adze 0.8.0

Define tree-sitter grammars alongside Rust logic with AST-first parsing
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
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
408
409
410
411
412
413
414
415
416
417
418
419
420
421
422
423
424
425
426
427
428
429
430
431
432
433
434
435
436
437
438
439
440
441
442
443
444
445
446
447
448
449
450
451
452
453
454
455
456
457
458
459
460
461
462
463
464
465
466
467
468
469
470
471
472
473
474
475
476
477
478
479
480
481
482
483
484
485
486
487
488
489
490
491
492
493
494
495
496
497
498
499
500
501
502
503
504
505
506
507
508
509
510
511
512
513
514
515
516
517
518
519
520
521
522
523
524
525
526
527
528
529
530
531
532
533
534
535
536
537
538
539
540
541
542
543
544
545
546
547
548
549
550
551
552
553
554
555
556
557
558
559
560
561
562
563
564
565
566
567
568
569
570
571
572
573
574
575
576
577
578
579
580
581
582
583
584
585
586
587
588
589
590
591
592
593
594
595
596
597
598
599
600
601
602
603
604
605
606
607
608
609
610
611
612
613
614
615
616
617
618
619
620
621
622
623
624
625
626
627
628
629
630
631
632
633
634
635
636
637
638
639
640
641
642
643
644
645
646
647
648
649
650
651
652
653
654
655
656
657
658
659
660
661
662
663
664
665
666
667
668
669
670
671
672
673
674
675
676
677
678
679
680
681
682
683
684
685
686
687
688
689
690
691
692
693
694
695
696
697
698
699
700
701
702
703
704
705
706
707
708
709
710
711
712
713
714
715
716
717
718
719
720
721
722
723
724
725
726
727
728
729
730
731
732
733
734
735
736
737
738
739
740
741
742
743
744
745
746
747
748
749
750
751
752
753
754
755
756
757
758
759
760
761
762
763
764
765
766
767
768
769
770
771
772
773
774
775
776
777
778
779
780
781
782
783
784
785
786
787
788
789
790
791
792
793
794
795
796
797
798
799
800
801
802
803
804
805
806
807
808
809
810
811
812
813
814
815
816
817
818
819
820
821
822
823
824
825
826
827
828
829
830
831
832
833
834
835
836
837
838
839
840
841
842
843
844
845
846
847
848
849
850
851
852
853
854
855
856
857
858
859
860
861
862
863
864
865
866
867
868
869
870
871
872
873
874
875
876
877
878
879
880
881
882
883
884
885
886
887
888
889
890
891
892
893
894
895
896
897
898
899
900
901
902
903
904
905
906
907
908
909
910
911
912
913
914
915
916
917
918
919
920
921
922
923
924
925
926
927
928
929
930
931
932
933
934
935
936
937
938
939
940
941
942
943
944
945
946
947
948
949
950
951
952
953
954
955
956
957
958
959
960
961
962
963
964
965
966
967
968
969
970
971
972
973
974
975
976
977
978
979
980
981
982
983
984
985
986
987
988
989
990
991
992
993
994
995
996
997
998
999
1000
1001
1002
1003
1004
1005
1006
1007
1008
1009
1010
1011
// Pure-Rust Tree-sitter compatible parser runtime
// This implements the core parsing algorithm with GLR support

use crate::external_scanner_ffi::TSLexer;
use crate::linecol::LineCol;
use crate::{InputEdit, Node, Point, Range, Tree, TreeCursor};
use std::os::raw::c_void;

/// Parser state for incremental parsing
#[derive(Debug)]
pub struct Parser {
    language: Option<Language>,
    stack: Vec<StackEntry>,
    /// Syntax trees from previous parses for incremental parsing
    old_trees: Vec<Tree>,
    /// Timeout in microseconds (0 means no timeout)
    timeout_micros: u64,
    /// Cancellation flag for parsing
    cancellation_flag: Option<*const std::sync::atomic::AtomicBool>,
}

/// Language definition with parse tables
#[derive(Debug, Clone, Copy)]
pub struct Language {
    ptr: *const ffi::TSLanguage,
}

/// Stack entry for LR parsing
#[derive(Debug, Clone)]
struct StackEntry {
    state: u16,
    node: Option<Subtree>,
    position: usize,
}

