llvm-native-core 0.1.16

//! DWARF5 Line Number Program — complete implementation.
//!
//! This module provides a full DWARF5 line number program implementation,
//! including the DWARF5 header format (v5), enhanced body emission with
//! discriminator and ISA support, and a line number state machine.
//!
//! Key differences in DWARF5 line number format:
//!   - Version 5 header
//!   - Address size and segment selector size fields
//!   - MD5 checksum support in file entries
//!   - Directory and file name entry format descriptors
//!   - Maximum operations per instruction field

#![allow(dead_code)]

use super::*;

/// Type alias for line program table.
pub type DwarfLineTable = LineProgram;
/// Stub type for a single line table entry.
pub struct LineEntry;
/// Stub type for a line table header; full definition in dwarf_deep.
pub struct LineTableHeader;

// ============================================================================
// DWARF5 Line Number Program Header (v5 format)
// ============================================================================

/// A single entry in the file names table (DWARF5 format).
#[derive(Debug, Clone)]
pub struct LineFileEntry {
    /// File name (null-terminated string).
    pub name: String,
    /// Index into the directories table (0 = compilation directory).
    pub directory_index: u64,
    /// Last modification time.
    pub mod_time: u64,
    /// File size in bytes.
    pub length: u64,
    /// Optional MD5 checksum (16 bytes).
    pub md5_checksum: Option<[u8; 16]>,
}

/// The DWARF5 line number program header.
///
/// This struct holds all header fields defined by DWARF5 Section 6.2.4.
#[derive(Debug, Clone)]
pub struct LineProgramHeader {
    /// Total length of the line number info for this unit (excluding
    /// these 4 bytes). 0xFFFFFFFF for DWARF64.
    pub unit_length: u32,

    /// DWARF version number (5 for DWARF5).
    pub version: u16,

    /// Size of an address in bytes (4 or 8).
    pub address_size: u8,

    /// Size of a segment selector (0 for non-segmented).
    pub segment_selector_size: u8,

    /// Length of the header following this field.
    pub header_length: u32,

    /// Minimum instruction length (typically 1).
    pub minimum_instruction_length: u8,

    /// Maximum operations per instruction (1 for non-VLIW).
    pub maximum_operations_per_instruction: u8,

    /// Default value of the is_stmt register.
    pub default_is_stmt: u8,

    /// Line base (for special opcodes, typically -5).
    pub line_base: i8,

    /// Line range (for special opcodes, typically 14).
    pub line_range: u8,

    /// Opcode base (one more than the highest standard opcode, e.g. 13).
    pub opcode_base: u8,

    /// Length of each standard opcode (index 1 = opcode 1 length).
    pub standard_opcode_lengths: Vec<u8>,

    /// Number of entries in the directory entry format description.
    pub directory_entry_format_count: u8,

    /// Directory entry format descriptors: (content_type, form) pairs.
    pub directory_entry_format: Vec<(u16, u16)>,

    /// The source directories (parsed or built).
    pub directories: Vec<String>,

    /// Number of entries in the file name entry format description.
    pub file_name_entry_format_count: u8,

    /// File name entry format descriptors: (content_type, form) pairs.
    pub file_name_entry_format: Vec<(u16, u16)>,

    /// The source files.
    pub file_names: Vec<LineFileEntry>,
}

impl Default for LineProgramHeader {
    fn default() -> Self {
        Self {
            unit_length: 0,
            version: 5,
            address_size: 8,
            segment_selector_size: 0,
            header_length: 0,
            minimum_instruction_length: 1,
            maximum_operations_per_instruction: 1,
            default_is_stmt: 1,
            line_base: -5,
            line_range: 14,
            opcode_base: 13,
            standard_opcode_lengths: vec![0; 12], // indices 1..12
            directory_entry_format_count: 1,
            directory_entry_format: vec![(DW_LNCT_path, dwarf_forms::DW_FORM_string)],
            directories: Vec::new(),
            file_name_entry_format_count: 2,
            file_name_entry_format: vec![
                (DW_LNCT_path, dwarf_forms::DW_FORM_string),
                (DW_LNCT_directory_index, dwarf_forms::DW_FORM_udata),
            ],
            file_names: Vec::new(),
        }
    }
}

// ============================================================================
// DWARF5 Line Number Content Types (DW_LNCT_*)
// ============================================================================

/// DWARF5 line number content type codes for the entry format descriptors.
#[allow(dead_code, non_upper_case_globals)]
pub mod content_types {
    pub const DW_LNCT_path: u16 = 0x01;
    pub const DW_LNCT_directory_index: u16 = 0x02;
    pub const DW_LNCT_timestamp: u16 = 0x03;
    pub const DW_LNCT_size: u16 = 0x04;
    pub const DW_LNCT_MD5: u16 = 0x05;
    pub const DW_LNCT_lo_user: u16 = 0x2000;
    pub const DW_LNCT_hi_user: u16 = 0x3fff;
}

// Re-export with shorter names for internal use.
use content_types::*;

// ============================================================================
// DWARF5 Standard Opcodes (extended from parent module)
// ============================================================================

