Skip to content

Commit 95a1e4e

Browse files
david-waltermiredavid-waltermire
committed
This PR adds the following new Metapath map features.
- Support for map construction. - Added support for maps in function call, postfix and unary lookups. - Added support and unit tests for the following Metapath functions: - map:get - map:merge - map:entry - map:size - map:keys - map:contains - map:find - map:put - map:remove - Added missing characteristics on a number of existing Metapath functions. - Cleaned up a bunch of PMD warnings.
1 parent 74698b1 commit 95a1e4e

File tree

103 files changed

+4453
-506
lines changed

Some content is hidden

Large Commits have some content hidden by default. Use the searchbox below for content that may be hidden.

103 files changed

+4453
-506
lines changed

‎core/src/main/antlr4/Metapath10.g4

Lines changed: 7 additions & 10 deletions
Original file line numberDiff line numberDiff line change
@@ -4,11 +4,8 @@ parser grammar Metapath10;
44

55
options { tokenVocab=Metapath10Lexer; superClass=Metapath10ParserBase; }
66

7-
// Metapath extensions
8-
metapath : expr EOF ;
9-
107
// [1]
11-
// xpath : expr EOF ;
8+
metapath : expr EOF ;
129
// paramlist : param ( COMMA param)* ;
1310
// param : DOLLAR eqname typedeclaration? ;
1411
// functionbody : enclosedexpr ;
@@ -81,7 +78,7 @@ keyspecifier : NCName | IntegerLiteral | parenthesizedexpr | STAR ;
8178
//arrowfunctionspecifier : eqname | varref | parenthesizedexpr ;
8279
arrowfunctionspecifier : eqname;
8380
// primaryexpr : literal | varref | parenthesizedexpr | contextitemexpr | functioncall | functionitemexpr | mapconstructor | arrayconstructor | unarylookup ;
84-
primaryexpr : literal | varref | parenthesizedexpr | contextitemexpr | functioncall | arrayconstructor | unarylookup;
81+
primaryexpr : literal | varref | parenthesizedexpr | contextitemexpr | functioncall | mapconstructor | arrayconstructor | unarylookup;
8582
literal : numericliteral | StringLiteral ;
8683
numericliteral : IntegerLiteral | DecimalLiteral | DoubleLiteral ;
8784
varref : DOLLAR varname ;
@@ -97,11 +94,11 @@ argument : exprsingle ;
9794
// functionitemexpr : namedfunctionref | inlinefunctionexpr ;
9895
// namedfunctionref : eqname POUND IntegerLiteral /* xgc: reserved-function-names */;
9996
// inlinefunctionexpr : KW_FUNCTION OP paramlist? CP ( KW_AS sequencetype)? functionbody ;
100-
// mapconstructor : KW_MAP OC (mapconstructorentry ( COMMA mapconstructorentry)*)? CC ;
97+
mapconstructor : KW_MAP OC (mapconstructorentry ( COMMA mapconstructorentry)*)? CC ;
10198
// [70]
102-
// mapconstructorentry : mapkeyexpr COLON mapvalueexpr ;
103-
// mapkeyexpr : exprsingle ;
104-
// mapvalueexpr : exprsingle ;
99+
mapconstructorentry : mapkeyexpr COLON mapvalueexpr ;
100+
mapkeyexpr : exprsingle ;
101+
mapvalueexpr : exprsingle ;
105102
arrayconstructor : squarearrayconstructor | curlyarrayconstructor ;
106103
squarearrayconstructor : OB (exprsingle ( COMMA exprsingle)*)? CB ;
107104
// [75]
@@ -152,7 +149,7 @@ unarylookup : QM keyspecifier ;
152149

153150

154151
// Error in the spec. EQName also includes acceptable keywords.
155-
eqname : QName | URIQualifiedName
152+
eqname : NCName | QName | URIQualifiedName
156153
| KW_ANCESTOR
157154
| KW_ANCESTOR_OR_SELF
158155
| KW_AND

‎core/src/main/antlr4/Metapath10Lexer.g4

Lines changed: 156 additions & 142 deletions
Original file line numberDiff line numberDiff line change
@@ -1,166 +1,180 @@
11
// This grammar is derived from the XPath 3.1 grammar produced by Ken Domino, et al (https://github.com/antlr/grammars-v4/blob/63359bd91593ece31a384acd507ae860d6cf7ff7/xpath/xpath31/XPath31Lexer.g4).
22