/// Internal node representation during parsing
#[derive(Debug, Clone)]
struct Subtree {
    symbol: u16,
    children: Vec<Subtree>,
    start_byte: usize,
    end_byte: usize,
    start_point: Point,
    end_point: Point,
    field_id: Option<u16>,
}

// FFI types to match Tree-sitter C API
mod ffi {
    use std::os::raw::{c_char, c_void};

    #[repr(C)]
    pub struct TSLanguage {
        pub version: u32,
        pub symbol_count: u32,
        pub alias_count: u32,
        pub token_count: u32,
        pub external_token_count: u32,
        pub state_count: u32,
        pub large_state_count: u32,
        pub production_id_count: u32,
        pub field_count: u32,
        pub max_alias_sequence_length: u16,
        pub production_id_map: *const u16,
        pub parse_table: *const u16,
        pub small_parse_table: *const u16,
        pub small_parse_table_map: *const u32,
        pub parse_actions: *const TSParseAction,
        pub symbol_names: *const *const c_char,
        pub field_names: *const *const c_char,
        pub field_map_slices: *const u16,
        pub field_map_entries: *const u16,
        pub symbol_metadata: *const u8,
        pub public_symbol_map: *const u16,
        pub alias_map: *const u16,
        pub alias_sequences: *const u16,
        pub lex_modes: *const TSLexState,
        pub lex_fn: Option<unsafe extern "C" fn(*mut c_void, u16) -> bool>,
        pub keyword_lex_fn: Option<unsafe extern "C" fn(*mut c_void, u16) -> u16>,
        pub keyword_capture_token: u16,
        pub external_scanner: TSExternalScanner,
        pub primary_state_ids: *const u16,
    }

    #[repr(C)]
    pub struct TSParseAction {
        pub action_type: u8,
        pub extra: u8,
        pub child_count: u8,
        pub dynamic_precedence: i8,
        pub symbol: u16,
    }

    #[repr(C)]
    pub struct TSLexState {
        pub lex_state: u16,
        pub external_lex_state: u16,
    }

    #[repr(C)]
    pub struct TSExternalScanner {
        pub states: *const bool,
        pub symbol_map: *const u16,
        pub create: Option<unsafe extern "C" fn() -> *mut c_void>,
        pub destroy: Option<unsafe extern "C" fn(*mut c_void)>,
        pub scan: Option<unsafe extern "C" fn(*mut c_void, *mut c_void, *const bool) -> bool>,
        pub serialize: Option<unsafe extern "C" fn(*mut c_void, *mut u8) -> u32>,
        pub deserialize: Option<unsafe extern "C" fn(*mut c_void, *const u8, u32)>,
    }
}

impl Parser {
    /// Create a new parser
    pub fn new() -> Self {
        Parser {
            language: None,
            stack: Vec::new(),
            old_trees: Vec::new(),
            timeout_micros: 0,
            cancellation_flag: None,
        }
    }

    /// Set the language for parsing
    pub fn set_language(&mut self, language: Language) -> Result<(), String> {
        // SAFETY: `language.ptr` was created via `Language::from_ptr` which requires the
        // caller to guarantee a valid, properly-aligned `*const TSLanguage` that lives
        // for the duration of use. We only read the `version` field here.
        unsafe {
            let lang = &*language.ptr;
            if lang.version != tree_sitter::LANGUAGE_VERSION {
                return Err(format!(
                    "Incompatible language version. Expected {}, got {}",
                    tree_sitter::LANGUAGE_VERSION,
                    lang.version
                ));
            }
        }

        self.language = Some(language);
        self.reset();
        Ok(())
    }

    /// Get the current language
    pub fn language(&self) -> Option<Language> {
        self.language
    }

    /// Set timeout for parsing in microseconds
    pub fn set_timeout_micros(&mut self, timeout: u64) {
        self.timeout_micros = timeout;
    }

    /// Set cancellation flag for parsing
    pub fn set_cancellation_flag(&mut self, flag: Option<*const std::sync::atomic::AtomicBool>) {
        self.cancellation_flag = flag;
    }

    /// Reset parser state
    pub fn reset(&mut self) {
        self.stack.clear();
        self.old_trees.clear();
    }

