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

//! 1:1 port of `dart-pdf/packages/pdf_graphics/test/streaming_interpreter_test.dart`.
//!
//! Streaming (op-by-op) and materialized (parse-all-then-run)
//! interpretation must produce identical command lists.

use nigig_pdf_document::PdfDocument;
use nigig_pdf_graphics::content::{interpret_streaming, parse_content_stream};
use nigig_pdf_graphics::recording::{format_commands, RecordingDevice};

const PDFJS: &str = "/Users/aok/Projects/rustdev/CratesCode/dart-pdf/test_corpora/pdfjs";

#[test]
fn streaming_and_materialized_interpretation_agree() {
    let fixtures = [
        "operator-in-TJ-array.pdf",
        "rotation.pdf",
        "xobject-image.pdf",
        "gradientfill.pdf",
        "tiling-pattern-box.pdf",
        "pattern_text_embedded_font.pdf",
        "blendmode.pdf",
        "knockout_smask.pdf",
    ];
    let mut ran = 0;
    for name in fixtures {
        let path = format!("{PDFJS}/{name}");
        let Ok(bytes) = std::fs::read(&path) else {
            eprintln!("SKIP (missing corpus file): {path}");
            continue;
        };
        let Ok(mut doc) = PdfDocument::parse(&bytes) else {
            eprintln!("SKIP (unparseable — likely compressed xref): {path}");
            continue;
        };
        let pages = doc.page_count().min(2);
        for page_index in 0..pages {
            let Ok(page) = doc.page(page_index) else { continue };
            let content = page.content_data.clone();

            let ops = parse_content_stream(&content).expect("content parses");
            let materialized = RecordingDevice::from_ops(&ops);

            let mut streamed = RecordingDevice::new();
            // A malformed tail ends the stream; from_ops shares the same
            // prefix behaviour, so both sides stay comparable.
            let _ = interpret_streaming(&content, &mut streamed);

            assert!(
                !materialized.is_empty(),
                "{name} page {page_index} recorded nothing"
            );
            assert_eq!(
                format_commands(&streamed.commands),
                format_commands(&materialized),
                "{name} page {page_index} diverged"
            );
            ran += 1;
        }
    }
    eprintln!("streaming/materialized compared {ran} pages");
    assert!(ran > 0, "no corpus pages ran — checkout missing?");
}