cambridge_asm/parse/
lexer.rs1use crate::inst::Op;
7use logos::{Lexer, Logos};
8use std::{collections::HashMap, fmt::Debug, num::ParseIntError, ops::Range};
9use thiserror::Error;
10
11fn parse_num(lex: &mut Lexer<Token>) -> Result<usize, ErrorKind> {
12 let src = if lex.slice().as_bytes()[0] == b'#' {
13 &lex.slice()[1..]
14 } else {
15 lex.slice()
16 };
17
18 let res = match src.as_bytes()[0] {
19 b'b' | b'B' => usize::from_str_radix(&src[1..], 2),
20 b'x' | b'X' | b'&' => usize::from_str_radix(&src[1..], 16),
21 b'o' | b'O' => usize::from_str_radix(&src[1..], 8),
22 _ => src.parse(),
23 }?;
24
25 Ok(res)
26}
27
28fn pop_parens(lex: &mut Lexer<Token>) -> String {
29 let mut chars = lex.slice().chars();
30 chars.next();
31 chars.next_back();
32 chars.collect()
33}
34
35#[derive(Default, Error, Debug, Clone, PartialEq)]
36pub enum ErrorKind {
37 #[error("Invalid integer format")]
38 ParseIntError(#[from] ParseIntError),
39 #[error("Syntax error")]
40 #[default]
41 SyntaxError,
42 #[error("Invalid opcode `{0}`")]
43 InvalidOpcode(String),
44 #[error("Invalid operand")]
45 InvalidOperand,
46}
47
48pub type ErrorMap = HashMap<Span, ErrorKind>;
49
50pub type ParseError = WithSpan<ErrorKind>;
51
52#[derive(Debug, Clone, Copy, PartialEq, Eq)]
53pub struct LinearMemory {
54 pub init: usize,
55 pub len: usize,
56}
57
58impl LinearMemory {
59 pub(self) fn from_lexer(lexer: &mut Lexer<Token>) -> Self {
60 Self::from_str(lexer.slice())
61 }
62
63 pub(self) fn from_str(s: &str) -> Self {
64 let mut decl = s.trim_matches(|c| c == '[' || c == ']').split(';');
65
66 let init = decl.next().unwrap().parse().unwrap();
67 let len = decl.next().unwrap().parse().unwrap();
68
69 Self { init, len }
70 }
71}
72
73#[derive(Logos, Debug, Clone, PartialEq, Eq)]
74#[logos(skip r"[ \t]")]
75#[logos(error = ErrorKind)]
76pub enum Token {
77 #[regex(r"//[^\r\n]*", logos::skip)]
78 Comment,
79
80 #[regex(r"\w*", |lex| lex.slice().to_string(), priority = 0)]
81 Text(String),
82
83 #[token(":")]
84 Colon,
85
86 #[token(",")]
87 Comma,
88
89 #[regex("r[0-9][0-9]?", |lex| lex.slice()[1..].parse())]
90 Gpr(usize),
91
92 #[regex("#[&xXoObB][0-9a-fA-F]+", parse_num)]
93 #[regex("#[0-9]+", parse_num)]
94 Literal(usize),
95
96 #[regex("[xXoObB][0-9a-fA-F]+", parse_num)]
97 #[regex("[0-9]+", parse_num)]
98 BareNumber(usize),
99
100 #[regex(r"\(\w*\)", pop_parens)]
101 Indirect(String),
102
103 #[regex(r"(?:\r\n)|\n")]
104 Newline,
105
106 #[regex(r"\[[0-9]+;[0-9]+\]", LinearMemory::from_lexer)]
107 LinearMemory(LinearMemory),
108}
109
110impl From<Token> for Op {
111 fn from(t: Token) -> Self {
112 match t {
113 Token::BareNumber(addr) => Op::Addr(addr),
114 Token::Gpr(r) => Op::Gpr(r),
115 Token::Literal(lit) => Op::Literal(lit),
116 Token::Text(txt) => match txt.to_lowercase().as_str() {
117 "acc" => Op::Acc,
118 "cmp" => Op::Cmp,
119 "ix" => Op::Ix,
120 "ar" => Op::Ar,
121 _ => Op::Fail(txt),
122 },
123 Token::Indirect(s) => Op::Indirect(Box::new(Op::from(s))),
124 _ => unreachable!(),
125 }
126 }
127}
128
129pub type Span = Range<usize>;
130
131pub type WithSpan<T> = (Span, T);
132
133#[derive(Debug, Clone)]
134pub struct TokensWithError<'a>(pub Lexer<'a, Token>);
135
136impl TokensWithError<'_> {
137 pub fn lines(mut self) -> (Vec<Vec<WithSpan<Token>>>, ErrorMap) {
138 let mut errors = ErrorMap::new();
139 let acc = self.by_ref().fold(vec![Vec::new()], |mut acc, (r, t)| {
140 match t {
141 Ok(Token::Newline) => {
142 acc.push(Vec::new());
143 }
144 Ok(t) => {
145 acc.last_mut().unwrap().push((r, t));
146 }
147 Err(e) => {
148 errors.entry(r).or_insert(e);
149 }
150 }
151
152 acc
153 });
154
155 (acc, errors)
156 }
157}
158
159impl Iterator for TokensWithError<'_> {
160 type Item = WithSpan<Result<Token, ErrorKind>>;
161
162 fn next(&mut self) -> Option<Self::Item> {
163 self.0.next().map(|token| (self.0.span(), token))
164 }
165}