write_pdfa2u.rs

//! Writes a deed of assignment claiming the second part of the archival
//! standard, at level U.
//!
//! Level U adds one promise to level B: every mark of ink drawn on the page
//! says which letter it stands for, so a registry searching thirty thousand
//! deeds finds this one, and a clerk copying a clause out of it gets the
//! clause. Nothing about the layout changes for it.
//!
//! Whether the result really is PDF/A-2U 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_pdfa2u -- tmp/pdfa2u.pdf [font.ttf]`
//!        `HQF_PDF_LANG=fr cargo run --example write_pdfa2u -- tmp/pdfa2u_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 record was 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 = "2026-04-17T10:15:00+02:00";

/// 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 deed 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.
    body: &'static str,
    /// 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 deed is written in, in `language`.
    fn of(language: Language) -> &'static Self {
        language::pick(&WORDS, language)
    }
}

/// The deed in English.
const ENGLISH: Words = Words {
    title: "Deed of assignment, ACME/2026/0417",
    subject: "A record whose every word can be searched for and copied out",
    heading: "Deed of assignment",
    reference: "Reference ACME/2026/0417, filed 17 April 2026",
    body: "This deed claims the second part of the archival standard at the level above the \
        plain one: on top of the promise that the page keeps its looks, every mark of ink drawn \
        on it says which letter it stands for. A registry searching thirty thousand deeds for a \
        reference finds this one; a clerk copying a clause out of it gets the clause, not a row \
        of question marks. That is the whole of the difference between the two levels. Nothing \
        about the layout changes, and nothing is added to the page.",
    labels: "Assignor\nAssignee\nEffective",
    answers: "Ateliers du Nord, Lille\nACME Industries, Rouen\n1 May 2026",
};

/// The deed in French.
const FRENCH: Words = Words {
    title: "Acte de cession, ACME/2026/0417",
    subject: "Un document dont chaque mot peut être recherché et recopié",
    heading: "Acte de cession",
    reference: "Référence ACME/2026/0417, enregistrée le 17 avril 2026",
    body: "Cet acte déclare suivre la deuxième partie de la norme d’archivage, au niveau \
situé juste au-dessus du plus simple : en plus de la promesse que la page \
garde son apparence, chaque trace d’encre dit quelle lettre elle représente. \
Un greffe qui cherche une référence parmi trente mille actes trouve celui-ci ; \
un clerc qui en recopie une clause obtient la clause, et non une rangée de \
points d’interrogation. C’est toute la différence entre les deux niveaux : la \
mise en page ne change en rien, et rien n’est ajouté à la page.",
    labels: "Cédant\nCessionnaire\nPrise d’effet",
    answers: "Ateliers du Nord, Lille\nACME Industries, Rouen\n1er mai 2026",
};

/// The deed in German.
const GERMAN: Words = Words {
    title: "Abtretungsurkunde, ACME/2026/0417",
    subject: "Ein Dokument, in dem sich jedes Wort suchen und herauskopieren lässt",
    heading: "Abtretungsurkunde",
    reference: "Aktenzeichen ACME/2026/0417, eingereicht am 17. April 2026",
    body: "Diese Urkunde erklärt Konformität mit dem zweiten Teil der Archivnorm, auf der \
Stufe über der einfachen: Zusätzlich zur Zusage, dass die Seite ihr Aussehen \
behält, gibt jede darauf gesetzte Druckspur an, für welchen Buchstaben sie \
steht. Ein Register, das dreißigtausend Urkunden nach einem Aktenzeichen \
durchsucht, findet diese; wer eine Klausel daraus herauskopiert, erhält die \
Klausel und keine Reihe von Fragezeichen. Das ist der ganze Unterschied \
zwischen den beiden Stufen. Am Layout ändert sich nichts, und der Seite wird \
nichts hinzugefügt.",
    labels: "Zedent\nZessionar\nWirksam ab",
    answers: "Ateliers du Nord, Lille\nACME Industries, Rouen\n1. Mai 2026",
};

