use std::sync::Arc;

use oxide_compiler::compiler::Compiler;
use oxide_types::value::JsValue;
use oxide_vm::vm::Vm;

fn eval(vm: &mut Vm, source: &str) -> Result<JsValue, String> {
    let allocator = oxide_parser::Allocator::default();
    let program = oxide_parser::parse(&allocator, source).map_err(|e| format!("Parse error: {:?}", e))?;
    let module = Compiler::new().compile(&program).map_err(|e| format!("Compile error: {}", e))?;
    vm.run(&Arc::new(module))
}

fn to_str(vm: &Vm, val: JsValue) -> String {
    vm.lookup_str(val).unwrap_or_default()
}

#[test]
fn regexp_survives_gc_and_still_matches() {
    // 回归：正则存入全局对象触发晋升/GC 搬移后，native_fn 槽的已编译 Box
    // 必须深拷贝到新对象（而非共享指针），否则 epoch 释放与 teardown 双重释放。
    let mut vm = Vm::new();
    let result = eval(&mut vm, "var r = /ab+c/g; globalThis.r = r; r.test('xxabbcx')").unwrap();
    assert!(result.as_bool());

    // 触发完整收集：存活正则克隆进新 arena，旧 Box 由 sweep 释放。
    vm.reset();

    let result = eval(&mut vm, "globalThis.r.test('xxabbcx')").unwrap();
    assert!(result.as_bool(), "GC 后正则应仍可匹配");
    let result = eval(&mut vm, "globalThis.r.exec('xabbbcx')[0]").unwrap();
    assert_eq!(to_str(&vm, result), "abbbc");
}

// --- RegExp constructor ---

#[test]
fn regexp_constructor_creates_object() {
    let mut vm = Vm::new();
    let result = eval(&mut vm, "typeof new RegExp('abc')").unwrap();
    assert_eq!(to_str(&vm, result), "object");
}

#[test]
fn regexp_constructor_with_flags() {
    let mut vm = Vm::new();
    let result = eval(&mut vm, "new RegExp('abc', 'gi').ignoreCase").unwrap();
    assert!(result.as_bool());
}

#[test]
fn regexp_constructor_invalid_pattern_syntax_error() {
    let mut vm = Vm::new();
    let result = eval(&mut vm, "new RegExp('[')");
    assert!(result.is_err() || (result.is_ok() && result.unwrap().is_string()));
}

#[test]
fn regexp_source_property() {
    let mut vm = Vm::new();
    let result = eval(&mut vm, "new RegExp('hello', 'g').source").unwrap();
    assert_eq!(to_str(&vm, result), "hello");
}

#[test]
fn regexp_flags_property() {
    let mut vm = Vm::new();
    let result = eval(&mut vm, "new RegExp('abc', 'gi').flags").unwrap();
    assert_eq!(to_str(&vm, result), "gi");
}

#[test]
fn regexp_global_property() {
    let mut vm = Vm::new();
    let result = eval(&mut vm, "new RegExp('abc', 'g').global").unwrap();
    assert!(result.as_bool());
}

#[test]
fn regexp_multiline_property() {
    let mut vm = Vm::new();
    let result = eval(&mut vm, "new RegExp('abc', 'm').multiline").unwrap();
    assert!(result.as_bool());
}

#[test]
fn regexp_last_index_property() {
    let mut vm = Vm::new();
    let result = eval(&mut vm, "new RegExp('abc', 'g').lastIndex").unwrap();
    assert_eq!(result.as_int(), 0);
}

#[test]
fn regexp_dotall_sticky_unicode_default_false() {
    let mut vm = Vm::new();
    let result = eval(&mut vm, "new RegExp('abc').dotAll").unwrap();
    assert!(!result.as_bool());
}

// --- RegExp.prototype.test ---

#[test]
fn regexp_test_basic() {
    let mut vm = Vm::new();
    let result = eval(&mut vm, "new RegExp('abc').test('abc')").unwrap();
    assert!(result.as_bool());
}

#[test]
fn regexp_test_no_match() {
    let mut vm = Vm::new();
    let result = eval(&mut vm, "new RegExp('abc').test('xyz')").unwrap();
    assert!(!result.as_bool());
}

#[test]
fn regexp_test_case_insensitive() {
    let mut vm = Vm::new();
    let result = eval(&mut vm, "new RegExp('abc', 'i').test('ABC')").unwrap();
    assert!(result.as_bool());
}

#[test]
fn regexp_test_global_flag_iteration() {
    let mut vm = Vm::new();
    let result = eval(&mut vm, "var re = /a/g; re.test('a')").unwrap();
    assert!(result.as_bool());
}

