wordcraft/crates/engine/src/cmd/citations.rs

//! References: citations and bibliography (APA, MLA, Chicago, IEEE, Harvard, ISO 690, Turabian,
//! GB/T 7714, GOST, SIST02), source manager, index, table of figures, cross-references, table
//! of authorities, footnote options.

use serde_json::{Value, json};
use wordcraft_doc::collate::{SortKey, heading_letter, sort_key};
use wordcraft_doc::para::InlineObject;
use wordcraft_doc::props::{CharProps, TabAlign, TabLeader, TabStop};
use wordcraft_doc::{Block, Document, Paragraph, Path, Pos, Source, StoryRef};

use super::references::FieldCode;
use super::{delete_selection, sel_result};
use crate::{CmdError, CmdResult, CommandSpec, Selection, Session, p};

/// How a style orders its bibliography (and, for numbered styles, numbers its citations).
#[derive(Clone, Copy, Debug, PartialEq, Eq)]
pub enum Order {
    /// Alphabetical by the first author's surname (the title when there is no author).
    Author,
    /// Alphabetical by title.
    Title,
    /// In order of first citation in the document.
    Cited,
}

/// How one author's name is written in a bibliography entry.
#[derive(Clone, Copy, Debug, PartialEq, Eq)]
pub enum Name {
    /// Rivera, Alex
    LastFirst,
    /// Alex Rivera
    FirstLast,
    /// Rivera, A.
    LastInitials,
    /// A. Rivera
    InitialsLast,
    /// Rivera A.
    LastSpaceInitials,
    /// RIVERA, Alex
    UpperLastFirst,
    /// RIVERA A
    UpperLastBareInitials,
}

/// The author list of a bibliography entry.
#[derive(Clone, Copy, Debug)]
pub struct Names {
    /// The first author's form, and every later author's.
    pub first: Name,
    pub rest: Name,
    /// Between names; between exactly two; before the last of three or more.
    pub sep: &'static str,
    pub two: &'static str,
    pub last: &'static str,
    /// List at most this many authors, then `et_al` (0: list them all).
    pub max: usize,
    pub et_al: &'static str,
}

/// The in-text citation.
#[derive(Clone, Copy, Debug)]
pub enum Cite {
    /// `(Who<year_sep>Year<pages_sep>Pages)`.
    AuthorDate { year_sep: &'static str, pages_sep: &'static str },
    /// `(Who Pages)`.
    AuthorPage,
    /// `<open>n<close>`; pages follow `pages_sep` inside the brackets (left out when `None`).
    Number { open: &'static str, close: &'static str, pages_sep: Option<&'static str> },
}

/// A citation and bibliography style, as data. Entry templates fill `{n}`, `{names}`, `{year}`,
/// `{title}`, `{city}`, `{publisher}`, `{pub}` (city: publisher), `{journal}`, `{volume}`,
/// `{pages}`, `{url}` and `{type}` (GB/T 7714 document type code); text between `⟦` and `⟧` is
/// left out when a field in it is empty, and a full stop never doubles up.
#[derive(Debug)]
pub struct CiteStyle {
    /// The name shown in the style picker and stored in the session.
    pub name: &'static str,
    /// Lowercase letter/digit prefixes of Word's style names and style-sheet file names that mean
    /// this style (`gostname` matches "GOST - Name Sort"); the longest match wins.
    pub aliases: &'static [&'static str],
    pub cite: Cite,
    /// Joins two authors in a citation ("&" or "and").
    pub and: &'static str,
    /// Cite this many authors or more as "First et al.".
    pub et_al_from: usize,
    pub order: Order,
    pub names: Names,
    /// Entry template for a book (or anything without a journal), and for a journal article.
    pub book: &'static str,
    pub article: &'static str,
    /// Default title of the bibliography.
    pub bib_title: &'static str,
}

impl CiteStyle {
    /// Numbered styles cite `[1]`/`(1)` and list entries without a hanging indent.
    pub fn numbered(&self) -> bool {
        matches!(self.cite, Cite::Number { .. })
    }
}

const fn names(first: Name, rest: Name, sep: &'static str, two: &'static str, last: &'static str) -> Names {
    Names { first, rest, sep, two, last, max: 0, et_al: "" }
}

const AUTHOR_DATE: Cite = Cite::AuthorDate { year_sep: " ", pages_sep: ", p. " };

/// Every style, in picker order (alphabetical). Formats follow the published style guides'
/// rules for the source types WordCraft records, written in our own words.
pub static STYLES: [CiteStyle; 12] = [
    CiteStyle {
        name: "APA",
        aliases: &["apa"],
        cite: Cite::AuthorDate { year_sep: ", ", pages_sep: ", p. " },
        and: "&",
        et_al_from: 3,
        order: Order::Author,
        names: names(Name::LastInitials, Name::LastInitials, ", ", ", & ", ", & "),
        book: "⟦{names} ⟧({year}). {title}.⟦ {publisher}.⟧⟦ {url}⟧",
        article: "⟦{names} ⟧({year}). {title}. {journal}⟦, {volume}⟧⟦, {pages}⟧.⟦ {url}⟧",
        bib_title: "References",
    },
    CiteStyle {
        name: "Chicago",
        aliases: &["chicago"],
        cite: Cite::AuthorDate { year_sep: " ", pages_sep: ", " },
        and: "and",
        et_al_from: 3,
        order: Order::Author,
        names: names(Name::LastFirst, Name::FirstLast, ", ", ", and ", ", and "),
        book: "⟦{names}. ⟧{year}. {title}.⟦ {pub}.⟧⟦ {url}⟧",
        article: "⟦{names}. ⟧{year}. {title}. {journal}⟦, {volume}⟧⟦, {pages}⟧.⟦ {url}⟧",
        bib_title: "References",
    },
    // GB/T 7714 sequence-coding: numbered by first citation, surnames in capitals with bare
    // initials, at most three authors, a document type code after the title.
    CiteStyle {
        name: "GB/T 7714",
        aliases: &["gb"],
        cite: Cite::Number { open: "[", close: "]", pages_sep: Some(", ") },
        and: "and",
        et_al_from: 3,
        order: Order::Cited,
        names: Names { max: 3, et_al: ", et al.", ..names(Name::UpperLastBareInitials, Name::UpperLastBareInitials, ", ", ", ", ", ") },
        book: "[{n}] ⟦{names}. ⟧{title}[{type}].⟦ {pub},⟧ {year}.⟦ {url}.⟧",
        article: "[{n}] ⟦{names}. ⟧{title}[{type}]. {journal}, {year}⟦, {volume}⟧⟦: {pages}⟧.⟦ {url}.⟧",
        bib_title: "References",
    },
    // GOST 7.1: a numbered list in alphabetical order (by author or by title); areas separated
    // by " – ", the container after "//".
    CiteStyle {
        name: "GOST (name sort)",
        aliases: &["gost", "gostname"],
        cite: Cite::Number { open: "[", close: "]", pages_sep: Some(", p. ") },
        and: "and",
        et_al_from: 3,
        order: Order::Author,
        names: names(Name::LastSpaceInitials, Name::LastSpaceInitials, ", ", ", ", ", "),
        book: GOST_BOOK,
        article: GOST_ARTICLE,
        bib_title: "References",
    },
    CiteStyle {
        name: "GOST (title sort)",
        aliases: &["gosttitle"],
        cite: Cite::Number { open: "[", close: "]", pages_sep: Some(", p. ") },
        and: "and",
        et_al_from: 3,
        order: Order::Title,
        names: names(Name::LastSpaceInitials, Name::LastSpaceInitials, ", ", ", ", ", "),
        book: GOST_BOOK,
        article: GOST_ARTICLE,
        bib_title: "References",
    },
    // Harvard as taught by Anglia Ruskin: "Surname, I., Year. Title. Place: Publisher."
    CiteStyle {
        name: "Harvard (Anglia)",
        aliases: &["harvard"],
        cite: Cite::AuthorDate { year_sep: ", ", pages_sep: ", p. " },
        and: "and",
        et_al_from: 4,
        order: Order::Author,
        names: names(Name::LastInitials, Name::LastInitials, ", ", " and ", " and "),
        book: "⟦{names}, ⟧{year}. {title}.⟦ {pub}.⟧⟦ Available at: <{url}>.⟧",
        article: "⟦{names}, ⟧{year}. {title}. {journal}⟦, {volume}⟧⟦, pp. {pages}⟧.⟦ Available at: <{url}>.⟧",
        bib_title: "References",
    },
    CiteStyle {
        name: "IEEE",
        aliases: &["ieee"],
        cite: Cite::Number { open: "[", close: "]", pages_sep: None },
        and: "and",
        et_al_from: 3,
        order: Order::Cited,
        names: names(Name::InitialsLast, Name::InitialsLast, ", ", ", ", ", "),
        book: "[{n}] ⟦{names}, ⟧“{title},”⟦ {pub},⟧ {year}.⟦ {url}⟧",
        article: "[{n}] ⟦{names}, ⟧“{title},” {journal}⟦, {volume}⟧⟦, {pages}⟧, {year}.⟦ {url}⟧",
        bib_title: "References",
    },
    // ISO 690 first element and date: surnames in capitals, the year after the names.
    CiteStyle {
        name: "ISO 690 (author-date)",
        aliases: &["iso690"],
        cite: AUTHOR_DATE,
        and: "and",
        et_al_from: 4,
        order: Order::Author,
        names: names(Name::UpperLastFirst, Name::UpperLastFirst, ", ", " and ", " and "),
        book: "⟦{names}, ⟧{year}. {title}.⟦ {pub}.⟧⟦ Available from: {url}⟧",
        article: "⟦{names}, ⟧{year}. {title}. {journal}⟦, {volume}⟧⟦, {pages}⟧.⟦ Available from: {url}⟧",
        bib_title: "References",
    },
    // ISO 690 numeric reference: numbered by first citation, the year with the imprint.
    CiteStyle {
        name: "ISO 690 (numerical)",
        aliases: &["iso690num", "iso690nmerical"],
        cite: Cite::Number { open: "(", close: ")", pages_sep: Some(", p. ") },
        and: "and",
        et_al_from: 4,
        order: Order::Cited,
        names: names(Name::UpperLastFirst, Name::UpperLastFirst, ", ", " and ", " and "),
        book: "{n}. ⟦{names}. ⟧{title}.⟦ {pub},⟧ {year}.⟦ Available from: {url}⟧",
        article: "{n}. ⟦{names}. ⟧{title}. {journal}. {year}⟦, {volume}⟧⟦, {pages}⟧.⟦ Available from: {url}⟧",
        bib_title: "References",
    },
    CiteStyle {
        name: "MLA",
        aliases: &["mla"],
        cite: Cite::AuthorPage,
        and: "and",
        et_al_from: 3,
        order: Order::Author,
        names: Names { max: 1, et_al: ", et al.", ..names(Name::LastFirst, Name::LastFirst, ", ", ", ", ", ") },
        book: "⟦{names}. ⟧“{title}.”⟦ {publisher},⟧ {year}.⟦ {url}⟧",
        article: "⟦{names}. ⟧“{title}.” {journal}⟦, {volume}⟧⟦, {pages}⟧, {year}.⟦ {url}⟧",
        bib_title: "Works Cited",
    },
    // SIST 02: numbered by first citation, full names separated by semicolons.
    CiteStyle {
        name: "SIST02",
        aliases: &["sist"],
        cite: Cite::Number { open: "(", close: ")", pages_sep: Some(", p. ") },
        and: "and",
        et_al_from: 3,
        order: Order::Cited,
        names: names(Name::LastFirst, Name::LastFirst, "; ", "; ", "; "),
        book: "{n}) ⟦{names}. ⟧{title}.⟦ {city},⟧⟦ {publisher},⟧ {year}.⟦ {url}⟧",
        article: "{n}) ⟦{names}. ⟧{title}. {journal}. {year}⟦, vol. {volume}⟧⟦, p. {pages}⟧.⟦ {url}⟧",
        bib_title: "References",
    },
    // Turabian: author-date citations, a bibliography with the year in the imprint.
    CiteStyle {
        name: "Turabian",
        aliases: &["turabian"],
        cite: Cite::AuthorDate { year_sep: " ", pages_sep: ", " },
        and: "and",
        et_al_from: 4,
        order: Order::Author,
        names: names(Name::LastFirst, Name::FirstLast, ", ", ", and ", ", and "),
        book: "⟦{names}. ⟧{title}.⟦ {pub},⟧ {year}.⟦ {url}.⟧",
        article: "⟦{names}. ⟧“{title}.” {journal}⟦ {volume}⟧ ({year})⟦: {pages}⟧.⟦ {url}.⟧",
        bib_title: "Bibliography",
    },
];

const GOST_BOOK: &str = "{n}. ⟦{names} ⟧{title}. –⟦ {city} :⟧⟦ {publisher},⟧ {year}.⟦ – URL: {url}.⟧";
const GOST_ARTICLE: &str = "{n}. ⟦{names} ⟧{title} // {journal}. – {year}.⟦ – Vol. {volume}.⟧⟦ – P. {pages}.⟧⟦ – URL: {url}.⟧";

