nigig-org/crates/apps/pdf/pdf-graphics/tests/phase5_exit_criterion.rs

//! Phase 5 exit criterion.
//!
//! > Select text across a line with mixed fonts. Verify selection rectangles
//! > align with actual glyph positions. Copy produces correct Unicode.
//! > Search finds a string and highlights the right page/position.
//!
//! These drive the real measurement path — `ResolvedFont` + `TextState` +
//! `GlyphAdvanceCache` — rather than the byte-proportional estimate the
//! review calls out.

use nigig_pdf_cos::{PdfDict, PdfObj};
use nigig_pdf_graphics::advance::{
    glyph_advances, measure_advance, GlyphAdvanceCache, ResolvedFont, TextState,
};
use nigig_pdf_graphics::cid::{parse_to_unicode, CidCMap, CidWidths};
use nigig_pdf_graphics::font::GlyphWidths;
use nigig_pdf_graphics::recording::RenderCommand;
use nigig_pdf_graphics::text::PageText;

/// A proportional font: every glyph a different width, so any code that
/// assumes even spacing produces visibly wrong positions.
fn proportional() -> ResolvedFont {
    let mut w = GlyphWidths::new(500.0, 500.0);
    for (ch, width) in [
        (b'H', 722.0),
        (b'e', 556.0),
        (b'l', 222.0),
        (b'o', 556.0),
        (b' ', 278.0),
        (b'W', 944.0),
        (b'r', 333.0),
        (b'd', 556.0),
        (b'i', 222.0),
    ] {
        w.set_width(ch, width);
    }
    ResolvedFont::simple("Helv", w).with_vertical_metrics(0.718, -0.207)
}

/// A monospaced font, for the mixed-font line.
fn monospace() -> ResolvedFont {
    let mut w = GlyphWidths::new(600.0, 600.0);
    for c in 32u8..=126 {
        w.set_width(c, 600.0);
    }
    ResolvedFont::simple("Cour", w).with_vertical_metrics(0.629, -0.157)
}

/// A CID font with a ToUnicode map, for the correct-Unicode assertions.
fn cid_with_unicode() -> ResolvedFont {
    let mut d = PdfDict::new();
    d.set("DW", PdfObj::Int(1000));
    d.set(
        "W",
        PdfObj::Array(vec![
            PdfObj::Int(1),
            PdfObj::Array(vec![PdfObj::Int(500), PdfObj::Int(500), PdfObj::Int(500)]),
        ]),
    );
    // Codes 1..3 map to "N", "a", "\u{00EF}" (a non-ASCII scalar).
    let to_unicode = parse_to_unicode(
        br#"3 beginbfchar
            <0001> <004E>
            <0002> <0061>
            <0003> <00EF>
            endbfchar"#,
    );
    ResolvedFont::composite("CID", CidCMap::identity_h(), CidWidths::from_cid_font(&d))
        .with_to_unicode(to_unicode)
}

fn show(text: &str, size: f64, advance: f64) -> RenderCommand {
    RenderCommand::ShowTextWithMetrics {
        text: text.as_bytes().to_vec(),
        font_size: size,
        advance_x: advance,
        advance_y: 0.0,
        decoded: text.to_string(),
        advance_is_measured: true,
    }
}

/// The exit criterion's first clause: select across a line of mixed fonts
/// and verify the rectangles track real glyph positions.
#[test]
fn selection_across_mixed_fonts_uses_real_advances() {
    let prop = proportional();
    let mono = monospace();
    let state = TextState::with_size(12.0);

    // "Hello " in a proportional font, then "World" in a monospaced one.
    let run_a = measure_advance(&prop, b"Hello ", &state);
    let run_b = measure_advance(&mono, b"World", &state);

