aboutsummaryrefslogtreecommitdiffstats
path: root/src/tokenizer.rs
blob: 7dbd1587c3cadf1eaa0d2c9c59f525288a05374f (plain) (blame)
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
//! A tokenizer glues states from the state machine together.
//!
//! It facilitates everything needed to turn bytes into events with a state
//! machine.
//! It also enables the logic needed for parsing markdown, such as an
//! [`attempt`][] to try and parse something, which can succeed or, when
//! unsuccessful, revert the attempt.
//!
//! [`attempt`]: Tokenizer::attempt

use crate::event::{Content, Event, Kind, Link, Name, Point, VOID_EVENTS};
use crate::parser::ParseState;
use crate::resolve::{call as call_resolve, Name as ResolveName};
use crate::state::{call, State};
use crate::util::{constant::TAB_SIZE, edit_map::EditMap};
use alloc::{
    boxed::Box,
    format,
    string::{String, ToString},
    vec,
    vec::Vec,
};
use core::str;

/// Containers.
///
/// Containers are found when tokenizing
/// [document content][crate::construct::document].
/// They parse a portion at the start of one or more lines.
/// The rest of those lines is a different content type (specifically, flow),
/// which they “contain”.
#[derive(Debug, Eq, PartialEq)]
pub enum Container {
    /// [Block quote][crate::construct::block_quote].
    BlockQuote,
    /// [List item][crate::construct::list_item].
    ListItem,
    /// [GFM: Footnote definition][crate::construct::gfm_footnote_definition].
    GfmFootnoteDefinition,
}

/// Info used to tokenize a container.
///
/// Practically, these fields are only used for list items.
#[derive(Debug)]
pub struct ContainerState {
    /// Kind.
    pub kind: Container,
    /// Whether the first line was blank.
    pub blank_initial: bool,
    /// Size.
    pub size: usize,
}

/// How to handle a byte.
#[derive(Debug, PartialEq)]
enum ByteAction {
    /// This is a normal byte.
    ///
    /// Includes replaced bytes.
    Normal(u8),
    /// This byte must be ignored.
    Ignore,
    /// This is a new byte.
    Insert(u8),
}

/// Label start kind.
#[derive(Debug, PartialEq, Eq)]
pub enum LabelKind {
    /// Label (image) start.
    ///
    /// ```markdown
    /// > | a ![b] c
    ///       ^^
    /// ```
    ///
    /// Construct: [Label start (image)][crate::construct::label_start_image].
    Image,
    /// Label (image) link.
    ///
    /// ```markdown
    /// > | a [b] c
    ///       ^
    /// ```
    ///
    /// Construct: [Label start (link)][crate::construct::label_start_link].
    Link,
    /// GFM: Label (footnote) link.
    ///
    /// ```markdown
    /// > | a [^b] c
    ///       ^^
    /// ```
    ///
    /// Construct: [GFM: Label start (footnote)][crate::construct::gfm_label_start_footnote].
    GfmFootnote,
    /// GFM: Label (footnote) link, not matching a footnote definition, so
    /// handled as a label (link) start.
    ///
    /// ```markdown
    /// > | a [^b](c) d
    ///       ^^
    /// ```
    ///
    /// Construct: [Label end][crate::construct::label_end].
    GfmUndefinedFootnote,
}

/// Label start, looking for an end.
#[derive(Debug)]
pub struct LabelStart {
    /// Kind of start.
    pub kind: LabelKind,
    /// Indices of where the label starts and ends in `events`.
    pub start: (usize, usize),
    /// A boolean used internally to figure out if a (link) label start can’t
    /// be used anymore (because it would contain another link).
    /// That link start is still looking for a balanced closing bracket though,
    /// so we can’t remove it just yet.
    pub inactive: bool,
}

/// Valid label.
#[derive(Debug)]
pub struct Label {
    pub kind: LabelKind,
    /// Indices of label start.
    pub start: (usize, usize),
    /// Indices of label end.
    pub end: (usize, usize),
}

