solar_parse/lexer/cursor/
token.rs

1//! Raw, low-level tokens. Created using [`Cursor`](crate::Cursor).
2
3use solar_ast::{Base, StrKind};
4
5/// A raw token.
6///
7/// It doesn't contain information about data that has been parsed, only the type of the token and
8/// its size.
9#[derive(Clone, Debug, PartialEq, Eq)]
10pub struct RawToken {
11    /// The kind of token.
12    pub kind: RawTokenKind,
13    /// The length of the token in bytes.
14    pub len: u32,
15}
16
17impl RawToken {
18    /// The [`EOF`](RawTokenKind::Eof) token with length 0.
19    pub const EOF: Self = Self::new(RawTokenKind::Eof, 0);
20
21    /// Creates a new token.
22    #[inline]
23    pub const fn new(kind: RawTokenKind, len: u32) -> Self {
24        Self { kind, len }
25    }
26}
27
28/// Common lexeme types.
29#[derive(Clone, Copy, Debug, PartialEq, Eq)]
30pub enum RawTokenKind {
31    // Multi-char tokens:
32    /// `// comment`
33    ///
34    /// `/// doc comment`
35    LineComment { is_doc: bool },
36
37    /// `/* block comment */`
38    ///
39    /// `/** block doc comment */`
40    BlockComment { is_doc: bool, terminated: bool },
41
42    /// Any whitespace character sequence.
43    Whitespace,
44
45    /// `ident` or `continue`
46    ///
47    /// At this step, keywords are also considered identifiers.
48    Ident,
49
50    /// Examples: `123`, `0x123`, `hex"123"`. Note that `_` is an invalid
51    /// suffix, but may be present here on string and float literals. Users of
52    /// this type will need to check for and reject that case.
53    ///
54    /// See [`RawLiteralKind`] for more details.
55    Literal { kind: RawLiteralKind },
56
57    // One-char tokens:
58    /// `;`
59    Semi,
60    /// `,`
61    Comma,
62    /// `.`
63    Dot,
64    /// `(`
65    OpenParen,
66    /// `)`
67    CloseParen,
68    /// `{`
69    OpenBrace,
70    /// `}`
71    CloseBrace,
72    /// `[`
73    OpenBracket,
74    /// `]`
75    CloseBracket,
76    /// `~`
77    Tilde,
78    /// `?`
79    Question,
80    /// `:`
81    Colon,
82    /// `=`
83    Eq,
84    /// `!`
85    Bang,
86    /// `<`
87    Lt,
88    /// `>`
89    Gt,
90    /// `-`
91    Minus,
92    /// `&`
93    And,
94    /// `|`
95    Or,
96    /// `+`
97    Plus,
98    /// `*`
99    Star,
100    /// `/`
101    Slash,
102    /// `^`
103    Caret,
104    /// `%`
105    Percent,
106
107    /// Unknown token, not expected by the lexer, e.g. `№`
108    Unknown,
109
110    /// End of input.
111    Eof,
112}
113
114impl RawTokenKind {
115    /// Returns `true` if this token is EOF.
116    #[inline]
117    pub const fn is_eof(&self) -> bool {
118        matches!(self, Self::Eof)
119    }
120
121    /// Returns `true` if this token is a line comment or a block comment.
122    #[inline]
123    pub const fn is_comment(&self) -> bool {
124        matches!(self, Self::LineComment { .. } | Self::BlockComment { .. })
125    }
126
127    /// Returns `true` if this token is a whitespace, line comment, or block comment.
128    #[inline]
129    pub const fn is_trivial(&self) -> bool {
130        matches!(self, Self::Whitespace | Self::LineComment { .. } | Self::BlockComment { .. })
131    }
132}
133
134/// The literal types supported by the lexer.
135#[derive(Clone, Copy, Debug, PartialEq, Eq)]
136pub enum RawLiteralKind {
137    /// `123`, `0x123`; empty_int: `0x`
138    Int { base: Base, empty_int: bool },
139    /// `123.321`, `1.2e3`, `.2e3`; empty_exponent: `2e`, `2.3e`, `.3e`
140    Rational { base: Base, empty_exponent: bool },
141    /// `"abc"`, `"abc`; `unicode"abc"`, `unicode"abc`; `hex"abc"`, `hex"abc`
142    Str { kind: StrKind, terminated: bool },
143}
solar_parse/lexer/cursor/token.rs

solar_parse/lexer/cursor/
token.rs
