inkwell/core/src/local/derive.rs

//! Deriving structure from a note's body — the local mirror of what the server
//! computes on save. Pure string scanning (no regex dependency), kept in lockstep
//! with the frontend's inline rules (see frontend notes/markdown.ts):
//!
//! - `#tag`: `#` at the start of a line or after whitespace, then a letter, then
//!   letters, digits, `_` and `-`. On save these become labels attached with
//!   `via_tag = true`. The server and the web run the same cases
//!   (core/testdata/grammar.json).
//! - `- [ ] item`: a checklist item. The body IS the checklist — there is no
//!   table of items beside it, so a list can sit between two paragraphs instead of
//!   only after them.
//!
//! The two are the same idea at different strengths. Tags MATERIALISE into label
//! rows, because the board queries by label. Items materialise into nothing,
//! because nothing queries them: their only readers are the card, the editor and
//! `display_title`. So `extract_items` is the whole storage layer for a checklist,
//! and the rewriters below are how one is edited.
//!
//! Dedupes case-insensitively, preserving first-seen order.

/// Every `#tag` in ONE line, as `(start, end, name)` in char indices.
///
/// Char indices rather than byte offsets so the spans can be used to cut the tags
/// back out of the line without ever landing mid-codepoint — see
/// [`lift_standalone_tags`], which is the only reason the spans exist.
fn line_tags(chars: &[char]) -> Vec<(usize, usize, String)> {
    let mut out: Vec<(usize, usize, String)> = Vec::new();
    let mut i = 0;
    while i < chars.len() {
        if chars[i] == '#' {
            // Start of line or after whitespace, and nothing else, so `(#todo)` and a
            // pasted URL's `/#section` are not tags. All three implementations share
            // this rule.
            let boundary = i == 0 || chars[i - 1].is_whitespace();
            // A tag must start with a letter (so "#1" or a bare "#" is not a tag).
            if boundary && i + 1 < chars.len() && chars[i + 1].is_alphabetic() {
                let mut j = i + 1;
                while j < chars.len() && is_tag_char(chars[j]) {
                    j += 1;
                }
                out.push((i, j, chars[i + 1..j].iter().collect()));
                i = j;
                continue;
            }
        }
        i += 1;
    }
    out
}

/// One `#tag` and exactly where it sits, for a renderer drawing the body itself.
///
/// The card colours a tag where it was typed rather than repeating it as a chip.
/// To do that a renderer needs the
/// SPAN, not just the name, and asking it to find the name again would be a second
/// grammar quietly disagreeing with this one about what `##a` or `#1` is.
#[derive(Debug, Clone, PartialEq, Eq)]
pub struct DerivedTag {
    /// Which body line it sits on, like [`DerivedItem::line`].
    pub line: u32,
    /// Offsets into that line, in UTF-16 code units — INCLUDING the leading `#`.
    ///
    /// UTF-16 rather than chars or bytes because the two languages that consume this
    /// both index strings that way: Kotlin's `AnnotatedString` and JavaScript. A char
    /// index is right up until somebody puts an emoji before a tag, and then it lands
    /// mid-token with no error anywhere.
    pub start: u32,
    pub end: u32,
    pub name: String,
}

/// Every `#tag` in `body` with its position — the scan [`lift_standalone_tags`] does,
/// keeping the spans instead of throwing them away.
///
/// Not deduped: two mentions of `#todo` are two pieces of text to colour. Fences are
/// not skipped either, and that is deliberate — a `#tag` inside a code block stays an
/// inline label on the note, and a renderer that left it plain would be the only
/// surface disagreeing.
pub fn extract_tag_spans(body: &str) -> Vec<DerivedTag> {
    let mut out = Vec::new();
    for (n, line) in body.split('\n').enumerate() {
        let chars: Vec<char> = line.chars().collect();
        let spans = line_tags(&chars);
        if spans.is_empty() {
            continue;
        }
        // Prefix sums, built once per tagged line: char index -> UTF-16 offset.
        let mut units: Vec<u32> = Vec::with_capacity(chars.len() + 1);
        let mut total: u32 = 0;
        units.push(0);
        for c in &chars {
            total += c.len_utf16() as u32;
            units.push(total);
        }
        for (start, end, name) in spans {
            out.push(DerivedTag {
                line: n as u32,
                start: units[start],
                end: units[end],
                name,
            });
        }
    }
    out
}