/// The style called `name`: one of ours (any case), or Word's style name or style-sheet file
/// name (`Harvard - Anglia`, `\GostTitle.XSL`).
pub fn find_style(name: &str) -> Option<&'static CiteStyle> {
    if let Some(st) = STYLES.iter().find(|x| x.name.eq_ignore_ascii_case(name.trim())) {
        return Some(st);
    }
    let key: String = name.chars().filter(char::is_ascii_alphanumeric).map(|c| c.to_ascii_lowercase()).collect();
    let key = key.strip_suffix("xsl").unwrap_or(&key);
    STYLES
        .iter()
        .flat_map(|st| st.aliases.iter().map(move |a| (st, *a)))
        .filter(|(_, a)| key.starts_with(a))
        .max_by_key(|(_, a)| a.len())
        .map(|(st, _)| st)
}

/// The session's style; APA when the name is unknown.
pub fn style(name: &str) -> &'static CiteStyle {
    find_style(name).unwrap_or(&STYLES[0])
}

pub fn specs() -> Vec<CommandSpec> {
    vec![
        CommandSpec::new("references.sources", "Manage Sources", "References › Citations & Bibliography", sources)
            .params(r#"{"add"?: Source, "remove"?: tag, "import"?: path (a Word bibliography file, `b:Sources` / `Sources.xml`)} → list"#),
        CommandSpec::new("references.citation", "Insert Citation", "References › Citations & Bibliography", citation).params(r#"{"tag"?: string, "source"?: Source (added if new), "pages"?: string}"#),
        CommandSpec::new("references.citationStyle", "Style", "References › Citations & Bibliography", |s, v| {
            let st = p::str(v, "style").unwrap_or("APA");
            let st = find_style(st).ok_or_else(|| {
                CmdError::Params(format!("style must be one of {:?}", STYLES.iter().map(|x| x.name).collect::<Vec<_>>()))
            })?;
            s.bib_style = st.name.to_string();
            // Recorded in the document too, so a save keeps it.
            s.doc.settings.bib_style = Some(st.name.to_string());
            update_citations(s)?;
            Ok(json!({"style": s.bib_style}))
        })
        .params(
            r#"{"style": "APA|Chicago|GB/T 7714|GOST (name sort)|GOST (title sort)|Harvard (Anglia)|IEEE|ISO 690 (author-date)|ISO 690 (numerical)|MLA|SIST02|Turabian" (Word's style names work too)}"#,
        ),
        CommandSpec::new("references.bibliography", "Bibliography", "References › Citations & Bibliography", |s, v| {
            let title = p::str(v, "title").unwrap_or(style(&s.bib_style).bib_title).to_string();
            generated_list(s, "BIBLIOGRAPHY", &title)?;
            sel_result(s)
        })
        .params(r#"{"title"?: string}"#),
        CommandSpec::new("references.tableOfFigures", "Insert Table of Figures", "References › Captions", table_of_figures).params(
            r#"{"label"?: "Figure|Table|Equation" or a caption label of your own, "includeLabel"?: bool, "pageNumbers"?: bool, "rightAlign"?: bool, "tabLeader"?: "dot|hyphen|underscore|none"}"#,
        ),
        CommandSpec::new("references.updateFigures", "Update Table", "References › Captions", |s, _| {
            update_generated(s)?;
            sel_result(s)
        }),
        CommandSpec::new("references.markCitation", "Mark Citation", "References › Table of Authorities", |s, v| {
            let entry = p::str(v, "entry").map(str::to_string).unwrap_or_else(|| s.selected_text().trim().to_string());
            if entry.is_empty() {
                return Err(CmdError::Params("select a citation or give `entry`".into()));
            }
            let (_, b) = s.sel.ordered();
            s.doc.insert_object(&b, InlineObject::Field { instr: format!("TA \\l \"{}\"", entry.replace('"', "")), result: String::new(), locked: false, dirty: false, code: false, form: String::new() }, &CharProps { hidden: Some(true), ..Default::default() })?;
            Ok(json!({"entry": entry}))
        })
        .params(r#"{"entry"?: string}"#)
        .key(super::non_mac("Alt+Shift+I")),
        CommandSpec::new("references.tableOfAuthorities", "Insert Table of Authorities", "References › Table of Authorities", |s, _| {
            generated_list(s, "TOA", "Table of Authorities")?;
            sel_result(s)
        }),
        CommandSpec::new("references.updateAuthorities", "Update Table", "References › Table of Authorities", |s, _| {
            update_generated(s)?;
            sel_result(s)
        }),
        CommandSpec::new("insert.crossReference", "Cross-reference", "Insert › Links", cross_ref)
            .params(r#"{"to": "heading|bookmark|figure|table|equation|label:<Name> (a caption label of your own)", "target": string (text / name / number), "show"?: "text|page|number|aboveBelow"} or {} to list targets"#),
        CommandSpec::new("references.noteOptions", "Footnote and Endnote", "References › Footnotes", super::references::note_options).params(
            r#"{"kind"?: "footnote|endnote", "pos"?: "pageBottom|beneathText" (footnotes) | "sectEnd|docEnd" (endnotes), "numFmt"?: "decimal|lowerRoman|upperRoman|lowerLetter|upperLetter|…", "numStart"?: 1-32767, "numRestart"?: "continuous|eachSect|eachPage" (each page: footnotes only), "scope"?: "document|section", "footnoteFormat"?: …, "endnoteFormat"?: …} — no settings: the options in effect at the caret"#,
        ),
        CommandSpec::new("references.researcher", "Researcher", "References › Research", |s, _| {
            s.status = "Researcher needs an online service; WordCraft keeps your documents offline.".into();
            Ok(json!({"available": false}))
        })
        .pure(),
    ]
}

fn authors(a: &str) -> Vec<(String, String)> {
    a.split(';')
        .map(str::trim)
        .filter(|x| !x.is_empty())
        .map(|x| match x.split_once(',') {
            Some((last, first)) => (last.trim().to_string(), first.trim().to_string()),
            None => match x.rsplit_once(' ') {
                Some((first, last)) => (last.trim().to_string(), first.trim().to_string()),
                None => (x.to_string(), String::new()),
            },
        })
        .collect()
}

/// "Alex Mei" → "A. M." (`dots`), or "A M" without them.
fn initials(first: &str, dots: bool) -> String {
    let sep = if dots { ". " } else { " " };
    let mut out = first.split_whitespace().filter_map(|w| w.chars().next()).map(String::from).collect::<Vec<_>>().join(sep);
    if dots && !out.is_empty() {
        out.push('.');
    }
    out
}

fn name(form: Name, last: &str, first: &str) -> String {
    let join = |a: String, sep: &str, b: String| if a.is_empty() || b.is_empty() { format!("{a}{b}") } else { format!("{a}{sep}{b}") };
    match form {
        Name::LastFirst => join(last.to_string(), ", ", first.to_string()),
        Name::FirstLast => join(first.to_string(), " ", last.to_string()),
        Name::LastInitials => join(last.to_string(), ", ", initials(first, true)),
        Name::InitialsLast => join(initials(first, true), " ", last.to_string()),
        Name::LastSpaceInitials => join(last.to_string(), " ", initials(first, true)),
        Name::UpperLastFirst => join(last.to_uppercase(), ", ", first.to_string()),
        Name::UpperLastBareInitials => join(last.to_uppercase(), " ", initials(first, false)),
    }
}

fn name_list(au: &[(String, String)], form: &Names) -> String {
    let shown = if form.max > 0 { au.get(..form.max).unwrap_or(au) } else { au };
    let parts: Vec<String> = shown.iter().enumerate().map(|(i, (l, f))| name(if i == 0 { form.first } else { form.rest }, l, f)).collect();
    let mut out = match parts.as_slice() {
        [] => String::new(),
        [one] => one.clone(),
        [a, b] => format!("{a}{}{b}", form.two),
        [init @ .., z] => format!("{}{}{z}", init.join(form.sep), form.last),
    };
    if shown.len() < au.len() {
        out.push_str(form.et_al);
    }
    out
}

/// Append `s`, dropping its leading full stop when `out` already ends with one ("M." + ". ").
fn push_text(out: &mut String, s: &str) {
    match s.strip_prefix('.') {
        Some(rest) if out.ends_with('.') => out.push_str(rest),
        _ => out.push_str(s),
    }
}

/// Fill one template segment; `None` when `optional` and a field in it is empty.
fn fill_segment(seg: &str, field: &dyn Fn(&str) -> String, optional: bool) -> Option<String> {
    let mut out = String::new();
    let mut rest = seg;
    while let Some((before, after)) = rest.split_once('{') {
        push_text(&mut out, before);
        let Some((key, after)) = after.split_once('}') else {
            push_text(&mut out, after);
            return Some(out);
        };
        let value = field(key);
        if value.is_empty() && optional {
            return None;
        }
        push_text(&mut out, &value);
        rest = after;
    }
    push_text(&mut out, rest);
    Some(out)
}

/// Fill an entry template (see [`CiteStyle`]).
fn render(template: &str, field: &dyn Fn(&str) -> String) -> String {
    let mut out = String::new();
    let mut rest = template;
    while !rest.is_empty() {
        let (seg, optional, after) = match rest.strip_prefix('⟦') {
            Some(group) => match group.split_once('⟧') {
                Some((g, after)) => (g, true, after),
                None => (group, true, ""),
            },
            None => match rest.split_once('⟦') {
                Some((plain, _)) => (plain, false, rest.get(plain.len()..).unwrap_or("")),
                None => (rest, false, ""),
            },
        };
        if let Some(text) = fill_segment(seg, field, optional) {
            push_text(&mut out, &text);
        }
        rest = after;
    }
    out
}

/// In-text citation for a style; `n` is the source's number in numbered styles.
pub fn cite(src: &Source, style_name: &str, n: usize, pages: &str) -> String {
    let st = style(style_name);
    let au = authors(&src.author);
    let lasts: Vec<&str> = au.iter().map(|a| a.0.as_str()).collect();
    let who = match lasts.as_slice() {
        [] => src.title.clone(),
        [one] => one.to_string(),
        [first, ..] if lasts.len() >= st.et_al_from.max(2) => format!("{first} et al."),
        [a, b] => format!("{a} {} {b}", st.and),
        [init @ .., z] => format!("{} {} {z}", init.join(", "), st.and),
    };
    match st.cite {
        Cite::AuthorPage if pages.is_empty() => format!("({who})"),
        Cite::AuthorPage => format!("({who} {pages})"),
        Cite::AuthorDate { year_sep, .. } if pages.is_empty() => format!("({who}{year_sep}{})", src.year),
        Cite::AuthorDate { year_sep, pages_sep } => format!("({who}{year_sep}{}{pages_sep}{pages})", src.year),
        Cite::Number { open, close, pages_sep: Some(sep) } if !pages.is_empty() => format!("{open}{n}{sep}{pages}{close}"),
        Cite::Number { open, close, .. } => format!("{open}{n}{close}"),
    }
}

/// GB/T 7714 document type code: M book, J journal article, R report, EB/OL web page, Z other;
/// "/OL" marks a source read online.
fn gb_type(src: &Source) -> String {
    let code = match src.kind.to_ascii_lowercase().as_str() {
        "website" | "web" => return "EB/OL".into(),
        _ if !src.journal.is_empty() => "J",
        "book" | "" => "M",
        "article" => "J",
        "report" => "R",
        _ => "Z",
    };
    if src.url.is_empty() { code.into() } else { format!("{code}/OL") }
}

/// Bibliography entry for a style; `n` is the source's number in numbered styles.
pub fn entry(src: &Source, style_name: &str, n: usize) -> String {
    let st = style(style_name);
    let au = authors(&src.author);
    let field = |key: &str| -> String {
        match key {
            "n" => n.to_string(),
            "names" => name_list(&au, &st.names),
            "year" if src.year.is_empty() => "n.d.".into(),
            "year" => src.year.clone(),
            "title" => src.title.clone(),
            "city" => src.city.clone(),
            "publisher" => src.publisher.clone(),
            "pub" => [src.city.as_str(), src.publisher.as_str()].into_iter().filter(|x| !x.is_empty()).collect::<Vec<_>>().join(": "),
            "journal" => src.journal.clone(),
            "volume" => src.volume.clone(),
            "pages" => src.pages.clone(),
            "url" => src.url.clone(),
            "type" => gb_type(src),
            _ => String::new(),
        }
    };
    render(if src.journal.is_empty() { st.book } else { st.article }, &field)
}

