csveee 0.2.1

A very fast, parallel CSV parser.
Documentation
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
mod stepper;
mod table;

use self::stepper::DfaStepper;
use self::table::{Dfa, ERROR_STATE, FIRST_ACTION, RECORD_END_STATE, State};
use super::chunk::{Assumption, ChunkParser};
use super::driver::ChunkDriver;
use super::output::Output;
use crate::config::{Config, QuoteHandling};
use crate::io::ChunkReader;

/// A DFA-based chunk parser.
///
/// `dfa` implements the user's dialect and drives `parse_records`;
/// `find_dfa` is the same dialect forced to `QuoteHandling::Toggle` and
/// drives `find_record_start`.
#[derive(Debug)]
pub struct DfaChunkParser {
    dfa: Dfa,
    find_dfa: Dfa,
    config: Config,
}

impl DfaChunkParser {
    pub fn new(config: Config) -> Self {
        let dfa = Dfa::new(&config);
        let mut find_config = config.clone();
        find_config.quote_handling = QuoteHandling::Toggle;
        let find_dfa = Dfa::new(&find_config);
        Self {
            dfa,
            find_dfa,
            config,
        }
    }

    /// Map an assumption to the initial DFA state.
    fn assumption_to_state(&self, assumption: Assumption) -> State {
        match assumption {
            Assumption::OutOfQuotes => State::FieldStart,
            Assumption::InQuotes => State::InQuotedField,
            Assumption::InQuotesAfterEscape => {
                let doubled_quotes = matches!(
                    (self.config.quote, self.config.escape),
                    (Some(q), Some(e)) if q == e
                );
                if doubled_quotes {
                    State::AfterClosingQuote
                } else {
                    State::InEscapedQuote
                }
            }
        }
    }

    /// Scan for the first record boundary starting from `initial_state`.
    ///
    /// Bounded to the nominal chunk end via `reader.remaining_in_chunk()`
    /// — without the cap one iteration would scan past the chunk, since
    /// buffers extend to EOF. `Ok(None)` means no boundary inside the
    /// chunk; the scheduler's merge-time reparse covers that.
    ///
    /// `strict_probe` is only meaningful with `find_dfa`, which tags
    /// misplaced quotes instead of erroring on them.
    fn find_record_start_from<R: ChunkReader>(
        &self,
        dfa: &Dfa,
        strict_probe: bool,
        reader: &mut R,
        initial_state: State,
    ) -> crate::Result<Option<usize>> {
        let mut dfa_state = initial_state;
        let mut total_consumed: usize = 0;

        loop {
            reader.fill(1)?;
            let buf = reader.buffer();

            let buf_end = buf.len().min(reader.remaining_in_chunk());
            if buf_end == 0 {
                return Ok(None);
            }

            let mut pos = 0;
            while pos < buf_end {
                // Same scan fast path as the stepper, but the bytes
                // are discarded (they belong to the previous chunk's
                // record or the header row), so there is no write cursor
                // to maintain.
                let tail = &buf[pos..buf_end];
                pos += match dfa_state {
                    State::InField => dfa.in_field_needles().find(tail),
                    State::InQuotedField => dfa.in_quoted_needles().find(tail),
                    // Empty-line skip. Cold relative to the field scan,
                    // so the mask-based scan is fine here.
                    State::RecordStart => {
                        let mask = dfa.record_start_structural();
                        tail.iter().position(|&b| mask[b as usize])
                    }
                    _ => Some(0),
                }
                .unwrap_or(tail.len());
                if pos >= buf_end {
                    break;
                }

                let byte = buf[pos];
                let trans = dfa.transition(dfa_state, byte);
                pos += 1;
                let strict_rejected = strict_probe && trans.is_strict_error();
                dfa_state = trans.next_state();
                let s = dfa_state as u8;
                if s < FIRST_ACTION && !strict_rejected {
                    continue;
                }
                let anchor = pos - 1;
                let byte_offset = total_consumed + anchor;
                if s == ERROR_STATE || strict_rejected {
                    crate::trace::debug!(
                        scan_offset = byte_offset,
                        byte = byte,
                        byte_ch = format!("{:?}", byte as char),
                        "find_record_start: strict rejected byte",
                    );
                    reader.consume(pos);
                    return Err(crate::Error::InvalidQuote {
                        position: crate::error::Position { byte_offset },
                    });
                }
                if s == RECORD_END_STATE {
                    reader.consume(anchor);
                    return Ok(Some(byte_offset));
                }
                // FieldEnd — resume scanning the next field.
                dfa_state = State::FieldStart;
            }

