Joachim Bingel | dbbde77 | 2014-05-12 15:26:10 +0000 | [diff] [blame] | 1 | grammar CollectionQuery; |
| 2 | |
| 3 | @header {package de.ids_mannheim.korap.query.serialize.util;} |
| 4 | |
| 5 | /* |
| 6 | -- author: jbingel |
Joachim Bingel | 787836a | 2014-08-07 14:50:18 +0000 | [diff] [blame] | 7 | -- date: 2014-05-11 |
Joachim Bingel | dbbde77 | 2014-05-12 15:26:10 +0000 | [diff] [blame] | 8 | */ |
| 9 | |
| 10 | /* |
| 11 | * LEXER SECTION |
| 12 | */ |
Joachim Bingel | 787836a | 2014-08-07 14:50:18 +0000 | [diff] [blame] | 13 | /* |
| 14 | Regular expression |
| 15 | /x allows submatches like /^.*?RE.*?$/ |
| 16 | /X forces full matches |
| 17 | /i means case insensitivity |
| 18 | /I forces case sensitivity |
| 19 | */ |
| 20 | FLAG_xi : '/' ( ('x'|'X') ('i'|'I')? ); |
| 21 | FLAG_ix : '/' ( ('i'|'I') ('x'|'X')? ); |
| 22 | |
| 23 | |
Joachim Bingel | 43607ed | 2014-05-19 12:39:55 +0000 | [diff] [blame] | 24 | LRB : '('; |
| 25 | RRB : ')'; |
Joachim Bingel | 787836a | 2014-08-07 14:50:18 +0000 | [diff] [blame] | 26 | LB : '['; |
| 27 | RB : ']'; |
Joachim Bingel | 43607ed | 2014-05-19 12:39:55 +0000 | [diff] [blame] | 28 | LT : '<'; |
| 29 | GT : '>'; |
| 30 | LEQ : '<='; |
| 31 | GEQ : '>='; |
| 32 | EQ : '='; |
Joachim Bingel | 43607ed | 2014-05-19 12:39:55 +0000 | [diff] [blame] | 33 | AND : '&' | 'AND' | 'and' | 'UND' | 'und' ; |
| 34 | OR : '|' | 'OR' | 'or' | 'ODER' | 'oder' ; |
Joachim Bingel | 787836a | 2014-08-07 14:50:18 +0000 | [diff] [blame] | 35 | NEG : '!'; |
Joachim Bingel | a3f51f7 | 2014-07-22 14:45:31 +0000 | [diff] [blame] | 36 | QMARK : '?'; |
| 37 | SLASH : '/'; |
Joachim Bingel | 787836a | 2014-08-07 14:50:18 +0000 | [diff] [blame] | 38 | COLON : ':'; |
Joachim Bingel | 624854b | 2014-07-23 13:53:28 +0000 | [diff] [blame] | 39 | DASH : '-'; |
Joachim Bingel | 787836a | 2014-08-07 14:50:18 +0000 | [diff] [blame] | 40 | TILDE : '~'; |
Joachim Bingel | 3fa584b | 2014-12-17 13:35:43 +0000 | [diff] [blame] | 41 | NEGTILDE : '!~'; |
Joachim Bingel | 1f4c5ad | 2014-12-16 10:40:42 +0000 | [diff] [blame] | 42 | SINCE : 'since'; |
| 43 | UNTIL : 'until'; |
| 44 | IN : 'in'; |
| 45 | ON : 'on'; |
Joachim Bingel | dbbde77 | 2014-05-12 15:26:10 +0000 | [diff] [blame] | 46 | WS : ( ' ' | '\t' | '\r' | '\n' )+ -> skip ; |
Joachim Bingel | dbbde77 | 2014-05-12 15:26:10 +0000 | [diff] [blame] | 47 | fragment NO_RE : ~[ \t\/]; |
| 48 | fragment ALPHABET : ~('\t' | ' ' | '/' | '*' | '?' | '+' | '{' | '}' | '[' | ']' |
| 49 | | '(' | ')' | '|' | '"' | ',' | ':' | '\'' | '\\' | '!' | '=' | '~' | '&' | '^' | '<' | '>' ); |
Joachim Bingel | 4bf1ec5 | 2014-10-08 08:18:03 +0000 | [diff] [blame] | 50 | fragment ALPHA : [a-zA-Z]; |
Joachim Bingel | 4bf1ec5 | 2014-10-08 08:18:03 +0000 | [diff] [blame] | 51 | |
Joachim Bingel | 1f4c5ad | 2014-12-16 10:40:42 +0000 | [diff] [blame] | 52 | |
| 53 | DIGIT : [0-9]; |
Joachim Bingel | 4bf1ec5 | 2014-10-08 08:18:03 +0000 | [diff] [blame] | 54 | DATE |
| 55 | : DIGIT DIGIT DIGIT DIGIT (DASH DIGIT DIGIT (DASH DIGIT DIGIT)?)? |