/// The tag a new source gets, as Word makes it: the first three letters of the first author's
/// name and the last two digits of the year (`Smi20`), with `1`, `2`… added while it is taken.
fn auto_tag(doc: &Document, src: &Source) -> String {
    let base = auto_tag_base(src);
    let taken = |t: &str| doc.sources.iter().any(|x| x.tag == t);
    if !taken(&base) {
        return base;
    }
    (1u32..).map(|n| format!("{base}{n}")).find(|t| !taken(t)).unwrap_or(base)
}

/// The author of a source for matching untagged entries: the editor when there is no author.
#[cfg(not(target_arch = "wasm32"))]
fn by_author(x: &Source) -> &str {
    if x.author.is_empty() { x.editor.as_str() } else { x.author.as_str() }
}

/// The stem of a new source's tag, before `1`, `2`… are added while it is taken: the first three
/// alphanumerics of the first author's name (nothing of the author's, `Src`) and the last two
/// digits of the year.
fn auto_tag_base(src: &Source) -> String {
    let name: String =
        authors(&src.author).first().map(|x| x.0.clone()).unwrap_or_default().chars().filter(|c| c.is_alphanumeric()).take(3).collect();
    let name = if name.is_empty() { "Src".to_string() } else { name };
    let digits: Vec<char> = src.year.chars().filter(char::is_ascii_digit).collect();
    let year: String = digits.get(digits.len().saturating_sub(2)..).unwrap_or_default().iter().collect();
    format!("{name}{year}")
}

/// Manage Sources: applies `add`, `remove` and `import` (in that order) and returns the
/// document's sources. `import` reads a Word bibliography file (`b:Sources`, #2640).
fn sources(s: &mut Session, v: &Value) -> CmdResult {
    if let Some(a) = v.get("add") {
        let mut src: Source = serde_json::from_value(a.clone()).map_err(|e| CmdError::Params(e.to_string()))?;
        // A new source's tag is made from its author and year; an edit keeps the tag it has.
        if src.tag.is_empty() {
            src.tag = auto_tag(&s.doc, &src);
        }
        // Edit Source keeps the source's place in the list and refreshes its citations.
        match s.doc.sources.iter().position(|x| x.tag == src.tag) {
            Some(i) => {
                if let Some(x) = s.doc.sources.get_mut(i) {
                    *x = src;
                }
                update_citations(s)?;
            }
            None => s.doc.sources.push(src),
        }
        s.touch();
    }
    if let Some(t) = p::str(v, "remove") {
        s.doc.sources.retain(|x| x.tag != t);
    }
    if let Some(path) = p::str(v, "import") {
        import_sources(s, path)?;
    }
    Ok(serde_json::to_value(&s.doc.sources).unwrap_or(Value::Null))
}

/// The largest bibliography sources file imported from disk.
#[cfg(not(target_arch = "wasm32"))]
const MAX_SOURCES_FILE: u64 = 32 << 20;

/// Import the sources of a Word bibliography file (`b:Sources`: Word's own `Sources.xml`) into
/// this document (#2640). A tag already here is left alone — an import never overwrites an
/// edited source; a source with no tag is matched to an existing one by title and author, and
/// otherwise gets a tag the way a new source does.
fn import_sources(s: &mut Session, path: &str) -> Result<(), CmdError> {
    #[cfg(target_arch = "wasm32")]
    {
        let _ = (s, path);
        return Err(CmdError::Failed("reading files by path isn't available on the web".into()));
    }
    #[cfg(not(target_arch = "wasm32"))]
    {
        // Through `io`, which keeps automation inside its read root (#769).
        let bytes = crate::io::read_file_capped(path, MAX_SOURCES_FILE, "sources file").map_err(CmdError::Failed)?;
        let file = wordcraft_docx::read_sources(&bytes).map_err(|e| CmdError::Failed(e.to_string()))?;
        let mut added = false;
        // The tags and (title, author-or-editor) keys already here, once: an import of thousands
        // of untagged sources must not rescan the list per source. Tags follow `auto_tag`: the
        // stem, then 1, 2… while taken.
        let mut tags: std::collections::HashSet<String> = s.doc.sources.iter().map(|x| x.tag.clone()).collect();
        let mut keys: std::collections::HashSet<(String, String)> =
            s.doc.sources.iter().map(|x| (x.title.clone(), by_author(x).to_string())).collect();
        let mut next: std::collections::HashMap<String, u32> = std::collections::HashMap::new();
        for mut src in file.sources {
            if s.doc.sources.len() >= wordcraft_docx::MAX_SOURCES {
                break;
            }
            if src.tag.is_empty() {
                // A source with no tag is matched to one already here by its title and author (or
                // editor), as repeated parts of a document are; otherwise it gets a tag the way a
                // new source does.
                if keys.contains(&(src.title.clone(), by_author(&src).to_string())) {
                    continue;
                }
                let base = auto_tag_base(&src);
                let n = next.entry(base.clone()).or_insert(0);
                let mut tag = if *n == 0 { base.clone() } else { format!("{base}{}", *n) };
                while tags.contains(&tag) {
                    *n += 1;
                    tag = format!("{base}{}", *n);
                }
                *n = n.saturating_add(1);
                src.tag = tag;
            }
            if !tags.insert(src.tag.clone()) {
                continue;
            }
            keys.insert((src.title.clone(), by_author(&src).to_string()));
            s.doc.sources.push(src);
            added = true;
        }
        if added {
            s.touch();
        }
        Ok(())
    }
}

fn citation(s: &mut Session, v: &Value) -> CmdResult {
    if v.get("source").is_some() {
        sources(s, &json!({"add": v.get("source")}))?;
    }
    let tag = match p::str(v, "tag") {
        Some(t) => t.to_string(),
        None => match v.get("source") {
            Some(_) => s.doc.sources.last().map(|x| x.tag.clone()).unwrap_or_default(),
            None => return Err(CmdError::Params("give `tag` or `source`".into())),
        },
    };
    let src = s.doc.sources.iter().find(|x| x.tag == tag).cloned().ok_or_else(|| CmdError::Params(format!("no source `{tag}`")))?;
    let pages = p::str(v, "pages").unwrap_or("").to_string();
    let st = style(&s.bib_style);
    let n = number_of(&s.doc, &bib_order(&s.doc, st, super::default_lang(s)), &tag);
    let props = s.typing_props();
    let at = delete_selection(s)?;
    let instr = if pages.is_empty() { format!("CITATION {tag}") } else { format!("CITATION {tag} \\p {pages}") };
    let end = s.doc.insert_object(
        &at,
        InlineObject::Field { instr, result: cite(&src, st.name, n, &pages), locked: false, dirty: false, code: false, form: String::new() },
        &props,
    )?;
    s.sel = Selection::caret(end);
    // A new citation can renumber the others (numbered styles count in order of first citation).
    if st.numbered() {
        refresh_citations(s)?;
    }
    Ok(json!({"tag": tag}))
}

/// The tag and `\p` pages of a CITATION field instruction.
fn citation_field(instr: &str) -> Option<(&str, &str)> {
    let mut words = instr.split_whitespace();
    if words.next()? != "CITATION" {
        return None;
    }
    let tag = words.next()?;
    let pages = instr.split("\\p").nth(1).map(str::trim).unwrap_or("");
    Some((tag, pages))
}

/// Tags cited in the body, in order of first citation.
fn cited_tags(doc: &Document) -> Vec<String> {
    let mut seen = std::collections::HashSet::new();
    let mut out = Vec::new();
    for path in doc.para_paths(StoryRef::Body) {
        let Some(p) = doc.para(StoryRef::Body, &path) else { continue };
        for o in &p.objects {
            if let InlineObject::Field { instr, .. } = o
                && let Some((tag, _)) = citation_field(instr)
                && seen.insert(tag.to_string())
            {
                out.push(tag.to_string());
            }
        }
    }
    out
}

fn author_key(src: &Source) -> String {
    authors(&src.author).first().map(|a| a.0.to_lowercase()).unwrap_or_else(|| src.title.to_lowercase())
}

/// Indices into `doc.sources` in bibliography order: the cited sources (every source while
/// nothing is cited), sorted or in order of first citation as the style asks.
fn bib_order(doc: &Document, st: &CiteStyle, lang: Option<String>) -> Vec<usize> {
    let cited = cited_tags(doc);
    let mut idx: Vec<usize> = if cited.is_empty() {
        (0..doc.sources.len()).collect()
    } else if st.order == Order::Cited {
        cited.iter().filter_map(|t| doc.sources.iter().position(|x| &x.tag == t)).collect()
    } else {
        let cited: std::collections::HashSet<&str> = cited.iter().map(String::as_str).collect();
        doc.sources.iter().enumerate().filter(|(_, x)| cited.contains(x.tag.as_str())).map(|(i, _)| i).collect()
    };
    // Sorted in the document's language (Polish authors `Łukasiewicz` before `Mazur`).
    let key = |i: &usize, by_title: bool| {
        sort_key(&doc.sources.get(*i).map(|x| if by_title { x.title.clone() } else { author_key(x) }).unwrap_or_default(), lang.as_deref())
    };
    match st.order {
        Order::Author => idx.sort_by_cached_key(|i| key(i, false)),
        Order::Title => idx.sort_by_cached_key(|i| key(i, true)),
        Order::Cited => {}
    }
    idx
}

/// A source's number in a numbered style: its place in the bibliography order.
fn number_of(doc: &Document, order: &[usize], tag: &str) -> usize {
    order.iter().position(|&i| doc.sources.get(i).is_some_and(|x| x.tag == tag)).unwrap_or(order.len()).saturating_add(1)
}

/// Refresh CITATION results and generated lists.
pub fn update_citations(s: &mut Session) -> Result<(), CmdError> {
    refresh_citations(s)?;
    update_generated(s)
}

/// Refresh the results of the CITATION fields in the body.
fn refresh_citations(s: &mut Session) -> Result<(), CmdError> {
    let st = style(&s.bib_style);
    let order = bib_order(&s.doc, st, super::default_lang(s));
    for path in s.doc.para_paths(StoryRef::Body) {
        let Some(p) = s.doc.para(StoryRef::Body, &path) else { continue };
        let updates: Vec<(usize, String)> = p
            .objects
            .iter()
            .enumerate()
            .filter_map(|(k, o)| match o {
                InlineObject::Field { instr, .. } => {
                    let (tag, pages) = citation_field(instr)?;
                    let src = s.doc.sources.iter().find(|x| x.tag == tag)?;
                    Some((k, cite(src, st.name, number_of(&s.doc, &order, tag), pages)))
                }
                _ => None,
            })
            .collect();
        if updates.is_empty() {
            continue;
        }
        let para = s.doc.para_mut(StoryRef::Body, &path)?;
        for (k, r) in updates {
            if let Some(InlineObject::Field { result, .. }) = para.objects.get_mut(k) {
                *result = r;
            }
        }
        para.touch();
    }
    Ok(())
}

/// The TOC field for `references.tableOfFigures` (ECMA-376 §17.16.5.68): `\c "<label>"` lists the
/// captions that label's SEQ field numbers, `\a "<label>"` lists them without their label and
/// number; `\n` leaves page numbers out and `\p " "` puts them right after the text.
pub fn figures_instr(v: &Value) -> String {
    let label: String = p::str(v, "label").unwrap_or("Figure").chars().filter(|c| !matches!(c, '"' | '\\') && !c.is_control()).take(40).collect();
    let ident = seq_ident(&label);
    let ident = if ident.is_empty() { "Figure".to_string() } else { ident };
    let on = |k: &str| p::bool(v, k).unwrap_or(true);
    let mut s = format!("TOC \\{} \"{ident}\"", if on("includeLabel") { 'c' } else { 'a' });
    if !on("pageNumbers") {
        s.push_str(" \\n");
    } else if !on("rightAlign") {
        s.push_str(" \\p \" \"");
    }
    s
}

/// The SEQ identifier of a caption label: Word joins a multi-word label with underscores.
fn seq_ident(label: &str) -> String {
    label.split_whitespace().collect::<Vec<_>>().join("_")
}

fn table_of_figures(s: &mut Session, v: &Value) -> CmdResult {
    generated_list_with(s, &figures_instr(v), "", p::str(v, "tabLeader").map(super::references::leader))?;
    sel_result(s)
}

/// Generated lists (BIBLIOGRAPHY, INDEX, table of figures, TOA): a heading paragraph holding
/// the field, then entry paragraphs marked with the "GeneratedEntry" style family.
pub(crate) fn generated_list(s: &mut Session, instr: &str, title: &str) -> Result<(), CmdError> {
    generated_list_with(s, instr, title, None)
}