    /// Parse a string of source code
    pub fn parse(&mut self, text: &str, old_tree: Option<&Tree>) -> Option<Tree> {
        self.parse_with_callback(
            |offset, _position| {
                if offset < text.len() {
                    &text.as_bytes()[offset..]
                } else {
                    &[]
                }
            },
            old_tree,
        )
    }

    /// Parse with a callback function for reading input
    pub fn parse_with_callback<F>(
        &mut self,
        mut callback: F,
        old_tree: Option<&Tree>,
    ) -> Option<Tree>
    where
        F: FnMut(usize, Point) -> &[u8],
    {
        let language = self.language?;

        // Initialize parser state
        self.stack.clear();
        self.stack.push(StackEntry {
            state: 0,
            node: None,
            position: 0,
        });

        // Store old tree for incremental parsing
        if let Some(tree) = old_tree {
            self.old_trees.clear();
            self.old_trees.push(tree.clone());
        }

        // Main parsing loop
        let mut position = 0;
        let mut point = Point { row: 0, column: 0 };
        let start_time = std::time::Instant::now();

        loop {
            // Check timeout
            if self.timeout_micros > 0 {
                let elapsed = start_time.elapsed().as_micros() as u64;
                if elapsed > self.timeout_micros {
                    return None; // Timeout
                }
            }

            // Check cancellation
            if let Some(flag) = self.cancellation_flag {
                // SAFETY: The caller of `set_cancellation_flag` is responsible for
                // ensuring the `AtomicBool` pointer remains valid for the lifetime of
                // the parser. We perform only an atomic relaxed load.
                // TODO(safety): No lifetime tie between the flag and the Parser — a
                // dangling pointer here would be UB. Consider using `&AtomicBool`.
                unsafe {
                    if (*flag).load(std::sync::atomic::Ordering::Relaxed) {
                        return None; // Cancelled
                    }
                }
            }

            // Get current state
            let current_state = self.stack.last()?.state;

            // Lex next token
            let input = callback(position, point);
            let token = self.lex_token(language, current_state, input, position)?;

            // Get action for current state and token
            let action = self.get_action(language, current_state, token.symbol)?;

            match action {
                Action::Shift(next_state) => {
                    // Shift token onto stack
                    self.stack.push(StackEntry {
                        state: next_state,
                        node: Some(Subtree {
                            symbol: token.symbol,
                            children: Vec::new(),
                            start_byte: position,
                            end_byte: position + token.length,
                            start_point: point,
                            end_point: advance_point(point, &input[..token.length]),
                            field_id: None,
                        }),
                        position: position + token.length,
                    });

                    // Advance position
                    position += token.length;
                    point = advance_point(point, &input[..token.length]);
                }

                Action::Reduce(rule_id) => {
                    // Perform reduction
                    self.reduce(language, rule_id)?;
                }

                Action::Accept => {
                    // Parse successful
                    if let Some(entry) = self.stack.pop() {
                        if let Some(root) = entry.node {
                            return Some(Tree::new(root, language));
                        }
                    }
                    return None;
                }

                Action::Error => {
                    // Try error recovery
                    if !self.recover_from_error(language, &mut position, &mut point, &mut callback)
                    {
                        return None;
                    }
                }
            }
        }
    }

    /// Lex a token at the current position
    fn lex_token(
        &self,
        language: Language,
        state: u16,
        input: &[u8],
        position: usize,
    ) -> Option<Token> {
        // SAFETY: `language.ptr` is a valid `*const TSLanguage` (invariant of
        // `Language`). We read `state_count` and `lex_modes` which are immutable
        // static data generated by Tree-sitter. The `from_raw_parts` call is safe
        // because `lex_modes` has at least `state_count` elements (Tree-sitter ABI
        // guarantee). The `lex_fn` call passes a stack-local `Lexer` cast to
        // `*mut c_void`; the function only accesses it during the call.
        unsafe {
            let lang = &*language.ptr;

            // Get lex state for current parse state
            let lex_state = if state < lang.state_count as u16 {
                let lex_modes =
                    std::slice::from_raw_parts(lang.lex_modes, lang.state_count as usize);
                &lex_modes[state as usize]
            } else {
                return None;
            };

            // Try external scanner first if available
            if lang.external_token_count > 0 {
                if let Some(token) = self.lex_external_token(language, lex_state, input) {
                    return Some(token);
                }
            }