| 56 | ; |
Joachim Bingel | dbbde77 | 2014-05-12 15:26:10 +0000 | [diff] [blame] | 57 | |
| 58 | NL : [\r\n] -> skip; |
| 59 | ws : WS+; |
| 60 | |
Joachim Bingel | 1f4c5ad | 2014-12-16 10:40:42 +0000 | [diff] [blame] | 61 | WORD : ALPHABET+; |
| 62 | //WORD : ALPHABET* ALPHA ALPHABET*; // needs to have at least one alphabetical letter (non-numeric) |
| 63 | |
Joachim Bingel | 787836a | 2014-08-07 14:50:18 +0000 | [diff] [blame] | 64 | |
Joachim Bingel | 43607ed | 2014-05-19 12:39:55 +0000 | [diff] [blame] | 65 | /* |
Joachim Bingel | a3f51f7 | 2014-07-22 14:45:31 +0000 | [diff] [blame] | 66 | * Regular expressions |
Joachim Bingel | 43607ed | 2014-05-19 12:39:55 +0000 | [diff] [blame] | 67 | */ |
Joachim Bingel | a3f51f7 | 2014-07-22 14:45:31 +0000 | [diff] [blame] | 68 | fragment FOCC : '{' WS* ( [0-9]* WS* ',' WS* [0-9]+ | [0-9]+ WS* ','? ) WS* '}'; |
Joachim Bingel | 43607ed | 2014-05-19 12:39:55 +0000 | [diff] [blame] | 69 | fragment RE_char : ~('*' | '?' | '+' | '{' | '}' | '[' | ']' | '/' |
| 70 | | '(' | ')' | '|' | '"' | ':' | '\'' | '\\'); |
| 71 | fragment RE_alter : ((RE_char | ('(' REGEX ')') | RE_chgroup) '|' REGEX )+; |
| 72 | fragment RE_chgroup : '[' RE_char+ ']'; |
Joachim Bingel | a3f51f7 | 2014-07-22 14:45:31 +0000 | [diff] [blame] | 73 | fragment RE_quant : (RE_star | RE_plus | RE_occ) QMARK?; |
Joachim Bingel | 43607ed | 2014-05-19 12:39:55 +0000 | [diff] [blame] | 74 | fragment RE_opt : (RE_char | RE_chgroup | ( '(' REGEX ')')) '?'; |
| 75 | fragment RE_star : (RE_char | RE_chgroup | ( '(' REGEX ')')) '*'; |
| 76 | fragment RE_plus : (RE_char | RE_chgroup | ( '(' REGEX ')')) '+'; |
| 77 | fragment RE_occ : (RE_char | RE_chgroup | ( '(' REGEX ')')) FOCC; |
| 78 | fragment RE_group : '(' REGEX ')'; |
Joachim Bingel | a3f51f7 | 2014-07-22 14:45:31 +0000 | [diff] [blame] | 79 | REGEX : SLASH ('.' | RE_char | RE_alter | RE_chgroup | RE_opt | RE_quant | RE_group)* SLASH; |
Joachim Bingel | dbbde77 | 2014-05-12 15:26:10 +0000 | [diff] [blame] | 80 | |
| 81 | /* |
| 82 | * PARSER SECTION |
| 83 | */ |
| 84 | |
Joachim Bingel | 43607ed | 2014-05-19 12:39:55 +0000 | [diff] [blame] | 85 | regex |
Joachim Bingel | 624854b | 2014-07-23 13:53:28 +0000 | [diff] [blame] | 86 | : REGEX |
| 87 | ; |
| 88 | |
| 89 | date |
Joachim Bingel | 4bf1ec5 | 2014-10-08 08:18:03 +0000 | [diff] [blame] | 90 | : DATE |
Joachim Bingel | 43607ed | 2014-05-19 12:39:55 +0000 | [diff] [blame] | 91 | ; |
| 92 | |
Joachim Bingel | 1f4c5ad | 2014-12-16 10:40:42 +0000 | [diff] [blame] | 93 | dateOp |
| 94 | : SINCE |
| 95 | | UNTIL |
| 96 | | IN |
| 97 | | ON |
| 98 | ; |
| 99 | |
Joachim Bingel | dbbde77 | 2014-05-12 15:26:10 +0000 | [diff] [blame] | 100 | operator |
Joachim Bingel | 3fa584b | 2014-12-17 13:35:43 +0000 | [diff] [blame] | 101 | : (NEG? EQ) | LT | GT | LEQ | GEQ | TILDE | NEGTILDE; |
Joachim Bingel | dbbde77 | 2014-05-12 15:26:10 +0000 | [diff] [blame] | 102 | |
| 103 | expr |
Joachim Bingel | 1f4c5ad | 2014-12-16 10:40:42 +0000 | [diff] [blame] | 104 | : constraint |