            reader.consume(pos);
            total_consumed += pos;
        }
    }
}

impl ChunkParser for DfaChunkParser {
    fn config(&self) -> &Config {
        &self.config
    }

    fn supports(_config: &Config) -> Result<(), &'static str> {
        Ok(()) // DFA supports all configurations.
    }

    fn scan_record_start<R: ChunkReader>(
        &self,
        reader: &mut R,
        assumption: Assumption,
        strict_probe: bool,
    ) -> crate::Result<Option<usize>> {
        self.find_record_start_from(
            &self.find_dfa,
            strict_probe,
            reader,
            self.assumption_to_state(assumption),
        )
    }

    /// The user's dialect, not `find_dfa`: chunk 0's state is exact.
    /// `RecordStart` consumes leading comment lines before the header.
    fn scan_header_end<R: ChunkReader>(&self, reader: &mut R) -> crate::Result<Option<usize>> {
        self.find_record_start_from(&self.dfa, false, reader, State::RecordStart)
    }

    fn parse_from<S, A, R: ChunkReader, O: Output + ?Sized>(
        &self,
        reader: &mut R,
        state: &mut S,
        base: usize,
        acc: &A,
    ) -> crate::Result<Option<usize>>
    where
        A: Fn(&mut S, &mut [&mut O]) -> Result<(), Box<dyn std::error::Error + Send + Sync>>,
    {
        // The driver owns fill / consume / UTF-8 validation; the stepper
        // owns the per-byte DFA.
        let mut stepper = DfaStepper::new(&self.dfa, &self.config, base);
        ChunkDriver::new(reader, base).run(&mut stepper, state, acc)
    }
}

#[cfg(test)]
mod tests {
    use super::*;
    use crate::parser::chunk::{FindRecordStart, PassResult};

    fn default_parser() -> DfaChunkParser {
        DfaChunkParser::new(Config::default())
    }

    fn strict_parser() -> DfaChunkParser {
        let mut config = Config::default();
        config.quote_handling = QuoteHandling::Strict;
        DfaChunkParser::new(config)
    }

    /// Helper: parse entire input from the start (no find_record_start).
    fn parse_collect(parser: &DfaChunkParser, input: &[u8]) -> crate::Result<Vec<Vec<String>>> {
        let pr = chunk_collect_with(
            parser,
            input,
            input.len(),
            Assumption::OutOfQuotes,
            FindRecordStart::No,
        );
        pr.result
    }

    #[test]
    fn parse_simple_csv() {
        let p = default_parser();
        let records = parse_collect(&p, b"a,b,c\n1,2,3\n").unwrap();
        assert_eq!(records, vec![vec!["a", "b", "c"], vec!["1", "2", "3"],]);
    }

    #[test]
    fn parse_no_trailing_newline() {
        let p = default_parser();
        let records = parse_collect(&p, b"a,b\n1,2").unwrap();
        assert_eq!(records, vec![vec!["a", "b"], vec!["1", "2"],]);
    }

    #[test]
    fn parse_crlf() {
        let p = default_parser();
        let records = parse_collect(&p, b"a,b\r\n1,2\r\n").unwrap();
        assert_eq!(records, vec![vec!["a", "b"], vec!["1", "2"],]);
    }

    #[test]
    fn parse_quoted_fields() {
        let p = default_parser();
        let records = parse_collect(&p, b"\"hello\",world\n").unwrap();
        assert_eq!(records, vec![vec!["hello", "world"]]);
    }

    #[test]
    fn parse_doubled_quotes() {
        let p = default_parser();
        let records = parse_collect(&p, b"\"he said \"\"hi\"\"\",ok\n").unwrap();
        assert_eq!(records, vec![vec!["he said \"hi\"", "ok"]]);
    }

    #[test]
    fn parse_quoted_with_delimiter() {
        let p = default_parser();
        let records = parse_collect(&p, b"\"a,b\",c\n").unwrap();
        assert_eq!(records, vec![vec!["a,b", "c"]]);
    }

    #[test]
    fn parse_quoted_with_newline() {
        let p = default_parser();
        let records = parse_collect(&p, b"\"a\nb\",c\n").unwrap();
        assert_eq!(records, vec![vec!["a\nb", "c"]]);
    }

