Dekker1/tree-sitter-minizinc
Archived
1
0
Fork 0
Code

Add other literals

Browse Source
This commit is contained in:
Jip J. Dekker 2020-09-19 19:45:36 +10:00
parent 099bc7d9bf
commit 4fc5365d0a
6 changed files with 818 additions and 233 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

12
corpus/expressions.txt
View File

@ -1,12 +0,0 @@
================
Boolean Literals
================
boolean_id = false;
boolean_id = true;
---
(source_file
(assignment_item (identifier) (boolean_literal))
(assignment_item (identifier) (boolean_literal)))

66
corpus/literals.txt Normal file
View File

@ -0,0 +1,66 @@
==============
Absent Literal
==============
is_absent = <>;
---
(source_file
(assignment_item (identifier) (absent)))
================
Boolean Literals
================
boolean_id = false;
boolean_id = true;
---
(source_file
(assignment_item (identifier) (boolean_literal))
(assignment_item (identifier) (boolean_literal)))
===============
Number Literals
===============
simple_float = 3.65;
exponent_float = 4.5e10;
simple_exponent_float = 5E-10;
decimal_integer = 123;
hexadecimal_integer = 0xA23F;
octal_integer = 0o701;
binary_integer = 0b00110101;
---
(source_file
(assignment_item (identifier) (float_literal))
(assignment_item (identifier) (float_literal))
(assignment_item (identifier) (float_literal))
(assignment_item (identifier) (integer_literal))
(assignment_item (identifier) (integer_literal))
(assignment_item (identifier) (integer_literal))
(assignment_item (identifier) (integer_literal)))
===============
String Literals
===============
empty_string = "";
small_string = "test";
spaced_string = " Another test ";
escaped_string = "\t\n";
---
(source_file
(assignment_item (identifier) (string_literal))
(assignment_item (identifier) (string_literal))
(assignment_item (identifier) (string_literal))
(assignment_item (identifier) (string_literal
(escape_sequence) (escape_sequence))))

40
grammar.js
View File

@ -3,6 +3,8 @@ module.exports = grammar({
extras: $ => [/\s/, $.line_comment, $.block_comment],
word: $ => $.identifier,
rules: {
source_file: $ => seq(sepBy(';', $._items), optional(';')),
@ -25,15 +27,45 @@ module.exports = grammar({
_literal: $ => choice(
// TODO: absent,
$.absent,
$.boolean_literal,
// TODO: float_literal,
// TODO: integer_literal,
// TODO: string_literal,
$.float_literal,
$.integer_literal,
$.string_literal,
),
absent: $ => '<>',
boolean_literal: $ => choice('true', 'false'),
float_literal: $ => token(choice(
/\d+\.\d+/,
/\d+(\.\d+)?[Ee][+-]?\d+/,
// TODO: Hexadecimal floating point numbers
)),
integer_literal: $ => token(choice(
/[0-9]+/,
/0x[0-9a-fA-F]+/,
/0b[01]+/,
/0o[0-7]+/
)),
string_literal: $ => seq(
'"',
repeat(choice(
token.immediate(prec(1, /[^"\n\\]+/)),
$.escape_sequence
)),
'"'
),
escape_sequence: $ => token.immediate(seq(
'\\',
choice(
/[^xuU]/,
/\d{2,3}/,
/x[0-9a-fA-F]{2,}/,
/u[0-9a-fA-F]{4}/,
/U[0-9a-fA-F]{8}/
)
)),
identifier: $ => /[A-Za-z][A-Za-z0-9_]*/,

134
src/grammar.json
View File

@ -1,5 +1,6 @@
{
"name": "minizinc",
"word": "identifier",
"rules": {
"source_file": {
"type": "SEQ",
@ -101,12 +102,32 @@
"_literal": {
"type": "CHOICE",
"members": [
{
"type": "SYMBOL",
"name": "absent"
},
{
"type": "SYMBOL",
"name": "boolean_literal"
},
{
"type": "SYMBOL",
"name": "float_literal"
},
{
"type": "SYMBOL",
"name": "integer_literal"
},
{
"type": "SYMBOL",
"name": "string_literal"
}
]
},
"absent": {
"type": "STRING",
"value": "<>"
},
"boolean_literal": {
"type": "CHOICE",
"members": [
@ -120,6 +141,119 @@
}
]
},
"float_literal": {
"type": "TOKEN",
"content": {
"type": "CHOICE",
"members": [
{
"type": "PATTERN",
"value": "\\d+\\.\\d+"
},
{
"type": "PATTERN",
"value": "\\d+(\\.\\d+)?[Ee][+-]?\\d+"
}
]
}
},
"integer_literal": {
"type": "TOKEN",
"content": {
"type": "CHOICE",
"members": [
{
"type": "PATTERN",
"value": "[0-9]+"
},
{
"type": "PATTERN",
"value": "0x[0-9a-fA-F]+"
},
{
"type": "PATTERN",
"value": "0b[01]+"
},
{
"type": "PATTERN",
"value": "0o[0-7]+"
}
]
}
