Skip to content

Commit 25a9dcb

Browse files
author
Nick Bargnesi
committed
fixes #1
Add all ANTLR-based grammars and XML schema definitions.
1 parent 17877d1 commit 25a9dcb

File tree

11 files changed

+3691
-0
lines changed

11 files changed

+3691
-0
lines changed

‎version_1.0/ANTLR/BELScript.g

Lines changed: 333 additions & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -0,0 +1,333 @@
1+
grammar BELScript;
2+
3+
options {
4+
language = Java;
5+
output = AST;
6+
}
7+
8+
@header {
9+
package org.openbel.framework.common.bel.parser;
10+
11+
import java.util.List;
12+
import java.util.ArrayList;
13+
import java.util.Stack;
14+
15+
import org.openbel.bel.model.BELParseErrorException;
16+
}
17+
18+
@lexer::header {
19+
package org.openbel.framework.common.bel.parser;
20+
}
21+
22+
@members {
23+
private final List<BELParseErrorException> syntaxErrors = new ArrayList<BELParseErrorException>();
24+
private final Stack<String> paraphrases = new Stack<String>();
25+
26+
public List<BELParseErrorException> getSyntaxErrors() {
27+
return syntaxErrors;
28+
}
29+
30+
@Override
31+
public void emitErrorMessage(String msg) {
32+
}
33+
34+
@Override
35+
public void displayRecognitionError(String[] tokenNames, RecognitionException e) {
36+
String context = "";
37+
if (paraphrases.size() > 0) {
38+
context = paraphrases.peek();
39+
}
40+
syntaxErrors.add(new BELParseErrorException.SyntaxException(e.line, e.charPositionInLine, context, e));
41+
}
42+
}
43+
44+
document:
45+
(NEWLINE | DOCUMENT_COMMENT | record)+ EOF
46+
;
47+
48+
record:
49+
(define_namespace | define_annotation | set_annotation | set_document | set_statement_group | unset_statement_group | unset | statement)
50+
;
51+
52+
set_document
53+
@init { paraphrases.push("in set document."); }
54+
@after { paraphrases.pop(); }
55+
:
56+
('SET' DOCUMENT_KEYWORD) document_property '=' (OBJECT_IDENT | vl=VALUE_LIST | quoted_value)
57+
{
58+
// https://github.com/OpenBEL/openbel-framework/issues/14
59+
if ($vl != null) $vl.setText($vl.getText().replace("\\\\", "\\"));
60+
}
61+
;
62+
63+
set_statement_group
64+
@init { paraphrases.push("in set statement group."); }
65+
@after { paraphrases.pop(); }
66+
:
67+
'SET' STATEMENT_GROUP_KEYWORD '=' (quoted_value | OBJECT_IDENT)
68+
;
69+
70+
set_annotation
71+
@init { paraphrases.push("in set annotation."); }
72+
@after { paraphrases.pop(); }
73+
:
74+
'SET' OBJECT_IDENT '=' (quoted_value | vl=VALUE_LIST | OBJECT_IDENT)
75+
{
76+
// https://github.com/OpenBEL/openbel-framework/issues/14
77+
if ($vl != null) $vl.setText($vl.getText().replace("\\\\", "\\"));
78+
}
79+
;
80+
81+
unset_statement_group
82+
@init { paraphrases.push("in unset statement group."); }
83+
@after { paraphrases.pop(); }
84+
:
85+
'UNSET' STATEMENT_GROUP_KEYWORD
86+
;
87+
88+
unset
89+
@init { paraphrases.push("in unset."); }
90+
@after { paraphrases.pop(); }
91+
:
92+
'UNSET' (OBJECT_IDENT | IDENT_LIST)
93+
;
94+
95+
define_namespace
96+
@init { paraphrases.push("in define namespace."); }
97+
@after { paraphrases.pop(); }
98+
:
99+
('DEFINE' (('DEFAULT')? 'NAMESPACE')) OBJECT_IDENT 'AS' 'URL' quoted_value
100+
;
101+
102+
define_annotation
103+
@init { paraphrases.push("in define annotation."); }
104+
@after { paraphrases.pop(); }
105+
:
106+
('DEFINE' 'ANNOTATION') OBJECT_IDENT 'AS' ((('URL' | 'PATTERN') quoted_value) | ('LIST' vl=VALUE_LIST))
107+
{
108+