    #[test]
    fn parse_empty_fields() {
        let p = default_parser();
        let records = parse_collect(&p, b",,,\n").unwrap();
        assert_eq!(records, vec![vec!["", "", "", ""]]);
    }

    #[test]
    fn parse_empty_input() {
        let p = default_parser();
        let records = parse_collect(&p, b"").unwrap();
        assert!(records.is_empty());
    }

    #[test]
    fn parse_empty_lines_skipped() {
        let p = default_parser();
        let records = parse_collect(&p, b"\n\na,b\n\n").unwrap();
        assert_eq!(records, vec![vec!["a", "b"]]);
    }

    #[test]
    fn parse_semicolon_delimiter() {
        let mut config = Config::default();
        config.delimiter = b';';
        let p = DfaChunkParser::new(config);
        let records = parse_collect(&p, b"a;b;c\n").unwrap();
        assert_eq!(records, vec![vec!["a", "b", "c"]]);
    }

    #[test]
    fn parse_backslash_escape() {
        let mut config = Config::default();
        config.escape = Some(b'\\');
        let p = DfaChunkParser::new(config);
        let records = parse_collect(&p, b"\"a\\\"b\",c\n").unwrap();
        assert_eq!(records, vec![vec!["a\"b", "c"]]);
    }

    #[test]
    fn parse_comment_lines() {
        let mut config = Config::default();
        config.comment = Some(b'#');
        let p = DfaChunkParser::new(config);
        let records = parse_collect(&p, b"# comment\na,b\n# another\n1,2\n").unwrap();
        assert_eq!(records, vec![vec!["a", "b"], vec!["1", "2"],]);
    }

    #[test]
    fn parse_unclosed_quote_errors() {
        let p = default_parser();
        let result = parse_collect(&p, b"\"unclosed\n");
        assert!(result.is_err());
    }

    #[test]
    fn parse_unclosed_quote_at_eof_errors() {
        let p = default_parser();
        let result = parse_collect(&p, b"\"unclosed");
        assert!(result.is_err());
    }

    #[test]
    fn invalid_quote_reports_position() {
        // Strict mode: a quote inside an unquoted field is an error,
        // and the reported byte_offset must point at the offending quote.
        let p = strict_parser();
        let err = parse_collect(&p, b"a\"b,c\n").unwrap_err();
        match err {
            crate::Error::InvalidQuote { position } => {
                assert_eq!(position.byte_offset, 1);
            }
            e => panic!("expected InvalidQuote, got {e:?}"),
        }
    }

    #[test]
    fn unclosed_quote_reports_position() {
        // The unclosed quoted field starts at byte 8 (the `"` after
        // `foo,`). Verifies that `field_start` survives a record
        // boundary and the buffer-shift compaction.
        let p = default_parser();
        let err = parse_collect(&p, b"a,b\nfoo,\"bar\n").unwrap_err();
        match err {
            crate::Error::UnclosedQuote { position } => {
                assert_eq!(position.byte_offset, 8);
            }
            e => panic!("expected UnclosedQuote, got {e:?}"),
        }
    }

    #[test]
    fn unclosed_quote_reports_position_when_the_prefix_is_consumed() {
        // The opening quote produces no output, so the record in flight
        // leaves the write cursor aligned and nothing on `pending_spans`
        // — the all-clean consume. `field_in_start` still names the quote
        // at byte 2 and must survive being consumed past.
        let p = default_parser();
        let err = parse_collect(&p, b"x\n\"").unwrap_err();
        match err {
            crate::Error::UnclosedQuote { position } => {
                assert_eq!(position.byte_offset, 2);
            }
            e => panic!("expected UnclosedQuote, got {e:?}"),
        }
    }

    #[test]
    fn unclosed_quote_after_escape_reports_position_of_the_quote() {
        // The same, two bytes deep: quote then escape, both without
        // output.
        let mut config = Config::default();
        config.escape = Some(b'\\');
        let p = DfaChunkParser::new(config);
        let err = parse_collect(&p, b"x\n\"\\").unwrap_err();
        match err {
            crate::Error::UnclosedQuote { position } => {
                assert_eq!(position.byte_offset, 2);
            }
            e => panic!("expected UnclosedQuote, got {e:?}"),
        }
    }