},
"string_literal": {
"type": "SEQ",
"members": [
{
"type": "STRING",
"value": "\""
},
{
"type": "REPEAT",
"content": {
"type": "CHOICE",
"members": [
{
"type": "IMMEDIATE_TOKEN",
"content": {
"type": "PREC",
"value": 1,
"content": {
"type": "PATTERN",
"value": "[^\"\\n\\\\]+"
}
}
},
{
"type": "SYMBOL",
"name": "escape_sequence"
}
]
}
},
{
"type": "STRING",
"value": "\""
}
]
},
"escape_sequence": {
"type": "IMMEDIATE_TOKEN",
"content": {
"type": "SEQ",
"members": [
{
"type": "STRING",
"value": "\\"
},
{
"type": "CHOICE",
"members": [
{
"type": "PATTERN",
"value": "[^xuU]"
},
{
"type": "PATTERN",
"value": "\\d{2,3}"
},
{
"type": "PATTERN",
"value": "x[0-9a-fA-F]{2,}"
},
{
"type": "PATTERN",
"value": "u[0-9a-fA-F]{4}"
},
{
"type": "PATTERN",
"value": "U[0-9a-fA-F]{8}"
}
]
}
]
}
},
"identifier": {
"type": "PATTERN",
"value": "[A-Za-z][A-Za-z0-9_]*"

51
src/node-types.json
View File

@ -7,13 +7,29 @@
"multiple": false,
"required": true,
"types": [
{
"type": "absent",
"named": true
},
{
"type": "boolean_literal",
"named": true
},
{
"type": "float_literal",
"named": true
},
{
"type": "identifier",
"named": true
},
{
"type": "integer_literal",
"named": true
},
{
"type": "string_literal",
"named": true
}
]
},
@ -49,6 +65,25 @@
]
}
},
{
"type": "string_literal",
"named": true,
"fields": {},
"children": {
"multiple": true,
"required": false,
"types": [
{
"type": "escape_sequence",
"named": true
}
]
}
},
{
"type": "\"",
"named": false
},
{
"type": ";",
"named": false
@ -57,14 +92,30 @@
"type": "=",
"named": false
},
{
"type": "absent",
"named": true
},
{
"type": "escape_sequence",
"named": true
},
{
"type": "false",
"named": false
},
{
"type": "float_literal",
"named": true
},
{
"type": "identifier",
"named": true
},
{
"type": "integer_literal",
"named": true
},
{
"type": "true",
"named": false

748
src/parser.c
View File

@ -6,39 +6,53 @@
#endif
#define LANGUAGE_VERSION 11
#define STATE_COUNT 15
#define STATE_COUNT 20
#define LARGE_STATE_COUNT 3
#define SYMBOL_COUNT 15
#define SYMBOL_COUNT 23
#define ALIAS_COUNT 0
#define TOKEN_COUNT 8
#define TOKEN_COUNT 14
#define EXTERNAL_TOKEN_COUNT 0
#define FIELD_COUNT 2
#define MAX_ALIAS_SEQUENCE_LENGTH 3
enum {
anon_sym_SEMI = 1,
anon_sym_EQ = 2,
anon_sym_true = 3,
anon_sym_false = 4,
sym_identifier = 5,
sym_line_comment = 6,
sym_block_comment = 7,
sym_source_file = 8,
sym__items = 9,
sym_assignment_item = 10,
sym__expression = 11,
sym__literal = 12,
sym_boolean_literal = 13,
aux_sym_source_file_repeat1 = 14,
sym_identifier = 1,
anon_sym_SEMI = 2,
anon_sym_EQ = 3,
sym_absent = 4,
anon_sym_true = 5,
anon_sym_false = 6,
sym_float_literal = 7,
sym_integer_literal = 8,
anon_sym_DQUOTE = 9,
aux_sym_string_literal_token1 = 10,
sym_escape_sequence = 11,
sym_line_comment = 12,
sym_block_comment = 13,
sym_source_file = 14,
sym__items = 15,
sym_assignment_item = 16,
sym__expression = 17,
sym__literal = 18,
sym_boolean_literal = 19,
sym_string_literal = 20,
aux_sym_source_file_repeat1 = 21,
aux_sym_string_literal_repeat1 = 22,
};
static const char *ts_symbol_names[] = {
[ts_builtin_sym_end] = "end",
[sym_identifier] = "identifier",
[anon_sym_SEMI] = ";",
[anon_sym_EQ] = "=",
[sym_absent] = "absent",
[anon_sym_true] = "true",
[anon_sym_false] = "false",
[sym_identifier] = "identifier",
[sym_float_literal] = "float_literal",
[sym_integer_literal] = "integer_literal",
[anon_sym_DQUOTE] = "\"",
[aux_sym_string_literal_token1] = "string_literal_token1",
[sym_escape_sequence] = "escape_sequence",
[sym_line_comment] = "line_comment",
[sym_block_comment] = "block_comment",
[sym_source_file] = "source_file",
@ -47,16 +61,24 @@ static const char *ts_symbol_names[] = {
[sym__expression] = "_expression",
[sym__literal] = "_literal",
[sym_boolean_literal] = "boolean_literal",
[sym_string_literal] = "string_literal",
[aux_sym_source_file_repeat1] = "source_file_repeat1",
[aux_sym_string_literal_repeat1] = "string_literal_repeat1",
};
static TSSymbol ts_symbol_map[] = {
[ts_builtin_sym_end] = ts_builtin_sym_end,
