nigig-org/crates/apps/nigig-ocr/tests/real_ocr.rs

//! OCR-T1 end-to-end with a REAL engine: fixture image → system Tesseract →
//! text + boxes → export. Skips (with a loud message) when `tesseract` is
//! not installed; set `NIGIG_OCR_REQUIRE_TESSERACT=1` (CI) to make a missing
//! engine a failure instead of a skip.

use std::sync::Arc;

use nigig_ocr::providers::TesseractEngine;
use nigig_ocr::{
    run_item, FileItem, InMemoryClipboard, JobEvent, OcrEngine, OcrError, OcrJob, OcrLanguage,
    Settings, UiAction, UiState,
};

const FIXTURE: &[u8] = include_bytes!("fixtures/hello.png");

fn engine() -> Option<TesseractEngine> {
    let e = TesseractEngine::detect();
    if e.is_none() {
        if std::env::var_os("NIGIG_OCR_REQUIRE_TESSERACT").is_some() {
            panic!("tesseract required but not found");
        }
        eprintln!("SKIP: tesseract not installed — real OCR not exercised");
    }
    e
}

fn norm(s: &str) -> String {
    s.split_whitespace().collect::<Vec<_>>().join(" ")
}

#[test]
fn recognises_fixture_text_and_boxes() {
    let Some(e) = engine() else { return };
    let mut item = FileItem::new(FIXTURE.to_vec());
    let r = run_item(&e, &mut item).expect("ocr");
    assert_eq!(norm(&r.text), "Hello Nairobi 2026 Invoice total 4500");
    assert_eq!(r.boxes.len(), 6);
    for b in &r.boxes {
        assert!(
            b.x >= 0.0 && b.y >= 0.0 && b.x + b.w <= 1.0 && b.y + b.h <= 1.0,
            "{b:?}"
        );
    }
    // Top-left origin: first line's words sit above the second line's.
    assert!(r.boxes[0].y < r.boxes[3].y);
    assert_eq!(item.recognized_text.as_deref(), Some(r.text.as_str()));
}

#[test]
fn rotation_is_applied_before_recognition() {
    let Some(e) = engine() else { return };
    // Store the fixture rotated 90° clockwise, then ask the app to rotate it back.
    let img = image::load_from_memory(FIXTURE).unwrap().rotate90();
    let mut bytes = Vec::new();
    img.write_to(
        &mut std::io::Cursor::new(&mut bytes),
        image::ImageFormat::Png,
    )
    .unwrap();
    let mut item = FileItem::new(bytes);
    item.rotate_left();
    let r = run_item(&e, &mut item).unwrap();
    assert!(
        norm(&r.text).contains("Hello Nairobi 2026"),
        "got {:?}",
        r.text
    );
}

#[test]
fn uninstalled_language_and_barcode_are_refused() {
    let Some(e) = engine() else { return };
    let mut item = FileItem::new(FIXTURE.to_vec());
    if !e
        .supported_languages()
        .contains(&OcrLanguage::SimplifiedChinese)
    {
        item.options.language = OcrLanguage::SimplifiedChinese;
        assert!(matches!(
            run_item(&e, &mut item),
            Err(OcrError::UnsupportedLanguage(_))
        ));
    }
    item.options.language = OcrLanguage::English;
    item.options.barcode_mode = true;
    assert_eq!(
        e.recognize(&item).unwrap_err(),
        OcrError::BarcodeUnsupported
    );
}

#[test]
fn background_job_with_real_engine() {
    let Some(e) = engine() else { return };
    let items = vec![
        (0, FileItem::new(FIXTURE.to_vec())),
        (1, FileItem::new(b"junk".to_vec())),
    ];
    let evs = OcrJob::spawn(Arc::new(e), items).wait();
    let results: Vec<bool> = evs
        .iter()
        .filter_map(|ev| match ev {
            JobEvent::Finished { result, .. } => Some(result.is_ok()),
            _ => None,
        })
        .collect();
    assert_eq!(results, vec![true, false]);
}

#[test]
fn ui_flow_drop_ocr_copy_export_for_real() {
    let Some(e) = engine() else { return };
    let dir = nigig_ocr::ingest::resolved_dir(
        std::env::temp_dir().join(format!("ocr-real-{}", std::process::id())),
    );
    let mut ui = UiState::with_services(
        Settings::default(),
        Arc::new(e),
        Box::new(InMemoryClipboard::default()),
        Some(dir.clone()),
    );
    assert_eq!(
        ui.apply(UiAction::DropBytes(FIXTURE.to_vec())),
        "Added file 0"
    );
    assert!(ui.apply(UiAction::DoOcr).starts_with("OCR done: 1/1"));
    assert_eq!(ui.apply(UiAction::Copy), "Copied");
    assert!(norm(&ui.clipboard().last_copied().unwrap()).contains("Invoice total 4500"));
    let m = ui.apply(UiAction::SavePdf { all: false });
    let p = m.strip_prefix("PDF File Exported: ").expect(&m);
    assert!(std::fs::read(p).unwrap().starts_with(b"%PDF-1.4"));
    let _ = std::fs::remove_dir_all(&dir);
}