
-  ( iri = iri() { o = createURI(iri, token.beginLine, token.beginColumn) ; }
-  | o = BlankNode()
-  | o = QuotedTriple()
-  )
-  { return o; }
-}
-
-Node QuotedTripleObject(): { Node o ; String iri; }
-{
-  ( iri = iri() { o = createURI(iri, token.beginLine, token.beginColumn) ; }
-  | o = BlankNode()
-  | o = Literal()
-  | o = QuotedTriple()
-  )
-  { return o ; } 
-}
-
-// The syntax for RDF-star <<>>
-Node QuotedTriple() : { Node s , p , o ; Token t ; }
-{
-  t = <LT2>
-  s = QuotedTripleSubject()
-  p = Verb()
-  o = QuotedTripleObject()
-  <GT2>
-  { Node n = createQuotedTriple(s, p, o, t.beginLine, t.beginColumn);
-    return n;
-  }
-}
-
 // Turtle [13] literal
 Node Literal() : { Node n ;}
 {
@@ -302,7 +254,7 @@ Node RDFLiteral() : { Token t ; String lex = null ; }
 String LangTag() : { Token t ; }
 {
   // Enumerate the directives here because they look like language tags.
-  ( t = <LANGTAG> | t = AnyDirective() )
+  ( t = <LANG_DIR> | t = AnyDirective() )
   { String lang = stripChars(t.image, 1) ; return lang ; }
 }
 
@@ -369,6 +321,106 @@ String IRIREF() : { Token t ; }
   { return resolveQuotedIRI(t.image, t.beginLine, t.beginColumn) ; }
 }
 
+Node Reifier() : { Node n = null; String iriStr; Token t; }
+{
+  t = <TILDE>
+  (
+    iriStr = iri() { n = createURI(iriStr, token.beginLine, token.beginColumn) ; }
+  | n = BlankNode()
+  )?
+  { if ( n == null ) n = createBNode(t.beginLine, t.beginColumn) ; }
+  { return n ; }
+}
+
+Node ReifiedTriple() : { Node reifId = null ; Token tok ;  Node s; Node p ; Node o ; }
+{
+   tok = <LT2>
+   s = ReifiedTripleSubject()
+   p = Verb()
+   o = ReifiedTripleObject()
+   ( reifId = Reifier())?
+   { reifId = emitTripleReifier(tok.beginLine, tok.beginColumn, reifId, s, p, o); }
+   <GT2>
+   { return reifId ; }
+}
+
+// rtSubject
+Node ReifiedTripleSubject() : { Node s; String iri; } {
+  ( iri = iri() { s = createURI(iri, token.beginLine, token.beginColumn) ; }
+  | s = BlankNode()
+  | s = ReifiedTriple()
+  )
+  { return s; }
+}
+
+// rtObject
+Node ReifiedTripleObject() : { Node o; String iri; } {
+  ( iri = iri() { o = createURI(iri, token.beginLine, token.beginColumn) ; }
+  | o = BlankNode()
+  | o = Literal()
+  | o = TripleTerm()
+  | o = ReifiedTriple()
+  )
+  { return o; }
+}
+
+Node TripleTerm() : { Node n = null ; Token openToken ; String iri ; Node s , p , o ; }
+{
+  <L_TRIPLE>
+  s = TripleTermSubject()
+  p = Verb()
+  o = TripleTermObject()
+  <R_TRIPLE>
+  { return createTripleTerm(s, p, o, token.beginLine, token.beginColumn) ; }
+}
+
+// ttSubject
+Node TripleTermSubject(): { Node o ; String iri; }
+{
+  ( iri = iri() { o = createURI(iri, token.beginLine, token.beginColumn) ; }
+  | o = BlankNode()
+  )
+  { return o; }
+}
+
+// ttObject
+Node TripleTermObject(): { Node o ; String iri; }
+{
+  ( iri = iri() { o = createURI(iri, token.beginLine, token.beginColumn) ; }
+  | o = BlankNode()
+  | o = Literal()
+  | o = TripleTerm()
+  )
+  { return o ; } 
+}
+
+// RDF-star Annotation Syntax
+void Annotation(Node s, Node p, Node o) : { Node reifId = null; }
+{
+  (
+    reifId = Reifier()
+    { reifId = emitTripleReifier(token.beginLine, token.beginColumn, reifId, s, p, o) ; }
+    { setReifierId(reifId); }
+  |
+    { reifId = getOrAllocReifierId(s, p, o, token.beginLine, token.beginColumn); }
+    { clearReifierId(); }
+    AnnotationBlock(reifId)
+  ) *
+}
+
+void AnnotationBlock( Node reifId ) : {}
+{
+   <L_ANN>
+   PredicateObjectList(reifId)
+   <R_ANN>
+// (
+//     <L_ANN>
+//       { Node x = createQuotedTriple(s, p, o, token.beginLine, token.beginColumn); }
+//       PredicateObjectList(x)
+//     <R_ANN>
+//   )?
+}
+
 // ------------------------------------------
 // Tokens
 
@@ -457,18 +509,23 @@ TOKEN :
 | < SEMICOLON: ";" >
 | < COMMA:     "," >
 | < DOT:       "." >
+
+| < DATATYPE: "^^">
+| < AT: "@">
+
+| < L_TRIPLE:  "<<(" >
+| < R_TRIPLE:  ")>>" >
 | < LT2:       "<<" >
 | < GT2:       ">>" >
 | < L_ANN:     "{|" >
 | < R_ANN:     "|}" >
-
-| < DATATYPE: "^^">
-| < AT: "@">
+| < TILDE:     "~"  >
+| < VBAR:      "|" >
 
 |  <PNAME_NS: (<PN_PREFIX>)? ":" >
 |  <PNAME_LN: <PNAME_NS> <PN_LOCAL> >
 |  <BLANK_NODE_LABEL: "_:" (<PN_CHARS_U> | ["0"-"9"]) ((<PN_CHARS>|".")* <PN_CHARS>)?  >
-|  <LANGTAG: <AT> (<A2Z>)+("-" (<A2ZN>)+)* >
+|  <LANG_DIR: <AT> (<A2Z>)+("-" (<A2ZN>)+)* ( "--" (<A2Z>)* )? >
 |  <#A2Z: ["a"-"z","A"-"Z"]>
 |  <#A2ZN: ["a"-"z","A"-"Z","0"-"9"]>