[sym_identifier] = sym_identifier,
[anon_sym_SEMI] = anon_sym_SEMI,
[anon_sym_EQ] = anon_sym_EQ,
[sym_absent] = sym_absent,
[anon_sym_true] = anon_sym_true,
[anon_sym_false] = anon_sym_false,
[sym_identifier] = sym_identifier,
[sym_float_literal] = sym_float_literal,
[sym_integer_literal] = sym_integer_literal,
[anon_sym_DQUOTE] = anon_sym_DQUOTE,
[aux_sym_string_literal_token1] = aux_sym_string_literal_token1,
[sym_escape_sequence] = sym_escape_sequence,
[sym_line_comment] = sym_line_comment,
[sym_block_comment] = sym_block_comment,
[sym_source_file] = sym_source_file,
@ -65,7 +87,9 @@ static TSSymbol ts_symbol_map[] = {
[sym__expression] = sym__expression,
[sym__literal] = sym__literal,
[sym_boolean_literal] = sym_boolean_literal,
[sym_string_literal] = sym_string_literal,
[aux_sym_source_file_repeat1] = aux_sym_source_file_repeat1,
[aux_sym_string_literal_repeat1] = aux_sym_string_literal_repeat1,
};
static const TSSymbolMetadata ts_symbol_metadata[] = {
@ -73,6 +97,10 @@ static const TSSymbolMetadata ts_symbol_metadata[] = {
.visible = false,
.named = true,
},
[sym_identifier] = {
.visible = true,
.named = true,
},
[anon_sym_SEMI] = {
.visible = true,
.named = false,
@ -81,6 +109,10 @@ static const TSSymbolMetadata ts_symbol_metadata[] = {
.visible = true,
.named = false,
},
[sym_absent] = {
.visible = true,
.named = true,
},
[anon_sym_true] = {
.visible = true,
.named = false,
@ -89,7 +121,23 @@ static const TSSymbolMetadata ts_symbol_metadata[] = {
.visible = true,
.named = false,
},
[sym_identifier] = {
[sym_float_literal] = {
.visible = true,
.named = true,
},
[sym_integer_literal] = {
.visible = true,
.named = true,
},
[anon_sym_DQUOTE] = {
.visible = true,
.named = false,
},
[aux_sym_string_literal_token1] = {
.visible = false,
.named = false,
},
[sym_escape_sequence] = {
.visible = true,
.named = true,
},
@ -125,10 +173,18 @@ static const TSSymbolMetadata ts_symbol_metadata[] = {
.visible = true,
.named = true,
},
[sym_string_literal] = {
.visible = true,
.named = true,
},
[aux_sym_source_file_repeat1] = {
.visible = false,
.named = false,
},
[aux_sym_string_literal_repeat1] = {
.visible = false,
.named = false,
},
};
enum {
@ -161,137 +217,315 @@ static bool ts_lex(TSLexer *lexer, TSStateId state) {
eof = lexer->eof(lexer);
switch (state) {
case 0:
if (eof) ADVANCE(5);
if (lookahead == '%') ADVANCE(18);
if (lookahead == '/') ADVANCE(1);
if (lookahead == ';') ADVANCE(6);
if (lookahead == '=') ADVANCE(7);
if (lookahead == 'f') ADVANCE(10);
if (lookahead == 't') ADVANCE(14);
if (eof) ADVANCE(23);
if (lookahead == '"') ADVANCE(34);
if (lookahead == '%') ADVANCE(44);
if (lookahead == '/') ADVANCE(3);
if (lookahead == '0') ADVANCE(29);
if (lookahead == ';') ADVANCE(24);
if (lookahead == '<') ADVANCE(6);
if (lookahead == '=') ADVANCE(25);
if (lookahead == '\\') ADVANCE(7);
if (lookahead == '\t' ||
lookahead == '\n' ||
lookahead == '\r' ||
lookahead == ' ') SKIP(22)
if (('1' <= lookahead && lookahead <= '9')) ADVANCE(30);
if (('A' <= lookahead && lookahead <= 'Z') ||
('a' <= lookahead && lookahead <= 'z')) ADVANCE(43);
END_STATE();
case 1:
if (lookahead == '\n') SKIP(2)
if (lookahead == '"') ADVANCE(34);
if (lookahead == '%') ADVANCE(39);
if (lookahead == '/') ADVANCE(38);
if (lookahead == '\\') ADVANCE(7);
if (lookahead == '\t' ||
lookahead == '\r' ||
lookahead == ' ') ADVANCE(35);
if (lookahead != 0) ADVANCE(39);
END_STATE();
case 2:
if (lookahead == '"') ADVANCE(34);
if (lookahead == '%') ADVANCE(44);
if (lookahead == '/') ADVANCE(3);
if (lookahead == '\t' ||
lookahead == '\n' ||
lookahead == '\r' ||
lookahead == ' ') SKIP(2)
END_STATE();
case 3:
if (lookahead == '*') ADVANCE(5);
END_STATE();
case 4:
if (lookahead == '*') ADVANCE(4);
if (lookahead == '/') ADVANCE(45);
if (lookahead != 0) ADVANCE(5);
END_STATE();
case 5:
if (lookahead == '*') ADVANCE(4);
if (lookahead != 0) ADVANCE(5);
END_STATE();
case 6:
if (lookahead == '>') ADVANCE(26);
END_STATE();
case 7:
if (lookahead == 'U') ADVANCE(21);