/// Whether a line opens or closes a fenced code block.
fn is_fence(line: &str) -> bool {
    let trimmed = line.trim_start();
    trimmed.starts_with("```") || trimmed.starts_with("~~~")
}

/// Runs of three or more newlines become two, and the ends are trimmed.
///
/// Removing a line must not leave a hole where it was.
fn collapse_blank_runs(text: &str) -> String {
    let mut out = String::with_capacity(text.len());
    let mut run = 0;
    for c in text.chars() {
        if c == '\n' {
            run += 1;
            if run <= 2 {
                out.push(c);
            }
        } else {
            run = 0;
            out.push(c);
        }
    }
    out.trim_matches('\n').to_string()
}

/// Split a body's tags by whether the text around them can be taken away.
///
/// Returns `(standalone, inline, lifted_body)`.
///
/// THE RULE: a line containing nothing but tags and whitespace is removed. Anything
/// else is left exactly as written.
///
/// The MIRROR of `split_body_tags` in the server's `notes/tags.py`, and it has to stay
/// one: a note lifted differently here than there would change the moment it synced.
/// Same discipline, and the same reason, as `DerivedTint`.
///
/// The conservative reading of "standalone" is deliberate. A trailing tag is
/// ambiguous and the text does not say which it is — `buy milk #grocery` is filing,
/// `remember to call #mom` is the sentence's object, and lifting the second leaves
/// "remember to call". A tag sharing a line with words keeps its words.
///
/// `standalone` tags become ORDINARY labels (`via_tag = 0`): nothing is left to derive
/// them from, so the row becomes the record and the chip's × becomes the way to remove
/// one. `inline` tags stay derived. That is what `via_tag` means: backed by text still
/// in the body.
pub fn lift_standalone_tags(body: &str) -> (Vec<String>, Vec<String>, String) {
    let mut standalone: Vec<String> = Vec::new();
    let mut inline: Vec<String> = Vec::new();
    let mut kept: Vec<&str> = Vec::new();
    let mut in_fence = false;

    for line in body.split('\n') {
        if is_fence(line) {
            in_fence = !in_fence;
            kept.push(line);
            continue;
        }
        let chars: Vec<char> = line.chars().collect();
        let spans = line_tags(&chars);
        // Cut the tags out and see whether anything is left. That is what
        // "standalone" means, and it is the whole rule.
        let mut remainder = String::new();
        let mut pos = 0;
        for (start, end, _) in &spans {
            remainder.extend(chars[pos..*start].iter());
            pos = *end;
        }
        remainder.extend(chars[pos..].iter());

        // A fence's contents are CODE: a `#tag` there is a shell comment in somebody's
        // snippet, and deleting the line would eat part of their example.
        if in_fence || spans.is_empty() || !remainder.trim().is_empty() {
            for (_, _, name) in &spans {
                push_unique(&mut inline, name);
            }
            kept.push(line);
        } else {
            for (_, _, name) in &spans {
                push_unique(&mut standalone, name);
            }
        }
    }

    let lifted = collapse_blank_runs(&kept.join("\n"));
    if !body.trim().is_empty() && lifted.trim().is_empty() {
        // The note was NOTHING but tags. Lifting would leave a blank card, which is a
        // worse outcome than a duplicated chip — so leave it alone.
        let mut all = standalone;
        for name in &inline {
            push_unique(&mut all, name);
        }
        return (Vec::new(), all, body.to_string());
    }

    // A tag that ALSO appears in prose stays derived: the prose copy still backs it,
    // so deleting that copy should still detach the label.
    let inline_folded: Vec<String> = inline.iter().map(|n| fold(n)).collect();
    let standalone = standalone
        .into_iter()
        .filter(|n| !inline_folded.contains(&fold(n)))
        .collect();
    (standalone, inline, lifted)
}

fn is_tag_char(c: char) -> bool {
    c.is_alphanumeric() || c == '_' || c == '-'
}