/// Different kinds of attempts.
#[derive(Debug, PartialEq)]
enum AttemptKind {
    /// Discard what was tokenized when unsuccessful.
    Attempt,
    /// Discard always.
    Check,
}

/// How to handle [`State::Ok`][] or [`State::Nok`][].
#[derive(Debug)]
struct Attempt {
    /// Where to go to when successful.
    ok: State,
    /// Where to go to when unsuccessful.
    nok: State,
    /// Kind of attempt.
    kind: AttemptKind,
    /// If needed, the progress to revert to.
    ///
    /// It is not needed to discard an [`AttemptKind::Attempt`] that has a
    /// `nok` of [`State::Nok`][], because that means it is used in *another*
    /// attempt, which will receive that `Nok`, and has to handle it.
    progress: Option<Progress>,
}

/// The internal state of a tokenizer.
///
/// Not to be confused with states from the state machine, this instead is all
/// the information on where we currently are and what’s going on.
#[derive(Clone, Debug)]
struct Progress {
    /// Length of `events`.
    ///
    /// It’s not allowed to remove events, so reverting will just pop stuff off.
    events_len: usize,
    /// Length of the stack.
    ///
    /// It’s not allowed to decrease the stack in an attempt.
    stack_len: usize,
    /// Previous code.
    previous: Option<u8>,
    /// Current code.
    current: Option<u8>,
    /// Current place in the file.
    point: Point,
}

/// A lot of shared fields used to tokenize things.
#[allow(clippy::struct_excessive_bools)]
#[derive(Debug)]
pub struct TokenizeState<'a> {
    // Couple complex fields used to tokenize the document.
    /// Tokenizer, used to tokenize flow in document.
    pub document_child: Option<Box<Tokenizer<'a>>>,
    /// State, used to tokenize containers.
    pub document_child_state: Option<State>,
    /// Stack of currently active containers.
    pub document_container_stack: Vec<ContainerState>,
    /// How many active containers continued.
    pub document_continued: usize,
    /// Index of last `data`.
    pub document_data_index: Option<usize>,
    /// Container exits by line number.
    pub document_exits: Vec<Option<Vec<Event>>>,
    /// Whether the previous flow was a paragraph or a definition.
    pub document_lazy_accepting_before: bool,
    /// Whether this is the first paragraph (potentially after definitions) in
    /// a list item.
    /// Used for GFM task list items.
    pub document_at_first_paragraph_of_list_item: bool,

    // Couple of very frequent settings for parsing whitespace.
    pub space_or_tab_eol_content: Option<Content>,
    pub space_or_tab_eol_connect: bool,
    pub space_or_tab_eol_ok: bool,
    pub space_or_tab_connect: bool,
    pub space_or_tab_content: Option<Content>,
    pub space_or_tab_min: usize,
    pub space_or_tab_max: usize,
    pub space_or_tab_size: usize,
    pub space_or_tab_token: Name,

    // Couple of media related fields.
    /// List of usable label starts.
    ///
    /// Used when tokenizing [text content][crate::construct::text].
    pub label_starts: Vec<LabelStart>,
    /// List of unusable label starts.
    ///
    /// Used when tokenizing [text content][crate::construct::text].
    pub label_starts_loose: Vec<LabelStart>,
    /// Stack of images and links.
    ///
    /// Used when tokenizing [text content][crate::construct::text].
    pub labels: Vec<Label>,

    /// List of defined definition identifiers.
    pub definitions: Vec<String>,
    /// List of defined GFM footnote definition identifiers.
    pub gfm_footnote_definitions: Vec<String>,

