nigig-org/crates/apps/pdf/pdf-document/tests/acroform_roundtrip.rs

//! Phase 3 exit criterion, against a real parsed PDF rather than synthetic
//! dictionaries: parse an AcroForm file, modify a text field through
//! `DocumentFormEditor`, and verify `/V` changed and the dirty flag is set.
//!
//! The fixture is a hand-written PDF checked in beside this file. It has two
//! pages so that widget page resolution is genuinely exercised: everything
//! interactive lives on page index 1, so any code that silently defaults to
//! page 0 fails these tests.

use nigig_pdf_document::appearance::{generate_widget_appearance, regenerate_dirty};
use nigig_pdf_document::form::{apply_value_to_dict, DocumentFormEditor, FieldValue, FormError};
use nigig_pdf_document::{AnnotationAction, FieldType, PdfDocument};

const FIXTURE: &[u8] = include_bytes!("acroform.pdf");

#[test]
fn parses_the_page_tree_of_a_real_file() {
    let mut doc = PdfDocument::parse(FIXTURE).expect("fixture should parse");
    assert_eq!(doc.page_count(), 2);
    assert!(!doc.page(0).expect("page 0").content_data.is_empty());
    assert!(!doc.page(1).expect("page 1").content_data.is_empty());
}

#[test]
fn parses_the_acroform_with_inherited_keys_and_qualified_names() {
    let mut doc = PdfDocument::parse(FIXTURE).expect("parse");
    let form = doc
        .acroform()
        .expect("acroform lookup")
        .expect("fixture has an AcroForm");

    // "name" inherits /FT and /DA from its structural parent "personal", and
    // its fully qualified name is built from both levels.
    let name = form
        .find_field("personal.name")
        .expect("qualified name should resolve");
    assert_eq!(name.field_type, FieldType::Text);
    assert_eq!(name.value, FieldValue::Text("John".into()));
    assert_eq!(
        name.default_appearance.as_deref(),
        Some("/Helv 10 Tf 0 g"),
        "/DA must be inherited from the parent field"
    );
    assert_eq!(name.max_len, Some(20));

    let agree = form.find_field("agree").expect("checkbox");
    assert_eq!(agree.field_type, FieldType::Checkbox);
    assert!(!agree.is_checked());
}

#[test]
fn widget_pages_resolve_to_the_real_page_and_not_page_zero() {
    let mut doc = PdfDocument::parse(FIXTURE).expect("parse");
    let form = doc.acroform().expect("acroform").expect("present");

    for field in form.fields() {
        for widget in &field.widgets {
            assert_eq!(
                widget.page_index,
                Some(1),
                "field {:?} is on page 1 in the fixture",
                field.full_name
            );
        }
    }
    assert_eq!(form.widgets_on_page(1).len(), 2);
    assert!(
        form.widgets_on_page(0).is_empty(),
        "nothing in this fixture is on page 0"
    );
}

/// The Phase 3 exit criterion.
#[test]
fn editing_a_field_changes_v_and_sets_the_dirty_flag() {
    let mut doc = PdfDocument::parse(FIXTURE).expect("parse");
    let mut form = doc.acroform().expect("acroform").expect("present");

    let field_ref = form.find_field("personal.name").expect("field").obj_ref;
    let original = form.field(field_ref).expect("field").value.clone();
    assert_eq!(original, FieldValue::Text("John".into()));

    let mut editor = DocumentFormEditor::new(&mut form);
    editor
        .set_text(field_ref, "Jane")
        .expect("edit should apply");
    assert_eq!(
        editor.invalidated_pages(),
        &[1],
        "editing must invalidate the page the widget is on"
    );

    let edited = form.field(field_ref).expect("field");
    assert_eq!(edited.value, FieldValue::Text("Jane".into()));
    assert!(edited.dirty, "an applied edit must mark the field dirty");
    assert_eq!(form.dirty_fields().count(), 1);
}

#[test]
fn the_edit_survives_serialization_back_into_the_field_dictionary() {
    let mut doc = PdfDocument::parse(FIXTURE).expect("parse");
    let mut form = doc.acroform().expect("acroform").expect("present");
    let field_ref = form.find_field("personal.name").expect("field").obj_ref;

    DocumentFormEditor::new(&mut form)
        .set_text(field_ref, "Jane")
        .expect("edit");

    let field = form.field(field_ref).expect("field");
    let updated = apply_value_to_dict(field, &field.raw_dict_snapshot());
    assert_eq!(
        updated.get_str("V"),
        Some(&b"Jane"[..]),
        "/V must carry the new value"
    );
}

