llvm-native-core 0.1.9

//! C++ Preprocessor Extensions — adds C++-specific preprocessing
//! directives and predefined macros beyond the base C preprocessor.
//!
//! Implements:
//! - `__has_include` / `__has_cpp_attribute` (C++17)
//! - `#include_next` for header wrappers
//! - `import` / `export` for header units (C++20)
//! - C++ predefined macros: `__cplusplus`, `__STDCPP_DEFAULT_NEW_ALIGNMENT__`,
//!   `__cpp_*` feature test macros
//! - `#embed` directive (C++26 / C23)
//! - `#warning` directive
//!
//! Clean-room behavioral reconstruction from the C++ standards
//! (ISO/IEC 14882:1998–2024) and published Clang documentation.

use super::cpp_token::CppStandard;

// ═══════════════════════════════════════════════════════════════════════════════
// C++ Predefined Macros
// ═══════════════════════════════════════════════════════════════════════════════

/// Returns the `__cplusplus` value for a given C++ standard.
pub fn cplusplus_value(standard: CppStandard) -> i64 {
    match standard {
        CppStandard::Cxx98 | CppStandard::GnuXx98 => 199711,
        CppStandard::Cxx03 | CppStandard::GnuXx03 => 199711,
        CppStandard::Cxx11 | CppStandard::GnuXx11 => 201103,
        CppStandard::Cxx14 | CppStandard::GnuXx14 => 201402,
        CppStandard::Cxx17 | CppStandard::GnuXx17 => 201703,
        CppStandard::Cxx20 | CppStandard::GnuXx20 => 202002,
        CppStandard::Cxx23 | CppStandard::GnuXx23 => 202302,
    }
}

/// All C++ predefined macros for the given standard and target.
pub fn cpp_predefined_macros(standard: CppStandard) -> Vec<(String, String)> {
    let mut macros = Vec::new();

    // __cplusplus
    macros.push(("__cplusplus".into(), cplusplus_value(standard).to_string()));

    // __STDCPP_DEFAULT_NEW_ALIGNMENT__
    macros.push(("__STDCPP_DEFAULT_NEW_ALIGNMENT__".into(), "16".into()));

    // C++ standard feature test macros
    macros.push(("__cpp_raw_strings".into(), "200710".into()));
    macros.push(("__cpp_unicode_characters".into(), "200704".into()));
    macros.push(("__cpp_user_defined_literals".into(), "200809".into()));
    macros.push(("__cpp_lambdas".into(), "200907".into()));
    macros.push((
        "__cpp_constexpr".into(),
        if standard >= CppStandard::Cxx14 {
            "201304".into()
        } else {
            "200704".into()
        },
    ));
    macros.push((
        "__cpp_range_based_for".into(),
        if standard >= CppStandard::Cxx17 {
            "201603".into()
        } else {
            "200907".into()
        },
    ));
    macros.push((
        "__cpp_static_assert".into(),
        if standard >= CppStandard::Cxx17 {
            "201411".into()
        } else {
            "200410".into()
        },
    ));
    macros.push(("__cpp_decltype".into(), "200707".into()));
    macros.push(("__cpp_attributes".into(), "200809".into()));
    macros.push(("__cpp_rvalue_references".into(), "200610".into()));
    macros.push(("__cpp_variadic_templates".into(), "200704".into()));
    macros.push(("__cpp_initializer_lists".into(), "200806".into()));
    macros.push(("__cpp_delegating_constructors".into(), "200604".into()));
    macros.push(("__cpp_nsdmi".into(), "200809".into()));
    macros.push((
        "__cpp_inheriting_constructors".into(),
        if standard >= CppStandard::Cxx17 {
            "201511".into()
        } else {
            "200802".into()
        },
    ));
    macros.push(("__cpp_ref_qualifiers".into(), "200710".into()));
    macros.push(("__cpp_alias_templates".into(), "200704".into()));