/// [`generated_list`] whose entries take `leader`. The leader lives in the entries' tab stops (as
/// in a table of contents) and an update keeps the first entry's, so a placeholder entry carries
/// it through the rebuild, however the lists above it change.
fn generated_list_with(s: &mut Session, instr: &str, title: &str, leader: Option<TabLeader>) -> Result<(), CmdError> {
    let at = delete_selection(s)?;
    if at.story != StoryRef::Body || at.path.depth() > 0 {
        return Err(CmdError::Failed("lists go in the main text".into()));
    }
    let mut head = if title.is_empty() { Paragraph::new() } else { Paragraph::with_text(title, CharProps::default()).styled("Heading1") };
    let end = head.len();
    head.insert_object(
        end,
        InlineObject::Field { instr: instr.to_string(), result: String::new(), locked: false, dirty: false, code: false, form: String::new() },
        &CharProps::default(),
    )?;
    let i = if s.doc.para_at(&at).is_some_and(|p| p.is_empty()) { at.path.last() } else { s.doc.split_paragraph(&at)?.path.last() };
    s.doc.insert_block(StoryRef::Body, &Path::top(i), Block::Para(head))?;
    if let Some(leader) = leader {
        let mut seed = Paragraph::new().styled(ENTRY_STYLE);
        seed.props.tabs = Some(vec![TabStop { pos: 0.0, align: TabAlign::Right, leader }]);
        s.doc.insert_block(StoryRef::Body, &Path::top(i.saturating_add(1)), Block::Para(seed))?;
    }
    update_generated(s)
}

fn list_kind(p: &Paragraph) -> Option<String> {
    p.objects.iter().find_map(|o| match o {
        InlineObject::Field { instr, .. } => {
            let t = instr.trim_start();
            (t.starts_with("BIBLIOGRAPHY") || t.starts_with("INDEX") || t.starts_with("TOA") || figures_label(t).is_some()).then(|| t.to_string())
        }
        _ => None,
    })
}

/// The caption label (SEQ identifier) a table-of-figures field lists, `\c` or `\a`.
fn figures_label(instr: &str) -> Option<String> {
    let f = FieldCode::parse(instr);
    if !f.is_table_of_figures() {
        return None;
    }
    Some(f.arg('c').or_else(|| f.arg('a')).map_or_else(|| "Figure".to_string(), seq_ident))
}

/// Whether `p` is a caption numbered by SEQ identifier `ident` (any case, as Word matches it). A
/// caption typed by hand, without a SEQ field, goes by its label.
fn captioned_as(p: &Paragraph, ident: &str) -> bool {
    if p.props.style.as_deref() != Some("Caption") {
        return false;
    }
    let seq = p.objects.iter().find_map(|o| match o {
        InlineObject::Field { instr, .. } => Some(FieldCode::parse(instr)).filter(|f| f.name == "SEQ"),
        _ => None,
    });
    match seq {
        Some(f) => f.args.first().is_some_and(|id| id.eq_ignore_ascii_case(ident)),
        None => caption_label(p, &p.plain_text()).is_some_and(|l| seq_ident(&l).eq_ignore_ascii_case(ident)),
    }
}

/// A caption's text without its label and number (`\a`): what follows its SEQ field, or for a
/// caption typed by hand what follows the label and the number, less the separator.
fn caption_body(p: &Paragraph, label: &str) -> String {
    let seq_end = p.object_offsets().into_iter().find_map(|o| match p.object_at(o) {
        Some(InlineObject::Field { instr, .. }) if FieldCode::parse(instr).name == "SEQ" => Some(o + wordcraft_doc::para::OBJ.len_utf8()),
        _ => None,
    });
    let rest = match seq_end {
        Some(end) => {
            let mut k = 0;
            let mut rest = String::new();
            for (at, c) in p.text.char_indices() {
                if c == wordcraft_doc::para::OBJ {
                    if at >= end
                        && let Some(o) = p.objects.get(k)
                    {
                        rest.push_str(o.plain_text());
                    }
                    k += 1;
                } else if at >= end {
                    rest.push(c);
                }
            }
            rest
        }
        None => {
            let text = p.plain_text();
            // The label in any case, as captions are matched (ASCII case keeps the byte length).
            let start = text.trim_start();
            let after = start
                .get(..label.len())
                .filter(|head| head.eq_ignore_ascii_case(label))
                .and_then(|_| start.get(label.len()..))
                .unwrap_or(&text)
                .trim_start();
            // The number: the word after the label, when it starts with a digit.
            match after.find(char::is_whitespace) {
                Some(n) if after.starts_with(|c: char| c.is_ascii_digit()) => after[n..].to_string(),
                _ if after.starts_with(|c: char| c.is_ascii_digit()) => String::new(),
                _ => after.to_string(),
            }
        }
    };
    rest.trim_start_matches(|c: char| c.is_whitespace() || matches!(c, ':' | '.' | '-' | '\u{2013}' | '\u{2014}')).trim_end().to_string()
}

const ENTRY_STYLE: &str = "ListEntry";

/// Rebuild every generated list.
pub fn update_generated(s: &mut Session) -> Result<(), CmdError> {
    let mut i = 0;
    while i < s.doc.body.len() {
        let kind = s.doc.body.get(i).and_then(|b| b.as_para()).and_then(list_kind);
        let Some(kind) = kind else {
            i += 1;
            continue;
        };
        add_entry_style(s);
        if kind.starts_with("INDEX") {
            super::index::rebuild(s, i, None)?;
            i += 1;
            continue;
        }
        // A table of figures from Word can hold the list it saved as the field's result, in the
        // field's own paragraph: the rebuilt entries replace it, keeping its leader.
        let mut saved_leader = None;
        if figures_label(&kind).is_some()
            && let Ok(p) = s.doc.para_mut(StoryRef::Body, &Path::top(i))
        {
            let leader = p.props.tabs.as_ref().and_then(|t| t.last()).map(|t| t.leader);
            let mut cleared = false;
            for o in &mut p.objects {
                if let InlineObject::Field { instr, result, .. } = o
                    && !result.is_empty()
                    && figures_label(instr).is_some()
                {
                    result.clear();
                    cleared = true;
                }
            }
            if cleared {
                saved_leader = leader;
                p.touch();
            }
        }
        // Remove old entries, keeping their tab leader.
        let leader = s
            .doc
            .body
            .get(i + 1)
            .and_then(|b| b.as_para())
            .filter(|p| p.props.style.as_deref() == Some(ENTRY_STYLE))
            .and_then(|p| p.props.tabs.as_ref()?.last().map(|t| t.leader))
            .or(saved_leader)
            .unwrap_or(TabLeader::Dot);
        while s.doc.body.get(i + 1).and_then(|b| b.as_para()).is_some_and(|p| p.props.style.as_deref() == Some(ENTRY_STYLE)) {
            s.doc.remove_block(StoryRef::Body, &Path::top(i + 1))?;
        }
        let lines = entries_for(s, &kind);
        let width = super::page::sect(s).text_width();
        for (k, (text, hanging)) in lines.into_iter().enumerate() {
            let mut para = Paragraph::with_text(&text, CharProps::default()).styled(ENTRY_STYLE);
            if hanging {
                para.props.indent_left = Some(36.0);
                para.props.indent_first = Some(-36.0);
            }
            if text.contains('\t') {
                para.props.tabs = Some(vec![TabStop { pos: width - 0.5, align: TabAlign::Right, leader }]);
            }
            s.doc.insert_block(StoryRef::Body, &Path::top(i + 1 + k), Block::Para(para))?;
        }
        i += 1;
    }
    s.touch();
    Ok(())
}

/// Add the style generated list entries take, if the document doesn't have it yet. Only when a
/// list is there: a new style changes the formatting every paragraph is laid out with, so the
/// layout cache starts over (and the saved file would gain a style it doesn't use).
fn add_entry_style(s: &mut Session) {
    if s.doc.styles.get(ENTRY_STYLE).is_none() {
        s.doc.styles.upsert(wordcraft_doc::Style {
            id: ENTRY_STYLE.into(),
            name: "List Entry".into(),
            kind: wordcraft_doc::StyleKind::Paragraph,
            based_on: Some("Normal".into()),
            para: wordcraft_doc::ParaProps { space_after: Some(4.0), ..Default::default() },
            hidden: true,
            ..Default::default()
        });
    }
}

fn page_of(s: &mut Session, pos: &Pos) -> u32 {
    let l = s.layout();
    l.caret(pos).and_then(|c| l.pages.get(c.page)).map(|p| p.number).unwrap_or(1)
}

/// Entry lines (text, hanging indent) for a generated list.
fn entries_for(s: &mut Session, kind: &str) -> Vec<(String, bool)> {
    if kind.starts_with("BIBLIOGRAPHY") {
        let st = style(&s.bib_style);
        return bib_order(&s.doc, st, super::default_lang(s))
            .into_iter()
            .enumerate()
            .filter_map(|(k, i)| s.doc.sources.get(i).map(|x| (entry(x, st.name, k.saturating_add(1)), !st.numbered())))
            .collect();
    }
    // Collect marks with positions.
    let mut marks: Vec<(String, Pos)> = Vec::new();
    let figures = figures_label(kind);
    let bare = FieldCode::parse(kind).has('a');
    for path in s.doc.para_paths(StoryRef::Body) {
        let Some(p) = s.doc.para(StoryRef::Body, &path) else { continue };
        if let Some(ident) = &figures {
            if captioned_as(p, ident) {
                let text = if bare { caption_body(p, &ident.replace('_', " ")) } else { p.plain_text() };
                marks.push((text, Pos { story: StoryRef::Body, path: path.clone(), off: 0 }));
            }
            continue;
        }
        for off in p.object_offsets() {
            if let Some(InlineObject::Field { instr, .. }) = p.object_at(off) {
                let t = instr.trim_start();
                let want = if kind.starts_with("INDEX") { "XE" } else { "TA" };
                if t.starts_with(want) {
                    let e = t.split('"').nth(1).unwrap_or("").to_string();
                    marks.push((e, Pos { story: StoryRef::Body, path: path.clone(), off }));
                }
            }
        }
    }
    let mut with_pages: Vec<(String, u32)> = marks.into_iter().map(|(e, _)| (e, 0)).collect();
    // Pages (layout once).
    let positions: Vec<Pos> = {
        let mut v = Vec::new();
        for path in s.doc.para_paths(StoryRef::Body) {
            let Some(p) = s.doc.para(StoryRef::Body, &path) else { continue };
            if let Some(ident) = &figures {
                if captioned_as(p, ident) {
                    v.push(Pos { story: StoryRef::Body, path: path.clone(), off: 0 });
                }
                continue;
            }
            for off in p.object_offsets() {
                if let Some(InlineObject::Field { instr, .. }) = p.object_at(off) {
                    let want = if kind.starts_with("INDEX") { "XE" } else { "TA" };
                    if instr.trim_start().starts_with(want) {
                        v.push(Pos { story: StoryRef::Body, path: path.clone(), off });
                    }
                }
            }
        }
        v
    };
    for (i, pos) in positions.iter().enumerate() {
        let pg = page_of(s, pos);
        if let Some(x) = with_pages.get_mut(i) {
            x.1 = pg;
        }
    }
    if figures.is_some() {
        // `\n` without page numbers; `\p` the separator before them (a tab by default).
        let f = FieldCode::parse(kind);
        let sep = f.arg('p').unwrap_or("\t");
        return with_pages.into_iter().map(|(t, p)| (if f.has('n') { t } else { format!("{t}{sep}{p}") }, false)).collect();
    }
    // Index / TOA: group by entry, sort, list pages.
    let lang = super::default_lang(s);
    let mut grouped: std::collections::BTreeMap<(SortKey, String), Vec<u32>> = Default::default();
    for (e, p) in with_pages {
        let pages = grouped.entry((sort_key(&e, lang.as_deref()), e)).or_default();
        if !pages.contains(&p) {
            pages.push(p);
        }
    }
    let mut out = Vec::new();
    let mut letter = None;
    for ((_, e), pages) in grouped {
        let first = heading_letter(&e, lang.as_deref());
        if kind.starts_with("INDEX") && first != letter {
            letter = first.clone();
            if let Some(l) = first {
                out.push((l, false));
            }
        }
        let list = pages.iter().map(|p| p.to_string()).collect::<Vec<_>>().join(", ");
        out.push(if kind.starts_with("INDEX") { (format!("{e}, {list}"), true) } else { (format!("{e}\t{list}"), false) });
    }
    out
}

fn cross_ref(s: &mut Session, v: &Value) -> CmdResult {
    let to = p::str(v, "to").unwrap_or("heading");
    let targets = ref_targets(s, to);
    let Some(target) = p::str(v, "target") else {
        return Ok(Value::Array(targets.iter().map(|(t, _)| json!(t)).collect()));
    };
    let find = |ts: Vec<(String, Pos)>| ts.into_iter().find(|(t, _)| t == target || t.starts_with(target));
    let (text, pos) = find(targets).ok_or_else(|| CmdError::Params(format!("no {to} `{target}`")))?;
    let show = p::str(v, "show").unwrap_or("text");
    let page = page_of(s, &pos);
    let result = match show {
        "page" => page.to_string(),
        "number" => text.split(':').next().unwrap_or(&text).trim().to_string(),
        "aboveBelow" => {
            if pos < s.sel.focus {
                "above".into()
            } else {
                "below".into()
            }
        }
        _ if to == "bookmark" => super::references::bookmark_text(&s.doc, &text).unwrap_or_default(),
        _ => text.clone(),
    };
    let props = s.typing_props();
    let at = delete_selection(s)?;
    // The field points at a bookmark: the named one, or a hidden `_Ref` bookmark on the target
    // paragraph (as Word does), so Update Field and Word can resolve it.
    let (name, at) = if to == "bookmark" {
        (text, at)
    } else {
        // Deleting the selection can move paragraphs: look the target up again.
        let (_, pos) = find(ref_targets(s, to)).ok_or_else(|| CmdError::Params(format!("no {to} `{target}`")))?;
        ref_bookmark(s, &pos, show == "number", at)?
    };
    let instr = format!("{} {name} \\h", if show == "page" { "PAGEREF" } else { "REF" });
    let end = s.doc.insert_object(
        &at,
        InlineObject::Field { instr, result: result.clone(), locked: false, dirty: false, code: false, form: String::new() },
        &props,
    )?;
    s.sel = Selection::caret(end);
    Ok(json!({"inserted": result}))
}