if (lookahead == 'u') ADVANCE(17);
if (lookahead == 'x') ADVANCE(15);
if (('0' <= lookahead && lookahead <= '9')) ADVANCE(42);
if (lookahead != 0) ADVANCE(40);
END_STATE();
case 8:
if (lookahead == '+' ||
lookahead == '-') ADVANCE(12);
if (('0' <= lookahead && lookahead <= '9')) ADVANCE(28);
END_STATE();
case 9:
if (lookahead == '0' ||
lookahead == '1') ADVANCE(31);
END_STATE();
case 10:
if (('0' <= lookahead && lookahead <= '7')) ADVANCE(32);
END_STATE();
case 11:
if (('0' <= lookahead && lookahead <= '9')) ADVANCE(27);
END_STATE();
case 12:
if (('0' <= lookahead && lookahead <= '9')) ADVANCE(28);
END_STATE();
case 13:
if (('0' <= lookahead && lookahead <= '9') ||
('A' <= lookahead && lookahead <= 'F') ||
('a' <= lookahead && lookahead <= 'f')) ADVANCE(40);
END_STATE();
case 14:
if (('0' <= lookahead && lookahead <= '9') ||
('A' <= lookahead && lookahead <= 'F') ||
('a' <= lookahead && lookahead <= 'f')) ADVANCE(33);
END_STATE();
case 15:
if (('0' <= lookahead && lookahead <= '9') ||
('A' <= lookahead && lookahead <= 'F') ||
('a' <= lookahead && lookahead <= 'f')) ADVANCE(13);
END_STATE();
case 16:
if (('0' <= lookahead && lookahead <= '9') ||
('A' <= lookahead && lookahead <= 'F') ||
('a' <= lookahead && lookahead <= 'f')) ADVANCE(15);
END_STATE();
case 17:
if (('0' <= lookahead && lookahead <= '9') ||
('A' <= lookahead && lookahead <= 'F') ||
('a' <= lookahead && lookahead <= 'f')) ADVANCE(16);
END_STATE();
case 18:
if (('0' <= lookahead && lookahead <= '9') ||
('A' <= lookahead && lookahead <= 'F') ||
('a' <= lookahead && lookahead <= 'f')) ADVANCE(17);
END_STATE();
case 19:
if (('0' <= lookahead && lookahead <= '9') ||
('A' <= lookahead && lookahead <= 'F') ||
('a' <= lookahead && lookahead <= 'f')) ADVANCE(18);
END_STATE();
case 20:
if (('0' <= lookahead && lookahead <= '9') ||
('A' <= lookahead && lookahead <= 'F') ||
('a' <= lookahead && lookahead <= 'f')) ADVANCE(19);
END_STATE();
case 21:
if (('0' <= lookahead && lookahead <= '9') ||
('A' <= lookahead && lookahead <= 'F') ||
('a' <= lookahead && lookahead <= 'f')) ADVANCE(20);
END_STATE();
case 22:
if (eof) ADVANCE(23);
if (lookahead == '"') ADVANCE(34);
if (lookahead == '%') ADVANCE(44);
if (lookahead == '/') ADVANCE(3);
if (lookahead == '0') ADVANCE(29);
if (lookahead == ';') ADVANCE(24);
if (lookahead == '<') ADVANCE(6);
if (lookahead == '=') ADVANCE(25);
if (lookahead == '\t' ||
lookahead == '\n' ||
lookahead == '\r' ||
lookahead == ' ') SKIP(22)
if (('1' <= lookahead && lookahead <= '9')) ADVANCE(30);
if (('A' <= lookahead && lookahead <= 'Z') ||
('a' <= lookahead && lookahead <= 'z')) ADVANCE(43);
END_STATE();
case 23:
ACCEPT_TOKEN(ts_builtin_sym_end);
END_STATE();
case 24:
ACCEPT_TOKEN(anon_sym_SEMI);
END_STATE();
case 25:
ACCEPT_TOKEN(anon_sym_EQ);
END_STATE();
case 26:
ACCEPT_TOKEN(sym_absent);
END_STATE();
case 27:
ACCEPT_TOKEN(sym_float_literal);
if (lookahead == 'E' ||
lookahead == 'e') ADVANCE(8);
if (('0' <= lookahead && lookahead <= '9')) ADVANCE(27);
END_STATE();
case 28:
ACCEPT_TOKEN(sym_float_literal);
if (('0' <= lookahead && lookahead <= '9')) ADVANCE(28);
END_STATE();
case 29:
ACCEPT_TOKEN(sym_integer_literal);
if (lookahead == '.') ADVANCE(11);
if (lookahead == 'b') ADVANCE(9);
if (lookahead == 'o') ADVANCE(10);
if (lookahead == 'x') ADVANCE(14);
if (lookahead == 'E' ||
lookahead == 'e') ADVANCE(8);
if (('0' <= lookahead && lookahead <= '9')) ADVANCE(30);
END_STATE();
case 30:
ACCEPT_TOKEN(sym_integer_literal);
if (lookahead == '.') ADVANCE(11);
if (lookahead == 'E' ||
lookahead == 'e') ADVANCE(8);
if (('0' <= lookahead && lookahead <= '9')) ADVANCE(30);
END_STATE();
case 31:
ACCEPT_TOKEN(sym_integer_literal);
if (lookahead == '0' ||
lookahead == '1') ADVANCE(31);
END_STATE();
case 32:
ACCEPT_TOKEN(sym_integer_literal);
if (('0' <= lookahead && lookahead <= '7')) ADVANCE(32);
END_STATE();
case 33:
ACCEPT_TOKEN(sym_integer_literal);
if (('0' <= lookahead && lookahead <= '9') ||
('A' <= lookahead && lookahead <= 'F') ||
