gedcomkit 0.1.10

//! Converting a document between GEDCOM 5.5.1 and 7.0.
//!
//! Reading spans 5.5 to 7.1 and a file should be written in the version it
//! declares, so nothing here happens because a file was opened: conversion is
//! an explicit operation, it can be previewed before it runs, and it reports
//! every construct it touched and every one it could not carry.
//!
//! The report is the deliverable as much as the document is. A conversion that
//! silently drops what the target version cannot hold is how a tree loses
//! decades of work one release at a time, so anything unsupported is named,
//! counted, and — wherever the alternative is losing it — left in the document
//! as an extension rather than deleted.
//!
//! An application with extension tags of its own passes them to
//! [`to_version_7_with`], because version 7 wants every extension declared and
//! only the application knows the URIs its own tags are documented at.

use crate::dates::GedcomDate;
use crate::{Document, Node, schema, tags};
use std::fmt::Write as _;

/// What happened to one construct.
#[derive(Clone, Copy, Debug, Eq, PartialEq)]
#[non_exhaustive]
#[cfg_attr(feature = "serde", derive(serde::Serialize))]
pub enum ConversionKind {
    /// Rewritten into the target version's own spelling, losing nothing.
    Converted,
    /// Removed because the target version has no place for it and the
    /// information is carried elsewhere — a character set declaration in a
    /// format that is always UTF-8.
    Dropped,
    /// Kept as written because dropping it would lose something, even though
    /// the target version does not define it.
    Kept,
}

impl ConversionKind {
    /// The word the interface uses.
    #[must_use]
    pub const fn label(self) -> &'static str {
        match self {
            Self::Converted => "converted",
            Self::Dropped => "dropped",
            Self::Kept => "kept as written",
        }
    }
}

/// One construct the conversion touched.
#[derive(Clone, Debug, Eq, PartialEq)]
#[non_exhaustive]
#[cfg_attr(feature = "serde", derive(serde::Serialize))]
#[cfg_attr(feature = "serde", serde(rename_all = "camelCase"))]
pub struct ConversionNote {
    /// The record it sat in, when that record has an identifier.
    pub record: Option<String>,
    /// The tag as it was before the conversion.
    pub tag: String,
    /// What happened to it.
    pub kind: ConversionKind,
    /// Why, in the words a person would use.
    pub detail: String,
}

/// A conversion: the document it would produce, and everything it did.
#[derive(Clone, Debug)]
#[non_exhaustive]
pub struct Conversion {
    /// The converted document. Nothing is written anywhere by this module.
    pub document: Document,
    /// The version the document declared.
    pub from: String,
    /// The version it now declares.
    pub to: String,
    /// Every construct touched, in document order.
    pub notes: Vec<ConversionNote>,
}

impl Conversion {
    /// How many notes of each kind, for a summary line.
    #[must_use]
    pub fn counts(&self) -> (usize, usize, usize) {
        let mut converted = 0;
        let mut dropped = 0;
        let mut kept = 0;
        for note in &self.notes {
            match note.kind {
                ConversionKind::Converted => converted += 1,
                ConversionKind::Dropped => dropped += 1,
                ConversionKind::Kept => kept += 1,
            }
        }
        (converted, dropped, kept)
    }
}

/// The URI an extension nobody documented is declared against.
///
/// Version 7 requires every extension tag to be declared, and a file full of
/// somebody else's undeclared extensions cannot be made conformant without
/// saying *something*. This says exactly what is true: the tag came from
/// another application and the converting one does not know what it means.
/// The domain is RFC 2606's `.example`, which can never be registered and so
/// can never be mistaken for real documentation.
const UNDOCUMENTED: &str = "https://gedcom.example/undocumented";

struct Pass {
    notes: Vec<ConversionNote>,
    record: Option<String>,
}

impl Pass {
    fn note(&mut self, tag: &str, kind: ConversionKind, detail: impl Into<String>) {
        self.notes.push(ConversionNote {
            record: self.record.clone(),
            tag: tag.to_owned(),
            kind,
            detail: detail.into(),
        });
    }
}

/// Converts a document to GEDCOM 7.0.
///
/// Every extension found is declared as undocumented; an application with
/// extension tags of its own should use [`to_version_7_with`] so they are
/// declared against their real URIs.
#[must_use]
pub fn to_version_7(document: &Document) -> Conversion {
    to_version_7_with(document, &[])
}

/// Converts a document to GEDCOM 7.0.
///
/// `owned` names the extension tags the caller writes itself, each with the
/// URI it is documented at. Those are declared by name; any other extension in
/// the file is declared as undocumented rather than dropped, because it is
/// somebody's data either way.
#[must_use]
pub fn to_version_7_with(document: &Document, owned: &[(&str, &str)]) -> Conversion {
    let from = document.version().unwrap_or_else(|| "unstated".to_owned());
    let mut converted = document.clone();
    let mut pass = Pass {
        notes: Vec::new(),
        record: None,
    };