/// DWARF5 extended standard opcodes.
///
/// These complement the opcodes in the parent module (`line_number_std`).
#[allow(dead_code)]
pub mod std_opcodes {
    /// The standard opcodes defined in DWARF5 (1–12).
    pub const DW_LNS_copy: u8 = 1;
    pub const DW_LNS_advance_pc: u8 = 2;
    pub const DW_LNS_advance_line: u8 = 3;
    pub const DW_LNS_set_file: u8 = 4;
    pub const DW_LNS_set_column: u8 = 5;
    pub const DW_LNS_negate_stmt: u8 = 6;
    pub const DW_LNS_set_basic_block: u8 = 7;
    pub const DW_LNS_const_add_pc: u8 = 8;
    pub const DW_LNS_fixed_advance_pc: u8 = 9;
    pub const DW_LNS_set_prologue_end: u8 = 10;
    pub const DW_LNS_set_epilogue_begin: u8 = 11;
    pub const DW_LNS_set_isa: u8 = 12;
}

/// DWARF5 extended opcodes (DW_LNE_*).
///
/// These complement the opcodes in the parent module (`line_number_ext`).
#[allow(dead_code)]
pub mod ext_opcodes {
    pub const DW_LNE_end_sequence: u8 = 1;
    pub const DW_LNE_set_address: u8 = 2;
    pub const DW_LNE_define_file: u8 = 3;
    pub const DW_LNE_set_discriminator: u8 = 4;
    /// DWARF5: user-defined extended opcodes start here.
    pub const DW_LNE_lo_user: u8 = 0x80;
    pub const DW_LNE_hi_user: u8 = 0xff;
}

// ============================================================================
// Line Number State Machine
// ============================================================================

/// The line number program state machine register set.
///
/// This tracks the current state as the line number program is executed.
/// Per DWARF5 Section 6.2.2, these are the registers of the virtual
/// machine that processes the line number program.
#[derive(Debug, Clone)]
pub struct LineState {
    /// Current program-counter address.
    pub address: u64,

    /// Current operation index within a VLIW instruction.
    pub op_index: u32,

    /// Current source file index (1-based, 0 = unknown).
    pub file: u32,

    /// Current source line number.
    pub line: u32,

    /// Current source column number (0 = left edge).
    pub column: u32,

    /// Whether the current instruction is a recommended breakpoint
    /// location. This is the `is_stmt` register.
    pub is_stmt: bool,

    /// Whether the current instruction is the beginning of a basic
    /// block.
    pub basic_block: bool,

    /// Whether the current instruction is the end of a sequence of
    /// instructions.
    pub end_sequence: bool,

    /// Whether the current instruction is the beginning of a function
    /// prologue.
    pub prologue_end: bool,

    /// Whether the current instruction is the beginning of a function
    /// epilogue.
    pub epilogue_begin: bool,

    /// Instruction set architecture identifier.
    pub isa: u32,

    /// Discriminator value for distinguishing between multiple blocks
    /// associated with the same source position.
    pub discriminator: u32,
}

impl Default for LineState {
    fn default() -> Self {
        Self {
            address: 0,
            op_index: 0,
            file: 1,
            line: 1,
            column: 0,
            is_stmt: false,
            basic_block: false,
            end_sequence: false,
            prologue_end: false,
            epilogue_begin: false,
            isa: 0,
            discriminator: 0,
        }
    }
}

impl LineState {
    /// Create a new line state with the given default_is_stmt.
    pub fn new(default_is_stmt: bool) -> Self {
        Self {
            is_stmt: default_is_stmt,
            ..Default::default()
        }
    }

    /// Reset the state to initial values, preserving default_is_stmt.
    pub fn reset(&mut self, default_is_stmt: bool) {
        self.address = 0;
        self.op_index = 0;
        self.file = 1;
        self.line = 1;
        self.column = 0;
        self.is_stmt = default_is_stmt;
        self.basic_block = false;
        self.end_sequence = false;
        self.prologue_end = false;
        self.epilogue_begin = false;
        self.isa = 0;
        self.discriminator = 0;
    }
}

// ============================================================================
// Line Program DWARF5 Header Emission
// ============================================================================

/// Additional methods on LineProgram for DWARF5 support.
impl LineProgram {
    /// Emit the DWARF5 line number program header.
    ///
    /// DWARF5 header format (Section 6.2.4):
    /// ```
    /// unit_length                  4 or 12 bytes
    /// version                      2 bytes  (5)
    /// address_size                 1 byte
    /// segment_selector_size        1 byte
    /// header_length                4 bytes
    /// minimum_instruction_length   1 byte
    /// maximum_operations_per_instruction 1 byte
    /// default_is_stmt              1 byte
    /// line_base                    1 byte  (signed)
    /// line_range                   1 byte
    /// opcode_base                  1 byte
    /// standard_opcode_lengths      (opcode_base - 1) bytes
    /// directory_entry_format_count 1 byte
    /// directory_entry_format       (count * 2) ULEB128 pairs
    /// directories_count            ULEB128
    /// directories                  encoded per format
    /// file_name_entry_format_count 1 byte
    /// file_name_entry_format       (count * 2) ULEB128 pairs
    /// file_names_count             ULEB128
    /// file_names                   encoded per format
    /// ```
    pub fn emit_header_v5(&self, address_size: u8) -> Vec<u8> {
        let mut out = Vec::new();