    /// Whether to connect events.
    pub connect: bool,
    /// Marker.
    pub marker: u8,
    /// Secondary marker.
    pub marker_b: u8,
    /// Several markers.
    pub markers: &'static [u8],
    /// Whether something was seen.
    pub seen: bool,
    /// Size.
    pub size: usize,
    /// Secondary size.
    pub size_b: usize,
    /// Tertiary size.
    pub size_c: usize,
    /// Index.
    pub start: usize,
    /// Index.
    pub end: usize,
    /// Slot for an event name.
    pub token_1: Name,
    /// Slot for an event name.
    pub token_2: Name,
    /// Slot for an event name.
    pub token_3: Name,
    /// Slot for an event name.
    pub token_4: Name,
    /// Slot for an event name.
    pub token_5: Name,
    /// Slot for an event name.
    pub token_6: Name,
}

/// A tokenizer itself.
#[allow(clippy::struct_excessive_bools)]
#[derive(Debug)]
pub struct Tokenizer<'a> {
    /// Jump between line endings.
    column_start: Vec<(usize, usize)>,
    // First line where this tokenizer starts.
    first_line: usize,
    /// Current point after the last line ending (excluding jump).
    line_start: Point,
    /// Track whether the current byte is already consumed (`true`) or expected
    /// to be consumed (`false`).
    ///
    /// Tracked to make sure everything’s valid.
    consumed: bool,
    /// Stack of how to handle attempts.
    attempts: Vec<Attempt>,
    /// Current byte.
    pub current: Option<u8>,
    /// Previous byte.
    pub previous: Option<u8>,
    /// Current relative and absolute place in the file.
    pub point: Point,
    /// Semantic labels.
    pub events: Vec<Event>,
    /// Hierarchy of semantic labels.
    ///
    /// Tracked to make sure everything’s valid.
    pub stack: Vec<Name>,
    /// Edit map, to batch changes.
    pub map: EditMap,
    /// List of resolvers.
    pub resolvers: Vec<ResolveName>,
    /// Shared parsing state across tokenizers.
    pub parse_state: &'a ParseState<'a>,
    /// A lot of shared fields used to tokenize things.
    pub tokenize_state: TokenizeState<'a>,
    /// Whether we would be interrupting something.
    ///
    /// Used when tokenizing [flow content][crate::construct::flow].
    pub interrupt: bool,
    /// Whether containers cannot “pierce” into the current construct.
    ///
    /// Used when tokenizing [document content][crate::construct::document].
    pub concrete: bool,
    /// Whether this row is piercing into the current construct with more
    /// containers.
    ///
    /// Used when tokenizing [document content][crate::construct::document].
    pub pierce: bool,
    /// Whether this line is lazy: there are less containers than before.
    pub lazy: bool,
}

impl<'a> Tokenizer<'a> {
    /// Create a new tokenizer.
    pub fn new(point: Point, parse_state: &'a ParseState) -> Tokenizer<'a> {
        Tokenizer {
            previous: None,
            current: None,
            // To do: reserve size when feeding?
            column_start: vec![],
            first_line: point.line,
            line_start: point.clone(),
            consumed: true,
            attempts: vec![],
            point,
            stack: vec![],
            events: vec![],
            parse_state,
            tokenize_state: TokenizeState {
                connect: false,
                document_container_stack: vec![],
                document_exits: vec![],
                document_continued: 0,
                document_lazy_accepting_before: false,
                document_data_index: None,
                document_child_state: None,
                document_child: None,
                document_at_first_paragraph_of_list_item: false,
                definitions: vec![],
                gfm_footnote_definitions: vec![],
                end: 0,
                label_starts: vec![],
                label_starts_loose: vec![],
                marker: 0,
                marker_b: 0,
                markers: &[],
                labels: vec![],
                seen: false,
                size: 0,
                size_b: 0,
                size_c: 0,
                space_or_tab_eol_content: None,
                space_or_tab_eol_connect: false,
                space_or_tab_eol_ok: false,
                space_or_tab_connect: false,
                space_or_tab_content: None,
                space_or_tab_min: 0,
                space_or_tab_max: 0,
                space_or_tab_size: 0,
                space_or_tab_token: Name::SpaceOrTab,
                start: 0,
                token_1: Name::Data,
                token_2: Name::Data,
                token_3: Name::Data,
                token_4: Name::Data,
                token_5: Name::Data,
                token_6: Name::Data,
            },
            map: EditMap::new(),
            interrupt: false,
            pierce: true,
            concrete: false,
            lazy: false,
            resolvers: vec![],
        }
    }