    // A note record becomes a shared note record, and every pointer at one has
    // to follow — a `NOTE @N1@` in version 7 is an inline note whose payload
    // happens to look like a pointer, which is a different structure entirely.
    let shared: Vec<String> = converted
        .records_of("NOTE")
        .filter_map(|record| record.xref.clone())
        .collect();

    for record in &mut converted.records {
        pass.record = record.xref.clone();
        if record.tag == "HEAD" {
            head_to_7(record, &mut pass);
        }
        if record.tag == "NOTE" && record.xref.is_some() {
            "SNOTE".clone_into(&mut record.tag);
            record.forget_verbatim();
            pass.note(
                "NOTE",
                ConversionKind::Converted,
                "a note record is a shared note record in version 7",
            );
        }
        // The record itself, not only its substructures: a media object is a
        // record, and its `FORM` is the one that has to move.
        structures_to_7_node(record, &shared, &mut pass);
    }

    // Every extension the file carries has to be declared, ours by name and
    // anybody else's as what it is.
    let extension_tags = extension_tags(&converted);
    if let Some(header) = converted.records.iter_mut().find(|node| node.tag == "HEAD") {
        pass.record = None;
        for tag in extension_tags {
            if schema::is_declared(header, &tag) {
                continue;
            }
            if let Some((_, uri)) = owned.iter().find(|(own, _)| *own == tag) {
                schema::declare(header, &tag, uri);
                header.forget_verbatim();
                pass.note(
                    &tag,
                    ConversionKind::Converted,
                    "declared in the header, which version 7 requires of every extension",
                );
            } else {
                schema::declare(header, &tag, UNDOCUMENTED);
                header.forget_verbatim();
                pass.note(
                    &tag,
                    ConversionKind::Kept,
                    "another application's extension, declared as undocumented rather than dropped",
                );
            }
        }
    }

    Conversion {
        document: converted,
        from,
        to: "7.0".to_owned(),
        notes: pass.notes,
    }
}

fn head_to_7(header: &mut Node, pass: &mut Pass) {
    header.forget_verbatim();
    if let Some(gedc) = header.first_mut("GEDC") {
        gedc.forget_verbatim();
        if let Some(vers) = gedc.first_mut("VERS") {
            vers.set_value(Some("7.0".to_owned()));
        } else {
            gedc.push(Node::with_value("VERS", "7.0"));
        }
        // `FORM` under `GEDC` said the file was lineage-linked, which is the
        // only thing version 7 is.
        if remove_child(gedc, "FORM") {
            pass.note(
                "GEDC.FORM",
                ConversionKind::Dropped,
                "version 7 has one form and does not name it",
            );
        }
    } else {
        // A header with no `GEDC` at all — a minimal or careless 5.x file.
        // Version 7 requires the declaration, so it is created whole.
        header
            .children
            .insert(0, Node::new("GEDC").child(Node::with_value("VERS", "7.0")));
        pass.note(
            "GEDC",
            ConversionKind::Converted,
            "the header declared no GEDCOM version, and version 7 requires it",
        );
    }
    if remove_child(header, "CHAR") {
        pass.note(
            "CHAR",
            ConversionKind::Dropped,
            "version 7 is always UTF-8, so a character set declaration says nothing",
        );
    }
}

fn structures_to_7_node(node: &mut Node, shared: &[String], pass: &mut Pass) -> bool {
    let mut changed = false;

    if node.tag == "_UID" {
        "UID".clone_into(&mut node.tag);
        changed = true;
        pass.note(
            "_UID",
            ConversionKind::Converted,
            "version 7 defines UID, so the extension is no longer needed",
        );
    }

    // A pointer at what is now a shared note record has to say so.
    if node.tag == "NOTE"
        && let Some(pointer) = node.pointer()
        && shared.iter().any(|xref| xref.eq_ignore_ascii_case(pointer))
    {
        "SNOTE".clone_into(&mut node.tag);
        changed = true;
        pass.note(
            "NOTE",
            ConversionKind::Converted,
            "a pointer to a note record is a shared note pointer in version 7",
        );
    }

    if node.tag == "DATE" {
        changed |= date_to_7(node, pass);
    }

    // A multimedia record's files: `FORM` moves from the object to each
    // `FILE` (5.5), and each file's path, format and medium take version 7's
    // shape (5.5.1 already nests `FORM` under `FILE`, with `TYPE` beneath).
    if node.tag == "OBJE" && node.first("FILE").is_some() {
        changed |= media_form_to_7(node, pass);
    }

    for child in &mut node.children {
        changed |= structures_to_7_node(child, shared, pass);
    }
    if changed {
        node.forget_verbatim();
    }
    changed
}

