write_pdfa2b.rs

//! Writes the minutes of a works council claiming the second part of the
//! archival standard, at level B.
//!
//! Level B promises one thing and one thing only: the page will look the same
//! in thirty years as it does today. Every face the page draws with is carried
//! inside the file, so nothing on it waits on a font that happens to be
//! installed. It says nothing about the words being searchable — that is level
//! U — nor about the page saying what it is made of, which is level A.
//!
//! Whether the result really is PDF/A-2B is not for us to say. `veraPDF` says,
//! and `scripts/check_archival.sh` asks it.
//!
//! The words of the page and of its metadata are held in `Words`, once per
//! language, and `HQF_PDF_LANG` picks which set the record is written in.
//!
//! Usage: `cargo run --example write_pdfa2b -- tmp/pdfa2b.pdf [font.ttf]`
//!        `HQF_PDF_LANG=fr cargo run --example write_pdfa2b -- tmp/pdfa2b_fr.pdf`

use std::env;
use std::fs;
use std::path::{Path, PathBuf};

use hqf_pdf::content::Content;
use hqf_pdf::layout::Align;
use hqf_pdf::metadata::xmp::{Metadata, PdfA};
use hqf_pdf::{Document, Error, Font, FontHandle, Page, TextFlow};

#[path = "shared/out.rs"]
mod out;

#[path = "shared/licence.rs"]
mod licence;

#[path = "shared/language.rs"]
mod language;

#[path = "shared/failure.rs"]
mod failure;

use language::Language;

/// The font the example draws with when none is given on the command line.
fn default_font() -> PathBuf {
    Path::new(env!("CARGO_MANIFEST_DIR"))
        .join("tests")
        .join("fonts")
        .join("DejaVuSans.ttf")
}

/// When the minutes were drawn up. The library never reads the clock — a
/// document that stamped itself with the time would be a different file on
/// every build — so the caller says when.
const ISSUED: &str = "2026-03-12T18:05:00+01:00";

/// Who the file says wrote it, and what produced it, in every language alike.
const AUTHOR: &str = "HQF Development";
const PRODUCER: &str = "hqf-pdf-server";

/// The words the minutes is written in, one set per language.
#[derive(Debug)]
struct Words {
    /// What the file says it is, in its metadata.
    title: &'static str,
    /// What the file says it is about, in its metadata.
    subject: &'static str,
    /// The heading at the top of the page.
    heading: &'static str,
    /// The line under the heading saying what the record is filed under.
    reference: &'static str,
    /// What the page says of the part it claims, in two runs set end to end
    /// with a space between them.
    body: [&'static str; 2],
    /// What each line of the record asks, one a line.
    labels: &'static str,
    /// What each line of the record answers, beside its label.
    answers: &'static str,
}

impl Words {
    /// The words the minutes is written in, in `language`.
    fn of(language: Language) -> &'static Self {
        language::pick(&WORDS, language)
    }
}

/// The minutes in English.
const ENGLISH: Words = Words {
    title: "Minutes of the works council, 12 March 2026",
    subject: "A record kept for thirty years, carrying nothing but itself",
    heading: "Minutes of the works council",
    reference: "Thursday 12 March 2026, 17:00, meeting room 2",
    body: [
        "This sheet claims the second part of the archival standard, at the level that promises \
            one thing and one thing only: the page will look the same in thirty years as it does \
            today. Every face it draws with is carried inside the file, every colour says what \
            it means, and nothing on the page waits on a program, a network or a font that \
            happens to be installed.",
        "The second part names no carried file the way the third does, so a record that has to \
            travel with a spreadsheet or an invoice claims the third part instead.",
    ],
    labels: "Present\nApologies\nNext sitting",
    answers: "Nine members of the council and two of the board\nOne member, on leave\nThursday 9 \
        April 2026, 17:00",
};

/// The minutes in French.
const FRENCH: Words = Words {
    title: "Procès-verbal du comité social et économique, 12 mars 2026",
    subject: "Un document conservé trente ans, qui ne contient rien d’autre que lui-même",
    heading: "Procès-verbal du CSE",
    reference: "Jeudi 12 mars 2026, 17 h, salle de réunion 2",
    body: [
        "Cette feuille déclare suivre la deuxième partie de la norme d’archivage, au \
niveau qui ne promet qu’une chose : dans trente ans, la page aura le même \
aspect qu’aujourd’hui. Chaque police employée est incorporée au fichier, \
chaque couleur dit ce qu’elle signifie, et rien ne dépend d’un programme, d’un \
réseau ou une police installée par hasard.",
        "La deuxième partie n’admet pas les fichiers joints que prévoit la troisième : \
un document qui doit voyager avec un tableur ou une facture se réclame donc de \
la troisième.",
    ],
    labels: "Présents\nExcusé\nProchaine séance",
    answers: "Neuf membres du comité et deux de la direction\nUn membre, en congé\nJeudi 9 avril \
        2026, 17 h",
};