            // Use lexer function if available
            if let Some(lex_fn) = lang.lex_fn {
                // Create lexer context
                let mut lexer = Lexer::new(input, position);
                let lexer_ptr = &mut lexer as *mut _ as *mut c_void;

                if lex_fn(lexer_ptr, lex_state.lex_state) {
                    return Some(Token {
                        symbol: lexer.result_symbol,
                        length: lexer.result_length,
                    });
                }
            }

            // Fallback: simple lexer for testing
            self.simple_lex(input)
        }
    }

    /// Simple lexer for basic tokens (for testing)
    fn simple_lex(&self, input: &[u8]) -> Option<Token> {
        if input.is_empty() {
            return Some(Token {
                symbol: 0,
                length: 0,
            }); // EOF
        }

        // Skip whitespace
        let mut i = 0;
        while i < input.len() && input[i].is_ascii_whitespace() {
            i += 1;
        }

        if i > 0 {
            return Some(Token {
                symbol: 1,
                length: i,
            }); // Whitespace token
        }

        // Single character token
        Some(Token {
            symbol: input[0] as u16,
            length: 1,
        })
    }

    /// Lex external token
    fn lex_external_token(
        &self,
        language: Language,
        lex_state: &ffi::TSLexState,
        input: &[u8],
    ) -> Option<Token> {
        // SAFETY: `language.ptr` is a valid `*const TSLanguage`. The external scanner
        // functions (`create`, `scan`, `destroy`) are generated by Tree-sitter and
        // follow its C ABI contract. `create` returns a heap-allocated scanner that we
        // pass to `scan` and `destroy`. `valid_symbols` is a stack-local array whose
        // lifetime covers the `scan` call. The `TSLexer` wrapper is valid for the
        // duration of the `scan` call. `symbol_map` pointer arithmetic is bounded by
        // `external_token_count`.
        // TODO(safety): Scanner instance is leaked if `scan_fn` panics before
        // `destroy_fn` runs. Consider using a drop guard.
        unsafe {
            let lang = &*language.ptr;

            // Check if we have an external scanner and need to use it
            if lex_state.external_lex_state == 0 || lang.external_scanner.scan.is_none() {
                return None;
            }

            // Get the scan function
            let scan_fn = lang.external_scanner.scan?;

            // Create lexer interface for the scanner
            let mut lexer = ExternalLexer {
                input,
                position: 0,
                result_symbol: 0,
                line: 0,
                column: 0,
            };

            // Build valid symbols array based on external lex state
            let external_token_count = lang.external_token_count as usize;
            let mut valid_symbols = vec![false; external_token_count];

            // The external_lex_state is a bitset indicating which external tokens are valid
            for i in 0..external_token_count {
                if (lex_state.external_lex_state >> i) & 1 != 0 {
                    valid_symbols[i] = true;
                }
            }

            // Create scanner instance if needed
            let scanner_instance = if let Some(create_fn) = lang.external_scanner.create {
                create_fn()
            } else {
                std::ptr::null_mut()
            };

            // Call the external scanner
            let mut ts_lexer = create_ts_lexer(&mut lexer);
            let success = scan_fn(
                scanner_instance,
                &mut ts_lexer as *mut _ as *mut c_void,
                valid_symbols.as_ptr(),
            );

            // Clean up scanner instance
            if !scanner_instance.is_null() {
                if let Some(destroy_fn) = lang.external_scanner.destroy {
                    destroy_fn(scanner_instance);
                }
            }

            if success && lexer.result_symbol > 0 {
                // Map external symbol to actual symbol
                let symbol = if !lang.external_scanner.symbol_map.is_null() {
                    *lang
                        .external_scanner
                        .symbol_map
                        .add(lexer.result_symbol as usize)
                } else {
                    lexer.result_symbol
                };

                Some(Token {
                    symbol,
                    length: lexer.position,
                })
            } else {
                None
            }
        }
    }

    /// Get parse action for state and symbol
    fn get_action(&self, language: Language, state: u16, symbol: u16) -> Option<Action> {
        // SAFETY: `language.ptr` is a valid `*const TSLanguage`. State is bounds-checked
        // against `state_count` before any pointer arithmetic. `small_parse_table_map`
        // has `state_count + 1` entries and `parse_table` is sized to cover all offsets
        // referenced from the map (Tree-sitter ABI guarantee). All `.add()` calls are
        // within these bounds.
        // TODO(safety): The assumption that `small_parse_table_map` has `state_count + 1`
        // entries relies on the Tree-sitter ABI — a malformed language struct would cause
        // out-of-bounds reads.
        unsafe {
            let lang = &*language.ptr;