/// Rewrites a `DATE` payload into version 7's grammar wherever it is not
/// already written that way.
///
/// An empty payload is already valid version 7 (`DateValue` may be empty), so
/// it is left alone. Otherwise the payload is read and rendered in version
/// 7's own spelling ([`GedcomDate::to_gedcom_7`]); when that rendering is
/// exactly the payload already on the line, nothing is touched — the whole
/// point of checking is to leave a conformant file byte for byte as it was.
/// When the rendering differs, or none exists (an unparsed expression, or a
/// reading version 7's grammar cannot spell at all — a dual year, a calendar
/// this reader cannot place), nothing is lost: the reading's own phrase (a
/// 5.5.1 `INT date (phrase)`'s parenthetical, or the whole of an unparsed
/// expression) survives in a `PHRASE` substructure, and when there was no
/// such phrase to begin with, the original payload does instead — as the
/// date itself when a rendering exists, or as the whole of an otherwise-empty
/// `DATE` when none does. Either way the change is named in a note.
fn date_to_7(node: &mut Node, pass: &mut Pass) -> bool {
    let payload = node.logical_value();
    if payload.trim().is_empty() {
        return false;
    }
    let date = GedcomDate::parse(&payload);
    let kept = date.phrase.clone().unwrap_or_else(|| payload.clone());
    let reading = date.to_gedcom_7();
    if reading.as_deref() == Some(payload.as_str()) {
        return false;
    }
    // A `DATE` that already carries a `PHRASE` has nowhere to keep the
    // original payload: the grammar allows one `PHRASE`, and merging two
    // pieces of prose is a decision this conversion will not make for the
    // caller. It is left as it is, and said so.
    if node.first("PHRASE").is_some() {
        pass.note(
            "DATE",
            ConversionKind::Kept,
            format!(
                "{payload:?} is not version 7's date grammar, but the DATE already has a PHRASE, so the original cannot be kept beside a rewrite; it is left as it is"
            ),
        );
        return false;
    }
    // `set_logical_value` also takes away any `CONT`/`CONC` lines the old
    // payload was split across, which would otherwise be folded into the new
    // one again on the next read.
    node.set_logical_value(reading.as_deref().unwrap_or(""));
    node.push(Node::with_value("PHRASE", kept));
    let detail = if reading.is_some() {
        format!(
            "{payload:?} is not version 7's date grammar; the reading it converted to is kept as the date, and the original as a PHRASE"
        )
    } else {
        format!(
            "version 7's date grammar cannot hold {payload:?}; the DATE is left empty and the original kept as a PHRASE"
        )
    };
    pass.note("DATE", ConversionKind::Converted, detail);
    true
}

/// Brings a multimedia record's files into version 7's shape: a `FORM`
/// written on the object (5.5) moves to each file; then every file's path
/// becomes a URI reference, its `FORM` a media type, and the source medium
/// under it (`TYPE` in 5.5.1, `MEDI` in some producers) version 7's `MEDI`
/// enumeration. Whatever is already in version 7's shape is left byte for
/// byte, and every change is named in a note.
fn media_form_to_7(object: &mut Node, pass: &mut Pass) -> bool {
    let moved = object.first("FORM").cloned();
    let mut changed = moved.is_some();
    if let Some(form) = moved {
        remove_child(object, "FORM");
        for file in object.children.iter_mut().filter(|node| node.tag == "FILE") {
            if file.first("FORM").is_none() {
                file.push(form.clone());
                file.forget_verbatim();
            }
        }
        pass.note(
            "OBJE.FORM",
            ConversionKind::Converted,
            "version 7 records a file's format on the file, not on the object",
        );
    }
    for file in object.children.iter_mut().filter(|node| node.tag == "FILE") {
        changed |= file_to_7(file, pass);
    }
    changed
}

fn file_to_7(file: &mut Node, pass: &mut Pass) -> bool {
    let mut changed = false;
    let path = file.logical_value();
    let encoded = uri_reference(&path);
    if encoded != path {
        file.set_logical_value(&encoded);
        pass.note(
            "FILE",
            ConversionKind::Converted,
            format!(
                "{path:?} is not a URI reference, which version 7 requires of a file path; characters a URI cannot hold were percent-encoded"
            ),
        );
        changed = true;
    }
    if let Some(form) = file.first_mut("FORM") {
        let form_changed = form_to_7(form, pass);
        if form_changed {
            file.forget_verbatim();
        }
        changed |= form_changed;
    }
    changed
}

fn form_to_7(form: &mut Node, pass: &mut Pass) -> bool {
    let mut changed = false;
    let value = form.logical_value();
    let value = value.trim();
    // A media type has a slash; a 5.5.1 format is a bare extension.
    if !value.is_empty() && !value.contains('/') {
        match media_type_for_extension(value) {
            Some(media_type) => {
                form.set_logical_value(media_type);
                pass.note(
                    "FILE.FORM",
                    ConversionKind::Converted,
                    format!("the file extension {value:?} became the media type {media_type:?}"),
                );
                changed = true;
            }
            None => pass.note(
                "FILE.FORM",
                ConversionKind::Kept,
                format!(
                    "{value:?} has no known version 7 media type, so it is kept rather than guessed"
                ),
            ),
        }
    }
    for medium in form
        .children
        .iter_mut()
        .filter(|node| matches!(node.tag.as_str(), "TYPE" | "MEDI"))
    {
        changed |= medium_node_to_7(medium, pass);
    }
    changed
}

