Merge pull request #769 from fluree/feature/sparql-values

SPARQL VALUES patterns
fluree · Jun 17, 2024 · a74defb · a74defb
2 parents b39f11c + dee6c88
commit a74defb
Show file tree

Hide file tree

Showing 14 changed files with 788 additions and 106 deletions.
diff --git a/resources/sparql.bnf b/resources/sparql.bnf
@@ -3,16 +3,17 @@ UpdateUnit ::= Update
 Prologue ::= ( BaseDecl | PrefixDecl )*
 BaseDecl ::= <'BASE'> WS IRIREF
 PrefixDecl ::= <'PREFIX'> WS PNAME_NS WS IRIREF
-SelectQuery ::= WS SelectClause WS DatasetClause* WS WhereClause WS SolutionModifier WS
+SelectQuery ::= WS SelectClause WS DatasetClause WS WhereClause WS SolutionModifier WS
 SubSelect ::= SelectClause WhereClause SolutionModifier ValuesClause
 SelectClause ::= WS <'SELECT'> WS ( 'DISTINCT' | 'REDUCED')? ( ( WS ( Var | ( <'('> Expression WS 'AS' WS Var <')'> ) ) )+ | ( WS '*' ) )
-ConstructQuery	  ::=  	'CONSTRUCT' ( ConstructTemplate DatasetClause* WhereClause SolutionModifier | DatasetClause* 'WHERE' '{' TriplesTemplate? '}' SolutionModifier )
-DescribeQuery	  ::=  	'DESCRIBE' ( VarOrIri+ | '*' ) DatasetClause* WhereClause? SolutionModifier
-AskQuery	  ::=  	'ASK' DatasetClause* WhereClause SolutionModifier
-Modifiers ::= (ValuesClause? | PrettyPrint? ) (PrettyPrint? | ValuesClause? )
-DatasetClause	  ::=  	<'FROM'> WS ( DefaultGraphClause | NamedGraphClause )
+ConstructQuery	  ::=  	'CONSTRUCT' ( ConstructTemplate DatasetClause WhereClause SolutionModifier | DatasetClause 'WHERE' '{' TriplesTemplate? '}' SolutionModifier )
+DescribeQuery	  ::=  	'DESCRIBE' ( VarOrIri+ | '*' ) DatasetClause WhereClause? SolutionModifier
+AskQuery	  ::=  	'ASK' DatasetClause WhereClause SolutionModifier
+Modifiers         ::= (ValuesClause? | PrettyPrint? ) (PrettyPrint? | ValuesClause? )
+DatasetClause     ::= FromClause*
+<FromClause>      ::= <'FROM'> WS ( DefaultGraphClause | NamedGraphClause )
 DefaultGraphClause	  ::=  	SourceSelector
-NamedGraphClause	  ::=  	<'NAMED'> SourceSelector
+NamedGraphClause	  ::=  	<'NAMED'> WS SourceSelector
 <SourceSelector>	  ::=  	iri
 WhereClause	  ::=  	<'WHERE'?> WS GroupGraphPattern WS
 SolutionModifier	  ::=  	GroupClause? HavingClause? OrderClause? LimitOffsetClauses?
@@ -56,13 +57,15 @@ GroupGraphPatternSub	  ::=  WS TriplesBlock? ( GraphPatternNotTriples WS <'.'?>
 TriplesBlock	  ::=  WS TriplesSameSubjectPath WS ( <'.'> TriplesBlock? WS )?
 GraphPatternNotTriples	  ::=  	GroupOrUnionGraphPattern | OptionalGraphPattern | MinusGraphPattern | GraphGraphPattern | ServiceGraphPattern | Filter | Bind | InlineData
 OptionalGraphPattern	  ::=  	<'OPTIONAL'> GroupGraphPattern
-GraphGraphPattern	  ::=  	'GRAPH' VarOrIri GroupGraphPattern
-ServiceGraphPattern	  ::=  	'SERVICE' 'SILENT'? VarOrIri GroupGraphPattern
+GraphGraphPattern	  ::=  	<'GRAPH'> WS VarOrIri WS GroupGraphPattern
+ServiceGraphPattern	  ::=  	<'SERVICE'> WS 'SILENT'? WS VarOrIri GroupGraphPattern
 Bind	  ::=  	<'BIND' WS '(' WS>  Expression <WS 'AS' WS> Var <WS ')' WS>
 InlineData	  ::=  	<'VALUES'> WS DataBlock
 <DataBlock>	  ::=  	InlineDataOneVar | InlineDataFull
 InlineDataOneVar	  ::=  	Var <'{'> WS DataBlockValue* <'}'>
