Skip to main content

harn_lexer/
token.rs

1use std::fmt;
2
3/// A segment of an interpolated string.
4#[derive(Debug, Clone, PartialEq)]
5pub enum StringSegment {
6    Literal(String),
7    Expression(String),
8}
9
10impl fmt::Display for StringSegment {
11    fn fmt(&self, f: &mut fmt::Formatter<'_>) -> fmt::Result {
12        match self {
13            StringSegment::Literal(s) => write!(f, "{s}"),
14            StringSegment::Expression(e) => write!(f, "${{{e}}}"),
15        }
16    }
17}
18
19/// Source location for error reporting.
20#[derive(Debug, Clone, Copy, PartialEq, Eq)]
21pub struct Span {
22    /// Byte offset from start of source (inclusive).
23    pub start: usize,
24    /// Byte offset from start of source (exclusive).
25    pub end: usize,
26    /// 1-based line number of start position.
27    pub line: usize,
28    /// 1-based column number of start position.
29    pub column: usize,
30    /// 1-based line number of end position (for multiline span detection).
31    pub end_line: usize,
32}
33
34impl Span {
35    pub fn with_offsets(start: usize, end: usize, line: usize, column: usize) -> Self {
36        Self {
37            start,
38            end,
39            line,
40            column,
41            end_line: line,
42        }
43    }
44
45    /// Create a span covering two spans (from start of `a` to end of `b`).
46    pub fn merge(a: Span, b: Span) -> Span {
47        Span {
48            start: a.start,
49            end: b.end,
50            line: a.line,
51            column: a.column,
52            end_line: b.end_line,
53        }
54    }
55
56    /// A dummy span for synthetic/generated nodes.
57    pub fn dummy() -> Self {
58        Self {
59            start: 0,
60            end: 0,
61            line: 0,
62            column: 0,
63            end_line: 0,
64        }
65    }
66}
67
68impl fmt::Display for Span {
69    fn fmt(&self, f: &mut fmt::Formatter<'_>) -> fmt::Result {
70        write!(f, "{}:{}", self.line, self.column)
71    }
72}
73
74/// Token kinds produced by the lexer.
75#[derive(Debug, Clone, PartialEq)]