/// A source medium under `FORM`: 5.5.1's `TYPE` becomes `MEDI`, and its
/// value one of version 7's enumerated words — `OTHER` with the original kept
/// in a `PHRASE` when it is none of them.
fn medium_node_to_7(medium: &mut Node, pass: &mut Pass) -> bool {
    let changed = medium.tag == "TYPE";
    if changed {
        medium.set_tag("MEDI");
        pass.note(
            "FORM.MEDI",
            ConversionKind::Converted,
            "5.5.1's source medium under FORM is TYPE; version 7 calls it MEDI",
        );
    }
    let original = medium.logical_value();
    let (value, known) = medium_to_7(&original);
    if value == original {
        return changed;
    }
    if !known && medium.first("PHRASE").is_some() {
        pass.note(
            "FORM.MEDI",
            ConversionKind::Kept,
            format!(
                "the source medium {original:?} matches none of version 7's enumerated values, but it already has a PHRASE to keep the original in; it is left as it is"
            ),
        );
        return changed;
    }
    medium.set_logical_value(&value);
    if known {
        pass.note(
            "FORM.MEDI",
            ConversionKind::Converted,
            format!("the source medium {original:?} became the enumerated value {value}"),
        );
    } else {
        // The enumeration has no place for this source medium; `OTHER` says
        // so and `PHRASE` keeps what was actually written.
        medium.push(Node::with_value("PHRASE", original.trim()));
        pass.note(
            "FORM.MEDI",
            ConversionKind::Converted,
            format!(
                "the source medium {original:?} matches none of version 7's enumerated values; it became OTHER, with the original kept as a PHRASE"
            ),
        );
    }
    true
}

/// A file path as a URI reference: every byte a URI cannot hold as it stands —
/// a space, a backslash, anything outside ASCII — percent-encoded as UTF-8,
/// and a `%` that does not already begin an escape encoded as `%25`. A path
/// that is already a URI reference comes back unchanged.
fn uri_reference(path: &str) -> String {
    const ALLOWED: &[u8] = b"-._~:/?#[]@!$&'()*+,;=";
    let bytes = path.as_bytes();
    let mut out = String::with_capacity(path.len());
    for (index, &byte) in bytes.iter().enumerate() {
        let escape_begins = byte == b'%'
            && bytes.get(index + 1).is_some_and(u8::is_ascii_hexdigit)
            && bytes.get(index + 2).is_some_and(u8::is_ascii_hexdigit);
        if byte.is_ascii_alphanumeric() || ALLOWED.contains(&byte) || escape_begins {
            out.push(char::from(byte));
        } else {
            // Writing to a String cannot fail.
            let _ = write!(out, "%{byte:02X}");
        }
    }
    out
}

/// The version 7 media type for a 5.5.1 file extension, or `None` when this
/// reader does not know one — kept rather than guessed, per
/// [`media_form_to_7`].
fn media_type_for_extension(extension: &str) -> Option<&'static str> {
    match extension.trim().to_ascii_lowercase().as_str() {
        "jpg" | "jpeg" => Some("image/jpeg"),
        "png" => Some("image/png"),
        "gif" => Some("image/gif"),
        "bmp" => Some("image/bmp"),
        "tif" | "tiff" => Some("image/tiff"),
        "pdf" => Some("application/pdf"),
        "txt" => Some("text/plain"),
        "html" | "htm" => Some("text/html"),
        "wav" => Some("audio/x-wav"),
        "mp3" => Some("audio/mpeg"),
        "heic" => Some("image/heic"),
        "webp" => Some("image/webp"),
        "svg" => Some("image/svg+xml"),
        "json" => Some("application/json"),
        "xml" => Some("application/xml"),
        "eml" => Some("message/rfc822"),
        "mp4" => Some("video/mp4"),
        _ => None,
    }
}

/// A source medium in version 7's `FORM.MEDI` enumeration, upper case, and
/// whether the original value was one of the enumeration's own words —
/// `false` means it was folded into `OTHER`.
fn medium_to_7(medium: &str) -> (String, bool) {
    let upper = medium.trim().to_ascii_uppercase();
    let known = matches!(
        upper.as_str(),
        "AUDIO"
            | "BOOK"
            | "CARD"
            | "ELECTRONIC"
            | "FICHE"
            | "FILM"
            | "MAGAZINE"
            | "MANUSCRIPT"
            | "MAP"
            | "NEWSPAPER"
            | "PHOTO"
            | "TOMBSTONE"
            | "VIDEO"
    );
    if known {
        (upper, true)
    } else {
        ("OTHER".to_owned(), false)
    }
}

