nandi.uk
Fork of daniellemaywood.uk/gleam — Wasm codegen work
allow negative numbers in different bases
author
Giacomo Cavalieri
committer
Louis Pilfold
date
(Jul 15, 2026, 10:58 AM +0100)
commit
dfe7dd95
dfe7dd95d04cc49299923b3657b8e5e7efaf17f3
parent
ea9c598a
ea9c598a21401c439d284c90f5f58014f4207568
change-id
kzuwruox
kzuwruoxxlyrkuvlxvzutsqsknlloqwr
+323
-30
+4
CHANGELOG.md
+4
CHANGELOG.md
···
555
555
when finding the Gleam files of a package.
556
556
([Giacomo Cavalieri](https://github.com/giacomocavalieri))
557
557
558
+
- Fixed a bug where the compiler would not be able to correctly parse negative
559
+
numbers written in binary, octal, or hexadecimal base.
560
+
([Giacomo Cavalieri](https://github.com/giacomocavalieri))
561
+
558
562
- The formatter now properly formats binary operations in bit array size
559
563
segments.
560
564
([Andrey Kozhev](https://github.com/ankddev))
+61
-30
compiler-core/src/parse/lexer.rs
+61
-30
compiler-core/src/parse/lexer.rs
···
8
8
use crate::parse::error::{LexicalError, LexicalErrorType};
9
9
use crate::parse::token::Token;
10
10
use std::char;
11
+
use std::ops::Neg;
11
12
12
13
use super::error::InvalidUnicodeEscapeError;
13
14
···
879
880
880
881
fn lex_number(&mut self) -> LexResult {
881
882
let start_pos = self.get_pos();
883
+
884
+
// We call this function after making sure that what comes next starts
885
+
// with what seems to be a valid number. If we see that it starts with
886
+
// `-` we consume the token and record that the number is negative.
887
+
let is_negative = if self.chr0 == Some('-') {
888
+
let _ = self.next_char();
889
+
true
890
+
} else {
891
+
false
892
+
};
893
+
882
894
let num = if self.chr0 == Some('0') {
883
-
if self.chr1 == Some('x') || self.chr1 == Some('X') {
884
-
// Hex!
885
-
let _ = self.next_char();
886
-
let _ = self.next_char();
887
-
self.lex_number_radix(start_pos, 16, "0x")?
888
-
} else if self.chr1 == Some('o') || self.chr1 == Some('O') {
889
-
// Octal!
890
-
let _ = self.next_char();
891
-
let _ = self.next_char();
892
-
self.lex_number_radix(start_pos, 8, "0o")?
893
-
} else if self.chr1 == Some('b') || self.chr1 == Some('B') {
894
-
// Binary!
895
-
let _ = self.next_char();
896
-
let _ = self.next_char();
897
-
self.lex_number_radix(start_pos, 2, "0b")?
898
-
} else {
899
-
self.lex_decimal_number()?
895
+
match self.chr1 {
896
+
Some('x' | 'X') => {
897
+
// Hex!
898
+
let _ = self.next_char();
899
+
let _ = self.next_char();
900
+
self.lex_number_radix(start_pos, 16, is_negative, "0x")?
901
+
}
902
+
Some('o' | 'O') => {
903
+
// Octal!
904
+
let _ = self.next_char();
905
+
let _ = self.next_char();
906
+
self.lex_number_radix(start_pos, 8, is_negative, "0o")?
907
+
}
908
+
Some('b' | 'B') => {
909
+
// Binary!
910
+
let _ = self.next_char();
911
+
let _ = self.next_char();
912
+
self.lex_number_radix(start_pos, 2, is_negative, "0b")?
913
+
}
914
+
_ => self.lex_decimal_number(start_pos, is_negative)?,
900
915
}
901
916
} else {
902
-
self.lex_decimal_number()?
917
+
self.lex_decimal_number(start_pos, is_negative)?
903
918
};
904
919
905
920
if Some('_') == self.chr0 {
···
917
932
}
918
933
919
934
// Lex a hex/octal/decimal/binary number without a decimal point.
920
-
fn lex_number_radix(&mut self, start_pos: u32, radix: u32, prefix: &str) -> LexResult {
935
+
fn lex_number_radix(
936
+
&mut self,
937
+
start_pos: u32,
938
+
radix: u32,
939
+
is_negative: bool,
940
+
prefix: &str,
941
+
) -> LexResult {
921
942
let num = self.radix_run(radix);
922
943
if num.is_empty() {
923
944
let location = self.get_pos() - 1;
···
941
962
let value = format!("{prefix}{num}");
942
963
let int_value = super::parse_int_value(&value).expect("int value to parse as bigint");
943
964
let end_pos = self.get_pos();
965
+
966
+
let (value, int_value) = if is_negative {
967
+
(format!("-{value}"), int_value.neg())
968
+
} else {
969
+
(value, int_value)
970
+
};
971
+
944
972
Ok((
945
973
start_pos,
946
974
Token::Int {
···
954
982
955
983
// Lex a normal number, that is, no octal, hex or binary number.