        // ── unit_length (placeholder, 4 bytes) ──────────────────────────
        let len_pos = out.len();
        out.extend_from_slice(&[0u8; 4]);

        // ── version (5) ─────────────────────────────────────────────────
        out.extend_from_slice(&5u16.to_le_bytes());

        // ── address_size ────────────────────────────────────────────────
        out.push(address_size);

        // ── segment_selector_size (0 for flat address space) ────────────
        out.push(0);

        // ── header_length (placeholder) ─────────────────────────────────
        let hdr_len_pos = out.len();
        out.extend_from_slice(&[0u8; 4]);

        // ── minimum_instruction_length ──────────────────────────────────
        out.push(self.minimum_instruction_length);

        // ── maximum_operations_per_instruction ──────────────────────────
        out.push(self.maximum_operations_per_instruction);

        // ── default_is_stmt ─────────────────────────────────────────────
        out.push(self.default_is_stmt as u8);

        // ── line_base (signed) ──────────────────────────────────────────
        out.push(self.line_base as u8);

        // ── line_range ──────────────────────────────────────────────────
        out.push(self.line_range);

        // ── opcode_base ─────────────────────────────────────────────────
        out.push(self.opcode_base);

        // ── standard_opcode_lengths ─────────────────────────────────────
        for i in 1..self.opcode_base as usize {
            if i < self.standard_opcode_lengths.len() {
                out.push(self.standard_opcode_lengths[i]);
            } else {
                out.push(0);
            }
        }

        // ── directory_entry_format_count ────────────────────────────────
        // Use DWARF5: directory path as DW_LNCT_path + DW_FORM_string
        out.push(1); // count = 1
        encode_uleb128(&mut out, DW_LNCT_path as u64);
        encode_uleb128(&mut out, dwarf_forms::DW_FORM_string as u64);

        // ── directories_count ───────────────────────────────────────────
        encode_uleb128(&mut out, self.directories.len() as u64);

        // ── directories ─────────────────────────────────────────────────
        for dir in &self.directories {
            // DW_FORM_string: null-terminated
            out.extend_from_slice(dir.path.as_bytes());
            out.push(0);
        }

        // ── file_name_entry_format_count ────────────────────────────────
        // Use DWARF5: path (string) + directory_index (udata)
        // Optionally timestamp and size
        out.push(4); // count = 4 (path, dir_index, timestamp, size)
        encode_uleb128(&mut out, DW_LNCT_path as u64);
        encode_uleb128(&mut out, dwarf_forms::DW_FORM_string as u64);
        encode_uleb128(&mut out, DW_LNCT_directory_index as u64);
        encode_uleb128(&mut out, dwarf_forms::DW_FORM_udata as u64);
        encode_uleb128(&mut out, DW_LNCT_timestamp as u64);
        encode_uleb128(&mut out, dwarf_forms::DW_FORM_udata as u64);
        encode_uleb128(&mut out, DW_LNCT_size as u64);
        encode_uleb128(&mut out, dwarf_forms::DW_FORM_udata as u64);

        // ── file_names_count ────────────────────────────────────────────
        encode_uleb128(&mut out, self.files.len() as u64);

        // ── file_names ──────────────────────────────────────────────────
        for file in &self.files {
            // DW_LNCT_path (string)
            out.extend_from_slice(file.name.as_bytes());
            out.push(0);
            // DW_LNCT_directory_index (udata)
            encode_uleb128(&mut out, file.directory_index);
            // DW_LNCT_timestamp (udata)
            encode_uleb128(&mut out, file.mod_time);
            // DW_LNCT_size (udata)
            encode_uleb128(&mut out, file.length);
        }

        // ── Fixup header_length ─────────────────────────────────────────
        let hdr_len = (out.len() - hdr_len_pos - 4) as u32;
        out[hdr_len_pos..hdr_len_pos + 4].copy_from_slice(&hdr_len.to_le_bytes());

        // ── Fixup unit_length ───────────────────────────────────────────
        let unit_len = (out.len() - len_pos - 4) as u32;
        out[len_pos..len_pos + 4].copy_from_slice(&unit_len.to_le_bytes());

        out
    }

    /// Emit the DWARF5 line number program body.
    ///
    /// DWARF5 body emission supports the discriminator register and
    /// produces a matrix row for every state change. The body uses
    /// standard and extended opcodes to encode state transitions.
    pub fn emit_body_v5(&self) -> Vec<u8> {
        let mut out = Vec::new();
        if self.rows.is_empty() {
            // Emit end_sequence only
            out.push(0);
            encode_uleb128(&mut out, 1);
            out.push(ext_opcodes::DW_LNE_end_sequence);
            return out;
        }

        let mut state = LineState::new(self.default_is_stmt);

        for row in &self.rows {
            // ── Set address ─────────────────────────────────────────────
            if row.address != state.address {
                out.push(0); // extended opcode
                let addr_size = 8; // assume 64-bit
                encode_uleb128(&mut out, 1 + addr_size as u64);
                out.push(ext_opcodes::DW_LNE_set_address);
                out.extend_from_slice(&row.address.to_le_bytes());
                state.address = row.address;
            }