/// The deed in Spanish.
const SPANISH: Words = Words {
    title: "Escritura de cesión, ACME/2026/0417",
    subject: "Un documento en el que cada palabra se puede buscar y copiar",
    heading: "Escritura de cesión",
    reference: "Referencia ACME/2026/0417, registrada el 17 de abril de 2026",
    body: "Esta escritura declara cumplir la segunda parte de la norma de archivo, en el nivel \
        por encima del básico: además de prometer que la página conserva su aspecto, cada trazo \
        de tinta dice qué letra representa. Un registro que busca una referencia entre treinta \
        mil escrituras encuentra esta; un oficial que copia una cláusula obtiene la cláusula, no \
        una hilera de signos de interrogación. Esa es toda la diferencia entre los dos niveles. \
        La maquetación no cambia en nada y no se añade nada a la página.",
    labels: "Cedente\nCesionario\nFecha de efecto",
    answers: "Ateliers du Nord, Lille\nACME Industries, Rouen\n1 de mayo de 2026",
};

/// The deed in Italian.
const ITALIAN: Words = Words {
    title: "Atto di cessione, ACME/2026/0417",
    subject: "Un documento di cui ogni parola si può cercare e copiare",
    heading: "Atto di cessione",
    reference: "Riferimento ACME/2026/0417, depositato il 17 aprile 2026",
    body: "Quest’atto dichiara di rispettare la seconda parte della norma di \
archiviazione, al livello sopra quello di base: oltre alla promessa che la \
pagina conservi il suo aspetto, ogni segno d’inchiostro dice quale lettera \
rappresenta. Un ufficio che cerca un riferimento fra trentamila atti trova \
quest’atto; un impiegato che ne copia una clausola ottiene la clausola, non \
una fila di punti interrogativi. Tutta la differenza fra i due livelli è qui. \
L’impaginazione non cambia in nulla e alla pagina non si aggiunge niente.",
    labels: "Cedente\nCessionario\nDecorrenza",
    answers: "Ateliers du Nord, Lille\nACME Industries, Rouen\n1° maggio 2026",
};

/// The deed in Portuguese.
const PORTUGUESE: Words = Words {
    title: "Escritura de cessão, ACME/2026/0417",
    subject: "Um documento em que cada palavra pode ser pesquisada e copiada",
    heading: "Escritura de cessão",
    reference: "Referência ACME/2026/0417, registada a 17 de abril de 2026",
    body: "Esta escritura declara cumprir a segunda parte da norma de arquivo, no nível \
acima do básico: além da promessa de que a página mantém o seu aspeto, cada \
traço de tinta diz que letra representa. Uma conservatória que procure uma \
referência entre trinta mil escrituras encontra esta; um funcionário que copia \
uma cláusula obtém a cláusula, e não uma fila de pontos de interrogação. É \
essa toda a diferença entre os dois níveis. A paginação não muda em nada e \
nada é acrescentado à página.",
    labels: "Cedente\nCessionário\nEm vigor a partir de",
    answers: "Ateliers du Nord, Lille\nACME Industries, Rouen\n1 de maio de 2026",
};

/// The deed in Dutch.
const DUTCH: Words = Words {
    title: "Akte van overdracht, ACME/2026/0417",
    subject: "Een document waarvan elk woord doorzocht en overgenomen kan worden",
    heading: "Akte van overdracht",
    reference: "Kenmerk ACME/2026/0417, ingediend op 17 april 2026",
    body: "Deze akte verklaart te voldoen aan het tweede deel van de archiefnorm, op het \
niveau boven het eenvoudige: bovenop de belofte dat de pagina haar uiterlijk \
behoudt, zegt elk inktspoor voor welke letter het staat. Een register dat \
dertigduizend akten naar een kenmerk doorzoekt, vindt deze; een griffier die \
er een clausule uit overneemt, krijgt de clausule en geen rij vraagtekens. Dat \
is het hele verschil tussen de twee niveaus. Aan de opmaak verandert niets, en \
er komt niets bij op de pagina.",
    labels: "Overdrager\nVerkrijger\nIngangsdatum",
    answers: "Ateliers du Nord, Lille\nACME Industries, Rouen\n1 mei 2026",
};

/// 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("pdfa2u")));
    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::A2U);

    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: 576.0,
    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: 576.0,
    width: 323.0,
    set: Some((17.0, Align::Left)),
};

/// The one page, block by block.
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)?;
    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);
            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"
                );
            }
        }
    }
}