1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
// Lexer generation for pure-Rust parser
//
// NOTE: This module generates `quote!` token streams containing `unsafe` blocks.
// The generated unsafe code dereferences a `*mut TsLexer` raw pointer passed by the
// GLR runtime. The safety contract is: the runtime guarantees the pointer is valid
// and exclusively borrowed for the duration of the `lexer_fn` call.
use adze_ir::{Grammar, SymbolId, TokenPattern};
use proc_macro2::TokenStream;
use quote::quote;
use std::collections::BTreeMap;
/// Generate a simple lexer function for the grammar
pub fn generate_lexer(
grammar: &Grammar,
symbol_to_index: &BTreeMap<SymbolId, usize>,
) -> TokenStream {
// The lexer needs to return symbol indices that match the parse table.
// We use the symbol_to_index mapping from the parse table to ensure consistency.
// Collect all tokens and categorize them
let mut keywords = Vec::new();
let mut other_strings = Vec::new();
let mut regex_patterns = Vec::new();
let mut identifier_pattern = None;
// Track which patterns we've already seen to avoid duplicates
let mut seen_string_patterns = std::collections::HashSet::new();
let mut seen_regex_patterns = std::collections::HashSet::new();
// Sort tokens by name to process primary tokens (with meaningful names) first
let mut sorted_tokens: Vec<_> = grammar.tokens.iter().collect();
sorted_tokens.sort_by_key(|(_, token)| {
if token.name.starts_with('_') && token.name[1..].chars().all(|c| c.is_ascii_digit()) {
(1, token.name.clone())
} else {
(0, token.name.clone())
}
});
for (id, token) in sorted_tokens {
if let Some(&idx) = symbol_to_index.get(id) {
let symbol_index = idx as u16;
match &token.pattern {
TokenPattern::String(s) => {
if seen_string_patterns.contains(s) {
continue;
}
seen_string_patterns.insert(s.clone());
if s.chars().all(|c| c.is_ascii_alphabetic() || c == '_') && s.len() > 1 {
keywords.push((symbol_index, s.clone()));
} else {
other_strings.push((symbol_index, s.clone()));
}
}
TokenPattern::Regex(pattern) => {
if seen_regex_patterns.contains(pattern) {
continue;
}
seen_regex_patterns.insert(pattern.clone());
if pattern == r"[a-zA-Z_][a-zA-Z0-9_]*" {
identifier_pattern = Some(symbol_index);
} else {
regex_patterns.push((symbol_index, pattern.clone()));
}
}
}
}
}
// Sort keywords by length (longest first)
keywords.sort_by(|a, b| b.1.len().cmp(&a.1.len()));
let mut token_matches = Vec::new();
// First: Add keyword matching
for (symbol_index, keyword) in keywords {
let bytes = keyword.as_bytes();
let mut checks = Vec::new();
for byte in bytes {
let b = *byte as u32;
checks.push(quote! {
if ((*lexer).lookahead)(lexer) == #b {
((*lexer).advance)(lexer, false);
} else {
return false;
}
});
}
token_matches.push(quote! {
if (|| unsafe {
#(#checks)*
let next = ((*lexer).lookahead)(lexer);
if next != 0 && ((next as u8).is_ascii_alphanumeric() || next == b'_' as u32) {
return false;
}
true
})() {
unsafe {
(*lexer).result_symbol = #symbol_index;
((*lexer).mark_end)(lexer);
}
return true;
}
});
}
// Second: Add other string patterns
for (symbol_index, s) in other_strings {
if s.len() == 1 {
let ch = s.chars().next().expect("s.len() == 1 guarantees a char") as u32;
token_matches.push(quote! {
if unsafe { ((*lexer).lookahead)(lexer) == #ch } {
unsafe {