/// The label of a caption: the text before its first field (`Plate ` in `Plate `, `SEQ Plate`,
/// `: A rock`). When the caption starts with its chapter number, the text between that field and
/// the `SEQ` number, unless it is only a separator. Failing both, the `SEQ` identifier names it.
/// A caption without a `SEQ` field (typed by hand) has what comes before its first digit.
fn caption_label(p: &Paragraph, text: &str) -> Option<String> {
    let fields: Vec<(usize, &str)> = p
        .object_offsets()
        .into_iter()
        .filter_map(|o| match p.object_at(o) {
            Some(InlineObject::Field { instr, .. }) => Some((o, instr.as_str())),
            _ => None,
        })
        .collect();
    let seq = fields.iter().position(|(_, instr)| instr.split_whitespace().next().is_some_and(|w| w.eq_ignore_ascii_case("SEQ")));
    let label = match seq.and_then(|i| Some((i, fields.get(i)?))) {
        Some((i, (seq_at, instr))) => {
            // The text in `from..to` with the bookmark marks dropped, if it says anything.
            let said = |from: usize, to: usize| {
                let t: String = p.text.get(from..to).unwrap_or_default().chars().filter(|c| *c != wordcraft_doc::para::OBJ).collect();
                let t = t.trim();
                t.chars().any(char::is_alphanumeric).then(|| t.to_string())
            };
            let first = fields.first().map_or(0, |(o, _)| *o);
            let after_previous =
                i.checked_sub(1).and_then(|k| fields.get(k)).map_or(0, |(o, _)| o.saturating_add(wordcraft_doc::para::OBJ.len_utf8()));
            said(0, first).or_else(|| said(after_previous, *seq_at)).or_else(|| instr.split_whitespace().nth(1).map(|id| id.replace('_', " ")))?
        }
        None => text.get(..text.find(|c: char| c.is_ascii_digit())?)?.trim().to_string(),
    };
    (!label.is_empty() && label.chars().count() <= 40).then_some(label)
}

/// The caption paragraphs of the body: (text, position, label).
fn captions(s: &Session) -> Vec<(String, Pos, Option<String>)> {
    s.doc
        .para_paths(StoryRef::Body)
        .into_iter()
        .filter_map(|path| {
            let p = s.doc.para(StoryRef::Body, &path).filter(|p| p.props.style.as_deref() == Some("Caption"))?;
            let text = p.plain_text();
            let label = caption_label(p, &text);
            Some((text, Pos { story: StoryRef::Body, path, off: 0 }, label))
        })
        .collect()
}

/// The labels the body's captions use, first seen first: Figure, Table and Equation, and any the
/// person made with Insert Caption › New Label. Each is a cross-reference type, `label:<Name>` (#2940).
pub fn caption_labels(s: &Session) -> Vec<String> {
    let mut labels: Vec<String> = Vec::new();
    for l in captions(s).into_iter().filter_map(|(_, _, l)| l) {
        if !labels.contains(&l) {
            labels.push(l);
        }
    }
    labels
}

/// Cross-reference targets of a kind: (shown text or bookmark name, position).
pub fn ref_targets(s: &Session, to: &str) -> Vec<(String, Pos)> {
    let mut targets: Vec<(String, Pos)> = Vec::new();
    match to {
        "bookmark" => targets = s.doc.bookmarks(),
        // A caption label of the person's own: the captions that use exactly that label.
        t if t.starts_with("label:") => {
            let want = t.get("label:".len()..).unwrap_or_default();
            targets = captions(s)
                .into_iter()
                .filter(|(_, _, label)| !want.is_empty() && label.as_deref() == Some(want))
                .map(|(text, pos, _)| (text, pos))
                .collect();
        }
        "figure" | "table" | "equation" => {
            let label = match to {
                "table" => "Table",
                "equation" => "Equation",
                _ => "Figure",
            };
            for path in s.doc.para_paths(StoryRef::Body) {
                if let Some(p) = s.doc.para(StoryRef::Body, &path)
                    && p.props.style.as_deref() == Some("Caption")
                    && p.plain_text().starts_with(label)
                {
                    targets.push((p.plain_text(), Pos { story: StoryRef::Body, path: path.clone(), off: 0 }));
                }
            }
        }
        _ => {
            for path in s.doc.para_paths(StoryRef::Body) {
                if let Some(p) = s.doc.para(StoryRef::Body, &path)
                    && s.doc.styles.resolve_para(&p.props).outline_level.is_some()
                    && !p.plain_text().trim().is_empty()
                {
                    targets.push((p.plain_text().trim().to_string(), Pos { story: StoryRef::Body, path: path.clone(), off: 0 }));
                }
            }
        }
    }
    targets
}

/// The hidden `_Ref` bookmark around the target paragraph at `pos` (its label and number only
/// when `number_only`), reusing one already there. Returns its name and `at` moved past the
/// bookmark marks inserted before it.
fn ref_bookmark(s: &mut Session, pos: &Pos, number_only: bool, mut at: Pos) -> Result<(String, Pos), CmdError> {
    const MARK: usize = wordcraft_doc::para::OBJ.len_utf8();
    let para = s.doc.para(pos.story, &pos.path).ok_or_else(|| CmdError::Params("cross-reference target is gone".into()))?;
    let end = if number_only { para.text.find(':').unwrap_or(para.len()) } else { para.len() };
    if let Some(InlineObject::BookmarkStart { name }) = para.object_at(0)
        && name.starts_with("_Ref")
    {
        // Reuse it when it covers the same extent: up to the colon, or the whole paragraph.
        let same = para.object_offsets().into_iter().any(|o| {
            matches!(para.object_at(o), Some(InlineObject::BookmarkEnd { name: n }) if n == name)
                && para.text.get(o.saturating_add(MARK)..).is_some_and(|rest| if number_only { rest.starts_with(':') } else { rest.is_empty() })
        });
        if same {
            return Ok((name.clone(), at));
        }
    }
    let next =
        s.doc.bookmarks().iter().filter_map(|(n, _)| n.strip_prefix("_Ref").and_then(|d| d.parse::<u64>().ok())).max().unwrap_or(0).saturating_add(1);
    let name = format!("_Ref{next:09}");
    let props = CharProps::default();
    s.doc.insert_object(&Pos { off: end, ..pos.clone() }, InlineObject::BookmarkEnd { name: name.clone() }, &props)?;
    s.doc.insert_object(&Pos { off: 0, ..pos.clone() }, InlineObject::BookmarkStart { name: name.clone() }, &props)?;
    if at.story == pos.story && at.path == pos.path {
        if at.off >= end {
            at.off = at.off.saturating_add(2 * MARK);
        } else if at.off > 0 {
            at.off = at.off.saturating_add(MARK);
        }
    }
    Ok((name, at))
}

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

    #[test]
    fn citations_and_bibliography() {
        let mut s = Session::new(wordcraft_doc::Document::new());
        s.run("text.insert", &json!({"text": "Studios matter "})).unwrap();
        s.run("references.citation", &json!({"source": {"author": "Rivera, Alex; Chen, Mei", "title": "Shared Spaces", "year": "2021", "publisher": "Harbor Books", "city": "Portland"}})).unwrap();
        assert!(s.doc.plain_text(StoryRef::Body).contains("(Rivera & Chen, 2021)"), "{}", s.doc.plain_text(StoryRef::Body));
        s.run("text.newParagraph", &json!({})).unwrap();
        s.run("references.bibliography", &json!({})).unwrap();
        let t = s.doc.plain_text(StoryRef::Body);
        assert!(t.contains("Rivera, A., & Chen, M. (2021). Shared Spaces. Harbor Books."), "{t}");
        s.run("references.citationStyle", &json!({"style": "MLA"})).unwrap();
        let t = s.doc.plain_text(StoryRef::Body);
        assert!(t.contains("(Rivera and Chen)"), "{t}");
        s.run("references.citationStyle", &json!({"style": "IEEE"})).unwrap();
        assert!(s.doc.plain_text(StoryRef::Body).contains("[1]"));
    }

    /// Updating fields in a document without generated lists leaves its styles alone: a new
    /// style would make the next layout start over from nothing.
    #[test]
    fn update_fields_adds_the_entry_style_only_with_a_list() {
        let mut s = Session::new(wordcraft_doc::Document::from_text("plain text"));
        s.run("select.all", &json!({})).unwrap();
        s.run("references.updateFields", &json!({})).unwrap();
        assert!(s.doc.styles.get(ENTRY_STYLE).is_none());
        s.run("caret.docEnd", &json!({})).unwrap();
        s.run("text.newParagraph", &json!({})).unwrap();
        s.run("references.citation", &json!({"source": {"author": "Rivera, Alex", "title": "Shared Spaces", "year": "2021"}})).unwrap();
        s.run("text.newParagraph", &json!({})).unwrap();
        s.run("references.bibliography", &json!({})).unwrap();
        assert!(s.doc.styles.get(ENTRY_STYLE).is_some());
    }

    #[test]
    fn bibliography_round_trip_docx() {
        let mut s = Session::new(wordcraft_doc::Document::new());
        s.run("text.insert", &json!({"text": "Studios matter "})).unwrap();
        s.run(
            "references.citation",
            &json!({
                "source": {
                    "author": "Rivera, Alex; Chen, Mei",
                    "title": "Shared Spaces",
                    "year": "2021",
                    "publisher": "Harbor Books",
                    "city": "Portland",
                    "editor": "Johnson, Robert",
                    "edition": "2nd",
                    "issue": "4",
                    "month": "10",
                    "day": "15",
                    "accessed": "2024-05-15",
                    "doi": "10.1000/182"
                }
            }),
        )
        .unwrap();
        s.run("text.newParagraph", &json!({})).unwrap();
        s.run("references.bibliography", &json!({})).unwrap();

        // Serialize to DOCX and back — sources must survive.
        let bytes = wordcraft_docx::write(&s.doc).expect("docx write");
        let doc = wordcraft_docx::read(&bytes).expect("docx read");
        assert_eq!(doc.sources.len(), 1, "round-tripped sources: {:?}", doc.sources);
        assert_eq!(doc.sources[0].author, "Rivera, Alex; Chen, Mei");
        assert_eq!(doc.sources[0].title, "Shared Spaces");
        assert_eq!(doc.sources[0].editor, "Johnson, Robert");
        assert_eq!(doc.sources[0].edition, "2nd");
        assert_eq!(doc.sources[0].issue, "4");
        assert_eq!(doc.sources[0].month, "10");
        assert_eq!(doc.sources[0].day, "15");
        assert_eq!(doc.sources[0].accessed, "2024-05-15");
        assert_eq!(doc.sources[0].doi, "10.1000/182");

        // Round-tripped document in a new session works with citation styles.
        let mut s2 = Session::new(doc);
        s2.run("references.citationStyle", &json!({"style": "Harvard (Anglia)"})).unwrap();
        let t = s2.doc.plain_text(StoryRef::Body);
        assert!(t.contains("(Rivera and Chen, 2021)"), "{t}");
        assert!(t.contains("Rivera, A. and Chen, M., 2021. Shared Spaces. Portland: Harbor Books."), "{t}");
    }

