write_pdfa4f.rs

//! Writes a statement of account claiming the fourth part of the archival
//! standard, at the level that carries files.
//!
//! The fourth part on its own carries nothing; the level named `f` after it is
//! what lets a record hold a file of any kind. Here that file is the same
//! statement the page states, in the form a machine reads.
//!
//! Whether the result really is PDF/A-4F 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_pdfa4f -- tmp/pdfa4f.pdf [font.ttf]`
//!        `HQF_PDF_LANG=fr cargo run --example write_pdfa4f -- tmp/pdfa4f_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::attachment::{Attachment, Relationship};
use hqf_pdf::metadata::xmp::{Metadata, PdfA};
use hqf_pdf::{Document, Error, Font, FontHandle, Page, TextFlow, Version};

#[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, and when the file it carries was last changed.
/// The library never reads the clock, so the caller says when.
const ISSUED: &str = "2026-06-02T08:15:00+02:00";
const CHANGED: &str = "2026-06-02T08:10:00+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 file the record carries. What it is to the record is said in `Words`.
const CARRIED_NAME: &str = "statement.xml";
const CARRIED_TYPE: &str = "text/xml";
const CARRIED: &str = "<?xml version=\"1.0\" encoding=\"UTF-8\"?>\n<statement period=\"2026-05\">\n  <line \
     name=\"Carriage\" amount=\"1240.00\"/>\n  <line name=\"Handling\" amount=\"186.50\"/>\n  \
     <due amount=\"1426.50\"/>\n</statement>\n";

/// The amounts themselves, beside what each one is for. Figures are the same in
/// every language.
const AMOUNTS: &str = "1 240.00\n186.50\n1 426.50";

/// The words the statement 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,
    /// What the carried file says it is to the record.
    carried_what: &'static str,
    /// The heading at the top of the page.
    heading: &'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,
}

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

/// The statement in English.
const ENGLISH: Words = Words {
    title: "Statement of account, May 2026",
    subject: "A record on the second version of PDF, carrying the figures a machine reads",
    carried_what: "The lines this page states, as a machine reads them",
    heading: "Statement of account, May 2026",
    body: "This record claims the fourth part of the archival standard at the level that carries \
        files. The part it is built on takes a file of any kind, whatever the file itself \
        conforms to, so the same record answers a person and a program. The lines below are \
        carried inside the file as well, in the form a machine reads.",
    labels: "Carriage\nHandling\nDue",
};

/// The statement in French.
const FRENCH: Words = Words {
    title: "Relevé de compte, mai 2026",
    subject: "Un document sur la deuxième version de PDF, qui contient les chiffres que lit une \
        machine",
    carried_what: "Les lignes que cette page indique, sous la forme que lit une machine",
    heading: "Relevé de compte, mai 2026",
    body: "Ce document déclare suivre la quatrième partie de la norme d’archivage, au \
niveau qui admet des fichiers joints. La partie sur laquelle il repose accepte \
un fichier de toute nature, quelle que soit la norme que ce fichier suit \
lui-même : le même document répond ainsi à une personne et à un programme. Les \
lignes ci-dessous sont aussi contenues dans le fichier, sous la forme que lit \
une machine.",
    labels: "Transport\nManutention\nÀ payer",
};

/// The statement in German.
const GERMAN: Words = Words {
    title: "Kontoauszug, Mai 2026",
    subject: "Ein Dokument in der zweiten PDF-Version, das die Zahlen für eine Maschine enthält",
    carried_what: "Die Posten dieser Seite, so wie eine Maschine sie liest",
    heading: "Kontoauszug, Mai 2026",
    body: "Dieses Dokument erklärt Konformität mit dem vierten Teil der Archivnorm, auf \
der Stufe, die Dateien mitführt. Der zugrunde liegende Teil nimmt eine Datei \
beliebiger Art auf, gleich welcher Norm sie selbst folgt, und so antwortet \
dasselbe Dokument einem Menschen und einem Programm. Die Posten unten stecken \
außerdem in der Datei, in der Form, die eine Maschine liest.",
    labels: "Fracht\nUmschlag\nFällig",
};