    // The two runs must measure differently: five monospaced glyphs at 600
    // are not the same as the proportional string.
    let expected_b = 5.0 * 600.0 / 1000.0 * 12.0;
    assert!((run_b - expected_b).abs() < 1e-9, "monospace advance");
    assert!(
        (run_a - run_b).abs() > 0.1,
        "a byte-proportional estimate would make these equal"
    );

    let commands = vec![
        RenderCommand::BeginText,
        RenderCommand::SetTextMatrix(1.0, 0.0, 0.0, 1.0, 100.0, 700.0),
        show("Hello ", 12.0, run_a),
        show("World", 12.0, run_b),
        RenderCommand::EndText,
    ];
    let page = PageText::from_commands(&commands);
    assert_eq!(page.segments().len(), 2);

    // The second run must begin exactly where the first ends.
    let first = &page.segments()[0];
    let second = &page.segments()[1];
    assert!(
        (second.origin[0] - (first.origin[0] + run_a)).abs() < 1e-9,
        "the second run must start at the first run's true end"
    );

    // Both runs sit on one line despite the different fonts.
    assert_eq!(page.plain_text(), "Hello World");

    // Selecting across the boundary picks up text from both runs.
    let mid_a = first.origin[0] + run_a * 0.5;
    let mid_b = second.origin[0] + run_b * 0.5;
    let selected = page.text_in_range((mid_a, 702.0), (mid_b, 702.0));
    assert!(
        selected.contains("lo ") && selected.contains("Wo"),
        "selection across the font boundary was {selected:?}"
    );
}

#[test]
fn selection_rectangles_align_with_glyph_positions() {
    let prop = proportional();
    let state = TextState::with_size(10.0);
    let text = b"Hello";
    let advance = measure_advance(&prop, text, &state);
    let per_glyph = glyph_advances(&prop, text, &state);

    // Per-glyph advances must sum to the run advance, or a caret placed by
    // accumulating them drifts away from the run's end.
    let sum: f64 = per_glyph.iter().sum();
    assert!((sum - advance).abs() < 1e-9);

    // 'H' is wider than 'l': even spacing would make these equal.
    assert!(
        per_glyph[0] > per_glyph[2],
        "H ({}) must be wider than l ({})",
        per_glyph[0],
        per_glyph[2]
    );

    let commands = vec![
        RenderCommand::BeginText,
        RenderCommand::SetTextMatrix(1.0, 0.0, 0.0, 1.0, 50.0, 400.0),
        show("Hello", 10.0, advance),
        RenderCommand::EndText,
    ];
    let page = PageText::from_commands(&commands);
    let seg = &page.segments()[0];

    // The run rectangle must span exactly the measured advance.
    let rect = seg.rect();
    assert!((rect[0] - 50.0).abs() < 1e-9);
    assert!((rect[2] - (50.0 + advance)).abs() < 1e-9);

    // Real ascent and descent straddle the baseline.
    let with_metrics = seg.rect_with_metrics(0.718, -0.207);
    assert!(with_metrics[3] > 400.0 && with_metrics[1] < 400.0);
    assert!((with_metrics[3] - (400.0 + 0.718 * 10.0)).abs() < 1e-9);
}

/// The exit criterion's "copy produces correct Unicode" clause.
#[test]
fn copying_composite_text_produces_correct_unicode() {
    let font = cid_with_unicode();
    // Three two-byte codes.
    let bytes = [0x00, 0x01, 0x00, 0x02, 0x00, 0x03];

    assert_eq!(font.decode_codes(&bytes), vec![1, 2, 3]);
    assert_eq!(
        font.decode_text(&bytes).as_deref(),
        Some("Na\u{00EF}"),
        "ToUnicode must drive the copied text, including non-ASCII"
    );

    // And the advance is three glyphs, not six bytes.
    let advance = measure_advance(&font, &bytes, &TextState::with_size(10.0));
    assert!((advance - 3.0 * 500.0 / 1000.0 * 10.0).abs() < 1e-9);
}