            // Validate state
            if state >= lang.state_count as u16 {
                return Some(Action::Error);
            }

            // The parse table is stored in compressed format
            // All states use small_parse_table_map for offsets
            let state_offset = *lang.small_parse_table_map.add(state as usize) as usize;

            // Find the next state's offset to know where this state's entries end
            let next_offset = if (state + 1) < lang.state_count as u16 {
                *lang.small_parse_table_map.add((state + 1) as usize) as usize
            } else {
                // For the last state, use the last entry in the map
                // The map has state_count + 1 entries
                *lang.small_parse_table_map.add(lang.state_count as usize) as usize
            };

            // The parse table stores entries as pairs: (symbol, action)
            let mut offset = state_offset;
            let end_offset = next_offset;

            while offset + 1 < end_offset {
                let entry_symbol = *lang.parse_table.add(offset);
                let action_value = *lang.parse_table.add(offset + 1);

                // Check if this is a default reduce entry
                // In Tree-sitter's format, reduce entries have the high bit set in the symbol field
                if entry_symbol & 0x8000 != 0 {
                    // This is a default reduce action that applies to all lookahead symbols
                    if action_value != 0 {
                        return Some(decode_action(action_value));
                    }
                }

                // Check if this entry matches our symbol
                if entry_symbol == symbol {
                    return Some(decode_action(action_value));
                }

                offset += 2;
            }

            // Default action (usually Error)
            Some(Action::Error)
        }
    }

    /// Perform a reduction
    fn reduce(&mut self, language: Language, rule_id: u16) -> Option<()> {
        // SAFETY: `language.ptr` is a valid `*const TSLanguage`. `parse_actions` has
        // `production_id_count` entries (Tree-sitter ABI). `rule_id` is bounds-checked
        // before indexing. `field_map_slices` and `field_map_entries` are accessed only
        // after null checks and with bounds validation on slice indices.
        unsafe {
            let lang = &*language.ptr;

            // Parse actions contain the full reduction information
            let parse_actions =
                std::slice::from_raw_parts(lang.parse_actions, lang.production_id_count as usize);

            if rule_id >= lang.production_id_count as u16 {
                return None;
            }

            let action = &parse_actions[rule_id as usize];
            let lhs_symbol = action.symbol;
            let rule_length = action.child_count as usize;

            // Pop rule_length items from stack
            let mut children = Vec::new();
            let mut start_byte = usize::MAX;
            let mut end_byte = 0;
            let mut start_point = Point {
                row: usize::MAX,
                column: usize::MAX,
            };
            let mut end_point = Point { row: 0, column: 0 };

            for _ in 0..rule_length {
                if let Some(entry) = self.stack.pop() {
                    if let Some(node) = entry.node {
                        if node.start_byte < start_byte {
                            start_byte = node.start_byte;
                            start_point = node.start_point;
                        }
                        if node.end_byte > end_byte {
                            end_byte = node.end_byte;
                            end_point = node.end_point;
                        }
                        children.push(node);
                    }
                }
            }

            children.reverse();

            // Extract field names for this production
            let mut field_names = vec![None; children.len()];
            if lang.field_count > 0
                && !lang.field_map_slices.is_null()
                && !lang.field_map_entries.is_null()
            {
                // Each production can have a slice in the field map
                // The slice tells us which children have field names
                let field_map_slices = std::slice::from_raw_parts(
                    lang.field_map_slices,
                    lang.production_id_count as usize * 2,
                );

                if (rule_id as usize) * 2 + 1 < field_map_slices.len() {
                    let slice_start = field_map_slices[rule_id as usize * 2] as usize;
                    let slice_length = field_map_slices[rule_id as usize * 2 + 1] as usize;

                    if slice_length > 0 {
                        let field_map_entries = std::slice::from_raw_parts(
                            lang.field_map_entries,
                            (slice_start + slice_length) * 2,
                        );

                        // Process each field entry
                        for i in 0..slice_length {
                            let entry_offset = (slice_start + i) * 2;
                            if entry_offset + 1 < field_map_entries.len() {
                                let entry_low = field_map_entries[entry_offset];
                                let entry_high = field_map_entries[entry_offset + 1];