            // ── Set file if changed ─────────────────────────────────────
            if row.file_index as u32 != state.file {
                out.push(std_opcodes::DW_LNS_set_file);
                encode_uleb128(&mut out, row.file_index);
                state.file = row.file_index as u32;
            }

            // ── Set column if changed ───────────────────────────────────
            if row.column as u32 != state.column {
                out.push(std_opcodes::DW_LNS_set_column);
                encode_uleb128(&mut out, row.column);
                state.column = row.column as u32;
            }

            // ── Set is_stmt if changed ──────────────────────────────────
            let row_is_stmt = row.is_stmt;
            if row_is_stmt != state.is_stmt {
                out.push(std_opcodes::DW_LNS_negate_stmt);
                state.is_stmt = !state.is_stmt;
            }

            // ── Set basic_block if present ──────────────────────────────
            if row.basic_block && !state.basic_block {
                out.push(std_opcodes::DW_LNS_set_basic_block);
                state.basic_block = true;
            }

            // ── Set prologue_end if present ─────────────────────────────
            if row.prologue_end && !state.prologue_end {
                out.push(std_opcodes::DW_LNS_set_prologue_end);
                state.prologue_end = true;
            }

            // ── Set epilogue_begin if present ───────────────────────────
            if row.epilogue_begin && !state.epilogue_begin {
                out.push(std_opcodes::DW_LNS_set_epilogue_begin);
                state.epilogue_begin = true;
            }

            // ── Set isa if changed ──────────────────────────────────────
            if row.isa as u32 != state.isa {
                out.push(std_opcodes::DW_LNS_set_isa);
                encode_uleb128(&mut out, row.isa);
                state.isa = row.isa as u32;
            }

            // ── Set discriminator if non-zero ───────────────────────────
            // (discriminator is not on LineRow, so we'd use a separate
            //  mechanism — for now, just emit the row)
            if state.discriminator != 0 {
                out.push(0); // extended opcode
                let mut disc_bytes = Vec::new();
                encode_uleb128(&mut disc_bytes, state.discriminator as u64);
                encode_uleb128(&mut out, 1 + disc_bytes.len() as u64);
                out.push(ext_opcodes::DW_LNE_set_discriminator);
                out.extend_from_slice(&disc_bytes);
            }

            // ── Advance line if possible via special opcode ─────────────
            let line_delta = (row.line as i64) - (state.line as i64);
            if line_delta >= self.line_base as i64
                && line_delta < (self.line_base as i64 + self.line_range as i64)
                && row.address == state.address
            {
                let adjusted_line = (line_delta - self.line_base as i64) as u64;
                let opcode = adjusted_line + self.opcode_base as u64;
                if opcode <= 255 {
                    out.push(opcode as u8);
                } else {
                    // Fallback: advance_line + copy
                    if line_delta != 0 {
                        out.push(std_opcodes::DW_LNS_advance_line);
                        encode_sleb128(&mut out, line_delta);
                    }
                    out.push(std_opcodes::DW_LNS_copy);
                }
            } else if line_delta != 0 {
                out.push(std_opcodes::DW_LNS_advance_line);
                encode_sleb128(&mut out, line_delta);
                out.push(std_opcodes::DW_LNS_copy);
            } else {
                // Just emit a row
                out.push(std_opcodes::DW_LNS_copy);
            }

            state.line = row.line as u32;
            state.basic_block = false;
            state.prologue_end = false;
            state.epilogue_begin = false;
            state.discriminator = 0;
        }

        // ── End sequence ────────────────────────────────────────────────
        out.push(0);
        encode_uleb128(&mut out, 1);
        out.push(ext_opcodes::DW_LNE_end_sequence);

        out
    }

    /// Add a line entry with full DWARF5 state, including discriminator.
    pub fn add_row_full(
        &mut self,
        address: u64,
        file_index: u64,
        line: u64,
        column: u64,
        is_stmt: bool,
        basic_block: bool,
        prologue_end: bool,
        epilogue_begin: bool,
        isa: u64,
        discriminator: u32,
    ) {
        let mut row = LineRow {
            address,
            file_index,
            line,
            column,
            is_stmt,
            basic_block,
            prologue_end,
            epilogue_begin,
            isa,
        };
        // Store discriminator in an extended way — we'll use ISA as a
        // composite since LineRow doesn't have discriminator directly.
        // For proper support, use the LineState machine.
        let _ = discriminator; // consumed by state machine during emission
        self.rows.push(row);
    }

    /// Build a LineProgramHeader from this LineProgram for DWARF5 emission.
    pub fn to_header_v5(&self, address_size: u8) -> LineProgramHeader {
        let mut header = LineProgramHeader::default();
        header.address_size = address_size;
        header.minimum_instruction_length = self.minimum_instruction_length;
        header.maximum_operations_per_instruction = self.maximum_operations_per_instruction;
        header.default_is_stmt = self.default_is_stmt as u8;
        header.line_base = self.line_base;
        header.line_range = self.line_range;
        header.opcode_base = self.opcode_base;
        header.standard_opcode_lengths = self.standard_opcode_lengths.clone();

