nigig-org/crates/apps/pdf/pdf-graphics/tests/cff_port.rs

//! 1:1 port of `dart-pdf/packages/pdf_graphics/test/cff_test.dart`
//! (tranche 1: fixture-built font — structure, encoding, outlines, widths).
//!
//! `build_test_cff_font` is a line-for-line port of dart's
//! `buildTestCffFont()`: glyph 0 = .notdef, glyph 1 = an 800x800 square at
//! the origin mapped to code 65 ('A'), advance 660 via nominalWidthX 600 +
//! leading operand 60.

use nigig_pdf_graphics::cff::CffFont;
use nigig_pdf_graphics::path::PdfPathSegment;

fn u8(v: u8) -> Vec<u8> {
    vec![v]
}

fn u16(v: u16) -> Vec<u8> {
    vec![(v >> 8) as u8, (v & 0xff) as u8]
}

/// CFF DICT 5-byte integer (operator 29).
fn int5(v: i32) -> Vec<u8> {
    vec![
        29,
        ((v >> 24) & 0xff) as u8,
        ((v >> 16) & 0xff) as u8,
        ((v >> 8) & 0xff) as u8,
        (v & 0xff) as u8,
    ]
}

fn join(parts: Vec<Vec<u8>>) -> Vec<u8> {
    parts.into_iter().flatten().collect()
}

fn index(items: Vec<Vec<u8>>) -> Vec<u8> {
    if items.is_empty() {
        return u16(0);
    }
    let mut out = u16(items.len() as u16);
    out.push(2); // offSize 2
    let mut offset: u16 = 1;
    out.extend(u16(offset));
    for item in &items {
        offset += item.len() as u16;
        out.extend(u16(offset));
    }
    for item in items {
        out.extend(item);
    }
    out
}

fn build_test_cff_font() -> Vec<u8> {
    // charstrings: .notdef = endchar; square = 60 width, then contour.
    let notdef = vec![14u8];
    let square = vec![
        139 + 60, // width delta 60 over nominalWidthX
        139, 139, 21, // 0 0 rmoveto
        249, 180, 6, // 800 hlineto
        249, 180, 7, // 800 vlineto
        253, 180, 6, // -800 hlineto
        14, // endchar
    ];

    let header = vec![1u8, 0, 4, 2];
    let name_index = index(vec![b"Test".to_vec()]);
    let string_index = u16(0);
    let gsubr_index = u16(0);

    // encoding: format 0, one code: 65 -> gid 1.
    let encoding = vec![0u8, 1, 65];
    let charstrings_index = index(vec![notdef, square]);
    // private dict: defaultWidthX 500 (20), nominalWidthX 600 (21).
    let private_dict = join(vec![int5(500), u8(20), int5(600), u8(21)]);

    // top dict (fixed-size operands so offsets are computable):
    // CharStrings (17), Encoding (16), Private (18).
    let top_dict_with = |charstrings_at: i32, encoding_at: i32, private_at: i32| -> Vec<u8> {
        join(vec![
            int5(charstrings_at),
            u8(17),
            int5(encoding_at),
            u8(16),
            int5(private_dict.len() as i32),
            int5(private_at),
            u8(18),
        ])
    };

    let top_dict_size = top_dict_with(0, 0, 0).len();
    let top_dict_index_size = index(vec![vec![0u8; top_dict_size]]).len();
    let fixed_prefix =
        header.len() + name_index.len() + top_dict_index_size + string_index.len() + gsubr_index.len();
    let encoding_at = fixed_prefix;
    let charstrings_at = encoding_at + encoding.len();
    let private_at = charstrings_at + charstrings_index.len();

    join(vec![
        header,
        name_index,
        index(vec![top_dict_with(
            charstrings_at as i32,
            encoding_at as i32,
            private_at as i32,
        )]),
        string_index,
        gsubr_index,
        encoding,
        charstrings_index,
        private_dict,
    ])
}

fn test_font() -> CffFont {
    CffFont::parse(&build_test_cff_font()).expect("test CFF parses")
}

#[test]
fn parses_the_cff_structure() {
    let font = test_font();
    assert_eq!(font.num_glyphs(), 2);
    assert!(!font.is_cid_keyed());
}