/// How two tag names compare: equal once every letter is lowercased, accented ones
/// included, so "Café" and "CAFÉ" are one tag. The server's `labeling.named` folds
/// the same way (Postgres `lower`, Python `str.lower`), and the store's `fold` SQL
/// function is this.
pub(crate) fn fold(name: &str) -> String {
    name.to_lowercase()
}

fn push_unique(out: &mut Vec<String>, candidate: &str) {
    let folded = fold(candidate);
    if !out.iter().any(|x| fold(x) == folded) {
        out.push(candidate.to_string());
    }
}

// ── checklist items ─────────────────────────────────────────────────────────
//
// The grammar, in one place, because three languages implement it (here,
// `notes/checklist.py`, `notes/markdown.ts`) and a difference between any two of
// them is a checklist that changes shape when it syncs:
//
//   optional indent, `-` or `*`, one-or-more spaces, `[ ]`/`[x]`/`[X]`,
//   then either end-of-line or one-or-more spaces and the text.
//
// `*` is accepted because markdown.ts already accepts it for a plain bullet, and a
// grammar that takes `* item` but not `* [ ] item` would be a rule with no reason
// anyone could guess. `- [ ]` with nothing after it IS an item with empty text:
// that is exactly what pressing Enter on a list leaves behind, and refusing to
// parse it would make a half-typed list stop being a list.

/// A checklist item, as found in the body. Its position in the returned vector is
/// its identity; the wire carries text and checked, never an id.
#[derive(Debug, Clone, PartialEq, Eq)]
pub struct DerivedItem {
    pub text: String,
    pub checked: bool,
    /// Which body line it sits on.
    ///
    /// Carried here rather than offered as a second function, because every renderer
    /// that walks a body line by line — the Android card, the block editor — needs the
    /// text, the state AND the position together, and asking for them separately is
    /// how two calls come to disagree about a body that changed between them.
    pub line: u32,
}

/// One parsed task line, holding enough to put it back exactly as it was found.
struct TaskLine<'a> {
    indent: &'a str,
    /// Preserved rather than normalised to `-`: rewriting someone's `*` bullets
    /// because they ticked a box would be an edit they did not ask for.
    bullet: char,
    checked: bool,
    text: &'a str,
}

fn parse_task_line(line: &str) -> Option<TaskLine<'_>> {
    let indent_len = line.len() - line.trim_start().len();
    let (indent, rest) = line.split_at(indent_len);

    let bullet = rest.chars().next()?;
    if bullet != '-' && bullet != '*' {
        return None;
    }
    // At least one space after the bullet. `-[ ] x` is not a list item in any
    // markdown either, so it stays prose here too.
    let rest = &rest[bullet.len_utf8()..];
    let gap = rest.len() - rest.trim_start_matches(' ').len();
    if gap == 0 {
        return None;
    }
    let rest = &rest[gap..];

    let mut chars = rest.chars();
    if chars.next()? != '[' {
        return None;
    }
    let mark = chars.next()?;
    if chars.next()? != ']' {
        return None;
    }
    // Decided BEFORE the slice below, which is what guarantees `mark` is one byte
    // and `[?]` is exactly three.
    let checked = match mark {
        ' ' => false,
        'x' | 'X' => true,
        _ => return None,
    };
    let rest = &rest[3..];

    let text = if rest.is_empty() {
        // "- [ ]" — an empty item, which is what an unfinished list line is.
        rest
    } else {
        let gap = rest.len() - rest.trim_start_matches(' ').len();
        // "- [ ]x" is prose: without the space this is not a marker, it is a
        // sentence that happens to start with brackets.
        if gap == 0 {
            return None;
        }
        &rest[gap..]
    };

    Some(TaskLine {
        indent,
        bullet,
        checked,
        text,
    })
}

/// One item as the line that stores it, in canonical form.
///
/// Public because a block editor has to write a line back after someone edits it in a
/// widget that never showed them the marker. Rendering is trivial where PARSING is
/// not, but it still belongs here: this is the file that decides what canonical looks
/// like, and a caller inventing its own `- [x] ` would be a fourth opinion on it.
pub fn render_item(text: &str, checked: bool) -> String {
    render_task_line("", '-', checked, text)
}

fn render_task_line(indent: &str, bullet: char, checked: bool, text: &str) -> String {
    // Always lowercase `x`, whatever was parsed: one canonical output is what makes
    // a round trip stable, so `- [X]` normalises the first time it is touched and
    // never again.
    let mark = if checked { 'x' } else { ' ' };
    if text.is_empty() {
        format!("{indent}{bullet} [{mark}]")
    } else {
        format!("{indent}{bullet} [{mark}] {text}")
    }
}