// --- RegExp.prototype.exec ---

#[test]
fn regexp_exec_basic() {
    let mut vm = Vm::new();
    let result = eval(&mut vm, "/hello/.exec('hello world') !== null").unwrap();
    assert!(result.as_bool());
}

#[test]
fn regexp_exec_no_match_returns_null() {
    let mut vm = Vm::new();
    let result = eval(&mut vm, "/xyz/.exec('hello')").unwrap();
    assert!(result.is_null());
}

#[test]
fn regexp_exec_returns_index() {
    let mut vm = Vm::new();
    let result = eval(&mut vm, "/ell/.exec('hello')").unwrap();
    assert!(result.is_object());
}

#[test]
fn regexp_exec_returns_input() {
    let mut vm = Vm::new();
    let m = eval(&mut vm, "/ell/.exec('hello')").unwrap();
    assert!(m.is_object());
}

#[test]
fn regexp_exec_global_last_index_advances() {
    let mut vm = Vm::new();
    let re = eval(&mut vm, "var re = /a/g; re.exec('aba'); re.lastIndex").unwrap();
    assert_eq!(re.as_int(), 1);
}

#[test]
fn regexp_exec_global_resets_on_exhaustion() {
    let mut vm = Vm::new();
    let result = eval(&mut vm, "var re = /a/g; re.exec('a'); re.exec('a') === null").unwrap();
    assert!(result.as_bool());
}

// --- RegExp literal compilation ---

#[test]
fn regexp_literal_basic() {
    let mut vm = Vm::new();
    let result = eval(&mut vm, "/hello/").unwrap();
    assert!(result.is_object());
}

#[test]
fn regexp_literal_with_flags() {
    let mut vm = Vm::new();
    let result = eval(&mut vm, "/abc/i.test('ABC')").unwrap();
    assert!(result.as_bool());
}

#[test]
fn regexp_literal_global_exec() {
    let mut vm = Vm::new();
    let result = eval(&mut vm, "var re = /a/g; re.exec('aba'); re.lastIndex").unwrap();
    assert_eq!(result.as_int(), 1);
}

// --- String integration ---

#[test]
fn string_match_with_regexp() {
    let mut vm = Vm::new();
    let result = eval(&mut vm, "'hello world'.match(/ell/)").unwrap();
    assert!(result.is_object());
}

#[test]
fn string_replace_with_regexp() {
    let mut vm = Vm::new();
    let result = eval(&mut vm, "'hello'.replace(/ell/, 'ipp')").unwrap();
    assert_eq!(to_str(&vm, result), "hippo");
}

#[test]
fn string_replace_with_regexp_global() {
    let mut vm = Vm::new();
    let result = eval(&mut vm, "'aba'.replace(/a/g, 'c')").unwrap();
    assert_eq!(to_str(&vm, result), "cbc");
}

#[test]
fn string_search_with_regexp() {
    let mut vm = Vm::new();
    let result = eval(&mut vm, "'hello'.search(/ell/)").unwrap();
    assert_eq!(result.as_int(), 1);
}

#[test]
fn string_search_not_found() {
    let mut vm = Vm::new();
    let result = eval(&mut vm, "'hello'.search(/xyz/)").unwrap();
    assert_eq!(result.as_int(), -1);
}

#[test]
fn string_split_with_regexp() {
    let mut vm = Vm::new();
    let result = eval(&mut vm, "'a,b,c'.split(/,/)").unwrap();
    assert!(result.is_object());
    let obj = unsafe { &*result.as_js_object_ptr() };
    assert_eq!(obj.prop_count(), 3);
    assert_eq!(to_str(&vm, obj.get_prop_at(1)), "b");
}

// --- toString ---

#[test]
fn regexp_to_string() {
    let mut vm = Vm::new();
    let result = eval(&mut vm, "/abc/gi.toString()").unwrap();
    assert_eq!(to_str(&vm, result), "/abc/gi");
}

// --- Unicode property escapes：Script/Script_Extensions 的 Unknown（Zzzz）取值 ---
// vendor 的 Unicode 表须含 Unknown（未分配码点集），否则合法模式编译期 SyntaxError。

#[test]
fn regexp_unicode_property_script_unknown_matches_unassigned() {
    let mut vm = Vm::new();
    // U+038B 未分配：Script 与 Script_Extensions 均为 Unknown
    let result = eval(&mut vm, "/\\p{Script=Unknown}/u.test(String.fromCodePoint(0x038B))").unwrap();
    assert!(result.as_bool(), "未分配码点应匹配 Script=Unknown");
    let result = eval(&mut vm, "/\\p{Script_Extensions=Unknown}/u.test(String.fromCodePoint(0x038B))").unwrap();
    assert!(result.as_bool(), "未分配码点应匹配 Script_Extensions=Unknown");
}