3+
// This is a faithful implementation of the XPath version 3.1 grammar
4+
// from the spec at https://www.w3.org/TR/2017/REC-xpath-31-20170321/
5+
6+
// $antlr-format alignTrailingComments true, columnLimit 150, maxEmptyLinesToKeep 1, reflowComments false, useTab false
7+
// $antlr-format allowShortRulesOnASingleLine true, allowShortBlocksOnASingleLine true, minEmptyLines 0, alignSemicolons ownLine
8+
// $antlr-format alignColons trailing, singleLineOverrulesHangingColon true, alignLexerCommands true, alignLabels true, alignTrailers true
9+
310
lexer grammar Metapath10Lexer;
411

5-
AT : '@' ;
6-
BANG : '!' ;
7-
CB : ']' ;
8-
CC : '}' ;
9-
CEQ : ':=' ;
10-
COLON : ':' ;
11-
COLONCOLON : '::' ;
12-
COMMA : ',' ;
13-
CP : ')' ;
14-
CS : ':*' ;
15-
D : '.' ;
16-
DD : '..' ;
17-
DOLLAR : '$' ;
18-
EG : '=>' ;
19-
EQ : '=' ;
20-
GE : '>=' ;
21-
GG : '>>' ;
22-
GT : '>' ;
23-
LE : '<=' ;
24-
LL : '<<' ;
25-
LT : '<' ;
26-
MINUS : '-' ;
27-
NE : '!=' ;
28-
OB : '[' ;
29-
OC : '{' ;
30-
OP : '(' ;
31-
P : '|' ;
32-
PLUS : '+' ;
33-
POUND : '#' ;
34-
PP : '||' ;
35-
QM : '?' ;
36-
SC : '*:' ;
37-
SLASH : '/' ;
38-
SS : '//' ;
39-
STAR : '*' ;
12+
AT : '@';
13+
BANG : '!';
14+
CB : ']';
15+
CC : '}';
16+
CEQ : ':=';
17+
COLON : ':';
18+
COLONCOLON : '::';
19+
COMMA : ',';
20+
CP : ')';
21+
CS : ':*';
22+
D : '.';
23+
DD : '..';
24+
DOLLAR : '$';
25+
EG : '=>';
26+
EQ : '=';
27+
GE : '>=';
28+
GG : '>>';
29+
GT : '>';
30+
LE : '<=';
31+
LL : '<<';
32+
LT : '<';
33+
MINUS : '-';
34+
NE : '!=';
35+
OB : '[';
36+
OC : '{';
37+
OP : '(';
38+
P : '|';
39+
PLUS : '+';
40+
POUND : '#';
41+
PP : '||';
42+
QM : '?';
43+
SC : '*:';
44+
SLASH : '/';
45+
SS : '//';
46+
STAR : '*';
4047

