obadh_engine 0.9.1

use obadh_engine::{ObadhEngine, PhoneticUnitType, Tokenizer};

#[test]
fn test_real_world_phola_examples() {
    // Create engine for transliteration testing
    let engine = ObadhEngine::new();
    let tokenizer = Tokenizer::new();

    // Test case 1: sohy => সহ্য
    let units = tokenizer.tokenize_word("sohy");
    // Verify structure
    assert!(!units.is_empty());

    // Test the transliteration result
    let result = engine.transliterate("sohy");
    assert_eq!(result, "সহ্য");

    // Test case 2: biSw => বিশ্ব
    let units = tokenizer.tokenize_word("biSw");
    // Verify structure
    assert!(!units.is_empty());

    // Test the transliteration result
    let result = engine.transliterate("biSw");
    assert_eq!(result, "বিশ্ব");
}

#[test]
fn test_specific_jo_phola_cases() {
    let engine = ObadhEngine::new();

    // Test various jo-phola (য-ফলা) cases
    let examples = [
        ("sohy", "সহ্য"),    // consonant + terminator + jo-phola
        ("sohyo", "সহ্যো"),  // consonant + terminator + jo-phola + vowel
        ("kohya", "কহ্যা"),   // with different consonant
        ("bhujy", "ভুজ্য"),   // consonant + vowel + jo-phola
        ("kriy", "ক্রিয়"),   // conjunct + vowel + jo-phola
        ("odhyoy", "অধ্যয়"), // complex form with multiple jo-pholas
    ];

    for (input, expected) in examples {
        let result = engine.transliterate(input);
        assert_eq!(result, expected);
    }
}

#[test]
fn test_aspirated_ya_phola_derivation_cases() {
    let engine = ObadhEngine::new();
    let tokenizer = Tokenizer::new();

    let examples = [
        ("khy", "খ্য"),
        ("Khya", "খ্যা"),
        ("ghy", "ঘ্য"),
        ("jhy", "ঝ্য"),
        ("JhY", "ঝ্য"),
        ("JHya", "ঝ্যা"),
        ("Thy", "ঠ্য"),
        ("THya", "ঠ্যা"),
        ("Dhy", "ঢ্য"),
        ("thy", "থ্য"),
        ("dhy", "ধ্য"),
        ("phy", "ফ্য"),
        ("fy", "ফ্য"),
        ("fya", "ফ্যা"),
        ("bhy", "ভ্য"),
        ("vy", "ভ্য"),
        ("chy", "ছ্য"),
        ("Cy", "ছ্য"),
        ("chhy", "ছ্য"),
        ("Chhy", "ছ্য"),
        ("CHHy", "ছ্য"),
        ("chY", "ছ্য"),
        ("Chy", "ছ্য"),
        ("CHY", "ছ্য"),
        ("chya", "ছ্যা"),
        ("Cya", "ছ্যা"),
        ("chhya", "ছ্যা"),
        ("Chya", "ছ্যা"),
        ("Chhya", "ছ্যা"),
        ("CHHya", "ছ্যা"),
        ("CHyA", "ছ্যা"),
        ("Ch,,y", "ছ্য"),
    ];

    for (input, expected) in examples {
        assert_eq!(engine.transliterate(input), expected, "{input}");
    }

    let units = tokenizer.tokenize_word("Chy");
    assert_eq!(units.len(), 1);
    assert_eq!(units[0].unit_type, PhoneticUnitType::Conjunct);
    assert_eq!(units[0].text, "Ch,,y");

    let units = tokenizer.tokenize_word("jhy");
    assert_eq!(units.len(), 1);
    assert_eq!(units[0].unit_type, PhoneticUnitType::Conjunct);
    assert_eq!(units[0].text, "jh,,y");
}