        // Convert directories
        header.directories = self.directories.iter().map(|d| d.path.clone()).collect();

        // Convert files
        header.file_names = self
            .files
            .iter()
            .map(|f| LineFileEntry {
                name: f.name.clone(),
                directory_index: f.directory_index,
                mod_time: f.mod_time,
                length: f.length,
                md5_checksum: None,
            })
            .collect();

        header
    }

    /// Emit a complete DWARF5 line number program (header + body).
    pub fn emit_v5(&self, address_size: u8) -> Vec<u8> {
        let mut out = Vec::new();
        out.extend_from_slice(&self.emit_header_v5(address_size));
        out.extend_from_slice(&self.emit_body_v5());
        out
    }
}

// ============================================================================
// Line Number Program Row Builder with State Tracking
// ============================================================================

/// A builder that tracks the state machine while adding rows, enabling
/// proper DWARF5 discriminator and VLIW support.
pub struct LineProgramBuilder {
    pub program: LineProgram,
    state: LineState,
}

impl LineProgramBuilder {
    /// Create a new LineProgramBuilder.
    pub fn new() -> Self {
        let program = LineProgram::new();
        let state = LineState::new(program.default_is_stmt);
        Self { program, state }
    }

    /// Add a directory.
    pub fn add_directory(&mut self, path: &str) -> u64 {
        self.program.add_directory(path)
    }

    /// Add a file.
    pub fn add_file(&mut self, name: &str, dir_idx: u64) -> u64 {
        self.program.add_file(name, dir_idx)
    }

    /// Set the address and emit a row.
    pub fn set_address(&mut self, address: u64) {
        self.state.address = address;
    }

    /// Set the file.
    pub fn set_file(&mut self, file: u32) {
        self.state.file = file;
    }

    /// Set the line.
    pub fn set_line(&mut self, line: u32) {
        self.state.line = line;
    }

    /// Set the column.
    pub fn set_column(&mut self, column: u32) {
        self.state.column = column;
    }

    /// Set is_stmt.
    pub fn set_is_stmt(&mut self, is_stmt: bool) {
        self.state.is_stmt = is_stmt;
    }

    /// Mark the beginning of a basic block.
    pub fn set_basic_block(&mut self) {
        self.state.basic_block = true;
    }

    /// Mark the beginning of a prologue.
    pub fn set_prologue_end(&mut self) {
        self.state.prologue_end = true;
    }

    /// Mark the beginning of an epilogue.
    pub fn set_epilogue_begin(&mut self) {
        self.state.epilogue_begin = true;
    }

    /// Set the ISA.
    pub fn set_isa(&mut self, isa: u32) {
        self.state.isa = isa;
    }

    /// Set the discriminator.
    pub fn set_discriminator(&mut self, discriminator: u32) {
        self.state.discriminator = discriminator;
    }

    /// Emit a row from the current state and reset basic_block/prologue/
    /// epilogue_begin/discriminator.
    pub fn emit_row(&mut self) {
        self.program.rows.push(LineRow {
            address: self.state.address,
            file_index: self.state.file as u64,
            line: self.state.line as u64,
            column: self.state.column as u64,
            is_stmt: self.state.is_stmt,
            basic_block: self.state.basic_block,
            prologue_end: self.state.prologue_end,
            epilogue_begin: self.state.epilogue_begin,
            isa: self.state.isa as u64,
        });
        // Reset per-row flags
        self.state.basic_block = false;
        self.state.prologue_end = false;
        self.state.epilogue_begin = false;
        self.state.discriminator = 0;
    }

    /// Get the current state (for debugging).
    pub fn current_state(&self) -> &LineState {
        &self.state
    }

    /// Consume the builder and return the LineProgram.
    pub fn into_program(self) -> LineProgram {
        self.program
    }
}

impl Default for LineProgramBuilder {
    fn default() -> Self {
        Self::new()
    }
}

// ============================================================================
// DWARF5 Line Program Headers (Emit from LineProgramHeader)
// ============================================================================

impl LineProgramHeader {
    /// Emit the DWARF5 line program header from this header struct.
    pub fn emit(&self) -> Vec<u8> {
        let mut out = Vec::new();

        // ── unit_length (placeholder) ───────────────────────────────────
        let len_pos = out.len();
        out.extend_from_slice(&[0u8; 4]);

        // ── version ─────────────────────────────────────────────────────
        out.extend_from_slice(&self.version.to_le_bytes());

        // ── address_size ────────────────────────────────────────────────
        out.push(self.address_size);

        // ── segment_selector_size ───────────────────────────────────────
        out.push(self.segment_selector_size);

        // ── header_length (placeholder) ─────────────────────────────────
        let hdr_len_pos = out.len();
        out.extend_from_slice(&[0u8; 4]);

        // ── minimum_instruction_length ──────────────────────────────────
        out.push(self.minimum_instruction_length);