956
984
// This function cannot be reached without the head of the stream being either 0-9 or '-', 0-9
957
-
fn lex_decimal_number(&mut self) -> LexResult {
958
-
self.lex_decimal_or_int_number(true)
985
+
fn lex_decimal_number(&mut self, start_pos: u32, is_negative: bool) -> LexResult {
986
+
self.lex_decimal_or_int_number(start_pos, is_negative, true)
959
987
}
960
988
961
-
fn lex_int_number(&mut self) -> LexResult {
962
-
self.lex_decimal_or_int_number(false)
989
+
fn lex_int_number(&mut self, start_pos: u32, is_negative: bool) -> LexResult {
990
+
self.lex_decimal_or_int_number(start_pos, is_negative, false)
963
991
}
964
992
965
-
fn lex_decimal_or_int_number(&mut self, can_lex_decimal: bool) -> LexResult {
966
-
let start_pos = self.get_pos();
993
+
fn lex_decimal_or_int_number(
994
+
&mut self,
995
+
start_pos: u32,
996
+
is_negative: bool,
997
+
can_lex_decimal: bool,
998
+
) -> LexResult {
967
999
let mut value = String::new();
968
-
// consume negative sign
969
-
if self.chr0 == Some('-') {
970
-
value.push(self.next_char().expect("lex_normal_number negative"));
971
-
}
1000
+
if is_negative {
1001
+
value.push('-')
1002
+
};
972
1003
// consume first run of digits
973
1004
value.push_str(&self.radix_run(10));
974
1005
···
1025
1056
// It can be nested like: `tuple.1.2.3.4`
1026
1057
loop {
1027
1058
if matches!(self.chr0, Some('0'..='9')) {
1028
-
let number = self.lex_int_number()?;
1059
+
let number = self.lex_int_number(self.get_pos(), false)?;
1029
1060
self.emit(number);
1030
1061
} else {
1031
1062
break;
+81
compiler-core/src/parse/snapshots/gleam_core__parse__tests__parse_negative_binary_number.snap
+81
compiler-core/src/parse/snapshots/gleam_core__parse__tests__parse_negative_binary_number.snap
···
1
+
---
2
+
source: compiler-core/src/parse/tests.rs
3
+
expression: "pub fn main() { -0b101 }"
4
+
---
5
+
Parsed {
6
+
module: Module {
7
+
name: "",
8
+
documentation: [],
9
+
type_info: (),
10
+
definitions: [
11
+
TargetedDefinition {
12
+
definition: Function(
13
+
Function {
14
+
location: SrcSpan {
15
+
start: 0,
16
+
end: 13,
17
+
},
18
+
body_start: Some(
19
+
14,
20
+
),
21
+
end_position: 24,
22
+
name: Some(
23
+
(
24
+
SrcSpan {
25
+
start: 7,
26
+
end: 11,
27
+
},
28
+
"main",
29
+
),
30
+
),
31
+
arguments: [],
32
+
body: [
33
+
Expression(
34
+
Int {
35
+
location: SrcSpan {
36
+
start: 16,
37
+
end: 22,
38
+
},
39
+
value: "-0b101",
40
+
int_value: -5,
41
+
},
42
+
),
43
+
],
44
+
publicity: Public,
45
+
deprecation: NotDeprecated,
46
+
return_annotation: None,
47
+
return_type: (),
48
+
documentation: None,
49
+
external_erlang: None,
50
+
external_javascript: None,
51
+
implementations: Implementations {
52
+
gleam: true,
53
+
can_run_on_erlang: true,
54
+
can_run_on_javascript: true,
55
+
uses_erlang_externals: false,
56
+
uses_javascript_externals: false,
57
+
},
58
+
purity: Pure,
59
+
},
60
+
),
61
+
target: None,
62
+
},
63
+
],
64
+
names: Names {
65
+
local_types: {},
66
+
imported_modules: {},
67
+
type_variables: {},
68
+
local_value_constructors: {},
69
+
reexport_aliases: {},
70
+
},
71
+
unused_definition_positions: {},
72
+
},
73
+
extra: ModuleExtra {
74
+
module_comments: [],
75
+
doc_comments: [],
76
+
comments: [],
77
+
empty_lines: [],
78
+
new_lines: [],
79
+
trailing_commas: [],
80
+
},
81
+
}
+81
compiler-core/src/parse/snapshots/gleam_core__parse__tests__parse_negative_hexadecimal_number.snap
+81
compiler-core/src/parse/snapshots/gleam_core__parse__tests__parse_negative_hexadecimal_number.snap
···
1
+
---
2
+
source: compiler-core/src/parse/tests.rs
3
+
expression: "pub fn main() { -0x6f }"
4
+
---
5
+
Parsed {
6
+
module: Module {
7
+
name: "",
8
+
documentation: [],
9
+
type_info: (),
10
+