/// Converts a document to GEDCOM 5.5.1.
#[must_use]
pub fn to_version_5_5_1(document: &Document) -> Conversion {
    let from = document.version().unwrap_or_else(|| "unstated".to_owned());
    let mut converted = document.clone();
    let mut pass = Pass {
        notes: Vec::new(),
        record: None,
    };

    for record in &mut converted.records {
        pass.record = record.xref.clone();
        if record.tag == "HEAD" {
            head_to_5(record, &mut pass);
        }
        if record.tag == "SNOTE" {
            "NOTE".clone_into(&mut record.tag);
            record.forget_verbatim();
            pass.note(
                "SNOTE",
                ConversionKind::Converted,
                "5.5.1 has one kind of note record",
            );
        }
        structures_to_5_node(record, &mut pass);
    }

    Conversion {
        document: converted,
        from,
        to: "5.5.1".to_owned(),
        notes: pass.notes,
    }
}

fn head_to_5(header: &mut Node, pass: &mut Pass) {
    header.forget_verbatim();
    if let Some(gedc) = header.first_mut("GEDC") {
        gedc.forget_verbatim();
        if let Some(vers) = gedc.first_mut("VERS") {
            vers.set_value(Some("5.5.1".to_owned()));
        } else {
            gedc.push(Node::with_value("VERS", "5.5.1"));
        }
        if gedc.first("FORM").is_none() {
            gedc.push(Node::with_value("FORM", "LINEAGE-LINKED"));
        }
    } else {
        // 5.5.1 requires the declaration too, `FORM` included.
        header.children.insert(
            0,
            Node::new("GEDC")
                .child(Node::with_value("VERS", "5.5.1"))
                .child(Node::with_value("FORM", "LINEAGE-LINKED")),
        );
        pass.note(
            "GEDC",
            ConversionKind::Converted,
            "the header declared no GEDCOM version, and 5.5.1 requires it",
        );
    }
    if header.first("CHAR").is_none() {
        header.push(Node::with_value("CHAR", "UTF-8"));
        pass.note(
            "CHAR",
            ConversionKind::Converted,
            "5.5.1 requires a character set, and the file is written as UTF-8",
        );
    }
    // `SCHMA` declares extensions, which 5.5.1 has no mechanism for. The
    // extensions themselves stay; only the declaration goes.
    if remove_child(header, "SCHMA") {
        pass.note(
            "SCHMA",
            ConversionKind::Dropped,
            "5.5.1 has no way to declare an extension; the extensions themselves are kept",
        );
    }
}

fn structures_to_5_node(node: &mut Node, pass: &mut Pass) -> bool {
    let mut changed = false;

    if node.tag == "UID" {
        "_UID".clone_into(&mut node.tag);
        changed = true;
        pass.note(
            "UID",
            ConversionKind::Converted,
            "5.5.1 has no UID, and every producer writes _UID instead",
        );
    }

    if node.tag == "SNOTE" && node.pointer().is_some() {
        "NOTE".clone_into(&mut node.tag);
        changed = true;
        pass.note(
            "SNOTE",
            ConversionKind::Converted,
            "a shared note pointer is an ordinary note pointer in 5.5.1",
        );
    }

    // `DATE` with a `PHRASE` folds back into 5.5.1's interpreted form, which
    // is the same information in the older spelling.
    if node.tag == "DATE"
        && let Some(phrase) = node.value_of("PHRASE")
    {
        let payload = node.logical_value();
        remove_child(node, "PHRASE");
        node.set_value(Some(format!("INT {payload} ({phrase})")));
        changed = true;
        pass.note(
            "PHRASE",
            ConversionKind::Converted,
            "a dated phrase became 5.5.1's interpreted form",
        );
    }

    // Everything else version 7 added and 5.5.1 does not define is kept and
    // named. Deleting a translation or a creation date to satisfy a version
    // number would be the worst trade this application could make.
    if matches!(node.tag.as_str(), "TRAN" | "CREA" | "EXID" | "SDATE") {
        pass.note(
            &node.tag,
            ConversionKind::Kept,
            "version 7 defines this and 5.5.1 does not; it is kept rather than lost",
        );
    }

    for child in &mut node.children {
        changed |= structures_to_5_node(child, pass);
    }
    if changed {
        node.forget_verbatim();
    }
    changed
}

fn remove_child(node: &mut Node, tag: &str) -> bool {
    let before = node.children.len();
    node.children.retain(|child| child.tag != tag);
    if node.children.len() != before {
        node.forget_verbatim();
        return true;
    }
    false
}

fn extension_tags(document: &Document) -> Vec<String> {
    let mut tags: Vec<String> = Vec::new();
    for record in &document.records {
        if record.tag == "HEAD" {
            continue;
        }
        for node in record.walk() {
            if tags::is_extension(&node.tag) && !tags.contains(&node.tag) {
                tags.push(node.tag.clone());
            }
        }
    }
    tags.sort();
    tags
}