// https://github.com/OpenBEL/openbel-framework/issues/14
109+
if ($vl != null) $vl.setText($vl.getText().replace("\\\\", "\\"));
110+
}
111+
;
112+
113+
quoted_value
114+
: qv=QUOTED_VALUE
115+
{
116+
// https://github.com/OpenBEL/openbel-framework/issues/14
117+
$qv.setText($qv.getText().replace("\\\\", "\\"));
118+
}
119+
;
120+
121+
document_property:
122+
'Authors' |
123+
'ContactInfo' |
124+
'Copyright' |
125+
'Description' |
126+
'Disclaimer' |
127+
'Licenses' |
128+
'Name' |
129+
'Version'
130+
;
131+
132+
statement
133+
@init { paraphrases.push("in statement."); }
134+
@after { paraphrases.pop(); }
135+
:
136+
outer_term (relationship ((OPEN_PAREN outer_term relationship outer_term CLOSE_PAREN) | outer_term))? STATEMENT_COMMENT?
137+
;
138+
139+
outer_term:
140+
function OPEN_PAREN (','? argument)* CLOSE_PAREN
141+
;
142+
143+
argument:
144+
param | term
145+
;
146+
147+
term:
148+
function OPEN_PAREN (','? (term | param))* CLOSE_PAREN
149+
;
150+
151+
/* XXX OBJECT_IDENT is used for namespace value because otherwise parsing will fail using a token like (LETTER | DIGIT)+ */
152+
fragment param:
153+
NS_PREFIX? (OBJECT_IDENT | quoted_value)
154+
;
155+
156+
function returns [String r]:
157+
(
158+
fv='proteinAbundance' {$r = $fv.getText();} |
159+
fv='p' {$r = $fv.getText();} |
160+
fv='rnaAbundance' {$r = $fv.getText();} |
161+
fv='r' {$r = $fv.getText();} |
162+
fv='abundance' {$r = $fv.getText();} |
163+
fv='a' {$r = $fv.getText();} |
164+
fv='microRNAAbundance' {$r = $fv.getText();} |
165+
fv='m' {$r = $fv.getText();} |
166+
fv='geneAbundance' {$r = $fv.getText();} |
167+
fv='g' {$r = $fv.getText();} |
168+
fv='biologicalProcess' {$r = $fv.getText();} |
169+
fv='bp' {$r = $fv.getText();} |
170+
fv='pathology' {$r = $fv.getText();} |
171+
fv='path' {$r = $fv.getText();} |
172+
fv='complexAbundance' {$r = $fv.getText();} |
173+
fv='complex' {$r = $fv.getText();} |
174+
fv='translocation' {$r = $fv.getText();} |
175+
fv='tloc' {$r = $fv.getText();} |
176+
fv='cellSecretion' {$r = $fv.getText();} |
177+
fv='sec' {$r = $fv.getText();} |
178+
fv='cellSurfaceExpression' {$r = $fv.getText();} |
179+
fv='surf' {$r = $fv.getText();} |
180+
fv='reaction' {$r = $fv.getText();} |
181+
fv='rxn' {$r = $fv.getText();} |
182+
fv='compositeAbundance' {$r = $fv.getText();} |
183+
fv='composite' {$r = $fv.getText();} |
184+
fv='fusion' {$r = $fv.getText();} |
185+
fv='fus' {$r = $fv.getText();} |
186+
fv='degradation' {$r = $fv.getText();} |
187+
fv='deg' {$r = $fv.getText();} |
188+
fv='molecularActivity' {$r = $fv.getText();} |
189+
fv='act' {$r = $fv.getText();} |
190+
fv='catalyticActivity' {$r = $fv.getText();} |
191+
fv='cat' {$r = $fv.getText();} |
192+
fv='kinaseActivity' {$r = $fv.getText();} |
193+
fv='kin' {$r = $fv.getText();} |
194+
fv='phosphataseActivity' {$r = $fv.getText();} |
195+
fv='phos' {$r = $fv.getText();} |
196+
fv='peptidaseActivity' {$r = $fv.getText();} |
197+
fv='pep' {$r = $fv.getText();} |
198+
fv='ribosylationActivity' {$r = $fv.getText();} |
199+
fv='ribo' {$r = $fv.getText();} |
200+
fv='transcriptionalActivity' {$r = $fv.getText();} |
201+
fv='tscript' {$r = $fv.getText();} |
202+
fv='transportActivity' {$r = $fv.getText();} |
203+
fv='tport' {$r = $fv.getText();} |
204+
fv='gtpBoundActivity' {$r = $fv.getText();} |
205+
fv='gtp' {$r = $fv.getText();} |
206+
fv='chaperoneActivity' {$r = $fv.getText();} |
207+