/// The minutes in German.
const GERMAN: Words = Words {
    title: "Protokoll des Betriebsrats, 12. März 2026",
    subject: "Ein Dokument, das dreißig Jahre aufbewahrt wird und nichts als sich selbst enthält",
    heading: "Protokoll des Betriebsrats",
    reference: "Donnerstag, 12. März 2026, 17:00 Uhr, Besprechungsraum 2",
    body: [
        "Dieses Blatt erklärt Konformität mit dem zweiten Teil der Archivnorm, auf der \
Stufe, die genau eines zusagt: Die Seite sieht in dreißig Jahren so aus wie \
heute. Jede verwendete Schrift steckt in der Datei, jede Farbe gibt an, was \
sie bedeutet, und nichts auf der Seite wartet auf ein Programm, ein Netzwerk \
oder eine zufällig installierte Schrift.",
        "Der zweite Teil sieht mitgeführte Dateien nicht so vor wie der dritte; ein \
Dokument, das mit einer Tabelle oder einer Rechnung reisen muss, erklärt \
deshalb Konformität mit dem dritten Teil.",
    ],
    labels: "Anwesend\nEntschuldigt\nNächste Sitzung",
    answers: "Neun Ratsmitglieder und zwei aus der Leitung\nEin Mitglied, im \
Urlaub\nDonnerstag, 9. April 2026, 17:00 Uhr",
};

/// The minutes in Spanish.
const SPANISH: Words = Words {
    title: "Acta del comité de empresa, 12 de marzo de 2026",
    subject: "Un documento que se conserva treinta años, sin nada adjunto",
    heading: "Acta del comité de empresa",
    reference: "Jueves 12 de marzo de 2026, 17:00, sala de reuniones 2",
    body: [
        "Esta hoja declara cumplir la segunda parte de la norma de archivo, en el nivel \
que promete una sola cosa: dentro de treinta años la página se verá igual que \
hoy. Cada fuente que usa va dentro del fichero, cada color dice lo que \
significa y nada depende de un programa, de una red ni de una fuente instalada \
en el equipo.",
        "La segunda parte no admite adjuntar ficheros de cualquier tipo, como sí hace \
la tercera; por eso un documento que deba viajar con una hoja de cálculo o una \
factura declara la tercera.",
    ],
    labels: "Presentes\nExcusados\nPróxima sesión",
    answers: "Nueve miembros del comité y dos de la dirección\nUn miembro, de permiso\nJueves 9 \
        de abril de 2026, 17:00",
};

/// The minutes in Italian.
const ITALIAN: Words = Words {
    title: "Verbale del comitato aziendale, 12 marzo 2026",
    subject: "Un documento conservato trent’anni, che non contiene altro che sé stesso",
    heading: "Verbale del comitato aziendale",
    reference: "Giovedì 12 marzo 2026, ore 17:00, sala riunioni 2",
    body: [
        "Questo foglio dichiara di rispettare la seconda parte della norma di \
archiviazione, al livello che promette una cosa sola: tra trent’anni la pagina \
avrà lo stesso aspetto di oggi. Ogni carattere usato è contenuto nel file, \
ogni colore dice che cosa significa e nulla sulla pagina dipende da un \
programma, da una rete o da un font che si trovi installato.",
        "La seconda parte non prevede, come la terza, allegati di qualsiasi tipo: un \
documento che deve viaggiare con un foglio di calcolo o una fattura dichiara \
quindi la terza.",
    ],
    labels: "Presenti\nGiustificati\nProssima seduta",
    answers: "Nove membri del comitato e due della direzione\nUn membro, in ferie\nGiovedì 9 \
        aprile 2026, ore 17:00",
};