    #[test]
    fn utf8_error_reports_position() {
        // 0xFF at byte 6 is not valid UTF-8; the validate_chunk path
        // should report it before any DFA stepping.
        let p = default_parser();
        let err = parse_collect(&p, b"hello,\xFFworld\n").unwrap_err();
        match err {
            crate::Error::Utf8 { position } => {
                assert_eq!(position.byte_offset, 6);
            }
            e => panic!("expected Utf8, got {e:?}"),
        }
    }

    #[test]
    fn parse_trailing_comma() {
        let p = default_parser();
        let records = parse_collect(&p, b"a,b,\n").unwrap();
        assert_eq!(records, vec![vec!["a", "b", ""]]);
    }

    #[test]
    fn parse_single_field() {
        let p = default_parser();
        let records = parse_collect(&p, b"hello\nworld\n").unwrap();
        assert_eq!(records, vec![vec!["hello"], vec!["world"]]);
    }

    // ── ChunkReader tests ──────────────────────────────────────────

    /// A simple in-memory ChunkReader for testing, backed by owned mutable data.
    struct VecChunkReader {
        data: Vec<u8>,
        pos: usize,
        chunk_end: usize,
    }

    impl VecChunkReader {
        fn new(data: &[u8], chunk_end: usize) -> Self {
            Self {
                data: data.to_vec(),
                pos: 0,
                chunk_end,
            }
        }
    }

    impl ChunkReader for VecChunkReader {
        fn buffer(&self) -> &[u8] {
            &self.data[self.pos..]
        }

        fn buffer_mut(&mut self) -> &mut [u8] {
            &mut self.data[self.pos..]
        }

        fn fill(&mut self, _n: usize) -> std::io::Result<()> {
            Ok(())
        }

        fn consume(&mut self, n: usize) {
            self.pos += n;
        }

        fn remaining_in_chunk(&self) -> usize {
            self.chunk_end.saturating_sub(self.pos)
        }
    }

    /// A ChunkReader that exposes its data through a sliding window and
    /// returns an I/O error once the parser asks to read past `fail_at`.
    /// Used to verify that a mid-chunk read failure propagates instead of
    /// being silently swallowed (see `docs/swallowed-read-error.md`).
    struct FailAtReader {
        data: Vec<u8>,
        pos: usize,
        revealed: usize,
        fail_at: usize,
    }

    impl FailAtReader {
        fn new(data: &[u8], fail_at: usize) -> Self {
            Self {
                data: data.to_vec(),
                pos: 0,
                revealed: 0,
                fail_at,
            }
        }
    }

    impl ChunkReader for FailAtReader {
        fn buffer(&self) -> &[u8] {
            &self.data[self.pos..self.revealed]
        }

        fn buffer_mut(&mut self) -> &mut [u8] {
            &mut self.data[self.pos..self.revealed]
        }

        fn fill(&mut self, n: usize) -> std::io::Result<()> {
            let want = self.pos + n;
            if want > self.fail_at {
                // Simulate a real read() failure mid-chunk (bad sector,
                // NFS timeout, O_DIRECT EINVAL, ...).
                return Err(std::io::Error::other("simulated mid-chunk read failure"));
            }
            self.revealed = self.revealed.max(want.min(self.data.len()));
            Ok(())
        }

        fn consume(&mut self, n: usize) {
            self.pos += n;
        }

        fn remaining_in_chunk(&self) -> usize {
            // Treat the whole input as one chunk so the parser keeps
            // reading until it either finishes or hits the failure.
            self.data.len().saturating_sub(self.pos)
        }
    }

    /// Regression test for the swallowed-read-error bug
    /// (`docs/swallowed-read-error.md`). When a `ChunkReader::fill` fails
    /// after the chunk has already produced some records, the driver must
    /// surface the error rather than returning a truncated `Ok`.
    #[test]
    fn mid_chunk_read_error_propagates() {
        let p = default_parser();
        // Eight records; `fill` fails once the parser tries to read past
        // byte 10 — well after it has emitted the first records.
        let data = b"a,b\nc,d\ne,f\ng,h\ni,j\nk,l\nm,n\no,p\n";
        let mut reader = FailAtReader::new(data, 10);
        let pr = p.parse(
            &mut reader,
            0usize,
            Assumption::OutOfQuotes,
            &|count: &mut usize, _fields: &mut [&mut str]| {
                *count += 1;
                Ok(())
            },
            FindRecordStart::No,
        );
        // Pre-fix this returned `Ok` with a partial count; the underlying
        // io::Error must now propagate.
        assert!(
            pr.result.is_err(),
            "mid-chunk read error was swallowed instead of surfaced",
        );
    }