4148
// KEYWORDS
4249

43-
KW_ANCESTOR : 'ancestor' ;
44-
KW_ANCESTOR_OR_SELF : 'ancestor-or-self' ;
45-
KW_AND : 'and' ;
46-
KW_ARRAY : 'array' ;
47-
KW_AS : 'as' ;
48-
KW_ATTRIBUTE : 'attribute' ;
49-
KW_CAST : 'cast' ;
50-
KW_CASTABLE : 'castable' ;
51-
KW_CHILD : 'child' ;
52-
KW_COMMENT : 'comment' ;
53-
KW_DESCENDANT : 'descendant' ;
54-
KW_DESCENDANT_OR_SELF : 'descendant-or-self' ;
55-
KW_DIV : 'div' ;
56-
KW_DOCUMENT_NODE : 'document-node' ;
57-
KW_ELEMENT : 'element' ;
58-
KW_ELSE : 'else' ;
59-
KW_EMPTY_SEQUENCE : 'empty-sequence' ;
60-
KW_EQ : 'eq' ;
61-
KW_EVERY : 'every' ;
62-
KW_EXCEPT : 'except' ;
63-
KW_FOLLOWING : 'following' ;
64-
KW_FOLLOWING_SIBLING : 'following-sibling' ;
65-
KW_FOR : 'for' ;
66-
KW_FUNCTION : 'function' ;
67-
KW_GE : 'ge' ;
68-
KW_GT : 'gt' ;
69-
KW_IDIV : 'idiv' ;
70-
KW_IF : 'if' ;
71-
KW_IN : 'in' ;
72-
KW_INSTANCE : 'instance' ;
73-
KW_INTERSECT : 'intersect' ;
74-
KW_IS : 'is' ;
75-
KW_ITEM : 'item' ;
76-
KW_LE : 'le' ;
77-
KW_LET : 'let' ;
78-
KW_LT : 'lt' ;
79-
KW_MAP : 'map' ;
80-
KW_MOD : 'mod' ;
81-
KW_NAMESPACE : 'namespace' ;
82-
KW_NAMESPACE_NODE : 'namespace-node' ;
83-
KW_NE : 'ne' ;
84-
KW_NODE : 'node' ;
85-
KW_OF : 'of' ;
86-
KW_OR : 'or' ;
87-
KW_PARENT : 'parent' ;
88-
KW_PRECEDING : 'preceding' ;
89-
KW_PRECEDING_SIBLING : 'preceding-sibling' ;
90-
KW_PROCESSING_INSTRUCTION : 'processing-instruction' ;
91-
KW_RETURN : 'return' ;
92-
KW_SATISFIES : 'satisfies' ;
93-
KW_SCHEMA_ATTRIBUTE : 'schema-attribute' ;
94-
KW_SCHEMA_ELEMENT : 'schema-element' ;
95-
KW_SELF : 'self' ;
96-
KW_SOME : 'some' ;
97-
KW_TEXT : 'text' ;
98-
KW_THEN : 'then' ;
99-
KW_TO : 'to' ;
100-
KW_TREAT : 'treat' ;
101-
KW_UNION : 'union' ;
50+
KW_ANCESTOR : 'ancestor';
51+
KW_ANCESTOR_OR_SELF : 'ancestor-or-self';
52+
KW_AND : 'and';
53+
KW_ARRAY : 'array';
54+
KW_AS : 'as';
55+
KW_ATTRIBUTE : 'attribute';
56+
KW_CAST : 'cast';
57+
KW_CASTABLE : 'castable';
58+
KW_CHILD : 'child';
59+
KW_COMMENT : 'comment';
60+
KW_DESCENDANT : 'descendant';
61+
KW_DESCENDANT_OR_SELF : 'descendant-or-self';
62+
KW_DIV : 'div';
63+
KW_DOCUMENT_NODE : 'document-node';
64+
KW_ELEMENT : 'element';
65+
KW_ELSE : 'else';
66+
KW_EMPTY_SEQUENCE : 'empty-sequence';
67+
KW_EQ : 'eq';
68+
KW_EVERY : 'every';
69+
KW_EXCEPT : 'except';
70+
KW_FOLLOWING : 'following';
71+
KW_FOLLOWING_SIBLING : 'following-sibling';
72+
KW_FOR : 'for';
73+
KW_FUNCTION : 'function';
74+
KW_GE : 'ge';
75+
KW_GT : 'gt';
76+
KW_IDIV : 'idiv';
77+
KW_IF : 'if';
78+
KW_IN : 'in';
79+
KW_INSTANCE : 'instance';
80+
KW_INTERSECT : 'intersect';
81+
KW_IS : 'is';
82+
KW_ITEM : 'item';
83+
KW_LE : 'le';
84+
KW_LET : 'let';
85+
KW_LT : 'lt';
86+
KW_MAP : 'map';
87+
KW_MOD : 'mod';
88+
KW_NAMESPACE : 'namespace';
89+
KW_NAMESPACE_NODE : 'namespace-node';
90+
KW_NE : 'ne';
91+
KW_NODE : 'node';
92+
KW_OF : 'of';
93+
KW_OR : 'or';
94+
KW_PARENT : 'parent';
95+
KW_PRECEDING : 'preceding';
96+
KW_PRECEDING_SIBLING : 'preceding-sibling';
97+
KW_PROCESSING_INSTRUCTION : 'processing-instruction';
98+
KW_RETURN : 'return';
99+
KW_SATISFIES : 'satisfies';
100+
KW_SCHEMA_ATTRIBUTE : 'schema-attribute';
101+
KW_SCHEMA_ELEMENT : 'schema-element';
102+
KW_SELF : 'self';
103+
KW_SOME : 'some';
104+
KW_TEXT : 'text';
105+
KW_THEN : 'then';
106+
KW_TO : 'to';
107+
KW_TREAT : 'treat';
108+
KW_UNION : 'union';
102109