#[test]
fn encoding_maps_character_codes_to_glyphs() {
    let font = test_font();
    assert_eq!(font.gid_for_code(65), 1);
    assert_eq!(font.gid_for_code(66), 0);
}

#[test]
fn charstrings_interpret_to_em_unit_outlines() {
    let font = test_font();
    let square = font.outline_for_glyph(1).expect("square outline");
    let segs = square.segments();
    let first = &segs[0];
    assert!(
        matches!(first, PdfPathSegment::MoveTo { x, y } if x.abs() < 1e-9 && y.abs() < 1e-9),
        "first must be MoveTo(0,0), got {first:?}"
    );
    // 800 font units at the default 0.001 matrix = 0.8 em.
    let lines: Vec<(f64, f64)> = segs
        .iter()
        .filter_map(|s| match s {
            PdfPathSegment::LineTo { x, y } => Some((*x, *y)),
            _ => None,
        })
        .collect();
    assert_eq!(lines.len(), 3, "square has 3 lines, got {segs:?}");
    assert!((lines[0].0 - 0.8).abs() < 1e-9, "got {lines:?}");
    assert!((lines[1].1 - 0.8).abs() < 1e-9, "got {lines:?}");
    assert!(
        matches!(segs.last(), Some(PdfPathSegment::Close)),
        "last must be Close, got {segs:?}"
    );
}

#[test]
fn width_comes_from_the_leading_charstring_operand() {
    let font = test_font();
    font.outline_for_glyph(1);
    // nominalWidthX 600 + operand 60 = 660 units = 0.66 em.
    let advance = font.advance_for_glyph(1).expect("advance");
    assert!((advance - 0.66).abs() < 1e-9, "got {advance}");
}

#[test]
fn empty_glyphs_return_no_outline_but_a_default_width() {
    let font = test_font();
    assert!(font.outline_for_glyph(0).is_none());
    // .notdef has no width operand: defaultWidthX 500.
    let advance = font.advance_for_glyph(0).expect("advance");
    assert!((advance - 0.5).abs() < 1e-9, "got {advance}");
}

#[test]
fn garbage_input_parses_to_none() {
    assert!(CffFont::parse(b"not a font").is_none());
}

// ---------- tranche 2: seac composition (corpus) + OpenType CFF ----------

use nigig_pdf_cos::{PdfDict, PdfObj};
use nigig_pdf_document::PdfDocument;
use nigig_pdf_graphics::cff::OpenTypeCffFont;

/// The dart-pdf checkout this port tracks; corpus PDFs are read from it.
/// Tests that need a corpus file skip (with a notice) when it is absent so
/// the suite stays green on machines without the checkout.
const DART_CORPUS: &str = "/Users/aok/Projects/rustdev/CratesCode/dart-pdf/test_corpora/pdfjs";

fn corpus(relative: &str) -> Option<Vec<u8>> {
    let path = format!("{DART_CORPUS}/{relative}");
    match std::fs::read(&path) {
        Ok(bytes) => Some(bytes),
        Err(_) => {
            eprintln!("SKIP (no dart-pdf checkout): {path}");
            None
        }
    }
}

fn cff_from_resources(
    doc: &mut PdfDocument,
    resources: &PdfDict,
    font_name: &str,
) -> Option<CffFont> {
    let fonts = doc.resolve(resources.get("Font")?).ok()?;
    let pdf_font = doc.resolve(fonts.as_dict()?.get(font_name)?).ok()?;
    let descriptor = doc.resolve(pdf_font.as_dict()?.get("FontDescriptor")?).ok()?;
    let font_file = descriptor.as_dict()?.get("FontFile3")?.clone();
    let data = doc.resolve_stream(&font_file).ok()?;
    CffFont::parse(&data)
}

fn page_resources(doc: &mut PdfDocument) -> Option<PdfDict> {
    let page_ref = doc.page_object_ref(0)?;
    let page_obj = doc.resolve_ref(page_ref).ok()?;
    let resources = doc.resolve(page_obj.as_dict()?.get("Resources")?).ok()?;
    resources.as_dict().cloned()
}

fn cff_from_page_font(path: &str, font_name: &str) -> Option<CffFont> {
    let bytes = corpus(path)?;
    let mut doc = PdfDocument::parse(&bytes).ok()?;
    let resources = page_resources(&mut doc)?;
    cff_from_resources(&mut doc, &resources, font_name)
}