#[test]
fn regexp_unicode_property_script_unknown_rejects_assigned() {
    let mut vm = Vm::new();
    // 已分配拉丁字母不属于 Unknown
    let result = eval(&mut vm, "/\\p{Script=Unknown}/u.test('d')").unwrap();
    assert!(!result.as_bool(), "已分配码点不应匹配 Script=Unknown");
    let result = eval(&mut vm, "/\\P{Script=Unknown}/u.test('d')").unwrap();
    assert!(result.as_bool(), "取反后已分配码点应匹配");
}

#[test]
fn regexp_unicode_property_script_unknown_aliases() {
    let mut vm = Vm::new();
    // 单字母缩写与符号名须同义解析
    for pattern in [r"\p{sc=Unknown}", r"\p{scx=Unknown}", r"\p{Script=Zzzz}", r"\p{scx=Zzzz}"] {
        let source = format!("/{pattern}/u.test(String.fromCodePoint(0x038B))");
        let result = eval(&mut vm, &source).unwrap();
        assert!(result.as_bool(), "别名 {pattern} 应匹配未分配码点");
    }
}

#[test]
fn regexp_unicode_property_script_existing_values_unchanged() {
    // 补齐 Unknown 不得扰动既有取值：Adlam 等已分配 Script 行为保持
    let mut vm = Vm::new();
    let result = eval(&mut vm, "/\\p{Script=Adlam}/u.test(String.fromCodePoint(0x1E900))").unwrap();
    assert!(result.as_bool(), "既有 Script 取值行为应保持不变");
    let result = eval(&mut vm, "/\\p{Script=Adlam}/u.test('d')").unwrap();
    assert!(!result.as_bool(), "既有 Script 取值行为应保持不变");
}

// --- UTF-16 单元向匹配入口：gate 运行时化 + 惰性 units IR 回归钉 ---
// 引擎构建已启用 regress 的 utf16 feature（纯加性 API）。字节字面量 pass 改为
// 逐编译运行时 gate 后，str 路径编译与执行不变；单元向 IR 在首次
// find_from_utf16 时惰性编译（关字节 pass）并缓存。以下钉直接走 regress API，
// 不依赖 JS 侧单元视图接线（后续批次才落地）。

fn units_of(text: &str) -> Vec<u16> {
    text.encode_utf16().collect()
}

#[test]
fn regexp_utf16_entry_surrogate_property() {
    // 孤立 surrogate 单元命中 \p{General_Category=Surrogate}；普通字符不命中；
    // 代理对（单码点）不命中。
    let re = regress::Regex::with_flags(r"\p{General_Category=Surrogate}", "u").unwrap();
    let m = re.find_from_utf16(&[0xD800], 0).next();
    assert!(m.is_some(), "孤立 surrogate 单元应命中 GC=Surrogate");
    assert_eq!(m.unwrap().range(), 0..1);
    assert!(re.find_from_utf16(&[0x41], 0).next().is_none(), "A 不应命中");
    assert!(
        re.find_from_utf16(&[0xD834, 0xDE00], 0).next().is_none(),
        "代理对是单码点，不应按两个孤立 surrogate 命中"
    );
}

#[test]
fn regexp_utf16_entry_surrogate_pair_equals_str_path() {
    // U+1D11E = 代理对 [0xD834, 0xDE00]：单元路径按单码点匹配，与 str 路径等价。
    let ch = char::from_u32(0x1D11E).unwrap();
    let re = regress::Regex::with_flags(&ch.to_string(), "").unwrap();
    let text: String = [ch, 'a', 'b', 'c'].iter().copied().collect();
    let m_str = re.find(&text).expect("str 路径应命中");
    assert_eq!(m_str.range(), 0..4, "str 路径命中 1 个码点（4 字节）");

    let units = units_of(&text);
    let m_u16 = re.find_from_utf16(&units, 0).next().expect("单元路径应命中");
    assert_eq!(m_u16.range(), 0..2, "单元路径命中 1 个码点（2 个单元）");
}

#[test]
fn regexp_utf16_entry_well_formed_equals_str_path() {
    // well-formed 语料：单元路径与 str 路径匹配结果逐一相等（含非 ASCII）。
    for (pattern, text) in [(r"\d+", "abc123x456"), (r"a[bc]d", "xabcdyabd"), (r"\w+\s+\w+", "héllo wörld")] {
        let re = regress::Regex::new(pattern).unwrap();
        let str_m = re.find(text).map(|m| m.as_str(text).to_string());
        let units = units_of(text);
        let u16_m = re.find_from_utf16(&units, 0).next().map(|m| {
            units[m.range()]
                .iter()
                .flat_map(|&u| char::from_u32(u as u32).map(|c| c.to_string()))
                .collect::<String>()
        });
        assert_eq!(str_m, u16_m, "pattern {pattern}");
    }
}