    /// Register a resolver.
    pub fn register_resolver(&mut self, name: ResolveName) {
        if !self.resolvers.contains(&name) {
            self.resolvers.push(name);
        }
    }

    /// Register a resolver, before others.
    pub fn register_resolver_before(&mut self, name: ResolveName) {
        if !self.resolvers.contains(&name) {
            self.resolvers.insert(0, name);
        }
    }

    /// Define a jump between two places.
    ///
    /// This defines to which future index we move after a line ending.
    pub fn define_skip(&mut self, mut point: Point) {
        move_point_back(self, &mut point);

        let info = (point.index, point.vs);
        log::debug!("position: define skip: {:?} -> ({:?})", point.line, info);
        let at = point.line - self.first_line;

        if at >= self.column_start.len() {
            self.column_start.push(info);
        } else {
            self.column_start[at] = info;
        }

        self.account_for_potential_skip();
    }

    /// Increment the current positional info if we’re right after a line
    /// ending, which has a skip defined.
    fn account_for_potential_skip(&mut self) {
        let at = self.point.line - self.first_line;

        if self.point.column == 1 && at != self.column_start.len() {
            self.move_to(self.column_start[at]);
        }
    }

    /// Prepare for a next byte to get consumed.
    fn expect(&mut self, byte: Option<u8>) {
        debug_assert!(self.consumed, "expected previous byte to be consumed");
        self.consumed = false;
        self.current = byte;
    }

    /// Consume the current byte.
    /// Each state function is expected to call this to signal that this code is
    /// used, or call a next function.
    pub fn consume(&mut self) {
        debug_assert!(!self.consumed, "expected code to not have been consumed: this might be because `x(code)` instead of `x` was returned");
        self.move_one();

        self.previous = self.current;
        // While we’re not at eof, it is at least better to not have the
        // same current code as `previous` *and* `current`.
        self.current = None;
        // Mark as consumed.
        self.consumed = true;
    }

    /// Move to the next (virtual) byte.
    fn move_one(&mut self) {
        match byte_action(self.parse_state.bytes, &self.point) {
            ByteAction::Ignore => {
                self.point.index += 1;
            }
            ByteAction::Insert(byte) => {
                self.previous = Some(byte);
                self.point.column += 1;
                self.point.vs += 1;
            }
            ByteAction::Normal(byte) => {
                self.previous = Some(byte);
                self.point.vs = 0;
                self.point.index += 1;

                if byte == b'\n' {
                    self.point.line += 1;
                    self.point.column = 1;

                    if self.point.line - self.first_line + 1 > self.column_start.len() {
                        self.column_start.push((self.point.index, self.point.vs));
                    }

                    self.line_start = self.point.clone();

                    self.account_for_potential_skip();
                    log::debug!("position: after eol: `{:?}`", self.point);
                } else {
                    self.point.column += 1;
                }
            }
        }
    }

    /// Move (virtual) bytes.
    fn move_to(&mut self, to: (usize, usize)) {
        let (to_index, to_vs) = to;
        while self.point.index < to_index || self.point.index == to_index && self.point.vs < to_vs {
            self.move_one();
        }
    }

    /// Mark the start of a semantic label.
    pub fn enter(&mut self, name: Name) {
        enter_impl(self, name, None);
    }

    /// Enter with a link.
    pub fn enter_link(&mut self, name: Name, link: Link) {
        enter_impl(self, name, Some(link));
    }