fn cff_from_form_font(path: &str, form_name: &str, font_name: &str) -> Option<CffFont> {
    let bytes = corpus(path)?;
    let mut doc = PdfDocument::parse(&bytes).ok()?;
    let resources = page_resources(&mut doc)?;
    let xobjects = doc.resolve(resources.get("XObject")?).ok()?;
    let form = doc.resolve(xobjects.as_dict()?.get(form_name)?).ok()?;
    let form_resources = doc.resolve(form.as_stream()?.dict.get("Resources")?).ok()?;
    cff_from_resources(&mut doc, form_resources.as_dict()?, font_name)
}

#[test]
fn endchar_seac_composes_accented_glyphs() {
    let Some(cff) = cff_from_form_font("endchar.pdf", "Fm0", "T1_0") else {
        return;
    };
    let e = cff.outline_for_glyph(cff.gid_for_name("E")).expect("E");
    let acute = cff
        .outline_for_glyph(cff.gid_for_name("acute"))
        .expect("acute");
    let eacute = cff
        .outline_for_glyph(cff.gid_for_name("Eacute"))
        .expect("Eacute");
    assert!(
        eacute.segments().len() > e.segments().len() + acute.segments().len() - 2,
        "Eacute should compose E + acute"
    );
}

#[test]
fn endchar_seac_composes_tilde_accents() {
    let Some(cff) = cff_from_page_font("glyph_accent.pdf", "F1") else {
        return;
    };
    let a = cff.outline_for_glyph(cff.gid_for_name("a")).expect("a");
    let tilde = cff
        .outline_for_glyph(cff.gid_for_name("tilde"))
        .expect("tilde");
    let atilde = cff
        .outline_for_glyph(cff.gid_for_name("atilde"))
        .expect("atilde");
    assert!(
        atilde.segments().len() > a.segments().len() + tilde.segments().len() - 2,
        "atilde should compose a + tilde"
    );
}

// --- OpenType CFF container builders (port of _testOpenTypeCff etc.) ---

fn u16be(v: u16) -> [u8; 2] {
    [(v >> 8) as u8, (v & 0xff) as u8]
}

fn u32be(v: u32) -> [u8; 4] {
    [
        (v >> 24) as u8,
        ((v >> 16) & 0xff) as u8,
        ((v >> 8) & 0xff) as u8,
        (v & 0xff) as u8,
    ]
}

fn flat(parts: Vec<&[u8]>) -> Vec<u8> {
    parts.concat()
}

/// Port of dart `_testCmaps`: one subtable per Unicode cmap format.
fn test_cmaps() -> Vec<Vec<u8>> {
    let mut format0 = flat(vec![&u16be(0), &u16be(262), &u16be(0)]);
    let mut glyphs = vec![0u8; 256];
    glyphs[0x41] = 1;
    format0.extend(glyphs);

    // format 4: two segments; 0x42 maps through glyphIdArray[0] = 2 with
    // idDelta -1, yielding glyph 1.
    let format4 = flat(vec![
        &u16be(4),
        &u16be(34),
        &u16be(0),
        &u16be(4),
        &u16be(4),
        &u16be(1),
        &u16be(0),
        &u16be(0x42),
        &u16be(0xffff),
        &u16be(0),
        &u16be(0x42),
        &u16be(0xffff),
        &u16be(0xffff),
        &u16be(1),
        &u16be(4),
        &u16be(0),
        &u16be(2),
    ]);

    let format6 = flat(vec![
        &u16be(6),
        &u16be(12),
        &u16be(0),
        &u16be(0x400),
        &u16be(1),
        &u16be(1),
    ]);

    let format12 = flat(vec![
        &u16be(12),
        &u16be(0),
        &u32be(28),
        &u32be(0),
        &u32be(1),
        &u32be(0x1f600),
        &u32be(0x1f600),
        &u32be(1),
    ]);

    vec![format0, format4, format6, format12]
}

fn cmap_table(subtables: &[Vec<u8>]) -> Vec<u8> {
    let records_len = 4 + subtables.len() * 8;
    let mut offset = records_len;
    let mut out = flat(vec![&u16be(0), &u16be(subtables.len() as u16)]);
    for (index, sub) in subtables.iter().enumerate() {
        out.extend(u16be(0));
        out.extend(u16be(index as u16));
        out.extend(u32be(offset as u32));
        offset += sub.len();
    }
    for sub in subtables {
        out.extend(sub);
    }
    out
}