                                // Unpack the field entry
                                // Format: field_id (16 bits) | child_index (8 bits) | inherited (8 bits)
                                let packed_entry = ((entry_high as u32) << 16) | (entry_low as u32);
                                let field_id = (packed_entry & 0xFFFF) as u16;
                                let child_index = ((packed_entry >> 16) & 0xFF) as usize;
                                // let inherited = ((packed_entry >> 24) & 0xFF) as u8;

                                if child_index < field_names.len()
                                    && field_id < lang.field_count as u16
                                {
                                    field_names[child_index] = Some(field_id);
                                }
                            }
                        }
                    }
                }
            }

            // Create children with field information
            let mut children_with_fields = Vec::new();
            for (i, mut child) in children.into_iter().enumerate() {
                if let Some(field_id) = field_names[i] {
                    child.field_id = Some(field_id);
                }
                children_with_fields.push(child);
            }

            // Create new node for reduction
            let new_node = Subtree {
                symbol: lhs_symbol,
                children: children_with_fields,
                start_byte,
                end_byte,
                start_point,
                end_point,
                field_id: None, // Parent nodes don't have field IDs
            };

            // Get goto state
            let prev_state = self.stack.last()?.state;
            let goto_state = self.get_goto(language, prev_state, lhs_symbol)?;

            // Push new node
            self.stack.push(StackEntry {
                state: goto_state,
                node: Some(new_node),
                position: end_byte,
            });

            Some(())
        }
    }

    /// Get goto state
    fn get_goto(&self, language: Language, state: u16, symbol: u16) -> Option<u16> {
        // In Tree-sitter, goto states are encoded as shift actions in the parse table
        // When we look up an action for a non-terminal symbol, we get a Shift action
        // that tells us which state to go to
        match self.get_action(language, state, symbol)? {
            Action::Shift(goto_state) => Some(goto_state),
            _ => None,
        }
    }

    /// Error recovery
    fn recover_from_error<F>(
        &mut self,
        _language: Language,
        position: &mut usize,
        point: &mut Point,
        callback: &mut F,
    ) -> bool
    where
        F: FnMut(usize, Point) -> &[u8],
    {
        // Drop the current stack entry if possible to avoid being stuck in an
        // invalid state. If the stack is empty there's nothing we can do.
        if self.stack.pop().is_none() {
            return false;
        }

        // Inspect remaining input at the current position.
        let input = callback(*position, *point);
        if input.is_empty() {
            return false; // no input left – nothing to recover from
        }

        // Consume contiguous ASCII whitespace, treating it as an inserted
        // missing token.
        let ws_len = input
            .iter()
            .position(|c| !c.is_ascii_whitespace())
            .unwrap_or(input.len());
        if ws_len > 0 {
            *point = advance_point(*point, &input[..ws_len]);
            *position += ws_len;
            return true;
        }

        // Otherwise skip a single byte (or the next punctuation boundary) to
        // ensure the parser makes forward progress.
        let skip = input
            .iter()
            .position(|c| c.is_ascii_whitespace() || matches!(c, b',' | b';' | b')' | b'}' | b']'))
            .unwrap_or(1);
        *point = advance_point(*point, &input[..skip]);
        *position += skip;
        true
    }
}

/// Simple lexer interface for external scanners
struct ExternalLexer<'a> {
    input: &'a [u8],
    position: usize,
    result_symbol: u16,
    line: u32,
    line_start: usize, // byte offset of beginning of current line
    token_end: usize,
}

impl<'a> ExternalLexer<'a> {
    fn new(input: &'a [u8], position: usize) -> Self {
        let (line, line_start) = Self::calculate_line_info(input, position);
        ExternalLexer {
            input,
            position,
            result_symbol: 0,
            line,
            line_start,
            token_end: position,
        }
    }

    fn calculate_line_info(input: &[u8], position: usize) -> (u32, usize) {
        let tracker = LineCol::at_position(input, position);
        (tracker.line as u32, tracker.line_start)
    }

    fn get_column(&self) -> u32 {
        (self.position.saturating_sub(self.line_start)) as u32
    }
}