    /// Mark the end of a semantic label.
    pub fn exit(&mut self, name: Name) {
        let current = self.stack.pop().expect("cannot close w/o open tokens");

        debug_assert_eq!(current, name, "expected exit event to match current event");

        let previous = self.events.last().expect("cannot close w/o open event");
        let mut point = self.point.clone();

        debug_assert!(
            current != previous.name
                || previous.point.index != point.index
                || previous.point.vs != point.vs,
            "expected non-empty event"
        );

        if VOID_EVENTS.iter().any(|d| d == &name) {
            debug_assert!(
                current == previous.name,
                "expected event to be void (`{:?}`), instead of including `{:?}`",
                current,
                previous.name
            );
        }

        // A bit weird, but if we exit right after a line ending, we *don’t* want to consider
        // potential skips.
        if matches!(self.previous, Some(b'\n')) {
            point = self.line_start.clone();
        } else {
            move_point_back(self, &mut point);
        }

        log::debug!("exit:    `{:?}`", name);
        self.events.push(Event {
            kind: Kind::Exit,
            name,
            point,
            link: None,
        });
    }

    /// Capture the tokenizer progress.
    fn capture(&mut self) -> Progress {
        Progress {
            previous: self.previous,
            current: self.current,
            point: self.point.clone(),
            events_len: self.events.len(),
            stack_len: self.stack.len(),
        }
    }

    /// Apply tokenizer progress.
    fn free(&mut self, previous: Progress) {
        self.previous = previous.previous;
        self.current = previous.current;
        self.point = previous.point;
        debug_assert!(
            self.events.len() >= previous.events_len,
            "expected to restore less events than before"
        );
        self.events.truncate(previous.events_len);
        debug_assert!(
            self.stack.len() >= previous.stack_len,
            "expected to restore less stack items than before"
        );
        self.stack.truncate(previous.stack_len);
    }

    /// Stack an attempt, moving to `ok` on [`State::Ok`][] and `nok` on
    /// [`State::Nok`][], reverting in both cases.
    pub fn check(&mut self, ok: State, nok: State) {
        // Always capture (and restore) when checking.
        // No need to capture (and restore) when `nok` is `State::Nok`, because the
        // parent attempt will do it.
        let progress = Some(self.capture());

        self.attempts.push(Attempt {
            kind: AttemptKind::Check,
            progress,
            ok,
            nok,
        });
    }

    /// Stack an attempt, moving to `ok` on [`State::Ok`][] and `nok` on
    /// [`State::Nok`][], reverting in the latter case.
    pub fn attempt(&mut self, ok: State, nok: State) {
        // Always capture (and restore) when checking.
        // No need to capture (and restore) when `nok` is `State::Nok`, because the
        // parent attempt will do it.
        let progress = if nok == State::Nok {
            None
        } else {
            Some(self.capture())
        };

        self.attempts.push(Attempt {
            kind: AttemptKind::Attempt,
            progress,
            ok,
            nok,
        });
    }

    /// Tokenize.
    pub fn push(&mut self, from: (usize, usize), to: (usize, usize), state: State) -> State {
        push_impl(self, from, to, state, false)
    }

    /// Flush.
    pub fn flush(&mut self, state: State, resolve: bool) -> Result<(), String> {
        let to = (self.point.index, self.point.vs);
        let state = push_impl(self, to, to, state, true);
        let result = state.to_result();

        if resolve && result.is_ok() {
            let resolvers = self.resolvers.split_off(0);
            let mut index = 0;
            while index < resolvers.len() {
                call_resolve(self, resolvers[index]);
                index += 1;
            }

            self.map.consume(&mut self.events);
        }

        result
    }
}

/// Move back past ignored bytes.
fn move_point_back(tokenizer: &mut Tokenizer, point: &mut Point) {
    while point.index > 0 {
        point.index -= 1;
        let action = byte_action(tokenizer.parse_state.bytes, point);
        if !matches!(action, ByteAction::Ignore) {
            point.index += 1;
            break;
        }
    }
}