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

    const OLD: &str = "0 HEAD\n\
1 GEDC\n\
2 VERS 5.5.1\n\
2 FORM LINEAGE-LINKED\n\
1 CHAR UTF-8\n\
0 @I1@ INDI\n\
1 NAME Rian /Testwright/\n\
1 _UID 4E7C9F2A\n\
1 BIRT\n\
2 DATE INT 1 JAN 1900 (from the family bible)\n\
1 NOTE @N1@\n\
1 _SOMEBODY _ELSE\n\
0 @N1@ NOTE\n\
1 CONC Written by hand.\n\
0 @M1@ OBJE\n\
1 FILE media/photo.jpg\n\
1 FORM jpeg\n\
2 MEDI photo\n\
0 TRLR\n";

    fn parse(text: &str) -> Document {
        Document::parse(text).expect("parse")
    }

    #[test]
    fn five_five_one_becomes_version_seven() {
        let outcome = to_version_7(&parse(OLD));
        let text = outcome.document.to_text();

        assert_eq!(outcome.from, "5.5.1");
        assert_eq!(outcome.to, "7.0");
        assert!(text.contains("2 VERS 7.0"), "{text}");
        assert!(!text.contains("CHAR"), "the character set went: {text}");
        assert!(!text.contains("2 FORM LINEAGE-LINKED"), "{text}");

        assert!(text.contains("0 @N1@ SNOTE"), "the record: {text}");
        assert!(
            text.contains("1 SNOTE @N1@"),
            "and the pointer at it: {text}"
        );

        assert!(text.contains("1 UID 4E7C9F2A"), "{text}");
        assert!(text.contains("2 DATE 1 JAN 1900"), "{text}");
        assert!(
            text.contains("3 PHRASE from the family bible"),
            "the reading is kept as a phrase: {text}"
        );

        // The media format moved onto the file, the extension became a media
        // type, and the source medium became the version 7 enumeration.
        assert!(text.contains("2 FORM image/jpeg"), "{text}");
        assert!(text.contains("3 MEDI PHOTO"), "{text}");

        // Somebody else's extension is declared rather than dropped.
        assert!(text.contains("2 TAG _SOMEBODY "), "{text}");
        assert!(text.contains("undocumented"), "{text}");
        assert!(text.contains("1 _SOMEBODY _ELSE"), "and kept: {text}");
    }

    #[test]
    fn version_seven_becomes_five_five_one() {
        let seven = to_version_7(&parse(OLD)).document;
        let outcome = to_version_5_5_1(&seven);
        let text = outcome.document.to_text();

        assert!(text.contains("2 VERS 5.5.1"), "{text}");
        assert!(text.contains("2 FORM LINEAGE-LINKED"), "{text}");
        assert!(text.contains("1 CHAR UTF-8"), "{text}");
        assert!(!text.contains("SCHMA"), "the declarations went: {text}");
        assert!(text.contains("0 @N1@ NOTE"), "{text}");
        assert!(text.contains("1 NOTE @N1@"), "{text}");
        assert!(text.contains("1 _UID 4E7C9F2A"), "{text}");
        assert!(
            text.contains("2 DATE INT 1 JAN 1900 (from the family bible)"),
            "the phrase folds back into the interpreted form: {text}"
        );
        // The extension itself survived both directions.
        assert!(text.contains("1 _SOMEBODY _ELSE"), "{text}");
    }

    #[test]
    fn the_report_names_every_construct_it_touched() {
        let outcome = to_version_7(&parse(OLD));
        let names: Vec<&str> = outcome.notes.iter().map(|note| note.tag.as_str()).collect();

        for expected in [
            "CHAR",
            "GEDC.FORM",
            "_UID",
            "NOTE",
            "DATE",
            "OBJE.FORM",
            "FILE.FORM",
            "FORM.MEDI",
        ] {
            assert!(
                names.contains(&expected),
                "{expected} missing from {names:?}"
            );
        }
        let (converted, dropped, kept) = outcome.counts();
        assert!(converted > 0 && dropped > 0, "{:?}", outcome.notes);
        assert_eq!(
            kept, 1,
            "only the undeclared extension: {:?}",
            outcome.notes
        );

        // A note names the record it sat in, so a report can be read against
        // the file rather than only counted.
        let uid = outcome
            .notes
            .iter()
            .find(|note| note.tag == "_UID")
            .expect("_UID note");
        assert_eq!(uid.record.as_deref(), Some("@I1@"));
    }

    #[test]
    fn an_unparsed_date_becomes_an_empty_date_with_a_phrase() {
        // Version 7 has no way to write "Infant" as a DATE payload, but it
        // does allow an empty one with a PHRASE, so that is what an
        // unparsed expression becomes rather than staying invalid text.
        let document =
            parse("0 HEAD\n1 GEDC\n2 VERS 5.5.1\n0 @I1@ INDI\n1 BIRT\n2 DATE Infant\n0 TRLR\n");
        let outcome = to_version_7(&document);
        let text = outcome.document.to_text();
        assert!(text.contains("2 DATE\n3 PHRASE Infant\n"), "{text}");
        let note = outcome
            .notes
            .iter()
            .find(|note| note.tag == "DATE")
            .expect("a note about it");
        assert_eq!(note.kind, ConversionKind::Converted);
        assert!(note.detail.contains("Infant"), "{}", note.detail);
    }