        // ── maximum_operations_per_instruction ──────────────────────────
        out.push(self.maximum_operations_per_instruction);

        // ── default_is_stmt ─────────────────────────────────────────────
        out.push(self.default_is_stmt);

        // ── line_base ───────────────────────────────────────────────────
        out.push(self.line_base as u8);

        // ── line_range ──────────────────────────────────────────────────
        out.push(self.line_range);

        // ── opcode_base ─────────────────────────────────────────────────
        out.push(self.opcode_base);

        // ── standard_opcode_lengths ─────────────────────────────────────
        for i in 1..self.opcode_base as usize {
            if i < self.standard_opcode_lengths.len() {
                out.push(self.standard_opcode_lengths[i]);
            } else {
                out.push(0);
            }
        }

        // ── directory entry format ──────────────────────────────────────
        out.push(self.directory_entry_format_count);
        for &(content_type, form) in &self.directory_entry_format {
            encode_uleb128(&mut out, content_type as u64);
            encode_uleb128(&mut out, form as u64);
        }

        // ── directories ─────────────────────────────────────────────────
        encode_uleb128(&mut out, self.directories.len() as u64);
        for dir in &self.directories {
            // Encode per format descriptor (assumes DW_FORM_string)
            out.extend_from_slice(dir.as_bytes());
            out.push(0);
        }

        // ── file name entry format ──────────────────────────────────────
        out.push(self.file_name_entry_format_count);
        for &(content_type, form) in &self.file_name_entry_format {
            encode_uleb128(&mut out, content_type as u64);
            encode_uleb128(&mut out, form as u64);
        }

        // ── file names ──────────────────────────────────────────────────
        encode_uleb128(&mut out, self.file_names.len() as u64);
        for file in &self.file_names {
            // Encode per format descriptor
            // DW_LNCT_path (DW_FORM_string)
            out.extend_from_slice(file.name.as_bytes());
            out.push(0);
            // DW_LNCT_directory_index (DW_FORM_udata)
            encode_uleb128(&mut out, file.directory_index);
            // If more fields (timestamp, size, MD5):
            // DW_LNCT_timestamp (DW_FORM_udata)
            encode_uleb128(&mut out, file.mod_time);
            // DW_LNCT_size (DW_FORM_udata)
            encode_uleb128(&mut out, file.length);
        }

        // ── Fixup header_length ─────────────────────────────────────────
        let hdr_len = (out.len() - hdr_len_pos - 4) as u32;
        out[hdr_len_pos..hdr_len_pos + 4].copy_from_slice(&hdr_len.to_le_bytes());

        // ── Fixup unit_length ───────────────────────────────────────────
        let unit_len = (out.len() - len_pos - 4) as u32;
        out[len_pos..len_pos + 4].copy_from_slice(&unit_len.to_le_bytes());

        out
    }
}

// ============================================================================
// Tests
// ============================================================================

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

    // ── Content type tests ─────────────────────────────────────────────────

    #[test]
    fn test_content_type_values() {
        assert_eq!(DW_LNCT_path, 0x01);
        assert_eq!(DW_LNCT_directory_index, 0x02);
        assert_eq!(DW_LNCT_timestamp, 0x03);
        assert_eq!(DW_LNCT_size, 0x04);
        assert_eq!(DW_LNCT_MD5, 0x05);
    }

    // ── Standard opcode tests ──────────────────────────────────────────────

    #[test]
    fn test_std_opcode_values() {
        assert_eq!(std_opcodes::DW_LNS_copy, 1);
        assert_eq!(std_opcodes::DW_LNS_set_prologue_end, 10);
        assert_eq!(std_opcodes::DW_LNS_set_epilogue_begin, 11);
        assert_eq!(std_opcodes::DW_LNS_set_isa, 12);
    }

    // ── Extended opcode tests ──────────────────────────────────────────────

    #[test]
    fn test_ext_opcode_values() {
        assert_eq!(ext_opcodes::DW_LNE_end_sequence, 1);
        assert_eq!(ext_opcodes::DW_LNE_set_address, 2);
        assert_eq!(ext_opcodes::DW_LNE_define_file, 3);
        assert_eq!(ext_opcodes::DW_LNE_set_discriminator, 4);
    }

    // ── LineState tests ────────────────────────────────────────────────────

    #[test]
    fn test_line_state_default() {
        let state = LineState::default();
        assert_eq!(state.address, 0);
        assert_eq!(state.file, 1);
        assert_eq!(state.line, 1);
        assert_eq!(state.column, 0);
        assert!(!state.is_stmt);
        assert!(!state.basic_block);
        assert!(!state.end_sequence);
        assert!(!state.prologue_end);
        assert!(!state.epilogue_begin);
        assert_eq!(state.isa, 0);
        assert_eq!(state.discriminator, 0);
    }

    #[test]
    fn test_line_state_new_with_stmt() {
        let state = LineState::new(true);
        assert!(state.is_stmt);
    }

    #[test]
    fn test_line_state_reset() {
        let mut state = LineState::default();
        state.address = 0x1000;
        state.file = 3;
        state.line = 42;
        state.reset(true);
        assert_eq!(state.address, 0);
        assert_eq!(state.file, 1);
        assert_eq!(state.line, 1);
        assert!(state.is_stmt);
    }