    // C++14 feature test macros
    if standard >= CppStandard::Cxx14 {
        macros.push(("__cpp_return_type_deduction".into(), "201304".into()));
        macros.push(("__cpp_decltype_auto".into(), "201304".into()));
        macros.push(("__cpp_generic_lambdas".into(), "201304".into()));
        macros.push(("__cpp_variable_templates".into(), "201304".into()));
        macros.push(("__cpp_binary_literals".into(), "201304".into()));
        macros.push(("__cpp_digit_separators".into(), "201309".into()));
        macros.push(("__cpp_init_captures".into(), "201304".into()));
        macros.push(("__cpp_aggregate_nsdmi".into(), "201304".into()));
    }

    // C++17 feature test macros
    if standard >= CppStandard::Cxx17 {
        macros.push(("__cpp_structured_bindings".into(), "201606".into()));
        macros.push(("__cpp_deduction_guides".into(), "201703".into()));
        macros.push(("__cpp_if_constexpr".into(), "201606".into()));
        macros.push(("__cpp_inline_variables".into(), "201606".into()));
        macros.push(("__cpp_fold_expressions".into(), "201603".into()));
        macros.push(("__cpp_capture_star_this".into(), "201603".into()));
        macros.push(("__cpp_namespace_attributes".into(), "201411".into()));
        macros.push(("__cpp_nested_namespace_definitions".into(), "201411".into()));
        macros.push(("__cpp_noexcept_function_type".into(), "201510".into()));
        macros.push(("__cpp_template_auto".into(), "201606".into()));
        macros.push(("__cpp_variadic_using".into(), "201611".into()));
        macros.push(("__cpp_guaranteed_copy_elision".into(), "201606".into()));
        macros.push(("__cpp_hex_float".into(), "201603".into()));
        macros.push(("__cpp_constexpr_dynamic_alloc".into(), "201907".into()));
    }

    // C++20 feature test macros
    if standard >= CppStandard::Cxx20 {
        macros.push(("__cpp_concepts".into(), "202002".into()));
        macros.push(("__cpp_coroutines".into(), "201902".into()));
        macros.push(("__cpp_modules".into(), "201907".into()));
        macros.push(("__cpp_using_enum".into(), "201907".into()));
        macros.push(("__cpp_constexpr_in_virtual".into(), "202002".into()));
        macros.push(("__cpp_aggregate_paren_init".into(), "201902".into()));
        macros.push(("__cpp_char8_t".into(), "201811".into()));
        macros.push(("__cpp_conditional_explicit".into(), "201806".into()));
        macros.push(("__cpp_consteval".into(), "201811".into()));
        macros.push(("__cpp_constinit".into(), "201907".into()));
        macros.push(("__cpp_designated_initializers".into(), "201707".into()));
        macros.push(("__cpp_implicit_move".into(), "202002".into()));
        macros.push(("__cpp_nontype_template_args".into(), "201911".into()));
        macros.push((
            "__cpp_nontype_template_parameter_auto".into(),
            "201606".into(),
        ));
        macros.push(("__cpp_range_based_for".into(), "201603".into()));
        macros.push(("__cpp_three_way_comparison".into(), "201907".into()));
        macros.push(("__cpp_lib_concepts".into(), "202002".into()));
        macros.push(("__cpp_lib_coroutine".into(), "201902".into()));
    }

    // C++23 feature test macros
    if standard >= CppStandard::Cxx23 {
        macros.push(("__cpp_explicit_this_parameter".into(), "202110".into()));
        macros.push(("__cpp_if_consteval".into(), "202106".into()));
        macros.push(("__cpp_multidimensional_subscript".into(), "202110".into()));
        macros.push(("__cpp_static_call_operator".into(), "202207".into()));
        macros.push(("__cpp_auto_cast".into(), "202110".into()));
        macros.push(("__cpp_lib_expected".into(), "202202".into()));
    }

    macros
}

// ═══════════════════════════════════════════════════════════════════════════════
// __has_include / __has_cpp_attribute
// ═══════════════════════════════════════════════════════════════════════════════