103110
// A.2.1. TERMINAL SYMBOLS
104111
// This isn't a complete list of tokens in the language.
105112
// Keywords and symbols are terminals.
106113

107-
IntegerLiteral : FragDigits ;
108-
DecimalLiteral : '.' FragDigits | FragDigits '.' [0-9]* ;
109-
DoubleLiteral : ('.' FragDigits | FragDigits ('.' [0-9]*)?) [eE] [+-]? FragDigits ;
110-
StringLiteral : '"' (~["] | FragEscapeQuot)* '"' | '\'' (~['] | FragEscapeApos)* '\'' ;
111-
URIQualifiedName : BracedURILiteral NCName ;
112-
BracedURILiteral : 'Q' '{' [^{}]* '}' ;
114+
IntegerLiteral : FragDigits;
115+
DecimalLiteral : '.' FragDigits | FragDigits '.' [0-9]*;
116+
DoubleLiteral : ('.' FragDigits | FragDigits ('.' [0-9]*)?) [eE] [+-]? FragDigits;
117+
StringLiteral : '"' (~["] | FragEscapeQuot)* '"' | '\'' (~['] | FragEscapeApos)* '\'';
118+
URIQualifiedName : BracedURILiteral NCName;
119+
BracedURILiteral : 'Q' '{' [^{}]* '}';
113120
// Error in spec: EscapeQuot and EscapeApos are not terminals!
114-
fragment FragEscapeQuot : '""' ;
121+
fragment FragEscapeQuot : '""';
115122
fragment FragEscapeApos : '\'\'';
116123
// Error in spec: Comment isn't really a terminal, but an off-channel object.
117-
Comment : '(:' (Comment | CommentContents)*? ':)' -> skip ;
118-
QName : FragQName ;
119-
NCName : FragmentNCName ;
124+
Comment : '(:' (Comment | CommentContents)*? ':)' -> skip;
125+
NCName : FragmentNCName;
126+
QName : FragQName;
120127
// Error in spec: Char is not a terminal!
121-
fragment Char : FragChar ;
122-
fragment FragDigits : [0-9]+ ;
123-
fragment CommentContents : Char ;
128+
fragment Char : FragChar;
129+
fragment FragDigits : [0-9]+;
130+
fragment CommentContents : Char;
124131
// https://www.w3.org/TR/REC-xml-names/#NT-QName
125-
fragment FragQName : FragPrefixedName | FragUnprefixedName ;
126-
fragment FragPrefixedName : FragPrefix ':' FragLocalPart ;
127-
fragment FragUnprefixedName : FragLocalPart ;
128-
fragment FragPrefix : FragmentNCName ;
129-
fragment FragLocalPart : FragmentNCName ;
130-
fragment FragNCNameStartChar
131-
: 'A'..'Z'
132-
| '_'
133-
| 'a'..'z'
134-
| '\u00C0'..'\u00D6'
135-
| '\u00D8'..'\u00F6'
136-
| '\u00F8'..'\u02FF'
137-
| '\u0370'..'\u037D'
138-
| '\u037F'..'\u1FFF'
139-
| '\u200C'..'\u200D'
140-
| '\u2070'..'\u218F'
141-
| '\u2C00'..'\u2FEF'
142-
| '\u3001'..'\uD7FF'
143-
| '\uF900'..'\uFDCF'
144-
| '\uFDF0'..'\uFFFD'
145-
| '\u{10000}'..'\u{EFFFF}'
146-
;
147-
fragment FragNCNameChar
148-
: FragNCNameStartChar | '-' | '.' | '0'..'9'
149-
| '\u00B7' | '\u0300'..'\u036F'
150-
| '\u203F'..'\u2040'
151-
;
152-
fragment FragmentNCName : FragNCNameStartChar FragNCNameChar* ;
132+
fragment FragQName : FragPrefixedName | FragUnprefixedName;
133+
fragment FragPrefixedName : FragPrefix ':' FragLocalPart;
134+
fragment FragUnprefixedName : FragLocalPart;
135+
fragment FragPrefix : FragmentNCName;
136+
fragment FragLocalPart : FragmentNCName;
137+
fragment FragNCNameStartChar:
138+
'A' ..'Z'
139+
| '_'
140+
| 'a' ..'z'
141+
| '\u00C0' ..'\u00D6'
142+
| '\u00D8' ..'\u00F6'
143+
| '\u00F8' ..'\u02FF'
144+
| '\u0370' ..'\u037D'
145+
| '\u037F' ..'\u1FFF'
146+
| '\u200C' ..'\u200D'
147+
| '\u2070' ..'\u218F'
148+
| '\u2C00' ..'\u2FEF'
149+
| '\u3001' ..'\uD7FF'
150+
| '\uF900' ..'\uFDCF'
151+
| '\uFDF0' ..'\uFFFD'
152+
| '\u{10000}' ..'\u{EFFFF}'
153+
;
154+
fragment FragNCNameChar:
155+
FragNCNameStartChar
156+
| '-'
157+
| '.'
158+
| '0' ..'9'
159+
| '\u00B7'
160+
| '\u0300' ..'\u036F'
161+
| '\u203F' ..'\u2040'
162+
;
163+
fragment FragmentNCName: FragNCNameStartChar FragNCNameChar*;
153164