#[test]
fn copying_simple_text_decodes_win_ansi() {
    let font = proportional();
    // 0x92 is a right single quote in WinAnsi, not U+0092.
    assert_eq!(font.decode_text(b"it\x92s").as_deref(), Some("it\u{2019}s"));
}

/// The exit criterion's search clause.
#[test]
fn search_finds_a_string_and_reports_its_position() {
    let prop = proportional();
    let state = TextState::with_size(12.0);
    let line1 = "Hello World";
    let line2 = "World order";
    let a1 = measure_advance(&prop, line1.as_bytes(), &state);
    let a2 = measure_advance(&prop, line2.as_bytes(), &state);

    let commands = vec![
        RenderCommand::BeginText,
        RenderCommand::SetTextMatrix(1.0, 0.0, 0.0, 1.0, 72.0, 700.0),
        show(line1, 12.0, a1),
        RenderCommand::SetTextMatrix(1.0, 0.0, 0.0, 1.0, 72.0, 680.0),
        show(line2, 12.0, a2),
        RenderCommand::EndText,
    ];
    let page = PageText::from_commands(&commands);

    let hits = page.find("world");
    assert_eq!(hits.len(), 2, "case-insensitive across both lines");

    // The hits must be on different lines, at the right vertical positions.
    let first = &hits[0];
    let second = &hits[1];
    assert_eq!(first.segment_index, 0);
    assert_eq!(second.segment_index, 1);
    assert_eq!(first.text, "World");

    let r1 = first.rects[0];
    let r2 = second.rects[0];
    assert!(r1[1] > r2[1], "line 1 sits above line 2");

    // The first hit starts partway into its line; the second at the start.
    assert!(r1[0] > 72.0, "\"World\" is not at the start of line 1");
    assert!((r2[0] - 72.0).abs() < 1.0, "\"World\" starts line 2");

    // Every rectangle stays inside its own run.
    for (hit, seg) in hits.iter().zip(page.segments()) {
        let bounds = seg.rect();
        assert!(hit.rects[0][0] >= bounds[0] - 1e-9);
        assert!(hit.rects[0][2] <= bounds[2] + 1e-9);
    }
}

#[test]
fn the_advance_cache_serves_repeated_measurements_of_a_page() {
    let mut cache = GlyphAdvanceCache::new(1);
    cache.register_font(proportional());
    cache.register_font(monospace());
    let state = TextState::with_size(12.0);

    // Simulate several frames re-measuring the same page.
    let mut last = None;
    for _ in 0..5 {
        let a = cache.advance("Helv", b"Hello World", &state).expect("font");
        if let Some(prev) = last {
            assert_eq!(a, prev, "the same input must measure identically");
        }
        last = Some(a);
    }
    let (hits, misses) = cache.stats();
    assert_eq!(misses, 1, "measured once");
    assert_eq!(hits, 4, "the other four frames were served from cache");

    // Different fonts do not collide in the cache.
    let helv = cache.advance("Helv", b"iiiii", &state).expect("font");
    let cour = cache.advance("Cour", b"iiiii", &state).expect("font");
    assert!(
        (helv - cour).abs() > 0.1,
        "narrow proportional glyphs must not measure as monospaced"
    );

    // A new page revision rebuilds rather than serving stale values.
    assert!(cache.sync_generation(2));
    assert!(cache.is_empty());
}

#[test]
fn an_unmeasurable_run_is_flagged_rather_than_estimated() {
    // The review's complaint was a silent font_size * 0.6 fallback. A run
    // with no width table must be visibly unmeasured instead.
    let commands = vec![RenderCommand::ShowTextWithMetrics {
        text: b"unknown".to_vec(),
        font_size: 12.0,
        advance_x: 0.0,
        advance_y: 0.0,
        decoded: "unknown".into(),
        advance_is_measured: false,
    }];
    let page = PageText::from_commands(&commands);
    assert!(!page.segments()[0].advance_is_measured);
}