/// The longest a note's name runs, in characters (not bytes): the server's
/// `DISPLAY_TITLE_CAP`.
pub const DISPLAY_TITLE_CAP: usize = 200;

/// The note's NAME: the first line of its body that says anything, cut to
/// [`DISPLAY_TITLE_CAP`] characters.
///
/// Mirrors `derive_display_title` in the server's notes/helpers.py — one rule written
/// twice, and they have to agree or a synced note is called different things on either
/// side of the wire. Both run the `display_titles` cases in testdata/grammar.json.
///
/// The items ARE lines of the body, so it strips the task marker off: a list-only
/// note is named by its first item, and calling that note "- [ ] milk" would be showing
/// someone the storage rather than the note. An empty item is skipped rather than
/// naming the note "", which is what a half-typed list would otherwise do.
pub fn display_title(body: &str) -> String {
    for line in body.split('\n') {
        let text = strip_marker(line.trim()).trim();
        if !text.is_empty() {
            return text.chars().take(DISPLAY_TITLE_CAP).collect();
        }
    }
    String::new()
}

/// The text of a line with its task marker removed, or the line as it was.
///
/// For naming a note: a list-only note is named by its first item, and calling one
/// "- [ ] milk" would be showing someone the storage instead of the note.
pub fn strip_marker(line: &str) -> &str {
    match parse_task_line(line) {
        Some(t) => t.text,
        None => line,
    }
}

/// Every checklist item in `body`, in the order they appear.
pub fn extract_items(body: &str) -> Vec<DerivedItem> {
    let mut out = Vec::new();
    for (n, line) in body.split('\n').enumerate() {
        if let Some(t) = parse_task_line(line) {
            out.push(DerivedItem {
                text: t.text.to_string(),
                checked: t.checked,
                line: n as u32,
            });
        }
    }
    out
}

/// Tick or untick the `index`-th item, keeping its indent, bullet and text.
///
/// The only edit to an item that isn't typing in the body: adding, rewording and
/// deleting one are all text edits, which every client makes in its editor.
///
/// A body with fewer task lines than that is returned UNCHANGED rather than
/// panicking: the index comes from a UI that may be a moment behind the store, and
/// a stale tap should do nothing rather than take the app down.
pub fn set_item_checked(body: &str, index: usize, checked: bool) -> String {
    let mut lines: Vec<String> = body.split('\n').map(str::to_string).collect();
    let Some(line) = lines
        .iter_mut()
        .filter(|l| parse_task_line(l.as_str()).is_some())
        .nth(index)
    else {
        return body.to_string();
    };
    let Some(t) = parse_task_line(line.as_str()) else {
        return body.to_string();
    };
    let ticked = render_task_line(t.indent, t.bullet, checked, t.text);
    *line = ticked;
    lines.join("\n")
}

/// Add an item at the end of the body.
///
/// Spaced exactly as `import_export.py:_note_markdown` writes a list — a blank line
/// between prose and the list, and nothing between consecutive items. That is not
/// cosmetic: the migrations that fold checklist rows into bodies use the same layout,
/// so every surface lands on byte-identical bodies.
///
/// `checked` is a parameter rather than always false because the two migrations that
/// fold existing rows into bodies have to carry the state those rows were in. A new
/// item from the UI passes false.
pub fn append_item(body: &str, text: &str, checked: bool) -> String {
    let line = render_task_line("", '-', checked, text.trim());
    let trimmed = body.trim_end_matches('\n');
    if trimmed.trim().is_empty() {
        return line;
    }
    let follows_a_list = trimmed
        .split('\n')
        .next_back()
        .is_some_and(|l| parse_task_line(l).is_some());
    if follows_a_list {
        format!("{trimmed}\n{line}")
    } else {
        format!("{trimmed}\n\n{line}")
    }
}