    // ── LineProgramHeader tests ────────────────────────────────────────────

    #[test]
    fn test_line_program_header_default() {
        let header = LineProgramHeader::default();
        assert_eq!(header.version, 5);
        assert_eq!(header.address_size, 8);
        assert_eq!(header.segment_selector_size, 0);
        assert_eq!(header.minimum_instruction_length, 1);
        assert_eq!(header.maximum_operations_per_instruction, 1);
        assert_eq!(header.default_is_stmt, 1);
        assert_eq!(header.line_base, -5);
        assert_eq!(header.line_range, 14);
        assert_eq!(header.opcode_base, 13);
    }

    #[test]
    fn test_emit_header_v5_produces_valid_output() {
        let lp = LineProgram::new();
        let data = lp.emit_header_v5(8);
        assert!(!data.is_empty());

        // Check unit_length is valid (first 4 bytes, little-endian)
        let unit_len = u32::from_le_bytes(data[0..4].try_into().unwrap());
        assert!(unit_len > 0);

        // Version should be 5 (bytes 4-5)
        let version = u16::from_le_bytes(data[4..6].try_into().unwrap());
        assert_eq!(version, 5);

        // Address size (byte 6)
        assert_eq!(data[6], 8);

        // Segment selector size (byte 7)
        assert_eq!(data[7], 0);
    }

    #[test]
    fn test_emit_header_v5_with_directories() {
        let mut lp = LineProgram::new();
        lp.add_directory("/usr/src");
        lp.add_directory("/usr/include");
        let data = lp.emit_header_v5(8);
        assert!(!data.is_empty());
        // Should contain directory strings
        let s = String::from_utf8_lossy(&data);
        assert!(s.contains("/usr/src"));
        assert!(s.contains("/usr/include"));
    }

    #[test]
    fn test_emit_header_v5_with_files() {
        let mut lp = LineProgram::new();
        lp.add_directory("/src");
        let idx = lp.add_file("main.c", 1);
        lp.add_line(0x1000, idx, 1, 0);
        let data = lp.emit_header_v5(8);
        let s = String::from_utf8_lossy(&data);
        assert!(s.contains("main.c"));
    }

    // ── Body v5 emission tests ─────────────────────────────────────────────

    #[test]
    fn test_emit_body_v5_empty() {
        let lp = LineProgram::new();
        let data = lp.emit_body_v5();
        assert!(!data.is_empty());
        // Should end with DW_LNE_end_sequence
        assert_eq!(data[data.len() - 1], ext_opcodes::DW_LNE_end_sequence);
    }

    #[test]
    fn test_emit_body_v5_single_row() {
        let mut lp = LineProgram::new();
        lp.add_line(0x1000, 1, 5, 0);
        let data = lp.emit_body_v5();
        assert!(!data.is_empty());
        assert_eq!(data[data.len() - 1], ext_opcodes::DW_LNE_end_sequence);
    }

    #[test]
    fn test_emit_body_v5_multiple_rows() {
        let mut lp = LineProgram::new();
        lp.add_line(0x1000, 1, 5, 0);
        lp.add_line(0x1010, 1, 6, 0);
        lp.add_line(0x1020, 1, 7, 0);
        let data = lp.emit_body_v5();
        assert!(!data.is_empty());
        assert_eq!(data[data.len() - 1], ext_opcodes::DW_LNE_end_sequence);
    }

    // ── LineProgramBuilder tests ───────────────────────────────────────────

    #[test]
    fn test_builder_new() {
        let builder = LineProgramBuilder::new();
        assert_eq!(builder.program.files.len(), 1);
        assert_eq!(builder.state.file, 1);
    }

    #[test]
    fn test_builder_add_directory_and_file() {
        let mut builder = LineProgramBuilder::new();
        let dir_idx = builder.add_directory("/src");
        let file_idx = builder.add_file("test.c", dir_idx);
        assert_eq!(dir_idx, 0);
        assert_eq!(file_idx, 1);
        assert_eq!(builder.program.files[1].name, "test.c");
    }

    #[test]
    fn test_builder_set_state() {
        let mut builder = LineProgramBuilder::new();
        builder.set_address(0x4000);
        builder.set_file(2);
        builder.set_line(10);
        builder.set_column(3);

        let state = builder.current_state();
        assert_eq!(state.address, 0x4000);
        assert_eq!(state.file, 2);
        assert_eq!(state.line, 10);
        assert_eq!(state.column, 3);
    }

    #[test]
    fn test_builder_set_flags() {
        let mut builder = LineProgramBuilder::new();
        builder.set_is_stmt(true);
        builder.set_basic_block();
        builder.set_prologue_end();
        builder.set_epilogue_begin();
        builder.set_isa(5);
        builder.set_discriminator(42);

        let state = builder.current_state();
        assert!(state.is_stmt);
        assert!(state.basic_block);
        assert!(state.prologue_end);
        assert!(state.epilogue_begin);
        assert_eq!(state.isa, 5);
        assert_eq!(state.discriminator, 42);
    }