('a' <= lookahead && lookahead <= 'f')) ADVANCE(33);
END_STATE();
case 34:
ACCEPT_TOKEN(anon_sym_DQUOTE);
END_STATE();
case 35:
ACCEPT_TOKEN(aux_sym_string_literal_token1);
if (lookahead == '%') ADVANCE(39);
if (lookahead == '/') ADVANCE(38);
if (lookahead == '\t' ||
lookahead == '\r' ||
lookahead == ' ') ADVANCE(35);
if (lookahead != 0 &&
lookahead != '\n' &&
lookahead != '"' &&
lookahead != '\\') ADVANCE(39);
END_STATE();
case 36:
ACCEPT_TOKEN(aux_sym_string_literal_token1);
if (lookahead == '*') ADVANCE(36);
if (lookahead == '/') ADVANCE(39);
if (lookahead != 0 &&
lookahead != '\n' &&
lookahead != '"' &&
lookahead != '\\') ADVANCE(37);
END_STATE();
case 37:
ACCEPT_TOKEN(aux_sym_string_literal_token1);
if (lookahead == '*') ADVANCE(36);
if (lookahead != 0 &&
lookahead != '\n' &&
lookahead != '"' &&
lookahead != '\\') ADVANCE(37);
END_STATE();
case 38:
ACCEPT_TOKEN(aux_sym_string_literal_token1);
if (lookahead == '*') ADVANCE(37);
if (lookahead != 0 &&
lookahead != '\n' &&
lookahead != '"' &&
lookahead != '\\') ADVANCE(39);
END_STATE();
case 39:
ACCEPT_TOKEN(aux_sym_string_literal_token1);
if (lookahead != 0 &&
lookahead != '\n' &&
lookahead != '"' &&
lookahead != '\\') ADVANCE(39);
END_STATE();
case 40:
ACCEPT_TOKEN(sym_escape_sequence);
END_STATE();
case 41:
ACCEPT_TOKEN(sym_escape_sequence);
if (('0' <= lookahead && lookahead <= '9')) ADVANCE(40);
END_STATE();
case 42:
ACCEPT_TOKEN(sym_escape_sequence);
if (('0' <= lookahead && lookahead <= '9')) ADVANCE(41);
END_STATE();
case 43:
ACCEPT_TOKEN(sym_identifier);
if (('0' <= lookahead && lookahead <= '9') ||
('A' <= lookahead && lookahead <= 'Z') ||
lookahead == '_' ||
('a' <= lookahead && lookahead <= 'z')) ADVANCE(43);
END_STATE();
case 44:
ACCEPT_TOKEN(sym_line_comment);
if (lookahead != 0 &&
lookahead != '\n') ADVANCE(44);
END_STATE();
case 45:
ACCEPT_TOKEN(sym_block_comment);
END_STATE();
default:
return false;
}
}
static bool ts_lex_keywords(TSLexer *lexer, TSStateId state) {
START_LEXER();
eof = lexer->eof(lexer);
switch (state) {
case 0:
if (lookahead == 'f') ADVANCE(1);
if (lookahead == 't') ADVANCE(2);
if (lookahead == '\t' ||
lookahead == '\n' ||
lookahead == '\r' ||
lookahead == ' ') SKIP(0)
if (('A' <= lookahead && lookahead <= 'Z') ||
('a' <= lookahead && lookahead <= 'z')) ADVANCE(17);
END_STATE();
case 1:
if (lookahead == '*') ADVANCE(3);
if (lookahead == 'a') ADVANCE(3);
END_STATE();
case 2:
if (lookahead == '*') ADVANCE(2);
if (lookahead == '/') ADVANCE(19);
if (lookahead != 0) ADVANCE(3);
if (lookahead == 'r') ADVANCE(4);
END_STATE();
case 3:
if (lookahead == '*') ADVANCE(2);
if (lookahead != 0) ADVANCE(3);
if (lookahead == 'l') ADVANCE(5);
END_STATE();
case 4:
if (eof) ADVANCE(5);
if (lookahead == '%') ADVANCE(18);
if (lookahead == '/') ADVANCE(1);
if (lookahead == ';') ADVANCE(6);
if (lookahead == '\t' ||
lookahead == '\n' ||
lookahead == '\r' ||
lookahead == ' ') SKIP(4)
if (('A' <= lookahead && lookahead <= 'Z') ||
('a' <= lookahead && lookahead <= 'z')) ADVANCE(17);
if (lookahead == 'u') ADVANCE(6);
END_STATE();
case 5:
ACCEPT_TOKEN(ts_builtin_sym_end);
if (lookahead == 's') ADVANCE(7);
END_STATE();
case 6:
ACCEPT_TOKEN(anon_sym_SEMI);
if (lookahead == 'e') ADVANCE(8);
END_STATE();
case 7:
ACCEPT_TOKEN(anon_sym_EQ);
if (lookahead == 'e') ADVANCE(9);
END_STATE();
case 8:
ACCEPT_TOKEN(anon_sym_true);
if (('0' <= lookahead && lookahead <= '9') ||
('A' <= lookahead && lookahead <= 'Z') ||
lookahead == '_' ||
('a' <= lookahead && lookahead <= 'z')) ADVANCE(17);
END_STATE();
case 9:
ACCEPT_TOKEN(anon_sym_false);
if (('0' <= lookahead && lookahead <= '9') ||
('A' <= lookahead && lookahead <= 'Z') ||
lookahead == '_' ||
('a' <= lookahead && lookahead <= 'z')) ADVANCE(17);
END_STATE();
case 10:
ACCEPT_TOKEN(sym_identifier);
if (lookahead == 'a') ADVANCE(13);
if (('0' <= lookahead && lookahead <= '9') ||
('A' <= lookahead && lookahead <= 'Z') ||
lookahead == '_' ||
('b' <= lookahead && lookahead <= 'z')) ADVANCE(17);
END_STATE();
case 11:
ACCEPT_TOKEN(sym_identifier);
if (lookahead == 'e') ADVANCE(8);
if (('0' <= lookahead && lookahead <= '9') ||
('A' <= lookahead && lookahead <= 'Z') ||
lookahead == '_' ||
('a' <= lookahead && lookahead <= 'z')) ADVANCE(17);
END_STATE();
case 12:
ACCEPT_TOKEN(sym_identifier);
if (lookahead == 'e') ADVANCE(9);
if (('0' <= lookahead && lookahead <= '9') ||
('A' <= lookahead && lookahead <= 'Z') ||
lookahead == '_' ||
('a' <= lookahead && lookahead <= 'z')) ADVANCE(17);
END_STATE();
case 13:
ACCEPT_TOKEN(sym_identifier);
if (lookahead == 'l') ADVANCE(15);
if (('0' <= lookahead && lookahead <= '9') ||
('A' <= lookahead && lookahead <= 'Z') ||
lookahead == '_' ||
('a' <= lookahead && lookahead <= 'z')) ADVANCE(17);
END_STATE();
case 14:
ACCEPT_TOKEN(sym_identifier);
if (lookahead == 'r') ADVANCE(16);
if (('0' <= lookahead && lookahead <= '9') ||
('A' <= lookahead && lookahead <= 'Z') ||
lookahead == '_' ||
('a' <= lookahead && lookahead <= 'z')) ADVANCE(17);
END_STATE();
case 15:
ACCEPT_TOKEN(sym_identifier);
if (lookahead == 's') ADVANCE(12);
if (('0' <= lookahead && lookahead <= '9') ||
('A' <= lookahead && lookahead <= 'Z') ||
lookahead == '_' ||
('a' <= lookahead && lookahead <= 'z')) ADVANCE(17);
END_STATE();
case 16:
ACCEPT_TOKEN(sym_identifier);
if (lookahead == 'u') ADVANCE(11);
if (('0' <= lookahead && lookahead <= '9') ||
('A' <= lookahead && lookahead <= 'Z') ||
lookahead == '_' ||
('a' <= lookahead && lookahead <= 'z')) ADVANCE(17);
END_STATE();
case 17:
ACCEPT_TOKEN(sym_identifier);
if (('0' <= lookahead && lookahead <= '9') ||
('A' <= lookahead && lookahead <= 'Z') ||
lookahead == '_' ||
('a' <= lookahead && lookahead <= 'z')) ADVANCE(17);
END_STATE();
case 18:
ACCEPT_TOKEN(sym_line_comment);
if (lookahead != 0 &&
lookahead != '\n') ADVANCE(18);
END_STATE();
case 19:
ACCEPT_TOKEN(sym_block_comment);
END_STATE();
default:
return false;
@ -300,50 +534,65 @@ static bool ts_lex(TSLexer *lexer, TSStateId state) {
static TSLexMode ts_lex_modes[STATE_COUNT] = {
[0] = {.lex_state = 0},
[1] = {.lex_state = 4},
[1] = {.lex_state = 0},
[2] = {.lex_state = 0},
[3] = {.lex_state = 4},
[4] = {.lex_state = 4},
[3] = {.lex_state = 0},
[4] = {.lex_state = 1},
[5] = {.lex_state = 0},
[6] = {.lex_state = 0},
[7] = {.lex_state = 0},
[8] = {.lex_state = 4},
[6] = {.lex_state = 1},
[7] = {.lex_state = 1},
[8] = {.lex_state = 0},
[9] = {.lex_state = 0},
[10] = {.lex_state = 0},
[11] = {.lex_state = 0},
[12] = {.lex_state = 0},
[13] = {.lex_state = 0},
[14] = {.lex_state = 0},
[15] = {.lex_state = 0},
[16] = {.lex_state = 0},
[17] = {.lex_state = 0},
[18] = {.lex_state = 0},
[19] = {.lex_state = 0},
};
static uint16_t ts_parse_table[LARGE_STATE_COUNT][SYMBOL_COUNT] = {
[0] = {
[ts_builtin_sym_end] = ACTIONS(1),
[sym_identifier] = ACTIONS(1),
[anon_sym_SEMI] = ACTIONS(1),
[anon_sym_EQ] = ACTIONS(1),
[sym_absent] = ACTIONS(1),
[anon_sym_true] = ACTIONS(1),
[anon_sym_false] = ACTIONS(1),
[sym_identifier] = ACTIONS(1),
[sym_float_literal] = ACTIONS(1),
[sym_integer_literal] = ACTIONS(1),
[anon_sym_DQUOTE] = ACTIONS(1),
[sym_escape_sequence] = ACTIONS(1),
[sym_line_comment] = ACTIONS(3),
[sym_block_comment] = ACTIONS(3),
},
[1] = {
[sym_source_file] = STATE(14),
[sym__items] = STATE(5),
[sym_assignment_item] = STATE(5),
[sym_source_file] = STATE(19),
[sym__items] = STATE(8),
[sym_assignment_item] = STATE(8),
[ts_builtin_sym_end] = ACTIONS(5),
[anon_sym_SEMI] = ACTIONS(7),
[sym_identifier] = ACTIONS(9),
[sym_identifier] = ACTIONS(7),
[anon_sym_SEMI] = ACTIONS(9),
[sym_line_comment] = ACTIONS(3),
[sym_block_comment] = ACTIONS(3),
},
[2] = {
[sym__expression] = STATE(10),
[sym__literal] = STATE(10),
[sym_boolean_literal] = STATE(10),
[anon_sym_true] = ACTIONS(11),
[anon_sym_false] = ACTIONS(11),
[sym_identifier] = ACTIONS(13),
[sym__expression] = STATE(13),
[sym__literal] = STATE(13),
[sym_boolean_literal] = STATE(13),
[sym_string_literal] = STATE(13),
[sym_identifier] = ACTIONS(11),
[sym_absent] = ACTIONS(13),
[anon_sym_true] = ACTIONS(15),