definitions: [
11
+
TargetedDefinition {
12
+
definition: Function(
13
+
Function {
14
+
location: SrcSpan {
15
+
start: 0,
16
+
end: 13,
17
+
},
18
+
body_start: Some(
19
+
14,
20
+
),
21
+
end_position: 23,
22
+
name: Some(
23
+
(
24
+
SrcSpan {
25
+
start: 7,
26
+
end: 11,
27
+
},
28
+
"main",
29
+
),
30
+
),
31
+
arguments: [],
32
+
body: [
33
+
Expression(
34
+
Int {
35
+
location: SrcSpan {
36
+
start: 16,
37
+
end: 21,
38
+
},
39
+
value: "-0x6f",
40
+
int_value: -111,
41
+
},
42
+
),
43
+
],
44
+
publicity: Public,
45
+
deprecation: NotDeprecated,
46
+
return_annotation: None,
47
+
return_type: (),
48
+
documentation: None,
49
+
external_erlang: None,
50
+
external_javascript: None,
51
+
implementations: Implementations {
52
+
gleam: true,
53
+
can_run_on_erlang: true,
54
+
can_run_on_javascript: true,
55
+
uses_erlang_externals: false,
56
+
uses_javascript_externals: false,
57
+
},
58
+
purity: Pure,
59
+
},
60
+
),
61
+
target: None,
62
+
},
63
+
],
64
+
names: Names {
65
+
local_types: {},
66
+
imported_modules: {},
67
+
type_variables: {},
68
+
local_value_constructors: {},
69
+
reexport_aliases: {},
70
+
},
71
+
unused_definition_positions: {},
72
+
},
73
+
extra: ModuleExtra {
74
+
module_comments: [],
75
+
doc_comments: [],
76
+
comments: [],
77
+
empty_lines: [],
78
+
new_lines: [],
79
+
trailing_commas: [],
80
+
},
81
+
}
+81
compiler-core/src/parse/snapshots/gleam_core__parse__tests__parse_negative_octal_number.snap
+81
compiler-core/src/parse/snapshots/gleam_core__parse__tests__parse_negative_octal_number.snap
···
1
+
---
2
+
source: compiler-core/src/parse/tests.rs
3
+
expression: "pub fn main() { -0o157 }"
4
+
---
5
+
Parsed {
6
+
module: Module {
7
+
name: "",
8
+
documentation: [],
9
+
type_info: (),
10
+
definitions: [
11
+
TargetedDefinition {
12
+
definition: Function(
13
+
Function {
14
+
location: SrcSpan {
15
+
start: 0,
16
+
end: 13,
17
+
},
18
+
body_start: Some(
19
+
14,
20
+
),
21
+
end_position: 24,
22
+
name: Some(
23
+
(
24
+
SrcSpan {
25
+
start: 7,
26
+
end: 11,
27
+
},
28
+
"main",
29
+
),
30
+
),
31
+
arguments: [],
32
+
body: [
33
+
Expression(
34
+
Int {
35
+
location: SrcSpan {
36
+
start: 16,
37
+
end: 22,
38
+
},
39
+
value: "-0o157",
40
+
int_value: -111,
41
+
},
42
+
),
43
+
],
44
+
publicity: Public,
45
+
deprecation: NotDeprecated,
46
+
return_annotation: None,
47
+
return_type: (),
48
+
documentation: None,
49
+
external_erlang: None,
50
+
external_javascript: None,
51
+
implementations: Implementations {
52
+
gleam: true,
53
+
can_run_on_erlang: true,
54
+
can_run_on_javascript: true,
55
+
uses_erlang_externals: false,
56
+
uses_javascript_externals: false,
57
+
},
58
+
purity: Pure,
59
+
},
60
+
),
61
+
target: None,
62
+
},
63
+
],
64
+
names: Names {
65
+
local_types: {},
66
+
imported_modules: {},
67
+
type_variables: {},
68
+
local_value_constructors: {},
69
+
reexport_aliases: {},
70
+
},
71
+
unused_definition_positions: {},
72
+
},
73
+
extra: ModuleExtra {
74
+
module_comments: [],
75
+
doc_comments: [],
76
+
comments: [],
77
+
empty_lines: [],
78
+
new_lines: [],
79
+
trailing_commas: [],
80
+
},
81
+
}
+15
compiler-core/src/parse/tests.rs
+15
compiler-core/src/parse/tests.rs
···
2220
2220
}
2221
2221
2222
2222
#[test]
2223
+
fn parse_negative_binary_number() {
2224
+
assert_parse_module!("pub fn main() { -0b101 }");
2225
+
}
2226
+
2227
+
#[test]
2228
+
fn parse_negative_octal_number() {
2229
+
assert_parse_module!("pub fn main() { -0o157 }");
2230
+
}
2231
+
2232
+
#[test]
2233
+
fn parse_negative_hexadecimal_number() {
2234
+
assert_parse_module!("pub fn main() { -0x6f }");
2235
+
}
2236
+
2237
+
#[test]
2223
2238
fn append_to_const_list() {
2224
2239
assert_module_error!(
2225
2240
"