    #[test]
    fn citation_style_survives_docx_save_and_reopen() {
        let mut s = Session::new(wordcraft_doc::Document::new());
        s.run("references.citation", &json!({"source": {"author": "Rivera, Alex", "title": "Shared Spaces", "year": "2021"}})).unwrap();
        s.run("references.citationStyle", &json!({"style": "Harvard (Anglia)"})).unwrap();
        let bytes = wordcraft_docx::write(&s.doc).expect("docx write");
        let doc = wordcraft_docx::read(&bytes).expect("docx read");
        assert_eq!(doc.settings.bib_style.as_deref(), Some("Harvard (Anglia)"));
        assert_eq!(Session::new(doc.clone()).bib_style, "Harvard (Anglia)");
        let mut other = Session::new(wordcraft_doc::Document::new());
        other.set_document(doc);
        assert_eq!(other.bib_style, "Harvard (Anglia)");
        // A document without a recorded style is APA.
        assert_eq!(Session::new(wordcraft_doc::Document::new()).bib_style, "APA");
    }

    #[test]
    fn index_and_figures() {
        let mut s = Session::new(wordcraft_doc::Document::from_text("Kilns are hot.\nPresses are heavy.\nMore kilns."));
        s.run("select.text", &json!({"text": "Kilns"})).unwrap();
        s.run("references.markEntry", &json!({})).unwrap();
        s.run("select.text", &json!({"text": "Presses"})).unwrap();
        s.run("references.markEntry", &json!({})).unwrap();
        s.run("caret.docEnd", &json!({})).unwrap();
        s.run("text.newParagraph", &json!({})).unwrap();
        s.run("references.index", &json!({})).unwrap();
        let t = s.doc.plain_text(StoryRef::Body);
        assert!(t.contains("Kilns, 1") && t.contains("Presses, 1"), "{t}");
        s.run("caret.docStart", &json!({})).unwrap();
        s.run("references.caption", &json!({"label": "Figure", "text": "The kiln"})).unwrap();
        s.run("caret.docEnd", &json!({})).unwrap();
        s.run("references.tableOfFigures", &json!({})).unwrap();
        assert!(s.doc.plain_text(StoryRef::Body).contains("Figure 1: The kiln\t1"));
        let list = s.run("insert.crossReference", &json!({"to": "figure"})).unwrap();
        assert_eq!(list.as_array().unwrap().len(), 1);
        s.run("insert.crossReference", &json!({"to": "figure", "target": "Figure 1", "show": "number"})).unwrap();
        assert!(s.doc.plain_text(StoryRef::Body).contains("Figure 1"));
        // The number-only reference marks just "Figure 1" of the caption.
        let (name, _) = s.doc.bookmarks().into_iter().find(|(n, _)| n.starts_with("_Ref")).unwrap();
        assert_eq!(crate::cmd::references::bookmark_text(&s.doc, &name).as_deref(), Some("Figure 1"));
        assert!(fields(&s).contains(&(format!("REF {name} \\h"), "Figure 1".to_string())), "{:?}", fields(&s));
    }

    /// #2940: a caption with a label of the person's own (Insert Caption › New Label) is a
    /// cross-reference type of its own, `label:<Name>`, next to Figure, Table and Equation.
    #[test]
    fn cross_reference_to_a_custom_caption_label() {
        let mut s = Session::new(wordcraft_doc::Document::from_text("Rock\nSee "));
        s.run("select.text", &json!({"text": "Rock"})).unwrap();
        s.run("references.caption", &json!({"label": "Plate", "text": "A rock"})).unwrap();
        assert_eq!(caption_labels(&s), vec!["Plate".to_string()]);
        let list = s.run("insert.crossReference", &json!({"to": "label:Plate"})).unwrap();
        assert_eq!(list, json!(["Plate 1: A rock"]));
        assert!(ref_targets(&s, "label:Figure").is_empty(), "only that label's captions");
        assert!(ref_targets(&s, "label:").is_empty() && ref_targets(&s, "label:Pla").is_empty(), "the whole label");
        s.run("caret.docEnd", &json!({})).unwrap();
        s.run("insert.crossReference", &json!({"to": "label:Plate", "target": "Plate 1", "show": "number"})).unwrap();
        assert!(s.doc.plain_text(StoryRef::Body).ends_with("See Plate 1"), "{}", s.doc.plain_text(StoryRef::Body));
        assert!(fields(&s).iter().any(|(i, r)| i.starts_with("REF _Ref") && r == "Plate 1"), "{:?}", fields(&s));
    }

    /// #2940: the label comes from the caption's structure (the text before its `SEQ` number), so
    /// letter and roman numbering, which have no digit, still name it; a multi-word label too.
    #[test]
    fn caption_labels_do_not_need_a_digit() {
        let mut s = Session::new(wordcraft_doc::Document::from_text("Rock\nGem\nSand\nSee "));
        for (text, label, format) in [("Rock", "Plate", "ALPHABETIC"), ("Gem", "Photo Plate", "ROMAN"), ("Sand", "Chart", "ARABIC")] {
            s.run("select.text", &json!({"text": text})).unwrap();
            s.run("references.caption", &json!({"label": label, "text": text, "format": format})).unwrap();
        }
        assert_eq!(caption_labels(&s), ["Plate", "Photo Plate", "Chart"]);
        assert_eq!(ref_targets(&s, "label:Plate").len(), 1);
        assert_eq!(ref_targets(&s, "label:Photo Plate").len(), 1);
        assert_eq!(ref_targets(&s, "label:Chart").len(), 1);
        s.run("caret.docEnd", &json!({})).unwrap();
        s.run("insert.crossReference", &json!({"to": "label:Photo Plate", "target": "Photo Plate I", "show": "number"})).unwrap();
        assert!(s.doc.plain_text(StoryRef::Body).ends_with("See Photo Plate I"), "{}", s.doc.plain_text(StoryRef::Body));
    }

    /// The entries of every generated list, in order, with their last tab stop's leader.
    fn list_entries(s: &Session) -> Vec<(String, Option<TabLeader>)> {
        s.doc
            .body
            .iter()
            .filter_map(|b| b.as_para())
            .filter(|p| p.props.style.as_deref() == Some(ENTRY_STYLE))
            .map(|p| (p.plain_text(), p.props.tabs.as_ref().and_then(|t| t.last()).map(|t| t.leader)))
            .collect()
    }

    /// #3013: a table of figures for any caption label, so a document can have a list of figures
    /// and a list of tables; each keeps its own label and leader when updated.
    #[test]
    fn table_of_figures_for_each_caption_label() {
        let mut s = Session::new(wordcraft_doc::Document::from_text("Kiln\nPress\nRock\nTablet\n"));
        for (text, label) in [("Kiln", "Figure"), ("Press", "Table"), ("Rock", "Photo Plate"), ("Tablet", "Figure")] {
            s.run("select.text", &json!({"text": text})).unwrap();
            s.run("references.caption", &json!({"label": label, "text": text})).unwrap();
        }
        s.run("caret.docEnd", &json!({})).unwrap();
        s.run("references.tableOfFigures", &json!({"label": "Table", "tabLeader": "hyphen"})).unwrap();
        assert_eq!(list_entries(&s), [("Table 1: Press\t1".to_string(), Some(TabLeader::Hyphen))]);
        s.run("caret.docEnd", &json!({})).unwrap();
        s.run("text.newParagraph", &json!({})).unwrap();
        s.run("references.tableOfFigures", &json!({"label": "Photo Plate"})).unwrap();
        s.run("caret.docEnd", &json!({})).unwrap();
        s.run("text.newParagraph", &json!({})).unwrap();
        s.run("references.tableOfFigures", &json!({})).unwrap();
        let instrs: Vec<String> = fields(&s).into_iter().map(|(i, _)| i).filter(|i| i.starts_with("TOC")).collect();
        assert_eq!(instrs, [r#"TOC \c "Table""#, r#"TOC \c "Photo_Plate""#, r#"TOC \c "Figure""#]);
        let want = [
            ("Table 1: Press\t1", TabLeader::Hyphen),
            ("Photo Plate 1: Rock\t1", TabLeader::Dot),
            ("Figure 1: Kiln\t1", TabLeader::Dot),
            ("Figure 2: Tablet\t1", TabLeader::Dot),
        ];
        let want: Vec<(String, Option<TabLeader>)> = want.iter().map(|(t, l)| (t.to_string(), Some(*l))).collect();
        assert_eq!(list_entries(&s), want);
        // Update Table rebuilds each list with its own label and keeps the leader.
        s.run("references.updateFigures", &json!({})).unwrap();
        assert_eq!(list_entries(&s), want);
        // A table of figures is not the table of contents.
        assert!(s.run("references.updateToc", &json!({})).is_ok());
        assert_eq!(list_entries(&s), want);
    }

    /// #3013: Insert Table of Figures' options as TOC switches: `\a` without the label and number,
    /// `\n` without page numbers, `\p " "` with the page right after the text.
    #[test]
    fn table_of_figures_options() {
        assert_eq!(figures_instr(&json!({})), r#"TOC \c "Figure""#);
        assert_eq!(figures_instr(&json!({"label": "Table", "includeLabel": false})), r#"TOC \a "Table""#);
        assert_eq!(figures_instr(&json!({"pageNumbers": false})), r#"TOC \c "Figure" \n"#);
        assert_eq!(figures_instr(&json!({"rightAlign": false})), r#"TOC \c "Figure" \p " ""#);
        assert_eq!(figures_instr(&json!({"label": "\"\\ "})), r#"TOC \c "Figure""#, "an empty label is Figure");
        for (params, entry) in [
            (json!({"includeLabel": false}), "The kiln\t1"),
            (json!({"pageNumbers": false}), "Figure 1: The kiln"),
            (json!({"rightAlign": false, "includeLabel": false}), "The kiln 1"),
        ] {
            let mut s = Session::new(wordcraft_doc::Document::from_text("Kiln\n"));
            s.run("select.text", &json!({"text": "Kiln"})).unwrap();
            s.run("references.caption", &json!({"label": "Figure", "text": "The kiln"})).unwrap();
            s.run("caret.docEnd", &json!({})).unwrap();
            s.run("references.tableOfFigures", &params).unwrap();
            let entries = list_entries(&s);
            assert_eq!(entries.iter().map(|(t, _)| t.as_str()).collect::<Vec<_>>(), [entry], "{params}");
            assert_eq!(entries.first().and_then(|(_, l)| *l).is_some(), entry.contains('\t'), "a tab stop only for a tab");
        }
    }

    /// #3013: a table of figures from Word (`TOC \h \z \c "Figure"`, switches in any order) and
    /// hand-typed captions are listed (the label in any case); a caption whose label only starts
    /// with another's is not.
    #[test]
    fn table_of_figures_from_word_and_typed_captions() {
        let mut d = wordcraft_doc::Document::from_text("Figure 1: Typed\nFigures 2: Not a figure\nFigure 3 \u{2014} Dashed\nfigure 4: Lower\nList");
        for path in d.para_paths(StoryRef::Body).into_iter().take(4) {
            d.para_mut(StoryRef::Body, &path).unwrap().props.style = Some("Caption".into());
        }
        let mut s = Session::new(d);
        s.run("caret.docEnd", &json!({})).unwrap();
        s.run("text.newParagraph", &json!({})).unwrap();
        let at = s.sel.focus.clone();
        let field = InlineObject::Field {
            instr: r#"TOC \h \z \a "Figure""#.into(),
            result: String::new(),
            locked: false,
            dirty: false,
            code: false,
            form: String::new(),
        };
        s.doc.insert_object(&at, field, &CharProps::default()).unwrap();
        s.run("references.updateFigures", &json!({})).unwrap();
        let entries: Vec<String> = list_entries(&s).into_iter().map(|(t, _)| t).collect();
        assert_eq!(entries, ["Typed\t1", "Dashed\t1", "Lower\t1"]);
    }

    /// #3013 review: the leader asked for lands on the new list even when a list above it is out
    /// of date and gains entries (moving the new one down) while every list is rebuilt.
    #[test]
    fn table_of_figures_leader_survives_a_stale_list_above() {
        let mut s = Session::new(wordcraft_doc::Document::from_text("Kiln\nPress\nRock\n"));
        for (text, label) in [("Kiln", "Figure"), ("Press", "Table")] {
            s.run("select.text", &json!({"text": text})).unwrap();
            s.run("references.caption", &json!({"label": label, "text": text})).unwrap();
        }
        s.run("caret.docEnd", &json!({})).unwrap();
        s.run("references.tableOfFigures", &json!({})).unwrap();
        s.run("select.text", &json!({"text": "Rock"})).unwrap();
        s.run("references.caption", &json!({"label": "Figure", "text": "Rock"})).unwrap();
        assert_eq!(list_entries(&s).len(), 1, "the list of figures is out of date");
        s.run("caret.docEnd", &json!({})).unwrap();
        s.run("text.newParagraph", &json!({})).unwrap();
        s.run("references.tableOfFigures", &json!({"label": "Table", "tabLeader": "hyphen"})).unwrap();
        let want = [("Figure 1: Kiln\t1", TabLeader::Dot), ("Figure 2: Rock\t1", TabLeader::Dot), ("Table 1: Press\t1", TabLeader::Hyphen)];
        let want: Vec<(String, Option<TabLeader>)> = want.iter().map(|(t, l)| (t.to_string(), Some(*l))).collect();
        assert_eq!(list_entries(&s), want);
    }