    /// Helper: parse a chunk and collect records (strict find by default).
    fn chunk_collect(
        parser: &DfaChunkParser,
        data: &[u8],
        chunk_end: usize,
        assumption: Assumption,
    ) -> PassResult<Vec<Vec<String>>> {
        chunk_collect_with(parser, data, chunk_end, assumption, FindRecordStart::Strict)
    }

    /// Helper: parse a chunk with explicit find mode.
    fn chunk_collect_with(
        parser: &DfaChunkParser,
        data: &[u8],
        chunk_end: usize,
        assumption: Assumption,
        find: FindRecordStart,
    ) -> PassResult<Vec<Vec<String>>> {
        let mut reader = VecChunkReader::new(data, chunk_end);
        parser.parse(
            &mut reader,
            Vec::<Vec<String>>::new(),
            assumption,
            &|state: &mut Vec<Vec<String>>, fields: &mut [&mut str]| {
                state.push(fields.iter().map(|s| s.to_string()).collect());
                Ok(())
            },
            find,
        )
    }

    // ── find_record_start tests ────────────────────────────────────

    #[test]
    fn chunk_find_record_start_skips_partial() {
        let p = default_parser();
        let data = b"eld2\nrec2_f1,rec2_f2\n";
        let pr = chunk_collect(&p, data, data.len(), Assumption::OutOfQuotes);
        assert_eq!(pr.record_start, Some(4));
        let records = pr.result.unwrap();
        assert_eq!(records, vec![vec!["rec2_f1", "rec2_f2"]]);
    }

    #[test]
    fn chunk_find_record_start_in_quotes() {
        let p = default_parser();
        let data = b"oted\"\nrec2\n";
        let pr = chunk_collect(&p, data, data.len(), Assumption::InQuotes);
        let records = pr.result.unwrap();
        assert_eq!(records, vec![vec!["rec2"]]);
    }

    #[test]
    fn chunk_find_record_start_in_quotes_with_delimiter() {
        let p = strict_parser();
        let data = b", safe ice circulation is generally from mid-November to mid-April. Geography Toponymy At various times in history, this territory has been occupied by the Attikameks, the Algonquin and the Cree. The toponym \"\"Ventadour River\"\" was made official on December 5, 1968, at the Commission de toponymie du Qu\xE9bec, when it was created. Notes and references See also Rivers of Nord-du-Qu\xE9bec Jam\xE9sie Nottaway River drainage basin\"\nnext record";
        let pr = chunk_collect(&p, data, data.len(), Assumption::InQuotes);
        let records = pr.result.unwrap();
        assert_eq!(records, vec![vec!["next record"]]);
    }

    #[test]
    fn find_record_start_stops_at_chunk_end() {
        let p = default_parser();
        let mut data = vec![b'x'; 5000];
        data.push(b'\n');
        data.extend_from_slice(b"rec,data\n");
        let chunk_end = 1024;
        let mut reader = VecChunkReader::new(&data, chunk_end);

        let result = p.scan_record_start(&mut reader, Assumption::OutOfQuotes, true);
        assert!(
            matches!(result, Ok(None)),
            "expected Ok(None), got {result:?} — scan crossed chunk_end",
        );
        assert!(
            reader.pos <= chunk_end,
            "reader advanced past chunk_end: pos={}, chunk_end={}",
            reader.pos,
            chunk_end,
        );
    }

    #[test]
    fn chunk_find_record_start_in_quotes_after_escape() {
        let mut config = Config::default();
        config.escape = Some(b'\\');
        let p = DfaChunkParser::new(config);
        let data = b"x\",next\nrec2\n";
        let pr = chunk_collect(&p, data, data.len(), Assumption::InQuotesAfterEscape);
        let records = pr.result.unwrap();
        assert_eq!(records, vec![vec!["rec2"]]);
    }

    // ── parse_records tests ────────────────────────────────────────

    #[test]
    fn chunk_stops_past_boundary() {
        let p = default_parser();
        let data = b"a,b\nc,d\ne,f\n";
        let pr = chunk_collect(&p, data, 5, Assumption::OutOfQuotes);
        let records = pr.result.unwrap();
        assert_eq!(records, vec![vec!["c", "d"]]);
    }