/// Create a TSLexer interface for the external scanner
///
/// # Safety
/// `lexer` must be a valid mutable reference to an `ExternalLexer` that outlives
/// the returned `TSLexer`. The returned `TSLexer` stores `lexer` as a raw pointer
/// in its `context` field — the caller must ensure `lexer` is not moved or dropped
/// while the `TSLexer` is in use.
unsafe fn create_ts_lexer(lexer: &mut ExternalLexer) -> TSLexer {
    extern "C" fn lookahead(lexer_ptr: *mut TSLexer) -> u32 {
        // SAFETY: `lexer_ptr.context` was set to a valid `*mut ExternalLexer` in
        // `create_ts_lexer` and the `ExternalLexer` is alive for the duration of
        // the scan call. We take only a shared reference.
        unsafe {
            let lexer = &*((*lexer_ptr).context as *const ExternalLexer);
            if lexer.position < lexer.input.len() {
                lexer.input[lexer.position] as u32
            } else {
                0
            }
        }
    }

    extern "C" fn advance(lexer_ptr: *mut TSLexer, skip: bool) {
        // SAFETY: Same invariant as `lookahead`. We take a mutable reference;
        // Tree-sitter guarantees no concurrent calls to lexer callbacks.
        unsafe {
            let lexer = &mut *((*lexer_ptr).context as *mut ExternalLexer);
            if lexer.position < lexer.input.len() {
                let byte = lexer.input[lexer.position];
                lexer.position += 1;

                // Handle newlines (CR, LF, CRLF)
                if byte == b'\n' {
                    lexer.line += 1;
                    lexer.line_start = lexer.position;
                } else if byte == b'\r' {
                    // Handle CR and CRLF
                    if lexer.position < lexer.input.len() && lexer.input[lexer.position] == b'\n' {
                        lexer.position += 1; // Skip the LF in CRLF
                    }
                    lexer.line += 1;
                    lexer.line_start = lexer.position;
                }

                if !skip && lexer.token_end < lexer.position {
                    lexer.token_end = lexer.position;
                }
            }
        }
    }

    extern "C" fn mark_end(lexer_ptr: *mut TSLexer) {
        // SAFETY: Same invariant as `advance` — exclusive access to ExternalLexer.
        unsafe {
            let lexer = &mut *((*lexer_ptr).context as *mut ExternalLexer);
            lexer.token_end = lexer.position;
        }
    }

    extern "C" fn get_column(lexer_ptr: *mut TSLexer) -> u32 {
        // SAFETY: Same invariant as `lookahead` — shared access to ExternalLexer.
        unsafe {
            let lexer = &*((*lexer_ptr).context as *const ExternalLexer);
            lexer.get_column()
        }
    }

    extern "C" fn is_at_included_range_start(lexer_ptr: *const TSLexer) -> bool {
        false
    }

    extern "C" fn eof(lexer_ptr: *const TSLexer) -> bool {
        // SAFETY: Same invariant as `lookahead` — shared access to ExternalLexer.
        unsafe {
            let lexer = &*((*lexer_ptr).context as *const ExternalLexer);
            lexer.position >= lexer.input.len()
        }
    }

    TSLexer {
        lookahead,
        advance,
        mark_end,
        get_column,
        is_at_included_range_start,
        eof,
        context: (lexer as *mut ExternalLexer).cast(),
        result_symbol: 0,
    }
}

/// Token returned by lexer
#[derive(Debug, Clone, Copy)]
struct Token {
    symbol: u16,
    length: usize,
}

/// Parse action
#[derive(Debug, Clone, Copy)]
enum Action {
    Shift(u16),
    Reduce(u16),
    Accept,
    Error,
}

/// Decode action from compressed format
fn decode_action(encoded: u16) -> Action {
    match encoded {
        0xFFFF => Action::Accept,
        0xFFFE => Action::Error,
        _ if encoded & 0x8000 != 0 => {
            let rule_id = (encoded & 0x7FFF).saturating_sub(1);
            Action::Reduce(rule_id)
        }
        state => Action::Shift(state),
    }
}

/// Advance point by text
fn advance_point(mut point: Point, text: &[u8]) -> Point {
    for &byte in text {
        if byte == b'\n' {
            point.row += 1;
            point.column = 0;
        } else {
            point.column += 1;
        }
    }
    point
}