[anon_sym_false] = ACTIONS(15),
[sym_float_literal] = ACTIONS(13),
[sym_integer_literal] = ACTIONS(11),
[anon_sym_DQUOTE] = ACTIONS(17),
[sym_line_comment] = ACTIONS(3),
[sym_block_comment] = ACTIONS(3),
},
@ -351,101 +600,148 @@ static uint16_t ts_parse_table[LARGE_STATE_COUNT][SYMBOL_COUNT] = {
static uint16_t ts_small_parse_table[] = {
[0] = 4,
ACTIONS(9), 1,
ACTIONS(7), 1,
sym_identifier,
ACTIONS(15), 1,
ACTIONS(19), 1,
ts_builtin_sym_end,
ACTIONS(3), 2,
sym_line_comment,
sym_block_comment,
STATE(11), 2,
STATE(14), 2,
sym__items,
sym_assignment_item,
[15] = 4,
ACTIONS(9), 1,
sym_identifier,
ACTIONS(17), 1,
ts_builtin_sym_end,
ACTIONS(3), 2,
sym_line_comment,
sym_block_comment,
STATE(11), 2,
sym__items,
sym_assignment_item,
[30] = 4,
ACTIONS(19), 1,
ts_builtin_sym_end,
ACTIONS(21), 1,
anon_sym_SEMI,
anon_sym_DQUOTE,
STATE(6), 1,
aux_sym_source_file_repeat1,
ACTIONS(3), 2,
aux_sym_string_literal_repeat1,
ACTIONS(23), 2,
aux_sym_string_literal_token1,
sym_escape_sequence,
ACTIONS(25), 2,
sym_line_comment,
sym_block_comment,
[44] = 4,
ACTIONS(15), 1,
ts_builtin_sym_end,
ACTIONS(23), 1,
anon_sym_SEMI,
STATE(7), 1,
aux_sym_source_file_repeat1,
ACTIONS(3), 2,
sym_line_comment,
sym_block_comment,
[58] = 4,
ACTIONS(25), 1,
ts_builtin_sym_end,
[30] = 4,
ACTIONS(7), 1,
sym_identifier,
ACTIONS(27), 1,
anon_sym_SEMI,
ts_builtin_sym_end,
ACTIONS(3), 2,
sym_line_comment,
sym_block_comment,
STATE(14), 2,
sym__items,
sym_assignment_item,
[45] = 4,
ACTIONS(29), 1,
anon_sym_DQUOTE,
STATE(7), 1,
aux_sym_string_literal_repeat1,
ACTIONS(25), 2,
sym_line_comment,
sym_block_comment,
ACTIONS(31), 2,
aux_sym_string_literal_token1,
sym_escape_sequence,
[60] = 4,
ACTIONS(33), 1,
anon_sym_DQUOTE,
STATE(7), 1,
aux_sym_string_literal_repeat1,
ACTIONS(25), 2,
sym_line_comment,
sym_block_comment,
ACTIONS(35), 2,
aux_sym_string_literal_token1,
sym_escape_sequence,
[75] = 4,
ACTIONS(38), 1,
ts_builtin_sym_end,
ACTIONS(40), 1,
anon_sym_SEMI,
STATE(9), 1,
aux_sym_source_file_repeat1,
ACTIONS(3), 2,
sym_line_comment,
sym_block_comment,
[72] = 3,
ACTIONS(9), 1,
[89] = 4,
ACTIONS(19), 1,
ts_builtin_sym_end,
ACTIONS(42), 1,
anon_sym_SEMI,
STATE(10), 1,
aux_sym_source_file_repeat1,
ACTIONS(3), 2,
sym_line_comment,
sym_block_comment,
[103] = 4,
ACTIONS(44), 1,
ts_builtin_sym_end,
ACTIONS(46), 1,
anon_sym_SEMI,
STATE(10), 1,
aux_sym_source_file_repeat1,
ACTIONS(3), 2,
sym_line_comment,
sym_block_comment,
[117] = 3,
ACTIONS(7), 1,
sym_identifier,
ACTIONS(3), 2,
sym_line_comment,
sym_block_comment,
STATE(11), 2,
STATE(14), 2,
sym__items,
sym_assignment_item,
[84] = 2,
[129] = 2,
ACTIONS(3), 2,
sym_line_comment,
sym_block_comment,
ACTIONS(30), 2,
ACTIONS(49), 2,
ts_builtin_sym_end,
anon_sym_SEMI,
[93] = 2,
[138] = 2,
ACTIONS(3), 2,
sym_line_comment,
sym_block_comment,
ACTIONS(32), 2,
ACTIONS(51), 2,
ts_builtin_sym_end,
anon_sym_SEMI,
[102] = 2,
[147] = 2,
ACTIONS(3), 2,
sym_line_comment,
sym_block_comment,
ACTIONS(25), 2,
ACTIONS(44), 2,
ts_builtin_sym_end,
anon_sym_SEMI,
[111] = 2,
ACTIONS(19), 1,
[156] = 2,
ACTIONS(3), 2,
sym_line_comment,
sym_block_comment,
ACTIONS(53), 2,
ts_builtin_sym_end,
anon_sym_SEMI,
[165] = 2,
ACTIONS(3), 2,
sym_line_comment,
sym_block_comment,
ACTIONS(55), 2,
ts_builtin_sym_end,
anon_sym_SEMI,
[174] = 2,
ACTIONS(38), 1,
ts_builtin_sym_end,
ACTIONS(3), 2,
sym_line_comment,
sym_block_comment,
[119] = 2,
ACTIONS(34), 1,
[182] = 2,
ACTIONS(57), 1,
anon_sym_EQ,
ACTIONS(3), 2,
sym_line_comment,
sym_block_comment,
[127] = 2,
ACTIONS(36), 1,
[190] = 2,
ACTIONS(59), 1,
ts_builtin_sym_end,
ACTIONS(3), 2,
sym_line_comment,
@ -456,15 +752,20 @@ static uint32_t ts_small_parse_table_map[] = {
[SMALL_STATE(3)] = 0,
[SMALL_STATE(4)] = 15,
[SMALL_STATE(5)] = 30,
[SMALL_STATE(6)] = 44,
[SMALL_STATE(7)] = 58,
[SMALL_STATE(8)] = 72,
[SMALL_STATE(9)] = 84,
[SMALL_STATE(10)] = 93,
[SMALL_STATE(11)] = 102,
[SMALL_STATE(12)] = 111,
[SMALL_STATE(13)] = 119,
[SMALL_STATE(14)] = 127,
[SMALL_STATE(6)] = 45,
[SMALL_STATE(7)] = 60,
[SMALL_STATE(8)] = 75,