    #[test]
    fn chunk_finishes_record_past_boundary() {
        let p = default_parser();
        let data = b"a\nc,d\ne\n";
        let pr = chunk_collect(&p, data, 5, Assumption::OutOfQuotes);
        let records = pr.result.unwrap();
        assert_eq!(records, vec![vec!["c", "d"]]);
    }

    #[test]
    fn chunk_no_records_all_partial() {
        let p = default_parser();
        let data = b"still in prev record";
        let pr = chunk_collect(&p, data, data.len(), Assumption::OutOfQuotes);
        assert!(pr.record_start.is_none());
        assert!(pr.record_end.is_none());
        let records = pr.result.unwrap();
        assert!(records.is_empty());
    }

    #[test]
    fn chunk_record_end_is_the_input_end_not_the_write_cursor() {
        // Last field is quoted and unterminated by a newline, so the
        // compaction cursor ends two bytes short of the input.
        let p = default_parser();
        let data = b"a,b\nc,\"d\"";
        let pr = chunk_collect_with(
            &p,
            data,
            data.len(),
            Assumption::OutOfQuotes,
            FindRecordStart::No,
        );
        assert_eq!(pr.result.unwrap(), vec![vec!["a", "b"], vec!["c", "d"]]);
        assert_eq!(pr.record_end, Some(data.len()));
    }

    #[test]
    fn chunk_error_preserves_first_record_start() {
        let p = strict_parser();
        let data = b"a\nb\"c\n";
        let pr = chunk_collect(&p, data, data.len(), Assumption::OutOfQuotes);
        assert_eq!(pr.record_start, Some(1));
        assert!(pr.result.is_err());
    }

    /// Verdict per assumption, in `[OutOfQuotes, InQuotes]` order:
    /// `Ok(Some(n))` accepted with a boundary at `n`, `Ok(None)` accepted
    /// but no boundary, `Err(())` rejected.
    type ProbeVerdicts = [Result<Option<usize>, ()>; 2];

    fn probe(p: &DfaChunkParser, data: &[u8]) -> ProbeVerdicts {
        [Assumption::OutOfQuotes, Assumption::InQuotes].map(|a| {
            let pr = chunk_collect_with(p, data, data.len(), a, FindRecordStart::Strict);
            match (pr.record_start, pr.result) {
                (start, Ok(_)) => Ok(start),
                (_, Err(_)) => Err(()),
            }
        })
    }

    /// The probe picks the assumption, so it must reject exactly the
    /// misplaced quotes — an opener not at a field boundary, a closer not
    /// followed by one — and nothing else, in every dialect.
    #[test]
    fn strict_probe_rejects_exactly_the_misplaced_quotes() {
        let cases: &[(&[u8], ProbeVerdicts, &str)] = &[
            // Opener at offset 0: the chunk handoff counts as a boundary,
            // so this is the legal start of a quoted field.
            (
                b"\"m\nn\"\nt,u\n",
                [Ok(Some(5)), Err(())],
                "opener at offset 0",
            ),
            (
                b"a,\"m\nn\"\nt,u\n",
                [Ok(Some(7)), Err(())],
                "opener after a delimiter",
            ),
            // Opener inside an unquoted field.
            (b"ab\"m\nn\"\nt,u\n", [Err(()), Err(())], "opener mid-field"),
            // Closer not followed by a delimiter or terminator.
            (
                b"\"mn\"x,y\nt,u\n",
                [Err(()), Err(())],
                "closer before bare text",
            ),
            // No quotes: in-quotes is coherent but finds no boundary.
            (b"ab,cd\nt,u\n", [Ok(Some(5)), Ok(None)], "no quotes"),
            // Chunk opens inside a quoted field.
            (
                b"mn\",x\nt,u\n",
                [Err(()), Ok(Some(5))],
                "opens inside a quoted field",
            ),
        ];
        for qh in [
            QuoteHandling::Toggle,
            QuoteHandling::Literal,
            QuoteHandling::Strict,
        ] {
            let mut config = Config::default();
            config.quote_handling = qh;
            let p = DfaChunkParser::new(config);
            for (data, expected, label) in cases {
                assert_eq!(&probe(&p, data), expected, "{qh:?}: {label}");
            }
        }
    }
}