/// The statement in Spanish.
const SPANISH: Words = Words {
    title: "Extracto de cuenta, mayo de 2026",
    subject: "Un documento basado en la segunda versión de PDF que contiene las cifras que \
lee una máquina",
    carried_what: "Las líneas que indica esta página, tal como las lee una máquina",
    heading: "Extracto de cuenta, mayo de 2026",
    body: "Este documento declara cumplir la cuarta parte de la norma de archivo, en el \
nivel que admite ficheros adjuntos. La parte en la que se basa admite un \
fichero de cualquier tipo, cumpla la norma que cumpla ese fichero, así que el \
mismo documento sirve a la vez a una persona y a un programa. Las líneas de \
abajo van también dentro del fichero, en la forma que lee una máquina.",
    labels: "Transporte\nManipulación\nA pagar",
};

/// The statement in Italian.
const ITALIAN: Words = Words {
    title: "Estratto conto, maggio 2026",
    subject: "Un documento sulla seconda versione di PDF, che contiene le cifre lette da una \
        macchina",
    carried_what: "Le righe indicate in questa pagina, come le legge una macchina",
    heading: "Estratto conto, maggio 2026",
    body: "Questo documento dichiara di rispettare la quarta parte della norma di \
archiviazione, al livello che ammette file allegati. La parte su cui poggia \
accetta un file di qualsiasi tipo, a qualunque norma esso stesso si conformi, \
così lo stesso documento risponde a una persona e a un programma. Le righe qui \
sotto sono contenute anche nel file, nella forma che una macchina legge.",
    labels: "Trasporto\nMovimentazione\nDovuto",
};

/// The statement in Portuguese.
const PORTUGUESE: Words = Words {
    title: "Extrato de conta, maio de 2026",
    subject: "Um documento na segunda versão do PDF, que contém os números que uma máquina lê",
    carried_what: "As linhas que esta página indica, tal como uma máquina as lê",
    heading: "Extrato de conta, maio de 2026",
    body: "Este documento declara cumprir a quarta parte da norma de arquivo, no nível que \
        contém ficheiros. A parte em que assenta aceita um ficheiro de qualquer tipo, seja qual \
        for a norma que esse ficheiro cumpre, e assim o mesmo documento responde a uma pessoa e \
        a um programa. As linhas abaixo também estão dentro do ficheiro, na forma que uma \
        máquina lê.",
    labels: "Transporte\nManuseamento\nA pagar",
};

/// The statement in Dutch.
const DUTCH: Words = Words {
    title: "Rekeningoverzicht, mei 2026",
    subject: "Een document op de tweede versie van PDF, met de cijfers die een machine leest",
    carried_what: "De regels op deze pagina, zoals een machine ze leest",
    heading: "Rekeningoverzicht, mei 2026",
    body: "Dit document verklaart te voldoen aan het vierde deel van de archiefnorm, op \
het niveau dat bestanden meevoert. Het deel waarop het gebaseerd is, neemt een \
bestand van elke soort op, aan welke norm dat bestand zelf ook voldoet, zodat \
hetzelfde document zowel een mens als een programma bedient. De regels \
hieronder zitten ook in het bestand, in de vorm die een machine leest.",
    labels: "Vervoer\nBehandeling\nTe betalen",
};

/// 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("pdfa4f")));
    let font_path = args.next().map_or_else(default_font, PathBuf::from);

    let mut doc = Document::new();
    doc.set_license(licence::licensed());
    // The fourth part is built on the second version of PDF, and the standard
    // refuses a file that declares an earlier one.
    doc.set_version(Version::V2_0);
    doc.set_conformance(PdfA::A4F);

    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);

    doc.attach(Attachment::new(
        CARRIED_NAME,
        words.carried_what,
        CARRIED_TYPE,
        Relationship::Alternative,
        CHANGED,
        CARRIED.as_bytes().to_vec(),
    ));

    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 body is set in.
const BODY_BOX: Block = Block {
    size: 11.0,
    left: 72.0,
    top: 736.0,
    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: 616.0,
    width: 228.0,
    set: Some((17.0, Align::Left)),
};

/// The box the figures are set in, beside the labels.
const FIGURES_BOX: Block = Block {
    size: 11.0,
    left: 300.0,
    top: 616.0,
    width: 223.0,
    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, &BODY_BOX, words.body)?;
    block(&mut c, font, &LABELS_BOX, words.labels)?;
    block(&mut c, font, &FIGURES_BOX, AMOUNTS)?;

    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::{
        BODY_BOX, Block, FIGURES_BOX, HEADING_BOX, LABELS_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();
            assert_eq!(
                lines(&font, &HEADING_BOX, words.heading),
                1,
                "the {code} record breaks {:?} over two lines",
                words.heading
            );

            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 = FIGURES_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"
                );
            }
        }
    }
}