[SMALL_STATE(9)] = 89,
[SMALL_STATE(10)] = 103,
[SMALL_STATE(11)] = 117,
[SMALL_STATE(12)] = 129,
[SMALL_STATE(13)] = 138,
[SMALL_STATE(14)] = 147,
[SMALL_STATE(15)] = 156,
[SMALL_STATE(16)] = 165,
[SMALL_STATE(17)] = 174,
[SMALL_STATE(18)] = 182,
[SMALL_STATE(19)] = 190,
};
static TSParseActionEntry ts_parse_actions[] = {
@ -472,21 +773,32 @@ static TSParseActionEntry ts_parse_actions[] = {
[1] = {.entry = {.count = 1, .reusable = false}}, RECOVER(),
[3] = {.entry = {.count = 1, .reusable = true}}, SHIFT_EXTRA(),
[5] = {.entry = {.count = 1, .reusable = true}}, REDUCE(sym_source_file, 0),
[7] = {.entry = {.count = 1, .reusable = true}}, SHIFT(12),
[9] = {.entry = {.count = 1, .reusable = true}}, SHIFT(13),
[11] = {.entry = {.count = 1, .reusable = false}}, SHIFT(9),
[13] = {.entry = {.count = 1, .reusable = false}}, SHIFT(10),
[15] = {.entry = {.count = 1, .reusable = true}}, REDUCE(sym_source_file, 2),
[17] = {.entry = {.count = 1, .reusable = true}}, REDUCE(sym_source_file, 3),
[19] = {.entry = {.count = 1, .reusable = true}}, REDUCE(sym_source_file, 1),
[21] = {.entry = {.count = 1, .reusable = true}}, SHIFT(3),
[23] = {.entry = {.count = 1, .reusable = true}}, SHIFT(4),
[25] = {.entry = {.count = 1, .reusable = true}}, REDUCE(aux_sym_source_file_repeat1, 2),
[27] = {.entry = {.count = 2, .reusable = true}}, REDUCE(aux_sym_source_file_repeat1, 2), SHIFT_REPEAT(8),
[30] = {.entry = {.count = 1, .reusable = true}}, REDUCE(sym_boolean_literal, 1),
[32] = {.entry = {.count = 1, .reusable = true}}, REDUCE(sym_assignment_item, 3, .production_id = 1),
[34] = {.entry = {.count = 1, .reusable = true}}, SHIFT(2),
[36] = {.entry = {.count = 1, .reusable = true}}, ACCEPT_INPUT(),
[7] = {.entry = {.count = 1, .reusable = true}}, SHIFT(18),
[9] = {.entry = {.count = 1, .reusable = true}}, SHIFT(17),
[11] = {.entry = {.count = 1, .reusable = false}}, SHIFT(13),
[13] = {.entry = {.count = 1, .reusable = true}}, SHIFT(13),
[15] = {.entry = {.count = 1, .reusable = false}}, SHIFT(12),
[17] = {.entry = {.count = 1, .reusable = true}}, SHIFT(4),
[19] = {.entry = {.count = 1, .reusable = true}}, REDUCE(sym_source_file, 2),
[21] = {.entry = {.count = 1, .reusable = false}}, SHIFT(15),
[23] = {.entry = {.count = 1, .reusable = true}}, SHIFT(6),
[25] = {.entry = {.count = 1, .reusable = false}}, SHIFT_EXTRA(),
[27] = {.entry = {.count = 1, .reusable = true}}, REDUCE(sym_source_file, 3),
[29] = {.entry = {.count = 1, .reusable = false}}, SHIFT(16),
[31] = {.entry = {.count = 1, .reusable = true}}, SHIFT(7),
[33] = {.entry = {.count = 1, .reusable = false}}, REDUCE(aux_sym_string_literal_repeat1, 2),
[35] = {.entry = {.count = 2, .reusable = true}}, REDUCE(aux_sym_string_literal_repeat1, 2), SHIFT_REPEAT(7),
[38] = {.entry = {.count = 1, .reusable = true}}, REDUCE(sym_source_file, 1),
[40] = {.entry = {.count = 1, .reusable = true}}, SHIFT(3),
[42] = {.entry = {.count = 1, .reusable = true}}, SHIFT(5),
[44] = {.entry = {.count = 1, .reusable = true}}, REDUCE(aux_sym_source_file_repeat1, 2),
[46] = {.entry = {.count = 2, .reusable = true}}, REDUCE(aux_sym_source_file_repeat1, 2), SHIFT_REPEAT(11),
[49] = {.entry = {.count = 1, .reusable = true}}, REDUCE(sym_boolean_literal, 1),
[51] = {.entry = {.count = 1, .reusable = true}}, REDUCE(sym_assignment_item, 3, .production_id = 1),
[53] = {.entry = {.count = 1, .reusable = true}}, REDUCE(sym_string_literal, 2),
[55] = {.entry = {.count = 1, .reusable = true}}, REDUCE(sym_string_literal, 3),
[57] = {.entry = {.count = 1, .reusable = true}}, SHIFT(2),
[59] = {.entry = {.count = 1, .reusable = true}}, ACCEPT_INPUT(),
};
#ifdef __cplusplus
@ -518,6 +830,8 @@ extern const TSLanguage *tree_sitter_minizinc(void) {
.field_map_entries = (const TSFieldMapEntry *)ts_field_map_entries,
.max_alias_sequence_length = MAX_ALIAS_SEQUENCE_LENGTH,
.lex_fn = ts_lex,
.keyword_lex_fn = ts_lex_keywords,
.keyword_capture_token = sym_identifier,
.external_token_count = EXTERNAL_TOKEN_COUNT,
};
return &language;