/// Evaluate `__has_include("header")` or `__has_include(<header>)`.
/// Returns true if the header is available.
pub fn eval_has_include(header: &str, _standard: CppStandard) -> bool {
    // In a real implementation, this would check the include paths.
    // For now, check common standard library headers.
    let known_headers = [
        "<iostream>",
        "<vector>",
        "<string>",
        "<map>",
        "<set>",
        "<algorithm>",
        "<memory>",
        "<functional>",
        "<utility>",
        "<type_traits>",
        "<tuple>",
        "<array>",
        "<queue>",
        "<stack>",
        "<bitset>",
        "<deque>",
        "<list>",
        "<forward_list>",
        "<unordered_map>",
        "<unordered_set>",
        "<regex>",
        "<thread>",
        "<mutex>",
        "<future>",
        "<chrono>",
        "<random>",
        "<filesystem>",
        "<optional>",
        "<variant>",
        "<any>",
        "<string_view>",
        "<span>",
        "<concepts>",
        "<coroutine>",
        "<compare>",
        "<format>",
        "<ranges>",
        "<source_location>",
        "\"stdio.h\"",
        "\"stdlib.h\"",
        "\"string.h\"",
        "\"math.h\"",
    ];
    known_headers.iter().any(|&h| {
        header == h
            || header.trim_matches(&['<', '>', '"'] as &[_])
                == h.trim_matches(&['<', '>', '"'] as &[_])
    })
}

/// Evaluate `__has_cpp_attribute(attribute_name)`.
/// Returns the attribute version or 0 if not supported.
pub fn eval_has_cpp_attribute(attr: &str, standard: CppStandard) -> u32 {
    match attr {
        // C++11 attributes
        "noreturn" if standard >= CppStandard::Cxx11 => 200809,
        "carries_dependency" if standard >= CppStandard::Cxx11 => 200809,
        // C++14 attributes
        "deprecated" if standard >= CppStandard::Cxx14 => 201309,
        // C++17 attributes
        "fallthrough" if standard >= CppStandard::Cxx17 => 201603,
        "nodiscard" if standard >= CppStandard::Cxx17 => 201603,
        "maybe_unused" if standard >= CppStandard::Cxx17 => 201603,
        // C++20 attributes
        "likely" if standard >= CppStandard::Cxx20 => 201803,
        "unlikely" if standard >= CppStandard::Cxx20 => 201803,
        "no_unique_address" if standard >= CppStandard::Cxx20 => 201803,
        // C++23 attributes
        "assume" if standard >= CppStandard::Cxx23 => 202207,
        _ => 0,
    }
}

// ═══════════════════════════════════════════════════════════════════════════════
// #include_next support
// ═══════════════════════════════════════════════════════════════════════════════

/// Determines if include_next should be used for a given header,
/// skipping the current include directory and searching the rest of the path.
pub struct IncludeNextResolver {
    /// The include paths, in search order.
    include_paths: Vec<String>,
    /// The current include directory (to skip).
    current_dir: String,
}

impl IncludeNextResolver {
    pub fn new(include_paths: Vec<String>, current_dir: &str) -> Self {
        Self {
            include_paths,
            current_dir: current_dir.to_string(),
        }
    }

    /// Resolve an `#include_next "header"` directive.
    /// Returns the resolved file path, if found.
    pub fn resolve(&self, header: &str) -> Option<String> {
        let mut found_current = false;
        for path in &self.include_paths {
            if path == &self.current_dir {
                found_current = true;
                continue;
            }
            if found_current {
                let candidate = std::path::Path::new(path).join(header);
                if candidate.exists() {
                    return Some(candidate.to_string_lossy().to_string());
                }
            }
        }
        None
    }
}

// ═══════════════════════════════════════════════════════════════════════════════
// #warning directive
// ═══════════════════════════════════════════════════════════════════════════════

/// Process a `#warning "message"` directive.
pub fn process_warning_directive(message: &str) -> String {
    format!("#warning: {}", message)
}

// ═══════════════════════════════════════════════════════════════════════════════
// #embed directive (C++26 / C23)
// ═══════════════════════════════════════════════════════════════════════════════

/// Process a `#embed "filename"` directive.
/// Reads a file and embeds its binary content as an initializer list.
pub struct EmbedProcessor;

impl EmbedProcessor {
    /// Process an `#embed` directive.
    /// Returns the embedded data as a comma-separated list of byte values.
    pub fn process_embed(filename: &str, limit: Option<usize>) -> Result<String, String> {
        let data =
            std::fs::read(filename).map_err(|e| format!("cannot embed '{}': {}", filename, e))?;

        let limit = limit.unwrap_or(data.len());
        let bytes: Vec<String> = data.iter().take(limit).map(|b| format!("{}", b)).collect();