/// The minutes in Portuguese.
const PORTUGUESE: Words = Words {
    title: "Ata da comissão de trabalhadores, 12 de março de 2026",
    subject: "Um documento guardado durante trinta anos, que não contém nada além de si mesmo",
    heading: "Ata da comissão de trabalhadores",
    reference: "Quinta-feira, 12 de março de 2026, 17:00, sala de reuniões 2",
    body: [
        "Esta folha declara cumprir a segunda parte da norma de arquivo, no nível que promete \
            uma única coisa: daqui a trinta anos a página terá o mesmo aspeto que hoje. Cada \
            tipo de letra está dentro do ficheiro, cada cor diz o que significa e nada depende \
            de um programa, de uma rede ou de um tipo de letra instalado por acaso.",
        "A segunda parte não prevê os ficheiros anexos que a terceira prevê, por isso \
um documento que tenha de viajar com uma folha de cálculo ou uma fatura \
declara a terceira parte.",
    ],
    labels: "Presentes\nAusências\nPróxima reunião",
    answers: "Nove membros da comissão e dois da direção\nUm membro, de férias\nQuinta-feira, 9 \
        de abril de 2026, 17:00",
};

/// The minutes in Dutch.
const DUTCH: Words = Words {
    title: "Notulen van de ondernemingsraad, 12 maart 2026",
    subject: "Een document dat dertig jaar bewaard wordt en niets dan zichzelf bevat",
    heading: "Notulen van de ondernemingsraad",
    reference: "Donderdag 12 maart 2026, 17.00 uur, vergaderzaal 2",
    body: [
        "Dit blad verklaart te voldoen aan het tweede deel van de archiefnorm, op het \
niveau dat maar één ding belooft: over dertig jaar ziet de pagina er net zo \
uit als vandaag. Elk gebruikt lettertype zit in het bestand, elke kleur zegt \
wat ze betekent, en niets wacht op een programma, een netwerk of een toevallig \
geïnstalleerd lettertype.",
        "Het tweede deel kent geen meegevoerd bestand zoals het derde; een document dat \
met een spreadsheet of factuur moet meereizen, verklaart daarom te voldoen aan \
het derde deel.",
    ],
    labels: "Aanwezig\nAfgemeld\nVolgende zitting",
    answers: "Negen leden van de raad en twee van de directie\nEén lid, met verlof\nDonderdag 9 \
        april 2026, 17.00 uur",
};

/// Every language the example is written in. A language is added by writing its
/// own set of words and naming it here.
static WORDS: [(Language, &Words); 7] = [
    (Language::English, &ENGLISH),
    (Language::French, &FRENCH),
    (Language::German, &GERMAN),
    (Language::Spanish, &SPANISH),
    (Language::Italian, &ITALIAN),
    (Language::Portuguese, &PORTUGUESE),
    (Language::Dutch, &DUTCH),
];

fn main() -> std::process::ExitCode {
    failure::reported(run())
}

/// Writes the file, and says where it went.
fn run() -> Result<(), Box<dyn std::error::Error>> {
    let language = Language::from_environment()?;
    let words = Words::of(language);

    let mut args = env::args().skip(1);
    // A named file is written as named; the default one carries the language,
    // so the languages do not overwrite each other in `tmp/`.
    let path = args
        .next()
        .unwrap_or_else(|| language.file_name(&out::default_path("pdfa2b")));
    let font_path = args.next().map_or_else(default_font, PathBuf::from);

    let mut doc = Document::new();
    doc.set_license(licence::licensed());
    doc.set_conformance(PdfA::A2B);

    let mut facts = Metadata::default();
    facts.title = Some(words.title.to_owned());
    facts.author = Some(AUTHOR.to_owned());
    facts.subject = Some(words.subject.to_owned());
    facts.producer = Some(PRODUCER.to_owned());
    facts.created = Some(ISSUED.to_owned());
    doc.set_metadata(facts);

    let font = doc.add_font(Font::parse(fs::read(&font_path)?)?);
    doc.add_page(sheet(&font, words)?)?;

    let bytes = doc.to_bytes()?;
    if let Some(parent) = Path::new(&path).parent() {
        fs::create_dir_all(parent)?;
    }
    fs::write(&path, &bytes)?;
    println!("wrote {path}: {} bytes", bytes.len());
    Ok(())
}

/// The box the heading is set in.
const HEADING_BOX: Block = Block {
    size: 20.0,
    left: 72.0,
    top: 782.0,
    width: 451.0,
    set: None,
};

/// The box the line under the heading is set in.
const REFERENCE_BOX: Block = Block {
    size: 10.5,
    left: 72.0,
    top: 750.0,
    width: 451.0,
    set: None,
};

/// The box the body is set in.
const BODY_BOX: Block = Block {
    size: 11.0,
    left: 72.0,
    top: 716.0,
    width: 451.0,
    set: Some((16.0, Align::Justify)),
};

/// The box the labels of the record are set in, one a line.
const LABELS_BOX: Block = Block {
    size: 11.0,
    left: 72.0,
    top: 596.0,
    width: 451.0,
    set: Some((17.0, Align::Left)),
};