((*lexer).advance)(lexer, false);
(*lexer).result_symbol = #symbol_index;
((*lexer).mark_end)(lexer);
}
return true;
}
});
} else {
let bytes = s.as_bytes();
let mut checks = Vec::new();
for byte in bytes {
let b = *byte as u32;
checks.push(quote! {
if ((*lexer).lookahead)(lexer) == #b {
((*lexer).advance)(lexer, false);
} else {
return false;
}
});
}
token_matches.push(quote! {
if (|| unsafe {
#(#checks)*
true
})() {
unsafe {
(*lexer).result_symbol = #symbol_index;
((*lexer).mark_end)(lexer);
}
return true;
}
});
}
}
// Third: Add regex patterns
for (symbol_index, pattern) in regex_patterns {
if pattern == r"\d+" {
token_matches.push(quote! {
let first = unsafe { ((*lexer).lookahead)(lexer) };
if first != 0 && (first as u8).is_ascii_digit() {
unsafe {
((*lexer).advance)(lexer, false);
while {
let next = ((*lexer).lookahead)(lexer);
next != 0 && (next as u8).is_ascii_digit()
} {
((*lexer).advance)(lexer, false);
}
(*lexer).result_symbol = #symbol_index;
((*lexer).mark_end)(lexer);
}
return true;
}
});
} else if pattern == r"\w+" {
token_matches.push(quote! {
let first = unsafe { ((*lexer).lookahead)(lexer) };
if first != 0 && ((first as u8).is_ascii_alphanumeric() || first == b'_' as u32) {
unsafe {
((*lexer).advance)(lexer, false);
while {
let next = ((*lexer).lookahead)(lexer);
next != 0 && ((next as u8).is_ascii_alphanumeric() || next == b'_' as u32)
} {
((*lexer).advance)(lexer, false);
}
(*lexer).result_symbol = #symbol_index;
((*lexer).mark_end)(lexer);
}
return true;
}
});
} else if pattern == r"[-+*/]" {
token_matches.push(quote! {
let first = unsafe { ((*lexer).lookahead)(lexer) };
if first == b'-' as u32 || first == b'+' as u32 || first == b'*' as u32 || first == b'/' as u32 {
unsafe {
((*lexer).advance)(lexer, false);
(*lexer).result_symbol = #symbol_index;
((*lexer).mark_end)(lexer);
}
return true;
}
});
} else if pattern == r"\s" || pattern == r"\s+" || pattern == r"\s*" {
token_matches.push(quote! {
let first = unsafe { ((*lexer).lookahead)(lexer) };
if first != 0 && (first as u8).is_ascii_whitespace() {
unsafe {
((*lexer).advance)(lexer, false);
while {
let next = ((*lexer).lookahead)(lexer);
next != 0 && (next as u8).is_ascii_whitespace()
} {
((*lexer).advance)(lexer, false);
}
(*lexer).result_symbol = #symbol_index;
((*lexer).mark_end)(lexer);
}
return true;
}
});
}
}
// Fourth: Add identifier pattern last
if let Some(symbol_index) = identifier_pattern {
token_matches.push(quote! {
let first = unsafe { ((*lexer).lookahead)(lexer) };
if first != 0 && ((first as u8).is_ascii_alphabetic() || first == b'_' as u32) {
unsafe {
((*lexer).advance)(lexer, false);
while {
let next = ((*lexer).lookahead)(lexer);
next != 0 && ((next as u8).is_ascii_alphanumeric() || next == b'_' as u32)
} {
((*lexer).advance)(lexer, false);
}
(*lexer).result_symbol = #symbol_index;
((*lexer).mark_end)(lexer);
}
return true;
}
});
}
quote! {
// SAFETY: Called by the GLR runtime which guarantees `state_ptr` is a valid
// `*mut TsLexer` for the duration of the call. The null check below guards
// against the degenerate case.
unsafe extern "C" fn lexer_fn(state_ptr: *mut ::std::ffi::c_void, _lex_mode: adze::pure_parser::TSLexState) -> bool {
if state_ptr.is_null() {
return false;
}
let lexer = state_ptr as *mut adze::lex::TsLexer;
#(#token_matches)*
false
}
}
}