        Ok(bytes.join(", "))
    }

    /// Generate an `#embed` result as an LLVM IR constant array.
    pub fn embed_to_ir(filename: &str, limit: Option<usize>) -> Result<String, String> {
        let data =
            std::fs::read(filename).map_err(|e| format!("cannot embed '{}': {}", filename, e))?;
        let limit = limit.unwrap_or(data.len());
        let bytes: Vec<String> = data
            .iter()
            .take(limit)
            .map(|b| format!("i8 {}", b))
            .collect();
        Ok(format!("[{} x i8] [{}]", bytes.len(), bytes.join(", ")))
    }
}

// ═══════════════════════════════════════════════════════════════════════════════
// Header Unit Support (C++20)
// ═══════════════════════════════════════════════════════════════════════════════

/// Support for C++20 header units: `import <header>;` and `export import <header>;`
pub struct HeaderUnitProcessor {
    pub standard: CppStandard,
}

impl HeaderUnitProcessor {
    pub fn new(standard: CppStandard) -> Self {
        Self { standard }
    }

    /// Check if a header can be imported as a header unit.
    pub fn is_importable_header(&self, header: &str) -> bool {
        if self.standard < CppStandard::Cxx20 {
            return false;
        }
        // Common importable headers
        let importable = [
            "<iostream>",
            "<vector>",
            "<string>",
            "<algorithm>",
            "<memory>",
            "<map>",
            "<set>",
            "<utility>",
        ];
        importable.iter().any(|&h| h == header)
    }

    /// Generate the BMI (Binary Module Interface) filename for a header unit.
    pub fn bmi_filename(&self, header: &str) -> String {
        let clean = header
            .replace(&['<', '>', '"'] as &[_], "")
            .replace('/', "_");
        format!("{}.pcm", clean)
    }
}

// ═══════════════════════════════════════════════════════════════════════════════
// Preprocessor expression evaluator
// ═══════════════════════════════════════════════════════════════════════════════

/// Evaluate C++ preprocessor constant expressions.
pub struct CppPreprocessorExpr;

impl CppPreprocessorExpr {
    /// Evaluate a simple integer expression in #if context.
    pub fn eval(expr: &str, macros: &std::collections::HashMap<String, String>) -> Option<i64> {
        // Substitute known macros
        let mut substituted = expr.to_string();
        for (name, value) in macros {
            substituted = substituted.replace(name, value);
        }

        // Handle __has_include directives in #if
        if substituted.contains("__has_include") {
            return Some(1); // assume available
        }
        if substituted.contains("__has_cpp_attribute") {
            return Some(1); // assume available
        }

        // Try to parse as integer
        substituted.parse::<i64>().ok()
    }

    /// Check if an expression is defined (for #ifdef).
    pub fn is_defined(name: &str, macros: &std::collections::HashMap<String, String>) -> bool {
        macros.contains_key(name)
    }
}

// ═══════════════════════════════════════════════════════════════════════════════
// Feature Test Macro Database
// ═══════════════════════════════════════════════════════════════════════════════

/// A database of C++ feature test macros and their values per standard.
pub struct FeatureTestMacroDb;

impl FeatureTestMacroDb {
    /// Look up a feature test macro and return its value for the given standard.
    pub fn lookup(macro_name: &str, standard: CppStandard) -> Option<String> {
        let macros = cpp_predefined_macros(standard);
        macros
            .iter()
            .find(|(name, _)| name == macro_name)
            .map(|(_, value)| value.clone())
    }

    /// List all feature test macros for a standard.
    pub fn list_all(standard: CppStandard) -> Vec<&'static str> {
        let macros = cpp_predefined_macros(standard);
        macros.iter().map(|(name, _)| name.as_str()).collect()
    }
}

// ═══════════════════════════════════════════════════════════════════════════════
// Tests
// ═══════════════════════════════════════════════════════════════════════════════

#[cfg(test)]
mod tests {
    use super::*;