76pub enum TokenKind {
77    // Keywords
78    Pipeline,
79    Extends,
80    Override,
81    Let,
82    Var,
83    If,
84    Else,
85    For,
86    In,
87    Match,
88    Retry,
89    Parallel,
90    ParallelMap,
91    ParallelSettle,
92    Return,
93    Import,
94    True,
95    False,
96    Nil,
97    Try,
98    Catch,
99    Throw,
100    Finally,
101    Fn,
102    Spawn,
103    While,
104    TypeKw,
105    Enum,
106    Struct,
107    Interface,
108    Pub,
109    From,
110    Thru,
111    Upto,
112    Guard,
113    Ask,
114    Deadline,
115    Yield,
116    Mutex,
117    Break,
118    Continue,
119    Select,
120    Impl,
121
122    // Literals
123    Identifier(String),
124    StringLiteral(String),
125    InterpolatedString(Vec<StringSegment>),
126    IntLiteral(i64),
127    FloatLiteral(f64),
128    /// Duration literal in milliseconds: 500ms, 5s, 30m, 2h
129    DurationLiteral(u64),
130
131    // Two-character operators
132    Eq,            // ==
133    Neq,           // !=
134    And,           // &&
135    Or,            // ||
136    Pipe,          // |>
137    NilCoal,       // ??
138    QuestionDot,   // ?.
139    Arrow,         // ->
140    Lte,           // <=
141    Gte,           // >=
142    PlusAssign,    // +=
143    MinusAssign,   // -=
144    StarAssign,    // *=
145    SlashAssign,   // /=
146    PercentAssign, // %=
147
148    // Single-character operators
149    Assign,   // =
150    Not,      // !
151    Dot,      // .
152    Plus,     // +
153    Minus,    // -
154    Star,     // *
155    Slash,    // /
156    Percent,  // %
157    Lt,       // <
158    Gt,       // >
159    Question, // ?
160    Bar,      // |  (for union types)
161
162    // Delimiters
163    LBrace,    // {
164    RBrace,    // }
165    LParen,    // (
166    RParen,    // )
167    LBracket,  // [
168    RBracket,  // ]
169    Comma,     // ,
170    Colon,     // :
171    Semicolon, // ;
172
173    // Comments
174    LineComment(String),  // // text
175    BlockComment(String), // /* text */
176
177    // Special
178    Newline,
179    Eof,
180}
181
182impl fmt::Display for TokenKind {
183    fn fmt(&self, f: &mut fmt::Formatter<'_>) -> fmt::Result {
184        match self {
185            TokenKind::Pipeline => write!(f, "pipeline"),
186            TokenKind::Extends => write!(f, "extends"),
187            TokenKind::Override => write!(f, "override"),
188            TokenKind::Let => write!(f, "let"),
189            TokenKind::Var => write!(f, "var"),
190            TokenKind::If => write!(f, "if"),
191            TokenKind::Else => write!(f, "else"),
192            TokenKind::For => write!(f, "for"),
193            TokenKind::In => write!(f, "in"),
194            TokenKind::Match => write!(f, "match"),
195            TokenKind::Retry => write!(f, "retry"),
196            TokenKind::Parallel => write!(f, "parallel"),
197            TokenKind::ParallelMap => write!(f, "parallel_map"),
198            TokenKind::ParallelSettle => write!(f, "parallel_settle"),
199            TokenKind::Return => write!(f, "return"),
200            TokenKind::Import => write!(f, "import"),
201            TokenKind::True => write!(f, "true"),
202            TokenKind::False => write!(f, "false"),
203            TokenKind::Nil => write!(f, "nil"),
204            TokenKind::Try => write!(f, "try"),
205            TokenKind::Catch => write!(f, "catch"),
206            TokenKind::Throw => write!(f, "throw"),
207            TokenKind::Finally => write!(f, "finally"),
208            TokenKind::Fn => write!(f, "fn"),
209            TokenKind::Spawn => write!(f, "spawn"),
210            TokenKind::While => write!(f, "while"),
211            TokenKind::TypeKw => write!(f, "type"),
212            TokenKind::Enum => write!(f, "enum"),
213            TokenKind::Struct => write!(f, "struct"),
214            TokenKind::Interface => write!(f, "interface"),
215            TokenKind::Pub => write!(f, "pub"),
216            TokenKind::From => write!(f, "from"),
217            TokenKind::Thru => write!(f, "thru"),
218            TokenKind::Upto => write!(f, "upto"),
219            TokenKind::Guard => write!(f, "guard"),
220            TokenKind::Ask => write!(f, "ask"),
221            TokenKind::Deadline => write!(f, "deadline"),
222            TokenKind::Yield => write!(f, "yield"),
223            TokenKind::Mutex => write!(f, "mutex"),
224            TokenKind::Break => write!(f, "break"),
225            TokenKind::Continue => write!(f, "continue"),
226            TokenKind::Select => write!(f, "select"),
227            TokenKind::Impl => write!(f, "impl"),
228            TokenKind::Identifier(s) => write!(f, "id({s})"),
229            TokenKind::StringLiteral(s) => write!(f, "str({s})"),
230            TokenKind::InterpolatedString(_) => write!(f, "istr(...)"),
231            TokenKind::IntLiteral(n) => write!(f, "int({n})"),
232            TokenKind::FloatLiteral(n) => write!(f, "float({n})"),
233            TokenKind::DurationLiteral(ms) => write!(f, "duration({ms}ms)"),
234            TokenKind::Eq => write!(f, "=="),
235            TokenKind::Neq => write!(f, "!="),
236            TokenKind::And => write!(f, "&&"),
237            TokenKind::Or => write!(f, "||"),
238            TokenKind::Pipe => write!(f, "|>"),
239            TokenKind::NilCoal => write!(f, "??"),
240            TokenKind::QuestionDot => write!(f, "?."),
241            TokenKind::Arrow => write!(f, "->"),
242            TokenKind::Lte => write!(f, "<="),
243            TokenKind::Gte => write!(f, ">="),
244            TokenKind::PlusAssign => write!(f, "+="),
245            TokenKind::MinusAssign => write!(f, "-="),
246            TokenKind::StarAssign => write!(f, "*="),
247            TokenKind::SlashAssign => write!(f, "/="),
248            TokenKind::PercentAssign => write!(f, "%="),
249            TokenKind::Assign => write!(f, "="),
250            TokenKind::Not => write!(f, "!"),
251            TokenKind::Dot => write!(f, "."),
252            TokenKind::Plus => write!(f, "+"),
253            TokenKind::Minus => write!(f, "-"),
254            TokenKind::Star => write!(f, "*"),
255            TokenKind::Slash => write!(f, "/"),
256            TokenKind::Percent => write!(f, "%"),
257            TokenKind::Lt => write!(f, "<"),
258            TokenKind::Gt => write!(f, ">"),
259            TokenKind::Question => write!(f, "?"),
260            TokenKind::Bar => write!(f, "|"),
261            TokenKind::LBrace => write!(f, "{{"),
262            TokenKind::RBrace => write!(f, "}}"),
263            TokenKind::LParen => write!(f, "("),
264            TokenKind::RParen => write!(f, ")"),
265            TokenKind::LBracket => write!(f, "["),
266            TokenKind::RBracket => write!(f, "]"),
267            TokenKind::Comma => write!(f, ","),
268            TokenKind::Colon => write!(f, ":"),
269            TokenKind::Semicolon => write!(f, ";"),
270            TokenKind::LineComment(s) => write!(f, "// {s}"),
271            TokenKind::BlockComment(s) => write!(f, "/* {s} */"),
272            TokenKind::Newline => write!(f, "\\n"),
273            TokenKind::Eof => write!(f, "EOF"),
274        }
275    }
276}
277
278/// A token with its kind and source location.
279#[derive(Debug, Clone, PartialEq)]
280pub struct Token {
281    pub kind: TokenKind,
282    pub span: Span,
283}
284
285impl Token {
286    pub fn with_span(kind: TokenKind, span: Span) -> Self {
287        Self { kind, span }
288    }
289}