-InlineDataFull	  ::=  	( NIL | '(' Var* ')' ) '{' ( '(' DataBlockValue* ')' | NIL )* '}'
+InlineDataFull ::= ( NIL | VarList ) WS <'{'> WS ( ValueList WS | NIL )* <'}'>
+VarList ::= ( <'('> Var* <')'> )
+ValueList ::= ( <'('> WS DataBlockValue* <')'> )
 DataBlockValue	  ::=  	iri | RDFLiteral | NumericLiteral | BooleanLiteral | 'UNDEF' WS
 MinusGraphPattern	  ::=  	'MINUS' GroupGraphPattern
 GroupOrUnionGraphPattern	  ::=  	GroupGraphPattern ( <'UNION'> GroupGraphPattern )*
@@ -76,7 +79,7 @@ ConstructTriples	  ::=  	TriplesSameSubject ( '.' ConstructTriples? )?
 TriplesSameSubject	  ::=  	VarOrTerm PropertyListNotEmpty | TriplesNode PropertyList
 PropertyList	  ::=  	PropertyListNotEmpty?
 PropertyListNotEmpty	  ::=  	Verb ObjectList ( <';'>  WS ( Verb ObjectList )? )*
-Verb	  ::=  	VarOrIri | 'a'
+Verb	  ::=  	VarOrIri | Type
 ObjectList	  ::=  	Object ( <','> WS Object )*
 Object	  ::=  	GraphNode
 TriplesSameSubjectPath	  ::=  	VarOrTerm PropertyListPathNotEmpty | TriplesNodePath PropertyListPath WS
@@ -92,9 +95,9 @@ PathSequence	  ::=  	PathEltOrInverse ( <'/'> PathEltOrInverse )*
 <PathElt>	  ::=  	PathPrimary PathMod?
 <PathEltOrInverse>	  ::=  	PathElt | <'^'> PathElt
 PathMod	  ::=  	'?' | '*' | ('+' INTEGER?) WS
-PathPrimary	  ::=  	iri | 'a' | '!' PathNegatedPropertySet | '(' Path ')'
+PathPrimary	  ::=  	iri | Type | '!' PathNegatedPropertySet | '(' Path ')'
 PathNegatedPropertySet	  ::=  	PathOneInPropertySet | '(' ( PathOneInPropertySet ( '|' PathOneInPropertySet )* )? ')'
-PathOneInPropertySet	  ::=  	iri | 'a' | '^' ( iri | 'a' )
+PathOneInPropertySet	  ::=  	iri | Type | '^' ( iri | Type )
 Integer	  ::=  	INTEGER
 TriplesNode	  ::=  	Collection | BlankNodePropertyList
 BlankNodePropertyList	  ::=  	'[' PropertyListNotEmpty ']'
@@ -105,14 +108,15 @@ CollectionPath	  ::=  	'(' GraphNodePath+ ')'
 <GraphNode>	  ::=  	VarOrTerm | TriplesNode
 <GraphNodePath>	  ::=  	VarOrTerm | TriplesNodePath
 <VarOrTerm>	  ::=  	Var | GraphTerm WS
-VarOrIri	  ::=  	Var | iri WS
+<VarOrIri>	  ::=  	Var | iri WS
 Var	  ::=  	VAR1 WS | VAR2 WS
+<Type> ::= (WS 'a' WS)
 <GraphTerm>	  ::=  	iri | RDFLiteral | NumericLiteral | BooleanLiteral | BlankNode | NIL
 Expression	  ::=  	WS ConditionalOrExpression WS
-<ConditionalOrExpression>	  ::=  	ConditionalAndExpression ( <'||'> ConditionalAndExpression )*
-<ConditionalAndExpression>	  ::=  	ValueLogical ( <'&&'> ValueLogical )*
+ConditionalOrExpression ::= ConditionalAndExpression ( <'||'> ConditionalAndExpression )*
+ConditionalAndExpression ::= ValueLogical ( <'&&'> ValueLogical )*
 <ValueLogical>	  ::=  	RelationalExpression
-RelationalExpression	  ::=  	NumericExpression WS ( '=' NumericExpression | '!=' NumericExpression | '<' NumericExpression | '>' NumericExpression | '<=' NumericExpression | '>=' NumericExpression | 'IN' ExpressionList | 'NOT' 'IN' ExpressionList )?
+RelationalExpression	  ::=  	NumericExpression WS ( '=' NumericExpression | '!=' NumericExpression | '<' NumericExpression | '>' NumericExpression | '<=' NumericExpression | '>=' NumericExpression | 'IN' WS ExpressionList | 'NOT' WS 'IN' WS ExpressionList )?
 NumericExpression	  ::=  	WS AdditiveExpression WS
 