    /// #3013 review: a table of figures from Word whose saved list sits in the field's result (one
    /// paragraph) is replaced by the rebuilt entries, not listed twice, and keeps its leader.
    #[test]
    fn table_of_figures_from_word_replaces_its_saved_list() {
        let mut s = Session::new(wordcraft_doc::Document::from_text("Kiln\nList"));
        s.run("select.text", &json!({"text": "Kiln"})).unwrap();
        s.run("references.caption", &json!({"label": "Figure", "text": "Kiln"})).unwrap();
        let last = s.doc.body.len() - 1;
        let p = s.doc.para_mut(StoryRef::Body, &Path::top(last)).unwrap();
        p.props.style = Some("TableofFigures".into());
        p.props.tabs = Some(vec![TabStop { pos: 400.0, align: TabAlign::Right, leader: TabLeader::Hyphen }]);
        let end = p.len();
        p.delete(0, end).unwrap();
        let field = InlineObject::Field {
            instr: r#"TOC \h \z \c "Figure""#.into(),
            result: "Figure 1: Kiln\t1".into(),
            locked: false,
            dirty: false,
            code: false,
            form: String::new(),
        };
        p.insert_object(0, field, &CharProps::default()).unwrap();
        for _ in 0..2 {
            s.run("references.updateFigures", &json!({})).unwrap();
            let text = s.doc.plain_text(StoryRef::Body);
            assert_eq!(text.matches("Figure 1: Kiln\t1").count(), 1, "{text:?}");
            assert_eq!(list_entries(&s), [("Figure 1: Kiln\t1".to_string(), Some(TabLeader::Hyphen))]);
        }
    }

    /// #3013 review: `\c` picks captions by their SEQ identifier, as Word does, not by the text
    /// before the number (`STYLEREF`, " Specimen ", `SEQ Figure` is a figure).
    #[test]
    fn table_of_figures_goes_by_the_seq_identifier() {
        let mut s = Session::new(wordcraft_doc::Document::from_text("Rock\n"));
        let field = |instr: &str| InlineObject::Field {
            instr: instr.into(),
            result: String::new(),
            locked: false,
            dirty: false,
            code: false,
            form: String::new(),
        };
        let plain = CharProps::default();
        let mut imported = Paragraph::new().styled("Caption");
        imported.insert_object(0, field("STYLEREF 1 \\s"), &plain).unwrap();
        imported.insert_text(imported.len(), " Specimen ", &plain).unwrap();
        imported.insert_object(imported.len(), field("SEQ Figure \\* ARABIC"), &plain).unwrap();
        imported.insert_text(imported.len(), ": A rock", &plain).unwrap();
        s.doc.insert_block(StoryRef::Body, &Path::top(1), Block::Para(imported)).unwrap();
        for (label, listed) in [("Figure", true), ("Specimen", false)] {
            let mut s = Session::new(s.doc.clone());
            s.run("caret.docEnd", &json!({})).unwrap();
            s.run("text.newParagraph", &json!({})).unwrap();
            s.run("references.tableOfFigures", &json!({"label": label, "includeLabel": false})).unwrap();
            let entries: Vec<String> = list_entries(&s).into_iter().map(|(t, _)| t).collect();
            assert_eq!(entries.len(), usize::from(listed), "{label}: {entries:?}");
            if listed {
                assert_eq!(entries.first().map(|t| t.split('\t').next()), Some(Some("A rock")));
            }
        }
    }

    /// #2940: a caption may start with its chapter number (`STYLEREF`) and carry its label text
    /// between that and the `SEQ` number; the text is the label, not the `SEQ` identifier.
    #[test]
    fn caption_label_after_a_chapter_number_field() {
        let mut s = Session::new(wordcraft_doc::Document::from_text("Rock\nSee "));
        let field = |instr: &str| InlineObject::Field {
            instr: instr.into(),
            result: String::new(),
            locked: false,
            dirty: false,
            code: false,
            form: String::new(),
        };
        let plain = CharProps::default();
        // STYLEREF, " Specimen ", SEQ Figure, ": A rock"
        let mut imported = Paragraph::new().styled("Caption");
        imported.insert_object(0, field("STYLEREF 1 \\s"), &plain).unwrap();
        imported.insert_text(imported.len(), " Specimen ", &plain).unwrap();
        imported.insert_object(imported.len(), field("SEQ Figure \\* ARABIC"), &plain).unwrap();
        imported.insert_text(imported.len(), ": A rock", &plain).unwrap();
        s.doc.insert_block(StoryRef::Body, &Path::top(1), Block::Para(imported)).unwrap();
        // A generated caption with a chapter number: label, STYLEREF, separator, SEQ.
        s.run("select.text", &json!({"text": "Rock"})).unwrap();
        s.run("references.caption", &json!({"label": "Plate", "chapter": 1, "separator": "-"})).unwrap();
        // No label text at all: the SEQ identifier names it, after the chapter number's separator.
        let mut bare = Paragraph::new().styled("Caption");
        bare.insert_object(0, field("STYLEREF 1 \\s"), &plain).unwrap();
        bare.insert_text(bare.len(), "-", &plain).unwrap();
        bare.insert_object(bare.len(), field("SEQ Photo_Plate"), &plain).unwrap();
        let end = s.doc.body.len();
        s.doc.insert_block(StoryRef::Body, &Path::top(end), Block::Para(bare)).unwrap();
        let labels = caption_labels(&s);
        assert!(labels.contains(&"Specimen".to_string()), "{labels:?}");
        assert!(labels.contains(&"Plate".to_string()), "{labels:?}");
        assert!(labels.contains(&"Photo Plate".to_string()), "{labels:?}");
        assert!(!labels.iter().any(|l| l == "-" || l.is_empty() || l == "Figure"), "{labels:?}");
    }

    /// #661: References › Table of Authorities › Update Table picks up citations marked after the table.
    #[test]
    fn table_of_authorities_updates() {
        let mut s = Session::new(wordcraft_doc::Document::from_text("Smith v. Jones held it.\nDoe v. Roe agreed."));
        s.run("select.text", &json!({"text": "Smith v. Jones"})).unwrap();
        s.run("references.markCitation", &json!({})).unwrap();
        s.run("caret.docEnd", &json!({})).unwrap();
        s.run("text.newParagraph", &json!({})).unwrap();
        s.run("references.tableOfAuthorities", &json!({})).unwrap();
        let t = s.doc.plain_text(StoryRef::Body);
        assert!(t.contains("Smith v. Jones") && !t.contains("Doe v. Roe, "), "{t}");
        s.run("select.text", &json!({"text": "Doe v. Roe"})).unwrap();
        s.run("references.markCitation", &json!({})).unwrap();
        s.run("references.updateAuthorities", &json!({})).unwrap();
        let after = s.doc.plain_text(StoryRef::Body);
        assert!(after.matches("Doe v. Roe").count() > t.matches("Doe v. Roe").count(), "{after}");
    }

    /// Field instructions and results in the body, in order.
    fn fields(s: &Session) -> Vec<(String, String)> {
        let mut out = Vec::new();
        for path in s.doc.para_paths(StoryRef::Body) {
            for o in &s.doc.para(StoryRef::Body, &path).unwrap().objects {
                if let InlineObject::Field { instr, result, .. } = o {
                    out.push((instr.clone(), result.clone()));
                }
            }
        }
        out
    }

    #[test]
    fn cross_references_point_at_real_bookmarks() {
        let mut s = Session::new(wordcraft_doc::Document::from_text("Kilns\nSee here."));
        s.run("caret.docStart", &json!({})).unwrap();
        s.run("styles.apply", &json!({"style": "Heading 1"})).unwrap();
        s.run("caret.docEnd", &json!({})).unwrap();
        s.run("insert.crossReference", &json!({"to": "heading", "target": "Kilns"})).unwrap();
        // The heading gets a hidden bookmark, and the REF field names it (not a literal `_Ref`).
        let marks = s.doc.bookmarks();
        assert_eq!(marks.len(), 1, "{marks:?}");
        let name = marks[0].0.clone();
        assert!(name.starts_with("_Ref") && name.len() > 4, "{name}");
        assert_eq!(marks[0].1.path, s.doc.para_paths(StoryRef::Body)[0]);
        assert_eq!(fields(&s), vec![(format!("REF {name} \\h"), "Kilns".to_string())]);
        // A second reference to the same heading reuses the bookmark.
        s.run("insert.crossReference", &json!({"to": "heading", "target": "Kilns", "show": "page"})).unwrap();
        assert_eq!(s.doc.bookmarks().len(), 1);
        assert_eq!(fields(&s)[1], (format!("PAGEREF {name} \\h"), "1".to_string()));
        // Update Field follows the heading's new text.
        s.run("select.text", &json!({"text": "ilns"})).unwrap();
        s.run("text.insert", &json!({"text": "ILNS"})).unwrap();
        s.run("references.updateFields", &json!({})).unwrap();
        assert_eq!(fields(&s)[0].1, "KILNS");
        assert!(s.doc.plain_text(StoryRef::Body).ends_with("See here.KILNS1"), "{}", s.doc.plain_text(StoryRef::Body));
    }

    #[test]
    fn cross_reference_to_a_bookmark_shows_its_text() {
        let mut s = Session::new(wordcraft_doc::Document::from_text("Kilns are hot.\nSee "));
        s.run("select.text", &json!({"text": "Kilns"})).unwrap();
        s.run("insert.bookmark", &json!({"name": "Spot"})).unwrap();
        s.run("caret.docEnd", &json!({})).unwrap();
        s.run("insert.crossReference", &json!({"to": "bookmark", "target": "Spot"})).unwrap();
        assert_eq!(s.doc.bookmarks().len(), 1, "no extra bookmark for a named one");
        assert_eq!(fields(&s), vec![("REF Spot \\h".to_string(), "Kilns".to_string())]);
    }

    #[test]
    fn cross_reference_inside_its_own_target_paragraph() {
        let mut s = Session::new(wordcraft_doc::Document::from_text("Kilns"));
        s.run("styles.apply", &json!({"style": "Heading 1"})).unwrap();
        s.run("caret.docEnd", &json!({})).unwrap();
        s.run("insert.crossReference", &json!({"to": "heading", "target": "Kilns"})).unwrap();
        assert_eq!(s.doc.plain_text(StoryRef::Body), "KilnsKilns");
        // The field sits after the bookmark, so it doesn't mark itself.
        let name = s.doc.bookmarks()[0].0.clone();
        assert_eq!(crate::cmd::references::bookmark_text(&s.doc, &name).as_deref(), Some("Kilns"));
    }

    fn src_of(author: &str, year: &str) -> Value {
        json!({"author": author, "year": year, "title": "T"})
    }

    #[test]
    fn editing_author_and_year_keeps_the_tag_and_updates_the_citation_text() {
        let mut s = Session::new(wordcraft_doc::Document::new());
        s.run("references.sources", &json!({"add": src_of("Smith, Ann", "2020")})).unwrap();
        assert_eq!(s.doc.sources[0].tag, "Smi20");
        s.run("references.citation", &json!({"tag": "Smi20"})).unwrap();
        // What the Edit Source dialog sends: the unchanged tag with the new author and year.
        let mut edited = src_of("Jones, Bo", "2021");
        edited["tag"] = json!("Smi20");
        s.run("references.sources", &json!({"add": edited})).unwrap();
        assert_eq!(s.doc.sources.len(), 1);
        assert_eq!(s.doc.sources[0].tag, "Smi20");
        let f = fields(&s);
        assert_eq!(f[0].0, "CITATION Smi20");
        assert!(f[0].1.contains("Jones") && f[0].1.contains("2021"), "{:?}", f[0].1);
    }

    #[test]
    fn a_new_sources_tag_is_three_letters_and_two_year_digits() {
        let mut s = Session::new(wordcraft_doc::Document::new());
        for (a, y) in [
            ("Kim", "2000"),
            ("Kim", "2000"),
            ("Kim", "2000"),
            ("Lee", "2019"),
            ("Li", "1999"),
            ("", ""),
            ("", ""),
            ("Ünal, Ece", "7"),
            ("O'Neil", "20x5"),
        ] {
            s.run("references.sources", &json!({"add": src_of(a, y)})).unwrap();
        }
        let tags: Vec<_> = s.doc.sources.iter().map(|x| x.tag.as_str()).collect();
        assert_eq!(tags, ["Kim00", "Kim001", "Kim002", "Lee19", "Li99", "Src", "Src1", "Üna7", "ONe05"]);
        let mut s = Session::new(wordcraft_doc::Document::new());
        let mut a = src_of("Lee", "2019");
        a["tag"] = json!("MyTag");
        s.run("references.sources", &json!({"add": a})).unwrap();
        assert_eq!(s.doc.sources[0].tag, "MyTag");
    }

    fn two_sources(s: &mut Session, a_title: &str) {
        let a = json!({"tag": "A", "author": "Rivera, Alex; Chen, Mei", "title": a_title, "year": "2021", "publisher": "Harbor Books", "city": "Portland"});
        let b = json!({"tag": "B", "author": "Okafor, Ngozi", "title": "Kilns and Clay", "year": "2019", "journal": "Studio Quarterly", "volume": "4", "pages": "10-20"});
        s.run("references.sources", &json!({"add": a})).unwrap();
        s.run("references.sources", &json!({"add": b})).unwrap();
    }