/// Enter.
fn enter_impl(tokenizer: &mut Tokenizer, name: Name, link: Option<Link>) {
    let mut point = tokenizer.point.clone();
    move_point_back(tokenizer, &mut point);

    log::debug!("enter:   `{:?}`", name);
    tokenizer.stack.push(name.clone());
    tokenizer.events.push(Event {
        kind: Kind::Enter,
        name,
        point,
        link,
    });
}

/// Run the tokenizer.
fn push_impl(
    tokenizer: &mut Tokenizer,
    from: (usize, usize),
    to: (usize, usize),
    mut state: State,
    flush: bool,
) -> State {
    debug_assert!(
        from.0 > tokenizer.point.index
            || (from.0 == tokenizer.point.index && from.1 >= tokenizer.point.vs),
        "cannot move backwards"
    );

    tokenizer.move_to(from);

    loop {
        match state {
            State::Error(_) => break,
            State::Ok | State::Nok => {
                if let Some(attempt) = tokenizer.attempts.pop() {
                    if attempt.kind == AttemptKind::Check || state == State::Nok {
                        if let Some(progress) = attempt.progress {
                            tokenizer.free(progress);
                        }
                    }

                    tokenizer.consumed = true;

                    let next = if state == State::Ok {
                        attempt.ok
                    } else {
                        attempt.nok
                    };

                    log::debug!("attempt: `{:?}` -> `{:?}`", state, next);
                    state = next;
                } else {
                    break;
                }
            }
            State::Next(name) => {
                let action = if tokenizer.point.index < to.0
                    || (tokenizer.point.index == to.0 && tokenizer.point.vs < to.1)
                {
                    Some(byte_action(tokenizer.parse_state.bytes, &tokenizer.point))
                } else if flush {
                    None
                } else {
                    break;
                };

                if let Some(ByteAction::Ignore) = action {
                    tokenizer.move_one();
                } else {
                    let byte =
                        if let Some(ByteAction::Insert(byte) | ByteAction::Normal(byte)) = action {
                            Some(byte)
                        } else {
                            None
                        };

                    let visible = byte.map(|d| {
                        if (b' '..=b'~').contains(&d) {
                            str::from_utf8(&[d]).unwrap().to_string()
                        } else {
                            format!("0x{:x}", d)
                        }
                    });
                    log::debug!("feed:    `{:?}` to {:?}", visible, name);
                    tokenizer.expect(byte);
                    state = call(tokenizer, name);
                };
            }
            State::Retry(name) => {
                log::debug!("retry:   `{:?}`", name);
                state = call(tokenizer, name);
            }
        }
    }

    tokenizer.consumed = true;

    if flush {
        debug_assert!(matches!(state, State::Ok | State::Error(_)), "must be ok");
    } else {
        debug_assert!(
            matches!(state, State::Next(_) | State::Error(_)),
            "must have a next state"
        );
    }

    state
}

/// Figure out how to handle a byte.
fn byte_action(bytes: &[u8], point: &Point) -> ByteAction {
    if point.index < bytes.len() {
        let byte = bytes[point.index];

        if byte == b'\r' {
            // CRLF.
            if point.index < bytes.len() - 1 && bytes[point.index + 1] == b'\n' {
                ByteAction::Ignore
            }
            // CR.
            else {
                ByteAction::Normal(b'\n')
            }
        } else if byte == b'\t' {
            let remainder = point.column % TAB_SIZE;
            let vs = if remainder == 0 {
                0
            } else {
                TAB_SIZE - remainder
            };

            // On the tab itself, first send it.
            if point.vs == 0 {
                if vs == 0 {
                    ByteAction::Normal(byte)
                } else {
                    ByteAction::Insert(byte)
                }
            } else if vs == 0 {
                ByteAction::Normal(b' ')
            } else {
                ByteAction::Insert(b' ')
            }
        } else {
            ByteAction::Normal(byte)
        }
    } else {
        unreachable!("out of bounds")
    }
}