fv='chap' {$r = $fv.getText();} |
208+
fv='proteinModification' {$r = $fv.getText();} |
209+
fv='pmod' {$r = $fv.getText();} |
210+
fv='substitution' {$r = $fv.getText();} |
211+
fv='sub' {$r = $fv.getText();} |
212+
fv='truncation' {$r = $fv.getText();} |
213+
fv='trunc' {$r = $fv.getText();} |
214+
fv='reactants' {$r = $fv.getText();} |
215+
fv='products' {$r = $fv.getText();} |
216+
fv='list' {$r = $fv.getText();}
217+
)
218+
;
219+
220+
relationship returns [String r]:
221+
(
222+
rv='increases' { $r = $rv.getText(); } |
223+
rv='->' { $r = $rv.getText(); } |
224+
rv='decreases' { $r = $rv.getText(); } |
225+
rv='-|' { $r = $rv.getText(); } |
226+
rv='directlyIncreases' { $r = $rv.getText(); } |
227+
rv='=>' { $r = $rv.getText(); } |
228+
rv='directlyDecreases' { $r = $rv.getText(); } |
229+
rv='=|' { $r = $rv.getText(); } |
230+
rv='causesNoChange' { $r = $rv.getText(); } |
231+
rv='positiveCorrelation' { $r = $rv.getText(); } |
232+
rv='negativeCorrelation' { $r = $rv.getText(); } |
233+
rv='translatedTo' { $r = $rv.getText(); } |
234+
rv='>>' { $r = $rv.getText(); } |
235+
rv='transcribedTo' { $r = $rv.getText(); } |
236+
rv=':>' { $r = $rv.getText(); } |
237+
rv='isA' { $r = $rv.getText(); } |
238+
rv='subProcessOf' { $r = $rv.getText(); } |
239+
rv='rateLimitingStepOf' { $r = $rv.getText(); } |
240+
rv='biomarkerFor' { $r = $rv.getText(); } |
241+
rv='prognosticBiomarkerFor' { $r = $rv.getText(); } |
242+
rv='orthologous' { $r = $rv.getText(); } |
243+
rv='analogous' { $r = $rv.getText(); } |
244+
rv='association' { $r = $rv.getText(); } |
245+
rv='--' { $r = $rv.getText(); } |
246+
rv='hasMembers' { $r = $rv.getText(); } |
247+
rv='hasComponents' { $r = $rv.getText(); } |
248+
rv='hasMember' { $r = $rv.getText(); } |
249+
rv='hasComponent' { $r = $rv.getText(); }
250+
)
251+
;
252+
253+
DOCUMENT_COMMENT:
254+
'#' ~('\n' | '\r')* {$channel=HIDDEN;}
255+
;
256+
257+
STATEMENT_COMMENT:
258+
'//' (('\\\n') | ('\\\r\n') | ~('\n' | '\r'))*
259+
;
260+
261+
DOCUMENT_KEYWORD:
262+
'DOCUMENT'
263+
;
264+
265+
STATEMENT_GROUP_KEYWORD:
266+
'STATEMENT_GROUP'
267+
;
268+
269+
IDENT_LIST:
270+
'{' OBJECT_IDENT (COMMA OBJECT_IDENT)* '}'
271+
;
272+
273+
VALUE_LIST:
274+
'{' (OBJECT_IDENT | QUOTED_VALUE | VALUE_LIST)? (COMMA (OBJECT_IDENT | QUOTED_VALUE | VALUE_LIST)?)* '}'
275+
;
276+
277+
OBJECT_IDENT:
278+
('_' | LETTER | DIGIT)+
279+
;
280+
281+
QUOTED_VALUE:
282+
'"' ( EscapeSequence | '\\\n' | '\\\r\n' | ~('\\'|'"') )* '"'
283+
;
284+
285+
OPEN_PAREN:
286+
'('
287+
;
288+
289+
CLOSE_PAREN:
290+
')'
291+
;
292+
293+
NS_PREFIX:
294+
LETTER (LETTER | DIGIT)* ':'
295+
;
296+
297+
NEWLINE:
298+
'\u000d'? '\u000a' | '\u000d'
299+
;
300+
301+
WS : (' ' | '\t' | '\n' | '\r'| '\f' | '\\\n' | '\\\r\n')+ {$channel = HIDDEN;};
302+
303+
fragment COMMA:
304+
' '* ',' ' '*
305+
;
306+
307+
fragment LETTER:
308+
('a'..'z' | 'A'..'Z')
309+
;
310+
311+
fragment DIGIT:
312+
'0'..'9'
313+
;
314+
315+
fragment EscapeSequence:
316+
'\\' ('b'|'t'|'n'|'f'|'r'|'\"'|'\''|'\\')
317+
| UnicodeEscape
318+
| OctalEscape
319+
;
320+
321+
fragment OctalEscape:
322+
'\\' ('0'..'3') ('0'..'7') ('0'..'7')
323+
| '\\' ('0'..'7') ('0'..'7')
324+
| '\\' ('0'..'7')
325+
;
326+
327+
fragment UnicodeEscape:
328+
'\\' 'u' HexDigit HexDigit HexDigit HexDigit
329+
;
330+
331+
fragment HexDigit:
332+
('0'..'9'|'a'..'f'|'A'..'F')
333+
;

0 commit comments

Comments
 (0)