#[test]
fn test_aspirated_ya_phola_derivation_stays_narrow() {
    let engine = ObadhEngine::new();

    let examples = [
        ("rya", "রয়া"),
        ("Rya", "ড়য়া"),
        ("Rhya", "ঢ়য়া"),
        ("Ngya", "ঙয়া"),
        ("zoy", "যয়"),
        ("zy", "য্য"),
        // `q` maps to `k` (foreign-letter alias), so it takes ya-phola like ক: qya → ক্যা.
        ("qya", "ক্যা"),
    ];

    for (input, expected) in examples {
        assert_eq!(engine.transliterate(input), expected, "{input}");
    }
}

#[test]
fn test_specific_bo_phola_cases() {
    let engine = ObadhEngine::new();

    // Test various bo-phola (ব-ফলা) cases
    let examples = [
        ("biSw", "বিশ্ব"),    // consonant + vowel + bo-phola
        ("biSwas", "বিশ্বাস"), // consonant + vowel + bo-phola + vowel + consonant
        ("tw", "ত্ব"),        // simple bo-phola
        ("twa", "ত্বা"),       // bo-phola with vowel
        ("SwaSw", "শ্বাশ্ব"),   // multiple bo-pholas
    ];

    for (input, expected) in examples {
        let result = engine.transliterate(input);
        assert_eq!(result, expected);
    }
}

#[test]
fn test_both_phola_in_one_word() {
    let engine = ObadhEngine::new();

    // Test words that have both jo-phola and bo-phola
    let examples = [
        ("Swy", "শ্বয়"), // sequential pholas
    ];

    for (input, expected) in examples {
        let result = engine.transliterate(input);
        assert_eq!(result, expected);
    }
}

#[test]
fn test_composable_phola_orthography() {
    let engine = ObadhEngine::new();

    let examples = [("SwayottoSw", "শ্বায়ত্তশ্ব"), ("dwitiyw", "দ্বিতীয়")];

    for (input, expected) in examples {
        let result = engine.transliterate(input);
        assert_eq!(result, expected);
    }
}

#[test]
fn test_specific_phola_examples() {
    let engine = ObadhEngine::new();

    // Test case 1: sohy => সহ্য
    let result = engine.transliterate("sohy");
    assert_eq!(result, "সহ্য");

    // Test case 2: biSw => বিশ্ব
    let result = engine.transliterate("biSw");
    assert_eq!(result, "বিশ্ব");
}

#[test]
fn test_vocalic_r_case() {
    let engine = ObadhEngine::new();

    // Test krri => কৃ
    let result = engine.transliterate("krri");
    assert_eq!(result, "কৃ");
}

#[test]
fn test_productive_ya_phola_on_conjunct_bases() {
    let engine = ObadhEngine::new();

    // ya-phola composes onto any renderable conjunct base (not just listed
    // clusters), so loanwords such as প্ল্যান/ব্ল্যাক/ফ্ল্যাট are typeable. The bare
    // `y`/`Y` marker and the অ্যা (`aY`/`AY`) vowel path are equivalent.
    for (input, expected) in [
        ("ply", "প্ল্য"),
        ("plY", "প্ল্য"),
        ("plYa", "প্ল্যা"),
        ("plYan", "প্ল্যান"),
        ("plaYn", "প্ল্যান"),
        ("plAYn", "প্ল্যান"),
        ("blYak", "ব্ল্যাক"),
        ("glYad", "গ্ল্যাদ"),
        ("flYaT", "ফ্ল্যাট"),
        ("klYap", "ক্ল্যাপ"),
    ] {
        assert_eq!(engine.transliterate(input), expected, "{input}");
    }

    // Regressions preserved: ya-refusing bases (র/ড়/ঢ়/ঙ) and phola-terminated
    // bases stay decomposed, and enumerated/aspirated forms are unchanged.
    for (input, expected) in [
        ("rya", "রয়া"),
        ("Ngya", "ঙয়া"),
        ("Swy", "শ্বয়"),
        ("stwy", "স্ত্বয়"),
        ("khy", "খ্য"),
        ("sty", "স্ত্য"),
    ] {
        assert_eq!(engine.transliterate(input), expected, "{input}");
    }
}