    #[test]
    fn dates_already_in_version_7_grammar_are_left_untouched() {
        let document = parse(
            "0 HEAD\n1 GEDC\n2 VERS 7.0\n0 @I1@ INDI\n1 BIRT\n2 DATE 29 MAY 1823\n\
1 DEAT\n2 DATE ABT 1900\n3 PHRASE around the turn of the century\n0 TRLR\n",
        );
        let outcome = to_version_7(&document);
        assert_eq!(
            outcome.document.to_text(),
            document.to_text(),
            "an already-conformant file is untouched byte for byte"
        );
        assert!(
            outcome.notes.iter().all(|note| note.tag != "DATE"),
            "nothing to report about a date already in version 7's grammar: {:?}",
            outcome.notes
        );
    }

    #[test]
    fn dates_the_lenient_reader_understands_but_5_5_1_grammar_that_is_not_7_0_grammar() {
        // Every one of these is read by the lenient parser but is not
        // GEDCOM 7.0's own DATE grammar, so each must be rewritten (with the
        // original kept in a PHRASE) rather than passed through unreported.
        let cases: &[(&str, &str)] = &[
            ("29 May 1823", "29 MAY 1823"),
            ("about 1575", "ABT 1575"),
            ("13. september 1900", "13 SEP 1900"),
            ("<1873>", "ABT 1873"),
            ("ABT Oct 1803", "ABT OCT 1803"),
            ("44 B.C.", "44 BCE"),
            ("@#DJULIAN@ 10 MAR 1700", "JULIAN 10 MAR 1700"),
        ];
        for (original, rendered) in cases {
            let document = parse(&format!(
                "0 HEAD\n1 GEDC\n2 VERS 5.5.1\n0 @I1@ INDI\n1 BIRT\n2 DATE {original}\n0 TRLR\n"
            ));
            let outcome = to_version_7(&document);
            let text = outcome.document.to_text();
            assert!(
                text.contains(&format!("2 DATE {rendered}\n3 PHRASE {original}\n")),
                "{original:?} -> {text}"
            );
            let note = outcome
                .notes
                .iter()
                .find(|note| note.tag == "DATE")
                .unwrap_or_else(|| panic!("a note about {original:?}"));
            assert_eq!(note.kind, ConversionKind::Converted, "{original:?}");
        }
    }

    #[test]
    fn a_dual_year_has_no_version_7_reading_and_becomes_an_empty_date_with_a_phrase() {
        let document =
            parse("0 HEAD\n1 GEDC\n2 VERS 5.5.1\n0 @I1@ INDI\n1 BIRT\n2 DATE 1740/41\n0 TRLR\n");
        let outcome = to_version_7(&document);
        let text = outcome.document.to_text();
        assert!(text.contains("2 DATE\n3 PHRASE 1740/41\n"), "{text}");
    }

    #[test]
    fn an_unrecognized_extension_is_kept_rather_than_guessed() {
        let document = parse(
            "0 HEAD\n1 GEDC\n2 VERS 5.5.1\n0 @M1@ OBJE\n1 FILE media/document.ole\n\
1 FORM ole\n0 TRLR\n",
        );
        let outcome = to_version_7(&document);
        let text = outcome.document.to_text();
        assert!(text.contains("2 FORM ole"), "kept as written: {text}");
        let note = outcome
            .notes
            .iter()
            .find(|note| note.tag == "FILE.FORM")
            .expect("a note about the extension");
        assert_eq!(note.kind, ConversionKind::Kept);
        assert!(note.detail.contains("ole"), "{}", note.detail);
    }

    #[test]
    fn an_unrecognized_source_medium_becomes_other_with_a_phrase() {
        let document = parse(
            "0 HEAD\n1 GEDC\n2 VERS 5.5.1\n0 @M1@ OBJE\n1 FILE media/clipping.jpg\n\
1 FORM jpg\n2 MEDI newspaper clipping\n0 TRLR\n",
        );
        let outcome = to_version_7(&document);
        let text = outcome.document.to_text();
        assert!(
            text.contains("3 MEDI OTHER\n4 PHRASE newspaper clipping\n"),
            "{text}"
        );
        let note = outcome
            .notes
            .iter()
            .find(|note| note.tag == "FORM.MEDI")
            .expect("a note about the medium");
        assert_eq!(note.kind, ConversionKind::Converted);
        assert!(
            note.detail.contains("newspaper clipping"),
            "{}",
            note.detail
        );
    }