#[test]
fn max_len_from_the_file_is_enforced() {
    let mut doc = PdfDocument::parse(FIXTURE).expect("parse");
    let mut form = doc.acroform().expect("acroform").expect("present");
    let field_ref = form.find_field("personal.name").expect("field").obj_ref;

    let mut editor = DocumentFormEditor::new(&mut form);
    let too_long = "x".repeat(21);
    let err = editor
        .set_text(field_ref, &too_long)
        .expect_err("21 characters exceeds the fixture's /MaxLen of 20");
    assert!(matches!(err, FormError::TooLong { max_len: 20, .. }));

    // The refused edit must have changed nothing.
    assert_eq!(
        form.field(field_ref).expect("field").value,
        FieldValue::Text("John".into())
    );
    assert!(!form.field(field_ref).expect("field").dirty);
}

#[test]
fn checking_the_box_uses_the_on_state_declared_in_the_file() {
    let mut doc = PdfDocument::parse(FIXTURE).expect("parse");
    let mut form = doc.acroform().expect("acroform").expect("present");
    let field_ref = form.find_field("agree").expect("checkbox").obj_ref;

    DocumentFormEditor::new(&mut form)
        .set_check(field_ref, true)
        .expect("check");

    let field = form.field(field_ref).expect("field");
    assert_eq!(
        field.value,
        FieldValue::State("On".into()),
        "the fixture's appearance dictionary names the on state /On"
    );
    assert!(field.is_checked());
}

#[test]
fn appearance_regeneration_reflects_the_edited_value() {
    let mut doc = PdfDocument::parse(FIXTURE).expect("parse");
    let mut form = doc.acroform().expect("acroform").expect("present");
    let field_ref = form.find_field("personal.name").expect("field").obj_ref;

    DocumentFormEditor::new(&mut form)
        .set_text(field_ref, "Jane")
        .expect("edit");

    let (generated, errors) = regenerate_dirty(&form);
    assert!(errors.is_empty(), "unexpected errors: {errors:?}");
    assert_eq!(generated.len(), 1);
    let content = generated[0].1.content_str();
    assert!(content.contains("(Jane) Tj"), "got: {content}");
    assert!(!content.contains("(John)"), "stale value must not survive");
}

#[test]
fn a_regenerated_appearance_is_a_wellformed_form_xobject() {
    let mut doc = PdfDocument::parse(FIXTURE).expect("parse");
    let form = doc.acroform().expect("acroform").expect("present");
    let field = form.find_field("personal.name").expect("field");

    let ap = generate_widget_appearance(field, &field.widgets[0], &form).expect("appearance");
    let xobj = ap.to_form_xobject();
    assert_eq!(xobj.dict.get_name("Subtype"), Some("Form"));
    assert_eq!(xobj.dict.get_int("Length"), Some(xobj.data.len() as i64));
    // The widget is 200x20 in the fixture.
    assert_eq!(ap.bbox, [0.0, 0.0, 200.0, 20.0]);
}

#[test]
fn annotations_report_their_real_page_and_typed_action() {
    let mut doc = PdfDocument::parse(FIXTURE).expect("parse");

    assert!(
        doc.page_annotations(0).expect("page 0").is_empty(),
        "page 0 has no annotations in this fixture"
    );

    let annots = doc.page_annotations(1).expect("page 1 annotations");
    assert_eq!(annots.len(), 3, "two widgets and one link");
    for annot in &annots {
        assert_eq!(
            annot.page_index,
            Some(1),
            "every annotation must know its real page"
        );
    }

    let link = annots
        .iter()
        .find(|a| a.action().is_some())
        .expect("fixture has a link with a URI action");
    assert_eq!(
        link.action(),
        Some(AnnotationAction::OpenUri("https://example.org/".into())),
        "the document reports the intent; it never opens the URL itself"
    );
    assert!(link.contains_point(150.0, 510.0));
    assert!(!link.contains_point(50.0, 510.0));
}

#[test]
fn annotations_for_a_missing_page_are_an_error() {
    let mut doc = PdfDocument::parse(FIXTURE).expect("parse");
    assert!(doc.page_annotations(99).is_err());
}

#[test]
fn page_index_of_maps_references_back_and_rejects_strangers() {
    let doc = PdfDocument::parse(FIXTURE).expect("parse");
    let page1 = doc.page_object_ref(1).expect("page 1 ref");
    assert_eq!(doc.page_index_of(page1), Some(1));

    let stranger = nigig_pdf_cos::ObjRef { num: 9999, gen: 0 };
    assert_eq!(
        doc.page_index_of(stranger),
        None,
        "an unknown reference must be unknown, not page 0"
    );
}