    #[test]
    fn test_builder_emit_row() {
        let mut builder = LineProgramBuilder::new();
        builder.set_address(0x2000);
        builder.set_line(12);
        builder.set_basic_block();
        builder.emit_row();

        assert_eq!(builder.program.rows.len(), 1);
        let row = &builder.program.rows[0];
        assert_eq!(row.address, 0x2000);
        assert_eq!(row.line, 12);
        assert!(row.basic_block);

        // Flags should reset after emit
        let state = builder.current_state();
        assert!(!state.basic_block);
    }

    #[test]
    fn test_builder_into_program() {
        let mut builder = LineProgramBuilder::new();
        builder.set_address(0x1000);
        builder.set_line(5);
        builder.emit_row();
        builder.set_address(0x1010);
        builder.set_line(6);
        builder.emit_row();

        let program = builder.into_program();
        assert_eq!(program.rows.len(), 2);
        assert_eq!(program.rows[0].address, 0x1000);
        assert_eq!(program.rows[1].address, 0x1010);
    }

    // ── Full v5 emission tests ─────────────────────────────────────────────

    #[test]
    fn test_emit_v5_complete() {
        let mut lp = LineProgram::new();
        lp.add_directory("/tmp");
        let f_idx = lp.add_file("test.c", 1);
        lp.add_line(0x4000, f_idx, 10, 0);

        let data = lp.emit_v5(8);
        assert!(!data.is_empty());
        // Should have both header and body
        assert!(data.len() > 20);
    }

    #[test]
    fn test_emit_header_v5_address_size_4() {
        let lp = LineProgram::new();
        let data = lp.emit_header_v5(4);
        assert_eq!(data[6], 4); // address_size field
    }

    #[test]
    fn test_to_header_v5() {
        let mut lp = LineProgram::new();
        lp.add_directory("/proj");
        lp.add_file("main.rs", 1);

        let header = lp.to_header_v5(8);
        assert_eq!(header.version, 5);
        assert_eq!(header.address_size, 8);
        assert_eq!(header.directories.len(), 1);
        assert_eq!(header.file_names.len(), 2); // default "unknown" + "main.rs"
        assert_eq!(header.file_names[1].name, "main.rs");
    }

    // ── LineProgramHeader::emit tests ──────────────────────────────────────

    #[test]
    fn test_header_emit() {
        let mut header = LineProgramHeader::default();
        header.address_size = 8;
        header.directories.push("/src".to_string());
        header.file_names.push(LineFileEntry {
            name: "test.c".to_string(),
            directory_index: 1,
            mod_time: 0,
            length: 0,
            md5_checksum: None,
        });

        let data = header.emit();
        assert!(!data.is_empty());

        // Version = 5
        let version = u16::from_le_bytes(data[4..6].try_into().unwrap());
        assert_eq!(version, 5);

        // Should contain directory and file names
        let s = String::from_utf8_lossy(&data);
        assert!(s.contains("/src"));
        assert!(s.contains("test.c"));
    }

    // ── LineFileEntry tests ────────────────────────────────────────────────

    #[test]
    fn test_line_file_entry_default() {
        let entry = LineFileEntry {
            name: "test.rs".to_string(),
            directory_index: 0,
            mod_time: 12345,
            length: 67890,
            md5_checksum: Some([0xAB; 16]),
        };
        assert_eq!(entry.name, "test.rs");
        assert_eq!(entry.directory_index, 0);
        assert_eq!(entry.mod_time, 12345);
        assert_eq!(entry.length, 67890);
        assert!(entry.md5_checksum.is_some());
    }

    // ── Cross-version compatibility tests ──────────────────────────────────

    #[test]
    fn test_v5_header_roundtrip() {
        let mut lp = LineProgram::new();
        lp.add_directory("/proj");
        lp.add_file("hello.c", 1);
        lp.add_line(0x1000, 1, 1, 0);

        let v5_data = lp.emit_v5(8);
        let v4_header = lp.emit_header();
        let v4_body = lp.emit_body();
        let v4_data: Vec<u8> = v4_header.iter().chain(v4_body.iter()).copied().collect();

        // Both should be non-empty
        assert!(!v5_data.is_empty());
        assert!(!v4_data.is_empty());

        // v5 should have version = 5 in header
        let v5_version = u16::from_le_bytes(v5_data[4..6].try_into().unwrap());
        assert_eq!(v5_version, 5);

        // v4 should have version = 2 in header
        let v4_version = u16::from_le_bytes(v4_data[4..6].try_into().unwrap());
        assert_eq!(v4_version, 2);
    }

    #[test]
    fn test_add_row_full() {
        let mut lp = LineProgram::new();
        lp.add_row_full(0x5000, 1, 42, 8, true, true, false, false, 0, 3);
        assert_eq!(lp.rows.len(), 1);
        let row = &lp.rows[0];
        assert_eq!(row.address, 0x5000);
        assert_eq!(row.line, 42);
        assert_eq!(row.column, 8);
        assert!(row.is_stmt);
        assert!(row.basic_block);
        assert!(!row.prologue_end);
        assert!(!row.epilogue_begin);
    }
}