    #[test]
    fn the_5_5_1_media_layout_under_file_is_converted() {
        let document = parse(
            "0 HEAD\n1 GEDC\n2 VERS 5.5.1\n0 @M1@ OBJE\n1 FILE media/my photo.JPG\n\
2 FORM JPG\n3 TYPE manuscript\n2 TITL A page\n0 TRLR\n",
        );
        let outcome = to_version_7(&document);
        let text = outcome.document.to_text();
        assert!(
            text.contains(
                "1 FILE media/my%20photo.JPG\n2 FORM image/jpeg\n3 MEDI MANUSCRIPT\n2 TITL A page\n"
            ),
            "{text}"
        );
        for tag in ["FILE", "FILE.FORM", "FORM.MEDI"] {
            assert!(
                outcome.notes.iter().any(|note| note.tag == tag),
                "{tag} is reported: {:?}",
                outcome.notes
            );
        }
    }

    #[test]
    fn a_path_that_is_already_a_uri_reference_is_untouched() {
        assert_eq!(uri_reference("media/a%20b.jpg"), "media/a%20b.jpg");
        assert_eq!(uri_reference("../x/y_z-1.png"), "../x/y_z-1.png");
        assert_eq!(uri_reference("50% off.jpg"), "50%25%20off.jpg");
        assert_eq!(uri_reference("C:\\a\\é.jpg"), "C:%5Ca%5C%C3%A9.jpg");
    }

    #[test]
    fn a_date_that_already_has_a_phrase_is_kept_and_reported() {
        let document = parse(
            "0 HEAD\n1 GEDC\n2 VERS 5.5.1\n0 @I1@ INDI\n1 BIRT\n2 DATE 29 May 1823\n\
3 PHRASE late May, the family said\n0 TRLR\n",
        );
        let outcome = to_version_7(&document);
        let text = outcome.document.to_text();
        assert!(
            text.contains("2 DATE 29 May 1823\n3 PHRASE late May, the family said\n"),
            "{text}"
        );
        assert_eq!(text.matches("PHRASE").count(), 1);
        assert!(
            outcome
                .notes
                .iter()
                .any(|note| note.tag == "DATE" && note.kind == ConversionKind::Kept)
        );
    }

    #[test]
    fn a_rewritten_date_loses_its_old_continuation_lines() {
        let document = parse(
            "0 HEAD\n1 GEDC\n2 VERS 5.5.1\n0 @I1@ INDI\n1 BIRT\n2 DATE 29 May\n3 CONC  1823\n0 TRLR\n",
        );
        let outcome = to_version_7(&document);
        let text = outcome.document.to_text();
        assert!(!text.contains("CONC"), "{text}");
        let reread = Document::parse(&text).expect("reads back");
        let date = reread
            .record("@I1@")
            .and_then(|person| person.first("BIRT"))
            .and_then(|birth| birth.first("DATE"))
            .expect("date");
        assert_eq!(date.logical_value(), "29 MAY 1823");
        assert_eq!(date.value_of("PHRASE").as_deref(), Some("29 May 1823"));
    }

    #[test]
    fn media_already_in_version_7_shape_is_left_untouched() {
        let document = parse(
            "0 HEAD\n1 GEDC\n2 VERS 7.0\n0 @M1@ OBJE\n1 FILE media/photo.jpg\n\
2 FORM image/jpeg\n3 MEDI PHOTO\n0 TRLR\n",
        );
        let outcome = to_version_7(&document);
        assert_eq!(
            outcome.document.to_text(),
            document.to_text(),
            "a file whose media is already version 7's shape is untouched byte for byte"
        );
        assert!(
            outcome.notes.iter().all(|note| note.tag != "OBJE.FORM"
                && note.tag != "FILE.FORM"
                && note.tag != "FORM.MEDI"),
            "nothing to report: {:?}",
            outcome.notes
        );
    }

    #[test]
    fn an_extension_the_caller_owns_is_declared_against_its_own_uri() {
        let document = parse(
            "0 HEAD\n1 GEDC\n2 VERS 5.5.1\n0 @I1@ INDI\n1 _MINE value\n1 _THEIRS value\n0 TRLR\n",
        );
        let outcome = to_version_7_with(&document, &[("_MINE", "https://mine.example/MINE")]);
        let text = outcome.document.to_text();

        assert!(
            text.contains("2 TAG _MINE https://mine.example/MINE"),
            "{text}"
        );
        assert!(
            text.contains("2 TAG _THEIRS https://gedcom.example/undocumented"),
            "{text}"
        );
        let mine = outcome
            .notes
            .iter()
            .find(|note| note.tag == "_MINE")
            .expect("a note about the owned tag");
        assert_eq!(mine.kind, ConversionKind::Converted);
    }

    #[test]
    fn converting_to_the_version_a_file_already_declares_changes_nothing_it_holds() {
        let seven = to_version_7(&parse(OLD)).document;
        let again = to_version_7(&seven);
        assert_eq!(
            again.document.to_text(),
            seven.to_text(),
            "a second conversion is a no-op"
        );
    }
}