/// Lexer context for C callbacks
#[repr(C)]
struct Lexer {
    input: *const u8,
    input_length: usize,
    position: usize,
    result_symbol: u16,
    result_length: usize,
}

impl Lexer {
    fn new(input: &[u8], position: usize) -> Self {
        Lexer {
            input: input.as_ptr(),
            input_length: input.len(),
            position,
            result_symbol: 0,
            result_length: 0,
        }
    }
}

impl Language {
    /// Create a language from a pointer
    ///
    /// # Safety
    /// `ptr` must point to a valid, properly-initialized `TSLanguage` struct that
    /// lives for as long as any `Language` copy is used. The struct is typically
    /// generated by Tree-sitter and has `'static` lifetime.
    pub unsafe fn from_ptr(ptr: *const ffi::TSLanguage) -> Self {
        Language { ptr }
    }

    /// Get language version
    pub fn version(&self) -> u32 {
        // SAFETY: `self.ptr` is valid per the `from_ptr` contract.
        unsafe { (*self.ptr).version }
    }

    /// Get symbol count
    pub fn symbol_count(&self) -> u32 {
        // SAFETY: `self.ptr` is valid per the `from_ptr` contract.
        unsafe { (*self.ptr).symbol_count }
    }

    /// Get field count
    pub fn field_count(&self) -> u32 {
        // SAFETY: `self.ptr` is valid per the `from_ptr` contract.
        unsafe { (*self.ptr).field_count }
    }

    /// Get symbol name
    pub fn symbol_name(&self, symbol: u16) -> &str {
        // SAFETY: `self.ptr` is valid per the `from_ptr` contract. `symbol_names`
        // has `symbol_count` entries pointing to null-terminated C strings in static
        // storage. We bounds-check `symbol` before indexing. `CStr::from_ptr` is
        // safe because Tree-sitter guarantees the strings are null-terminated.
        unsafe {
            let lang = &*self.ptr;
            if symbol >= lang.symbol_count as u16 {
                return "";
            }

            let symbol_names =
                std::slice::from_raw_parts(lang.symbol_names, lang.symbol_count as usize);

            let name_ptr = symbol_names[symbol as usize];
            let name_cstr = std::ffi::CStr::from_ptr(name_ptr as *const i8);
            name_cstr.to_str().unwrap_or("")
        }
    }

    /// Get field name
    pub fn field_name(&self, field_id: u16) -> Option<&str> {
        // SAFETY: Same invariant as `symbol_name` — `field_names` has `field_count`
        // entries of null-terminated C strings. `field_id` is bounds-checked.
        unsafe {
            let lang = &*self.ptr;
            if field_id >= lang.field_count as u16 {
                return None;
            }

            let field_names =
                std::slice::from_raw_parts(lang.field_names, lang.field_count as usize);

            let name_ptr = field_names[field_id as usize];
            let name_cstr = std::ffi::CStr::from_ptr(name_ptr as *const i8);
            name_cstr.to_str().ok()
        }
    }
}

// Re-export Tree-sitter version constant
pub use tree_sitter::LANGUAGE_VERSION;

// SAFETY: `Language` wraps a `*const TSLanguage` pointing to static, immutable data
// generated by Tree-sitter at compile time. The data has `'static` lifetime and is
// never mutated, so it is safe to share across threads (`Send`) and access
// concurrently (`Sync`).
unsafe impl Send for Language {}
unsafe impl Sync for Language {}

#[cfg(test)]
mod tests {
    use super::*;

    #[test]
    fn test_parser_creation() {
        let parser = Parser::new();
        assert!(parser.language.is_none());
        assert_eq!(parser.timeout_micros, 0);
    }

    #[test]
    fn test_action_decoding() {
        assert!(matches!(decode_action(42), Action::Shift(42)));
        assert!(matches!(decode_action(0x8012), Action::Reduce(17)));
        assert!(matches!(decode_action(0xFFFF), Action::Accept));
        assert!(matches!(decode_action(0xFFFE), Action::Error));
    }

    #[test]
    fn test_point_advance() {
        let point = Point { row: 0, column: 0 };
        let point = advance_point(point, b"hello");
        assert_eq!(point.row, 0);
        assert_eq!(point.column, 5);

        let point = advance_point(point, b"\nworld");
        assert_eq!(point.row, 1);
        assert_eq!(point.column, 5);
    }
}