fn test_opentype_cff(face_offset: usize, cmaps: Option<Vec<Vec<u8>>>) -> Vec<u8> {
    let cff = build_test_cff_font();
    let cmap = cmap_table(&cmaps.unwrap_or_else(test_cmaps));
    const DIR_LEN: usize = 12 + 2 * 16;
    let cff_offset = face_offset + DIR_LEN;
    let cmap_offset = cff_offset + cff.len();
    let mut out = flat(vec![
        &u32be(0x4f54544f), // OTTO
        &u16be(2),
        &u16be(0),
        &u16be(0),
        &u16be(0),
        b"CFF ",
        &u32be(0),
        &u32be(cff_offset as u32),
        &u32be(cff.len() as u32),
        b"cmap",
        &u32be(0),
        &u32be(cmap_offset as u32),
        &u32be(cmap.len() as u32),
    ]);
    out.extend(cff);
    out.extend(cmap);
    out
}

fn test_opentype_cff_collection() -> Vec<u8> {
    const FIRST_OFFSET: usize = 20;
    let first = test_opentype_cff(FIRST_OFFSET, None);
    let second_offset = FIRST_OFFSET + first.len();
    let second = test_opentype_cff(second_offset, None);
    let mut out = flat(vec![
        b"ttcf",
        &u32be(0x00010000),
        &u32be(2),
        &u32be(FIRST_OFFSET as u32),
        &u32be(second_offset as u32),
    ]);
    out.extend(first);
    out.extend(second);
    out
}

#[test]
fn opentype_cff_reads_unicode_cmap_formats_and_collections() {
    let bytes = test_opentype_cff(0, None);
    let face = OpenTypeCffFont::parse(&bytes, 0).expect("OT CFF face");
    assert_eq!(face.num_glyphs(), 2);
    assert_eq!(face.gid_for_unicode(0x41), 1, "format 0");
    assert_eq!(face.gid_for_unicode(0x42), 1, "format 4");
    assert_eq!(face.gid_for_unicode(0x400), 1, "format 6");
    assert_eq!(face.gid_for_unicode(0x1f600), 1, "format 12");
    assert_eq!(face.gid_for_unicode(0x43), 0);
    assert!(face.outline_for_glyph(1).is_some());
    let advance = face.advance_for_glyph(1).expect("advance");
    assert!((advance - 0.66).abs() < 1e-9, "got {advance}");

    let collection = test_opentype_cff_collection();
    assert!(OpenTypeCffFont::parse(&collection, 1).is_some());
    assert!(OpenTypeCffFont::parse(&collection, -1).is_none());
    assert!(OpenTypeCffFont::parse(&collection, 2).is_none());
}

#[test]
fn opentype_cff_rejects_malformed_or_incomplete_containers() {
    let plain = test_opentype_cff(0, None);
    assert!(OpenTypeCffFont::parse(&plain, 1).is_none());
    assert!(OpenTypeCffFont::parse(b"not an sfnt", 0).is_none());

    let mut truetype = plain.clone();
    truetype[0..4].copy_from_slice(&[0, 1, 0, 0]);
    assert!(OpenTypeCffFont::parse(&truetype, 0).is_none());
    assert!(OpenTypeCffFont::parse(&test_opentype_cff(0, Some(vec![])), 0).is_none());
}

#[test]
fn opentype_cff_system_face_keeps_unicode_cmap() {
    // Port of the macOS-only dart test; runs only when the Hiragino face is
    // present, otherwise skips with a notice (no system-font dependency).
    let path = "/System/Library/Fonts/ヒラギノ明朝 ProN.ttc";
    let Ok(bytes) = std::fs::read(path) else {
        eprintln!("SKIP (no system face): {path}");
        return;
    };
    let face = OpenTypeCffFont::parse(&bytes, 0).expect("system face");
    let glyph = face.gid_for_unicode('日' as u32);
    assert!(glyph > 0);
    assert!(face.outline_for_glyph(glyph).is_some());
    assert!(face.advance_for_glyph(glyph).unwrap_or(0.0) > 0.0);
}