 <AdditiveExpression>	  ::=  	MultiplicativeExpression ( '+' MultiplicativeExpression | '-' MultiplicativeExpression | ( NumericLiteralPositive | NumericLiteralNegative ) ( ( '*' UnaryExpression ) | ( '/' UnaryExpression ) )* )*
@@ -188,7 +192,7 @@ RegexExpression   ::=   <'REGEX'> <'('> Expression <','> Expression ( <','> Expr
 SubstringExpression   ::=   <'SUBSTR'> <'('> Expression <','> Expression ( <','> Expression )? <')'>
 StrReplaceExpression    ::=   <'REPLACE'> <'('> Expression <','> Expression <','> Expression ( <','> Expression )? <')'>
 ExistsFunc    ::=   <'EXISTS'> GroupGraphPattern
-NotExistsFunc   ::=   <'NOT'> <'EXISTS'> GroupGraphPattern
+NotExistsFunc   ::=   <'NOT'> WS <'EXISTS'> GroupGraphPattern
 Aggregate   ::=     'COUNT' WS <'('> WS 'DISTINCT'? WS ( '*' | Expression ) WS <')'> WS
 | 'SUM' WS <'('> WS 'DISTINCT'? Expression <')'>
 | 'MIN' <'('>  WS 'DISTINCT'? Expression <')'>
@@ -214,7 +218,7 @@ IRIREF	  ::=  	#"<[^<>\"{}|^`\x00-\x20]*>" WS
 BLANK_NODE_LABEL	  ::=  	'_:' ( PN_CHARS_U | #"[0-9]" ) ((PN_CHARS|'.')* PN_CHARS)?
 <VAR1>	  ::=  	<'?'> VARNAME
 <VAR2>	  ::=  	<'$'> VARNAME
-LANGTAG	  ::=  	#"@[a-zA-Z]+-[a-zA-Z0-9]*" WS
+LANGTAG ::= #"@[a-zA-Z]+(-[a-zA-Z0-9]+)*" WS
 <INTEGER>	  ::=  	#"[0-9]+"
 <DECIMAL>	  ::=  #"[0-9]*\.[0-9]*"
 <DOUBLE>	  ::=  	#"[0-9]+\.[0-9]*|(\.[0-9]+)|([0-9]+)" EXPONENT

diff --git a/src/clj/fluree/db/constants.cljc b/src/clj/fluree/db/constants.cljc
@@ -71,6 +71,7 @@
 (def ^:const iri-rdf-type "http://www.w3.org/1999/02/22-rdf-syntax-ns#type")
 (def ^:const iri-class "http://www.w3.org/2000/01/rdf-schema#Class")
 (def ^:const iri-lang-string "http://www.w3.org/1999/02/22-rdf-syntax-ns#langString")
+(def ^:const iri-string "http://www.w3.org/2001/XMLSchema#string")
 
 ;; rdfs
 (def ^:const iri-rdfs:Class "http://www.w3.org/2000/01/rdf-schema#Class")

diff --git a/src/clj/fluree/db/query/exec/eval.cljc b/src/clj/fluree/db/query/exec/eval.cljc
@@ -285,9 +285,13 @@
   [x]
   (str x))
 
+(defn in
+  [term expressions]
+  (contains? (set expressions) term))
+
 (def allowed-scalar-fns
   '#{&& || ! > < >= <= = + - * / quot and bound coalesce if lang nil? as
-     not not= or re-find re-pattern
+     not not= or re-find re-pattern in
      ;; string fns
      strStarts strEnds subStr strLen ucase lcase contains strBefore strAfter concat regex replace
      ;; numeric fns
@@ -320,6 +324,7 @@
     count          clojure.core/count
     floor          fluree.db.query.exec.eval/floor
     groupconcat    fluree.db.query.exec.eval/groupconcat
+    in             fluree.db.query.exec.eval/in
     lang           fluree.db.query.exec.eval/lang
     lcase          fluree.db.query.exec.eval/lcase
     median         fluree.db.query.exec.eval/median

diff --git a/src/clj/fluree/db/query/exec/update.cljc b/src/clj/fluree/db/query/exec/update.cljc
@@ -75,7 +75,7 @@
                        where/get-datatype-iri
                        (->> (generate-sid! db-vol)))
                (dbproto/-p-prop @db-vol :datatype p-iri)
-               (datatype/infer v))
+               (datatype/infer v (:lang m)))
         v* (datatype/coerce-value v dt)]
     (flake/create sid pid v* dt t true m)))
 

diff --git a/src/clj/fluree/db/query/exec/where.cljc b/src/clj/fluree/db/query/exec/where.cljc
@@ -533,6 +533,20 @@
       (dataset/activate alias)
       (match-clause fuel-tracker solution clause error-ch)))
 