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

    /// The tag names in `body`, once each — what the shared fixture lists.
    fn tag_names(body: &str) -> Vec<String> {
        let mut out = Vec::new();
        for t in extract_tag_spans(body) {
            push_unique(&mut out, &t.name);
        }
        out
    }

    #[test]
    fn tags_basic() {
        assert_eq!(
            tag_names("a #todo and #Work-item_2 here"),
            vec!["todo", "Work-item_2"]
        );
    }

    #[test]
    fn tags_require_letter_start_and_boundary() {
        // "#1" (digit) and an in-word "#" (email-ish) are not tags.
        assert_eq!(tag_names("#1 nope a#b no but #Yes ##no"), vec!["Yes"]);
    }

    #[test]
    fn empty_body() {
        assert!(extract_tag_spans("").is_empty());
    }

    // ── tag spans, for the renderer that draws them in place ─────────────────

    #[test]
    fn tag_spans_carry_the_hash_and_the_line() {
        let spans = extract_tag_spans("buy milk #grocery\nand call #mom about #mom");
        assert_eq!(spans.len(), 3);
        assert_eq!((spans[0].line, spans[0].start, spans[0].end), (0, 9, 17));
        assert_eq!(spans[0].name, "grocery");
        // Not deduped: two mentions are two pieces of text to colour.
        assert_eq!(spans[1].line, 1);
        assert_eq!(spans[2].name, "mom");
        assert_eq!((spans[2].start, spans[2].end), (20, 24));
    }

    #[test]
    fn tag_spans_are_utf16_offsets_not_char_indices() {
        // The emoji is ONE char and TWO UTF-16 code units. Kotlin and JS both index
        // the second way, so a char index would highlight one character too early.
        let spans = extract_tag_spans("🎁 #gift");
        assert_eq!(spans.len(), 1);
        assert_eq!((spans[0].start, spans[0].end), (3, 8));
    }

    // ── lifting standalone tags ──────────────────────────────────────────────
    //
    // The MIRROR of `split_body_tags` in the server's notes/tags.py, case for case.
    // A note lifted differently here than there would change the moment it synced,
    // so these are the cases that file agrees to.

    #[test]
    fn lifts_a_line_that_is_nothing_but_tags() {
        let (standalone, inline, body) = lift_standalone_tags("#todo\nreorganize the homepage");
        assert_eq!(standalone, vec!["todo"]);
        assert!(inline.is_empty());
        assert_eq!(body, "reorganize the homepage");

        let (standalone, _, body) = lift_standalone_tags("needs a tauri app\n#todo");
        assert_eq!(standalone, vec!["todo"]);
        assert_eq!(body, "needs a tauri app");

        let (standalone, _, body) = lift_standalone_tags("#todo #work\nreal text");
        assert_eq!(standalone, vec!["todo", "work"]);
        assert_eq!(body, "real text");
    }

    /// The cases that must come back byte-identical. Getting any of these wrong
    /// destroys somebody's words, which is why the rule is the conservative one:
    /// a trailing tag is ambiguous and the text does not say which kind it is.
    #[test]
    fn leaves_a_tag_that_shares_its_line_with_words() {
        for prose in [
            "remember to call #mom tomorrow",
            "buy milk #grocery",
            "#2024\nreal",
        ] {
            let (standalone, _, body) = lift_standalone_tags(prose);
            assert!(standalone.is_empty(), "{prose}");
            assert_eq!(body, prose, "{prose}");
        }
    }

    #[test]
    fn removing_a_line_leaves_no_hole() {
        let (_, _, body) = lift_standalone_tags("foo\n\n#todo\n\nbar");
        assert_eq!(body, "foo\n\nbar");
    }

    /// A `#tag` in a fence is a shell comment in somebody's snippet. It still becomes
    /// a label — it always has — but the line is never touched.
    #[test]
    fn never_touches_a_fenced_line() {
        let fenced = "code:\n```\n#!/bin/sh\n#deploy\n```\ndone";
        let (standalone, inline, body) = lift_standalone_tags(fenced);
        assert!(standalone.is_empty());
        assert_eq!(inline, vec!["deploy"]);
        assert_eq!(body, fenced);
    }