    #[test]
    fn test_cplusplus_value() {
        assert_eq!(cplusplus_value(CppStandard::Cxx98), 199711);
        assert_eq!(cplusplus_value(CppStandard::Cxx11), 201103);
        assert_eq!(cplusplus_value(CppStandard::Cxx14), 201402);
        assert_eq!(cplusplus_value(CppStandard::Cxx17), 201703);
        assert_eq!(cplusplus_value(CppStandard::Cxx20), 202002);
        assert_eq!(cplusplus_value(CppStandard::Cxx23), 202302);
    }

    #[test]
    fn test_cpp_predefined_macros() {
        let macros = cpp_predefined_macros(CppStandard::Cxx17);
        assert!(macros.iter().any(|(n, _)| n == "__cplusplus"));
        assert!(macros.iter().any(|(n, _)| n == "__cpp_constexpr"));
        assert!(macros.iter().any(|(n, _)| n == "__cpp_fold_expressions"));
    }

    #[test]
    fn test_cpp_feature_macros_cxx14() {
        let macros = cpp_predefined_macros(CppStandard::Cxx14);
        assert!(macros.iter().any(|(n, _)| n == "__cpp_generic_lambdas"));
        assert!(macros.iter().any(|(n, _)| n == "__cpp_binary_literals"));
    }

    #[test]
    fn test_cpp_feature_macros_cxx20() {
        let macros = cpp_predefined_macros(CppStandard::Cxx20);
        assert!(macros.iter().any(|(n, _)| n == "__cpp_concepts"));
        assert!(macros.iter().any(|(n, _)| n == "__cpp_coroutines"));
        assert!(macros
            .iter()
            .any(|(n, _)| n == "__cpp_three_way_comparison"));
    }

    #[test]
    fn test_eval_has_include() {
        assert!(eval_has_include("<iostream>", CppStandard::Cxx17));
        assert!(eval_has_include("<vector>", CppStandard::Cxx17));
        assert!(!eval_has_include("<nonexistent>", CppStandard::Cxx17));
    }

    #[test]
    fn test_eval_has_cpp_attribute() {
        assert_eq!(
            eval_has_cpp_attribute("noreturn", CppStandard::Cxx11),
            200809
        );
        assert_eq!(
            eval_has_cpp_attribute("deprecated", CppStandard::Cxx14),
            201309
        );
        assert_eq!(
            eval_has_cpp_attribute("nodiscard", CppStandard::Cxx17),
            201603
        );
        assert_eq!(eval_has_cpp_attribute("likely", CppStandard::Cxx20), 201803);
        assert_eq!(eval_has_cpp_attribute("assume", CppStandard::Cxx23), 202207);
        assert_eq!(eval_has_cpp_attribute("noreturn", CppStandard::Cxx98), 0);
    }

    #[test]
    fn test_include_next_resolver() {
        let resolver = IncludeNextResolver::new(
            vec!["/usr/include".into(), "/usr/local/include".into()],
            "/usr/include",
        );
        // This is a unit test; we can't check actual filesystem in test
        // But the resolver structure should be sound
        assert_eq!(resolver.include_paths.len(), 2);
    }

    #[test]
    fn test_warning_directive() {
        let msg = process_warning_directive("deprecated header");
        assert!(msg.contains("deprecated header"));
    }

    #[test]
    fn test_header_unit_processor() {
        let hp = HeaderUnitProcessor::new(CppStandard::Cxx20);
        assert!(hp.is_importable_header("<iostream>"));
        assert!(hp.is_importable_header("<vector>"));
        assert!(!hp.is_importable_header("<nonexistent>"));
        assert_eq!(hp.bmi_filename("<iostream>"), "iostream.pcm");
    }

    #[test]
    fn test_feature_test_macro_db() {
        let val = FeatureTestMacroDb::lookup("__cpp_constexpr", CppStandard::Cxx17);
        assert!(val.is_some());
        let val = FeatureTestMacroDb::lookup("__cpp_concepts", CppStandard::Cxx17);
        assert!(val.is_none()); // not in C++17
        let val = FeatureTestMacroDb::lookup("__cpp_concepts", CppStandard::Cxx20);
        assert!(val.is_some()); // in C++20
    }
}