| 105 | | dateconstraint |
Joachim Bingel | 787836a | 2014-08-07 14:50:18 +0000 | [diff] [blame] | 106 | | token |
| 107 | ; |
| 108 | |
Joachim Bingel | 1f4c5ad | 2014-12-16 10:40:42 +0000 | [diff] [blame] | 109 | dateconstraint |
| 110 | : field dateOp date |
| 111 | //| date dateOp field dateOp date |
| 112 | ; |
| 113 | |
| 114 | constraint |
| 115 | : field operator value |
Joachim Bingel | 787836a | 2014-08-07 14:50:18 +0000 | [diff] [blame] | 116 | ; |
| 117 | |
| 118 | token |
| 119 | : LB (term|termGroup) RB |
| 120 | ; |
| 121 | |
| 122 | term |
| 123 | : NEG* (foundry SLASH)? layer termOp key (COLON value)? flag? |
| 124 | ; |
| 125 | |
| 126 | termOp |
| 127 | : (NEG? EQ? EQ | NEG? TILDE? TILDE) |
| 128 | ; |
| 129 | |
| 130 | termGroup |
| 131 | : (term | LRB termGroup RRB) booleanOp (term | LRB termGroup RRB | termGroup) |
| 132 | ; |
| 133 | |
| 134 | key |
| 135 | : WORD |
| 136 | | regex |
Joachim Bingel | 787836a | 2014-08-07 14:50:18 +0000 | [diff] [blame] | 137 | ; |
| 138 | |
| 139 | foundry |
| 140 | : WORD |
| 141 | ; |
| 142 | |
| 143 | layer |
| 144 | : WORD |
| 145 | ; |
| 146 | |
| 147 | booleanOp |
| 148 | : AND |
| 149 | | OR |
| 150 | ; |
| 151 | |
| 152 | flag |
| 153 | : FLAG_xi |
| 154 | | FLAG_ix |
Joachim Bingel | dbbde77 | 2014-05-12 15:26:10 +0000 | [diff] [blame] | 155 | ; |
| 156 | |
| 157 | field |
Joachim Bingel | 624854b | 2014-07-23 13:53:28 +0000 | [diff] [blame] | 158 | : WORD |
Joachim Bingel | a3f51f7 | 2014-07-22 14:45:31 +0000 | [diff] [blame] | 159 | ; |
Joachim Bingel | dbbde77 | 2014-05-12 15:26:10 +0000 | [diff] [blame] | 160 | |
| 161 | value |
Joachim Bingel | 4bf1ec5 | 2014-10-08 08:18:03 +0000 | [diff] [blame] | 162 | : WORD |
Joachim Bingel | 1f4c5ad | 2014-12-16 10:40:42 +0000 | [diff] [blame] | 163 | | DIGIT+ |
| 164 | | DATE |
Joachim Bingel | 82e4ca7 | 2014-10-27 11:03:38 +0000 | [diff] [blame] | 165 | | multiword |
Joachim Bingel | a3f51f7 | 2014-07-22 14:45:31 +0000 | [diff] [blame] | 166 | | regex |
| 167 | ; |
| 168 | |
Joachim Bingel | 82e4ca7 | 2014-10-27 11:03:38 +0000 | [diff] [blame] | 169 | multiword |
| 170 | : '"' WORD+ '"' |
| 171 | ; |
| 172 | |
Joachim Bingel | a3f51f7 | 2014-07-22 14:45:31 +0000 | [diff] [blame] | 173 | relation |
Joachim Bingel | 787836a | 2014-08-07 14:50:18 +0000 | [diff] [blame] | 174 | : (expr|exprGroup) booleanOp (expr|exprGroup|relation) |
Joachim Bingel | a3f51f7 | 2014-07-22 14:45:31 +0000 | [diff] [blame] | 175 | ; |
Joachim Bingel | dbbde77 | 2014-05-12 15:26:10 +0000 | [diff] [blame] | 176 | |
| 177 | exprGroup |
Joachim Bingel | a3f51f7 | 2014-07-22 14:45:31 +0000 | [diff] [blame] | 178 | : LRB (expr | exprGroup | relation) RRB |
Joachim Bingel | dbbde77 | 2014-05-12 15:26:10 +0000 | [diff] [blame] | 179 | ; |
| 180 | |
| 181 | start |
Joachim Bingel | 787836a | 2014-08-07 14:50:18 +0000 | [diff] [blame] | 182 | : expr EOF |
| 183 | | exprGroup EOF |
| 184 | | relation EOF |
Joachim Bingel | dbbde77 | 2014-05-12 15:26:10 +0000 | [diff] [blame] | 185 | ; |