/// The box the answers are set in, beside the labels.
const ANSWERS_BOX: Block = Block {
    size: 11.0,
    left: 200.0,
    top: 596.0,
    width: 323.0,
    set: Some((17.0, Align::Left)),
};

/// The one page: the heading, when the council sat, what the part promises, and
/// the record in two columns.
fn sheet(font: &FontHandle, words: &Words) -> Result<Page, Error> {
    let mut c = Content::new();

    block(&mut c, font, &HEADING_BOX, words.heading)?;
    block(&mut c, font, &REFERENCE_BOX, words.reference)?;
    block(&mut c, font, &BODY_BOX, &words.body.join(" "))?;
    block(&mut c, font, &LABELS_BOX, words.labels)?;
    block(&mut c, font, &ANSWERS_BOX, words.answers)?;

    let mut page = Page::a4();
    page.content = c.into_bytes();
    Ok(page)
}

/// The box one block of words goes in: its upper-left corner, how far it runs,
/// the size it is set at, and the leading and alignment where the block states
/// them.
struct Block {
    size: f64,
    left: f64,
    top: f64,
    width: f64,
    set: Option<(f64, Align)>,
}

/// Draws one block of words in the box it is given.
fn block(c: &mut Content, font: &FontHandle, at: &Block, text: &str) -> Result<(), Error> {
    let mut flow = TextFlow::new(font, at.size);
    if let Some((leading, align)) = at.set {
        flow = flow.leading(leading).align(align);
    }
    let lines = flow.break_lines(text, at.width);
    c.begin_text();
    flow.draw(c, &lines, at.left, at.top, at.width)?;
    c.end_text();
    Ok(())
}

#[cfg(test)]
mod tests {
    use hqf_pdf::{Document, Font, FontHandle, TextFlow};

    use super::{
        ANSWERS_BOX, BODY_BOX, Block, HEADING_BOX, LABELS_BOX, REFERENCE_BOX, WORDS, default_font,
        language,
    };

    /// The lines two languages are allowed to write the same way. No two
    /// languages write one alike.
    const SPARED: [&str; 0] = [];

    /// The least room left between a label and the column beside it.
    const GAP: f64 = 8.0;

    /// How many lines `text` takes in the box `at`.
    fn lines(font: &FontHandle, at: &Block, text: &str) -> usize {
        let mut flow = TextFlow::new(font, at.size);
        if let Some((leading, _)) = at.set {
            flow = flow.leading(leading);
        }
        flow.break_lines(text, at.width).len()
    }

    #[test]
    fn every_language_writes_the_record_in_its_own_words() {
        let untranslated = language::untranslated_lines(&WORDS, &SPARED);

        assert!(
            untranslated.is_empty(),
            "the record says these in more than one language: {untranslated:?}"
        );
    }

    /// Every box on the page holds its words: a line meant to stand alone stays
    /// on one line, and the body ends a line above the block under it. Nothing
    /// else on the page says so when a language runs over.
    #[test]
    fn every_language_fits_the_boxes_the_page_sets_it_in() {
        let mut doc = Document::new();
        let font = doc.add_font(
            Font::parse(std::fs::read(default_font()).expect("the committed font is there"))
                .expect("the committed font parses"),
        );

        for (named, words) in WORDS {
            let code = named.code();
            for (at, text) in [
                (HEADING_BOX, words.heading),
                (REFERENCE_BOX, words.reference),
            ] {
                assert_eq!(
                    lines(&font, &at, text),
                    1,
                    "the {code} record breaks {text:?} over two lines"
                );
            }

            let count = lines(&font, &BODY_BOX, &words.body.join(" "));
            let (leading, _) = BODY_BOX.set.expect("the body states its leading");
            let held = f64::from(u32::try_from(count).expect("a body is a few lines")) * leading;
            assert!(
                held <= BODY_BOX.top - LABELS_BOX.top,
                "the {code} body takes {count} lines from {}, and the next block \
                 starts at {}",
                BODY_BOX.top,
                LABELS_BOX.top
            );

            let room = ANSWERS_BOX.left - LABELS_BOX.left - GAP;
            for label in words.labels.lines() {
                let measured = font.measure(label, LABELS_BOX.size);
                assert!(
                    measured <= room,
                    "the {code} record sets {label:?} over {measured:.1} points, \
                     and it has {room:.1} before the column beside it"
                );
            }

            for answer in words.answers.lines() {
                assert_eq!(
                    lines(&font, &ANSWERS_BOX, answer),
                    1,
                    "the {code} record breaks {answer:?} over two lines"
                );
            }
        }
    }
}