    /// Lifting would leave a blank card, which is worse than the duplication this
    /// removes. So the note keeps its text and its tags stay derived.
    #[test]
    fn will_not_blank_a_note_that_is_only_tags() {
        let (standalone, inline, body) = lift_standalone_tags("#todo");
        assert!(standalone.is_empty());
        assert_eq!(inline, vec!["todo"]);
        assert_eq!(body, "#todo");
    }

    /// Appearing on its own line does NOT lift a tag also written in a sentence — the
    /// sentence still backs it, so deleting the sentence should still detach it.
    #[test]
    fn a_tag_still_in_prose_stays_derived() {
        let (standalone, inline, body) = lift_standalone_tags("#todo\nremember the #todo list");
        assert!(standalone.is_empty());
        assert_eq!(inline, vec!["todo"]);
        assert_eq!(body, "remember the #todo list");
    }

    #[test]
    fn lifting_an_empty_body_is_a_no_op() {
        let (standalone, inline, body) = lift_standalone_tags("");
        assert!(standalone.is_empty());
        assert!(inline.is_empty());
        assert_eq!(body, "");
    }

    // ── checklist items ─────────────────────────────────────────────────────

    fn item(text: &str, checked: bool, line: u32) -> DerivedItem {
        DerivedItem {
            text: text.to_string(),
            checked,
            line,
        }
    }

    #[test]
    fn items_basic() {
        let body = "shopping\n\n- [ ] milk\n- [x] eggs";
        assert_eq!(
            extract_items(body),
            vec![item("milk", false, 2), item("eggs", true, 3)]
        );
    }

    #[test]
    fn items_may_sit_between_paragraphs() {
        // The whole reason the body owns the list: a table of rows could only ever
        // render after the prose.
        let body = "before\n- [ ] middle\nafter";
        assert_eq!(extract_items(body), vec![item("middle", false, 1)]);
    }

    #[test]
    fn items_accept_star_bullets_and_indentation() {
        // `*` because markdown.ts already takes it for a plain bullet.
        let body = "* [ ] star\n    - [x] indented";
        assert_eq!(
            extract_items(body),
            vec![item("star", false, 0), item("indented", true, 1)]
        );
    }

    #[test]
    fn an_empty_item_is_still_an_item() {
        // What pressing Enter on a list leaves behind.
        assert_eq!(extract_items("- [ ]"), vec![item("", false, 0)]);
        assert_eq!(extract_items("- [ ] "), vec![item("", false, 0)]);
    }

    #[test]
    fn uppercase_x_parses_and_normalises_on_rewrite() {
        assert_eq!(extract_items("- [X] done"), vec![item("done", true, 0)]);
        // Touching it once canonicalises it, and never again.
        assert_eq!(set_item_checked("- [X] done", 0, true), "- [x] done");
    }

    #[test]
    fn checking_preserves_indent_bullet_and_text() {
        assert_eq!(set_item_checked("  * [ ] milk", 0, true), "  * [x] milk");
        assert_eq!(set_item_checked("- [x] milk", 0, false), "- [ ] milk");
    }

    #[test]
    fn checking_addresses_items_not_lines() {
        let body = "note\n- [ ] a\nprose\n- [ ] b";
        assert_eq!(
            set_item_checked(body, 1, true),
            "note\n- [ ] a\nprose\n- [x] b"
        );
    }

    #[test]
    fn append_spaces_like_the_exporter() {
        // Prose then a blank line then the list — byte-for-byte what
        // import_export.py:_note_markdown writes, which is the layout the server
        // migration folds rows into.
        assert_eq!(append_item("a note", "milk", false), "a note\n\n- [ ] milk");
        // Nothing between consecutive items.
        let one = "a note\n\n- [ ] milk";
        assert_eq!(
            append_item(one, "eggs", false),
            format!("{one}\n- [ ] eggs")
        );
        // A list-only note starts at the first line.
        assert_eq!(append_item("", "milk", false), "- [ ] milk");
        assert_eq!(append_item("\n\n", "milk", false), "- [ ] milk");
        // Carries state, which is what the two migrations need of it.
        assert_eq!(append_item("", "done", true), "- [x] done");
    }

    #[test]
    fn strip_marker_names_a_list_only_note() {
        assert_eq!(strip_marker("- [x] milk"), "milk");
        assert_eq!(strip_marker("just prose"), "just prose");
    }