    #[test]
    fn author_date_styles_cite_and_list() {
        let mut s = Session::new(wordcraft_doc::Document::new());
        two_sources(&mut s, "Shared Spaces");
        s.run("references.citation", &json!({"tag": "A", "pages": "12"})).unwrap();
        s.run("references.citation", &json!({"tag": "B"})).unwrap();
        s.run("text.newParagraph", &json!({})).unwrap();
        s.run("references.bibliography", &json!({})).unwrap();
        let cases = [
            (
                "Harvard (Anglia)",
                "(Rivera and Chen, 2021, p. 12)(Okafor, 2019)",
                "Okafor, N., 2019. Kilns and Clay. Studio Quarterly, 4, pp. 10-20.\nRivera, A. and Chen, M., 2021. Shared Spaces. Portland: Harbor Books.",
            ),
            (
                "ISO 690 - First Element and Date",
                "(Rivera and Chen 2021, p. 12)(Okafor 2019)",
                "OKAFOR, Ngozi, 2019. Kilns and Clay. Studio Quarterly, 4, 10-20.\nRIVERA, Alex and CHEN, Mei, 2021. Shared Spaces. Portland: Harbor Books.",
            ),
            (
                "turabian",
                "(Rivera and Chen 2021, 12)(Okafor 2019)",
                "Okafor, Ngozi. “Kilns and Clay.” Studio Quarterly 4 (2019): 10-20.\nRivera, Alex, and Mei Chen. Shared Spaces. Portland: Harbor Books, 2021.",
            ),
        ];
        for (style, cites, list) in cases {
            s.run("references.citationStyle", &json!({"style": style})).unwrap();
            let t = s.doc.plain_text(StoryRef::Body);
            assert!(t.starts_with(cites) && t.contains(list), "{style}: {t}");
        }
        assert!(s.run("references.citationStyle", &json!({"style": "Vancouver"})).unwrap_err().to_string().contains("Turabian"));
        assert_eq!(s.bib_style, "Turabian", "an unknown style keeps the current one");
    }

    /// #2640: Manage Sources › Import reads a Word bibliography file (`b:Sources`: Word's own
    /// `Sources.xml`) into the document. An untagged source gets a tag like a new one; a tag
    /// already here is left alone; the whole import is one undo step; and a file that isn't
    /// `b:Sources` is an error, not a silent success.
    #[test]
    fn import_sources_reads_a_word_bibliography_file() {
        let dir = std::env::temp_dir().join(format!("wc-import-sources-{}", std::process::id()));
        let _ = std::fs::remove_dir_all(&dir);
        std::fs::create_dir_all(&dir).unwrap();
        let path = dir.join("Sources.xml");
        let xml = br#"<?xml version="1.0" encoding="UTF-8" standalone="no"?>
<b:Sources xmlns:b="http://schemas.openxmlformats.org/officeDocument/2006/bibliography"
           xmlns="http://schemas.openxmlformats.org/officeDocument/2006/bibliography"
           SelectedStyle="\APA.XSL" StyleName="APA">
  <b:Source><b:Tag>Riv21</b:Tag><b:SourceType>Book</b:SourceType>
    <b:Author><b:Author><b:NameList><b:Person><b:Last>Rivera</b:Last><b:First>Alex</b:First></b:Person></b:NameList></b:Author></b:Author>
    <b:Title>Shared Spaces</b:Title><b:Year>2021</b:Year><b:City>Portland</b:City><b:Publisher>Harbor Books</b:Publisher>
  </b:Source>
  <b:Source><b:SourceType>Book</b:SourceType></b:Source>
  <b:Source><b:SourceType>JournalArticle</b:SourceType>
    <b:Author><b:Author><b:NameList><b:Person><b:Last>Okafor</b:Last><b:First>Ngozi</b:First></b:Person></b:NameList></b:Author></b:Author>
    <b:Title>Kilns and Clay</b:Title><b:JournalName>Studio Quarterly</b:JournalName><b:Year>2019</b:Year>
  </b:Source>
</b:Sources>"#;
        std::fs::write(&path, xml).unwrap();
        let path = path.to_string_lossy().to_string();

        let mut s = Session::new(wordcraft_doc::Document::new());
        s.run("references.sources", &json!({"import": path})).unwrap();
        assert_eq!(s.doc.sources.len(), 2, "the source with no content is dropped");
        assert_eq!(s.doc.sources[0].tag, "Riv21");
        assert_eq!((s.doc.sources[0].author.as_str(), s.doc.sources[0].title.as_str()), ("Rivera, Alex", "Shared Spaces"));
        assert_eq!(
            (s.doc.sources[1].tag.as_str(), s.doc.sources[1].kind.as_str()),
            ("Oka19", "article"),
            "an untagged source gets a tag like a new one"
        );
        assert!(s.dirty, "the document is marked edited");

        // The import is one undo step, and importing again adds nothing.
        s.run("edit.undo", &json!({})).unwrap();
        assert!(s.doc.sources.is_empty(), "one undo takes the whole import back");
        s.run("references.sources", &json!({"import": path})).unwrap();
        s.doc.sources[0].title = "Shared Spaces, revised".into();
        s.run("references.sources", &json!({"import": path})).unwrap();
        assert_eq!(s.doc.sources.len(), 2, "a tag already here is not added again");
        assert_eq!(s.doc.sources[0].title, "Shared Spaces, revised", "an import never overwrites an edited source");
        assert_eq!(s.doc.sources[1].tag, "Oka19", "the untagged source keeps the tag it was given");

        // Anything that isn't b:Sources is refused.
        let bad = dir.join("not-sources.xml");
        std::fs::write(&bad, "<html><body>no</body></html>").unwrap();
        let e = s.run("references.sources", &json!({"import": bad.to_string_lossy()})).unwrap_err();
        assert!(e.to_string().contains("not a Word document"), "{e}");
        let _ = std::fs::remove_dir_all(&dir);
    }

    /// #2640: untagged sources that share a tag stem get the tag, then `1`, `2`… in import order,
    /// and a second import adds none of them (the untagged one matches by title).
    #[test]
    fn an_import_tags_each_untagged_source_once() {
        let dir = std::env::temp_dir().join(format!("wc-import-tags-{}", std::process::id()));
        let _ = std::fs::remove_dir_all(&dir);
        std::fs::create_dir_all(&dir).unwrap();
        let path = dir.join("Sources.xml");
        std::fs::write(
            &path,
            r#"<b:Sources xmlns:b="http://schemas.openxmlformats.org/officeDocument/2006/bibliography">
  <b:Source><b:SourceType>Book</b:SourceType><b:Title>First Work</b:Title><b:Year>2021</b:Year>
    <b:Author><b:Author><b:NameList><b:Person><b:Last>Rivera</b:Last><b:First>Alex</b:First></b:Person></b:NameList></b:Author></b:Author></b:Source>
  <b:Source><b:SourceType>Book</b:SourceType><b:Title>Second Work</b:Title><b:Year>2021</b:Year>
    <b:Author><b:Author><b:NameList><b:Person><b:Last>Rivera</b:Last><b:First>Alex</b:First></b:Person></b:NameList></b:Author></b:Author></b:Source>
  <b:Source><b:SourceType>Book</b:SourceType><b:Title>Third Work</b:Title></b:Source>
</b:Sources>"#,
        )
        .unwrap();
        let path = path.to_string_lossy().to_string();
        let mut s = Session::new(wordcraft_doc::Document::new());
        s.run("references.sources", &json!({"import": path})).unwrap();
        let tags: Vec<_> = s.doc.sources.iter().map(|x| x.tag.as_str()).collect();
        assert_eq!(tags, ["Riv21", "Riv211", "Src"], "{tags:?}");
        s.run("references.sources", &json!({"import": path})).unwrap();
        assert_eq!(s.doc.sources.len(), 3, "the tags are here and the untagged one matches by title");
        let _ = std::fs::remove_dir_all(&dir);
    }

    #[test]
    fn numbered_styles_count_in_order_of_first_citation() {
        let mut s = Session::new(wordcraft_doc::Document::new());
        two_sources(&mut s, "Shared Spaces");
        s.run("references.citationStyle", &json!({"style": "ISO 690 (numerical)"})).unwrap();
        s.run("text.insert", &json!({"text": "Later "})).unwrap();
        s.run("references.citation", &json!({"tag": "A", "pages": "12"})).unwrap();
        assert_eq!(fields(&s)[0].1, "(1, p. 12)");
        // Citing B earlier in the text makes it number 1 and renumbers A.
        s.run("caret.docStart", &json!({})).unwrap();
        s.run("references.citation", &json!({"tag": "B"})).unwrap();
        s.run("caret.docEnd", &json!({})).unwrap();
        s.run("text.newParagraph", &json!({})).unwrap();
        s.run("references.bibliography", &json!({})).unwrap();
        let cases = [
            (
                "ISO 690 (numerical)",
                "(1)Later (2, p. 12)",
                "1. OKAFOR, Ngozi. Kilns and Clay. Studio Quarterly. 2019, 4, 10-20.\n2. RIVERA, Alex and CHEN, Mei. Shared Spaces. Portland: Harbor Books, 2021.",
            ),
            (
                "GB/T 7714",
                "[1]Later [2, 12]",
                "[1] OKAFOR N. Kilns and Clay[J]. Studio Quarterly, 2019, 4: 10-20.\n[2] RIVERA A, CHEN M. Shared Spaces[M]. Portland: Harbor Books, 2021.",
            ),
            (
                "SIST02",
                "(1)Later (2, p. 12)",
                "1) Okafor, Ngozi. Kilns and Clay. Studio Quarterly. 2019, vol. 4, p. 10-20.\n2) Rivera, Alex; Chen, Mei. Shared Spaces. Portland, Harbor Books, 2021.",
            ),
            (
                "IEEE",
                "[1]Later [2]",
                "[1] N. Okafor, “Kilns and Clay,” Studio Quarterly, 4, 10-20, 2019.\n[2] A. Rivera, M. Chen, “Shared Spaces,” Portland: Harbor Books, 2021.",
            ),
        ];
        for (style, cites, list) in cases {
            s.run("references.citationStyle", &json!({"style": style})).unwrap();
            let t = s.doc.plain_text(StoryRef::Body);
            assert!(t.starts_with(cites) && t.contains(list), "{style}: {t}");
        }
    }

    #[test]
    fn gost_numbers_follow_its_name_or_title_sort() {
        let mut s = Session::new(wordcraft_doc::Document::new());
        two_sources(&mut s, "Atlas of Studios");
        s.run("references.citation", &json!({"tag": "A"})).unwrap();
        s.run("references.citation", &json!({"tag": "B", "pages": "15"})).unwrap();
        s.run("text.newParagraph", &json!({})).unwrap();
        s.run("references.bibliography", &json!({})).unwrap();
        let okafor = "Okafor N. Kilns and Clay // Studio Quarterly. – 2019. – Vol. 4. – P. 10-20.";
        let rivera = "Rivera A., Chen M. Atlas of Studios. – Portland : Harbor Books, 2021.";
        // Word's style and style-sheet names pick the same styles.
        s.run("references.citationStyle", &json!({"style": "\\GostName.XSL"})).unwrap();
        assert_eq!(s.bib_style, "GOST (name sort)");
        let t = s.doc.plain_text(StoryRef::Body);
        assert!(t.starts_with("[2][1, p. 15]") && t.contains(&format!("1. {okafor}\n2. {rivera}")), "{t}");
        s.run("references.citationStyle", &json!({"style": "GOST - Title Sort"})).unwrap();
        let t = s.doc.plain_text(StoryRef::Body);
        assert!(t.starts_with("[1][2, p. 15]") && t.contains(&format!("1. {rivera}\n2. {okafor}")), "{t}");
        assert_eq!(find_style("ISO 690 - Numerical Reference").map(|x| x.name), Some("ISO 690 (numerical)"));
        assert_eq!(find_style("Harvard - Anglia 2008").map(|x| x.name), Some("Harvard (Anglia)"));
        assert!(find_style("").is_none() && find_style("Vancouver").is_none());
    }
}