+(defmethod match-pattern :exists
+  [ds fuel-tracker solution pattern error-ch]
+  (let [clause (pattern-data pattern)]
+    (go
+      (when (async/<! (match-clause ds fuel-tracker solution clause error-ch))
+        solution))))
+
+(defmethod match-pattern :not-exists
+  [ds fuel-tracker solution pattern error-ch]
+  (let [clause (pattern-data pattern)]
+    (go
+      (when-not (async/<! (match-clause ds fuel-tracker solution clause error-ch))
+        solution))))
+
 (defmethod match-pattern :graph
   [ds fuel-tracker solution pattern error-ch]
   (let [[g clause] (pattern-data pattern)]
@@ -566,6 +580,19 @@
                           clause-ch)
     out-ch))
 
+(defmethod match-pattern :values
+  [db fuel-tracker solution pattern error-ch]
+  (let [inline-solutions (pattern-data pattern)
+        ;; transform a match into its identity for equality checks
+        match-identity   (juxt get-iri get-value get-datatype-iri (comp get-meta :lang))
+        solution*        (update-vals solution match-identity)]
+    ;; filter out any inline solutions whose matches don't match the solution's matches
+    (->> inline-solutions
+         (filterv (fn [inline-solution] (= (select-keys solution* (keys inline-solution))
+                                           (update-vals inline-solution match-identity))))
+         (mapv (partial merge solution) inline-solutions)
+         (async/to-chan!))))
+
 (defn with-default
   "Return a transducer that transforms an input stream of solutions to include the
   `default-solution` if and only if the stream was empty."

diff --git a/src/clj/fluree/db/query/fql/parse.cljc b/src/clj/fluree/db/query/fql/parse.cljc
@@ -60,7 +60,9 @@
           (json-ld/expand-iri context)
           (where/anonymous-value dt-iri))
       (where/anonymous-value v dt-iri))
-    (where/anonymous-value v)))
+    (if-let [lang (get attrs const/iri-language)]
+      (where/anonymous-value v const/iri-lang-string {:lang lang})
+      (where/anonymous-value v))))
 
 (defn parse-value-attributes
   [v attrs context]
@@ -79,8 +81,10 @@
         (let [expanded (json-ld/expand-iri val context)]
           (where/match-iri var-match expanded))
         (where/match-value var-match val dt-iri))
-      (let [dt (datatype/infer-iri val)]
-        (where/match-value var-match val dt)))))
+      (if-let [lang (get attrs const/iri-language)]
+        (where/match-value var-match val const/iri-lang-string {:lang lang})
+        (let [dt (datatype/infer-iri val)]
+          (where/match-value var-match val dt))))))
 
 (defn match-value-binding
   [var-match value context]
@@ -98,11 +102,11 @@
     (zipmap vars binding)))
 
 (defn parse-values
-  [q context]
-  (when-let [values (:values q)]
+  [values context]
+  (when values
     (let [[vars vals] values
-          vars*     (keep parse-var-name (util/sequential vars))
-          vals*     (mapv util/sequential vals)
+          vars* (keep parse-var-name (util/sequential vars))
+          vals* (mapv util/sequential vals)
           var-count (count vars*)]
       (if (every? (fn [binding]
                     (= (count binding) var-count))
@@ -408,6 +412,19 @@
   (let [parsed (parse-bind-map binds)]
     [(where/->pattern :bind parsed)]))
 
+(defmethod parse-pattern :values
+  [[_ values] vars context]
+  (let [[_vars solutions] (parse-values values context)]
+    [(where/->pattern :values solutions)]))
+
+(defmethod parse-pattern :exists
+  [[_ patterns] vars context]
+  [(where/->pattern :exists (parse-where-clause patterns vars context))])
+
+(defmethod parse-pattern :not-exists
+  [[_ patterns] vars context]
+  [(where/->pattern :not-exists (parse-where-clause patterns vars context))])
+
 (defmethod parse-pattern :graph
   [[_ graph where] vars context]
   (let [graph* (or (parse-variable graph)
@@ -570,7 +587,7 @@
 (defn parse-analytical-query
   [q]
   (let [context       (context/extract q)
-        [vars values] (parse-values q context)
+        [vars values] (parse-values (:values q) context)
         where         (parse-where q vars context)
         grouping      (parse-grouping q)
         ordering      (parse-ordering q)]
@@ -684,8 +701,8 @@
 
 (defn parse-txn
   [txn context]
-  (let [vals-map      {:values (util/get-first-value txn const/iri-values)}
-        [vars values] (parse-values vals-map context)
+  (let [values        (util/get-first-value txn const/iri-values)
+        [vars values] (parse-values values context)
         where-map     {:where (util/get-first-value txn const/iri-where)}
         where         (parse-where where-map vars context)
         bound-vars    (-> where where/bound-variables (into vars))