    #[test]
    fn render_item_is_what_extract_reads_back() {
        assert_eq!(render_item("milk", false), "- [ ] milk");
        assert_eq!(render_item("done", true), "- [x] done");
        // An empty item has no trailing space, so a round trip does not grow it.
        assert_eq!(render_item("", false), "- [ ]");
        let line = render_item("milk", true);
        assert_eq!(extract_items(&line), vec![item("milk", true, 0)]);
    }

    #[test]
    fn items_carry_the_line_they_sit_on() {
        let found = extract_items("a\n- [ ] x\nb\n- [x] y");
        assert_eq!(found.iter().map(|i| i.line).collect::<Vec<_>>(), vec![1, 3]);
    }

    #[test]
    fn a_stale_index_does_nothing() {
        // The index comes from a UI that may be a moment behind the store. A tap
        // that arrives late should be inert, not fatal.
        let body = "- [ ] only";
        assert_eq!(set_item_checked(body, 7, true), body);
    }

    #[test]
    fn a_plain_body_is_returned_byte_identical() {
        let body = "just prose\nwith two lines";
        assert_eq!(set_item_checked(body, 0, true), body);
    }

    #[test]
    fn round_trip_is_stable() {
        let body = "- [ ] a\n- [x] b\n- [ ] c";
        let items = extract_items(body);
        // Ticking and unticking returns the original bytes.
        let touched = set_item_checked(&set_item_checked(body, 0, true), 0, false);
        assert_eq!(touched, body);
        assert_eq!(extract_items(&touched), items);
    }

    // ── the shared fixture ───────────────────────────────────────────────────
    //
    // core/testdata/grammar.json is the one set of cases the server (pytest), the web
    // (vitest) and this file all run. The cases above stay as this file's own
    // reasoning; these are the ones the other languages have agreed to.

    fn fixture() -> serde_json::Value {
        serde_json::from_str(include_str!("../../testdata/grammar.json"))
            .expect("grammar.json parses")
    }

    fn strings(v: &serde_json::Value) -> Vec<String> {
        v.as_array()
            .expect("an array")
            .iter()
            .map(|s| s.as_str().expect("a string").to_string())
            .collect()
    }

    #[test]
    fn fixture_task_lines() {
        for case in fixture()["task_lines"].as_array().unwrap() {
            let line = case["line"].as_str().unwrap();
            let got: Vec<(String, bool)> = extract_items(line)
                .into_iter()
                .map(|i| (i.text, i.checked))
                .collect();
            let want: Vec<(String, bool)> = match &case["item"] {
                serde_json::Value::Null => Vec::new(),
                item => vec![(
                    item["text"].as_str().unwrap().to_string(),
                    item["checked"].as_bool().unwrap(),
                )],
            };
            assert_eq!(got, want, "line {line:?}");
        }
    }

    #[test]
    fn fixture_rendered_items() {
        for case in fixture()["rendered_items"].as_array().unwrap() {
            let text = case["text"].as_str().unwrap();
            let checked = case["checked"].as_bool().unwrap();
            assert_eq!(render_item(text, checked), case["line"].as_str().unwrap());
        }
    }

    #[test]
    fn fixture_tags() {
        for case in fixture()["tags"].as_array().unwrap() {
            let body = case["body"].as_str().unwrap();
            assert_eq!(tag_names(body), strings(&case["tags"]), "body {body:?}");
        }
    }

    #[test]
    fn fixture_display_titles() {
        for case in fixture()["display_titles"].as_array().unwrap() {
            let body = case["body"].as_str().unwrap();
            let want = case["title"].as_str().unwrap();
            assert_eq!(display_title(body), want, "body {body:?}");
        }
    }

    #[test]
    fn fixture_lifts() {
        for case in fixture()["lifts"].as_array().unwrap() {
            let body = case["body"].as_str().unwrap();
            let (standalone, inline, lifted) = lift_standalone_tags(body);
            assert_eq!(
                standalone,
                strings(&case["standalone"]),
                "standalone, body {body:?}"
            );
            assert_eq!(inline, strings(&case["inline"]), "inline, body {body:?}");
            assert_eq!(
                lifted,
                case["lifted"].as_str().unwrap(),
                "lifted, body {body:?}"
            );
        }
    }
}