154165
// https://www.w3.org/TR/REC-xml/#NT-Char
155166

156-
fragment FragChar : '\u0009' | '\u000a' | '\u000d'
157-
| '\u0020'..'\ud7ff'
158-
| '\ue000'..'\ufffd'
159-
| '\u{10000}'..'\u{10ffff}'
160-
;
167+
fragment FragChar:
168+
'\u0009'
169+
| '\u000a'
170+
| '\u000d'
171+
| '\u0020' ..'\ud7ff'
172+
| '\ue000' ..'\ufffd'
173+
| '\u{10000}' ..'\u{10ffff}'
174+
;
161175

162176
// https://github.com/antlr/grammars-v4/blob/17d3db3fd6a8fc319a12176e0bb735b066ec0616/xpath/xpath31/XPath31.g4#L389
163-
Whitespace : ('\u000d' | '\u000a' | '\u0020' | '\u0009')+ -> skip ;
177+
Whitespace: ('\u000d' | '\u000a' | '\u0020' | '\u0009')+ -> skip;
164178

165179
// Not per spec. Specified for testing.
166-
SEMI : ';' ;
180+
SEMI: ';';

‎core/src/main/java/gov/nist/secauto/metaschema/core/datatype/markup/flexmark/impl/AbstractMarkupWriter.java

Lines changed: 1 addition & 2 deletions
Original file line numberDiff line numberDiff line change
@@ -63,9 +63,8 @@
6363
import com.vladsch.flexmark.util.sequence.BasedSequence;
6464
import com.vladsch.flexmark.util.sequence.Escaping;
6565

66-
import gov.nist.secauto.metaschema.core.datatype.markup.flexmark.IMarkupWriter;
6766
import gov.nist.secauto.metaschema.core.datatype.markup.flexmark.HtmlQuoteTagExtension.DoubleQuoteNode;
68-
import gov.nist.secauto.metaschema.core.datatype.markup.flexmark.IMarkupWriter.ChildHandler;
67+
import gov.nist.secauto.metaschema.core.datatype.markup.flexmark.IMarkupWriter;
6968
import gov.nist.secauto.metaschema.core.datatype.markup.flexmark.InsertAnchorExtension.InsertAnchorNode;
7069
import gov.nist.secauto.metaschema.core.util.CollectionUtil;
7170
import gov.nist.secauto.metaschema.core.util.ObjectUtils;

‎core/src/main/java/gov/nist/secauto/metaschema/core/metapath/ICollectionValue.java

Lines changed: 3 additions & 1 deletion
Original file line numberDiff line numberDiff line change
@@ -40,7 +40,9 @@ public interface ICollectionValue {
4040

4141
@NonNull
4242
static Stream<? extends IItem> normalizeAsItems(@NonNull ICollectionValue value) {
43-
return value instanceof IItem ? ObjectUtils.notNull(Stream.of((IItem) value)) : value.asSequence().stream();
43+
return value instanceof IItem
44+
? ObjectUtils.notNull(Stream.of((IItem) value))
45+
: value.asSequence().stream();
4446
}
4547

4648
@NonNull

0 commit comments

Comments
 (0)