jessebrault/deimos-lang
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

Compare commits

base: jessebrault:c5781114a5cf427744bdd415603428a5c0c8aaa2
Branches Tags
jessebrault:main
..
compare: jessebrault:916b6377ac6cce2706e27e732a5d1edc2a760d03
Branches Tags
jessebrault:main

No commits in common. "c5781114a5cf427744bdd415603428a5c0c8aaa2" and "916b6377ac6cce2706e27e732a5d1edc2a760d03" have entirely different histories.
c5781114a5 ... 916b6377ac

21 changed files with 424 additions and 679 deletions
Show Stats Expand all files Collapse all files

21
dm/src/main.rs
View File

@ -4,7 +4,6 @@ mod run;
use crate::repl::repl;
use crate::run::compile_and_run_script;
use clap::{Parser, Subcommand};
use std::io;
use std::path::PathBuf;
#[derive(Debug, Parser)]
@ -24,27 +23,29 @@ enum SubCommand {
#[arg(long)]
show_ir: bool,
#[arg(long, default_value = "8")]
register_count: usize,
},
Repl {
#[arg(long, default_value = "8")]
register_count: usize,
},
Repl,
}
fn main() {
let register_count = std::env::var("DVM_REGISTER_COUNT")
.map(|v| v.parse::<usize>())
.unwrap_or(Ok(8))
.unwrap();
let args = Cli::parse();
match &args.sub_command {
SubCommand::Run {
script,
show_asm,
show_ir,
register_count,
} => {
compile_and_run_script(script, *show_asm, *show_ir, register_count);
compile_and_run_script(script, *show_asm, *show_ir, *register_count);
}
SubCommand::Repl => {
repl(&mut io::stdin().lock(), register_count);
SubCommand::Repl { register_count } => {
repl(*register_count);
}
}
}

206
dm/src/repl.rs
View File

@ -4,44 +4,28 @@ use dmc_lib::diagnostic::Diagnostic;
use dmc_lib::ir::ir_function::IrFunction;
use dmc_lib::ir::ir_return::IrReturn;
use dmc_lib::ir::ir_statement::IrStatement;
use dmc_lib::ir::ir_variable::IrVariable;
use dmc_lib::lexer::Lexer;
use dmc_lib::offset_counter::OffsetCounter;
use dmc_lib::parser::{parse_expression, parse_let_statement};
use dmc_lib::symbol::variable_symbol::VariableSymbol;
use dmc_lib::parser::parse_expression;
use dmc_lib::symbol_table::SymbolTable;
use dmc_lib::token::TokenKind;
use dmc_lib::type_info::TypeInfo;
use dmc_lib::types_table::TypesTable;
use dvm_lib::vm::constant::{Constant, StringConstant};
use dvm_lib::vm::function::Function;
use dvm_lib::vm::operand::Operand;
use dvm_lib::vm::{CallStack, DvmContext, loop_instructions, prepare_for_instruction_loop};
use std::cell::RefCell;
use std::collections::HashMap;
use dvm_lib::vm::{CallStack, DvmContext, call};
use std::io;
use std::io::{BufRead, Write};
use std::rc::Rc;
use std::io::Write;
pub fn repl(read: &mut impl BufRead, register_count: usize) {
pub fn repl(register_count: usize) {
let mut buffer = String::new();
let mut symbol_table = SymbolTable::new();
let mut types_table = TypesTable::new();
let mut repl_fn_body_scope_id: Option<usize> = None;
let mut repl_fn_offset_counter = OffsetCounter::new();
let mut repl_fn_local_variables = HashMap::new();
let mut constants_table = ConstantsTable::new();
let mut context = DvmContext::new();
let mut repl_fn_stack_locals: Vec<Operand> = vec![];
'repl: loop {
print!("> ");
io::stdout().flush().unwrap();
read.read_line(&mut buffer).unwrap();
io::stdin().read_line(&mut buffer).unwrap();
let input = buffer.trim();
if input.is_empty() {
buffer.clear();
@ -68,40 +52,9 @@ pub fn repl(read: &mut impl BufRead, register_count: usize) {
todo!("Parse functions in repl")
}
TokenKind::Let => {
match compile_let_statement(
input,
register_count,
&mut symbol_table,
&mut repl_fn_body_scope_id,
&mut repl_fn_offset_counter,
&mut repl_fn_local_variables,
&mut types_table,
&mut constants_table,
) {
Ok(function) => {
context
.functions_mut()
.insert(function.name_owned(), function);
}
Err(diagnostics) => {
for diagnostic in diagnostics {
eprintln!("{}", diagnostic.message());
}
buffer.clear();
continue 'repl;
}
}
todo!("Parse let statements in repl")
}
_ => match compile_expression(
input,
register_count,
&mut symbol_table,
&mut types_table,
&mut repl_fn_body_scope_id,
&mut repl_fn_local_variables,
&mut repl_fn_offset_counter,
&mut constants_table,
) {
_ => match compile_expression(input, register_count, &mut constants_table) {
Ok(function) => {
context
.functions_mut()
@ -110,9 +63,9 @@ pub fn repl(read: &mut impl BufRead, register_count: usize) {
Err(diagnostics) => {
for diagnostic in &diagnostics {
eprintln!("{}", diagnostic.message());
buffer.clear();
continue 'repl;
}
buffer.clear();
continue 'repl;
}
},
}
@ -124,25 +77,10 @@ pub fn repl(read: &mut impl BufRead, register_count: usize) {
);
}
let mut registers = vec![Operand::Null; register_count];
let mut call_stack = CallStack::new();
prepare_for_instruction_loop(&context, "__repl", &mut call_stack, &[]);
// copy all old locals to current call frame's stack
// this has to be done with indexing because the preparation above creates space for them
for (i, operand) in repl_fn_stack_locals.iter().enumerate() {
let target_index = call_stack.top().fp() + i;
call_stack.top_mut().stack_mut()[target_index] = operand.clone();
}
let result = loop_instructions(
&context,
&mut vec![Operand::Null; register_count],
&mut call_stack,
);
// copy the top frame's stack locals back to OUR stack locals for next iteration
repl_fn_stack_locals = std::mem::take(call_stack.top_mut().stack_mut());
let result = call(&context, &mut registers, &mut call_stack, "__repl", &[]);
if let Some(value) = result {
println!("{}", value);
}
@ -151,58 +89,35 @@ pub fn repl(read: &mut impl BufRead, register_count: usize) {
}
}
fn prepare_scopes(symbol_table: &mut SymbolTable, fn_body_scope_id: &mut Option<usize>) -> usize {
if let Some(scope_id) = fn_body_scope_id {
symbol_table.change_scope(*scope_id);
*scope_id
} else {
symbol_table.push_module_scope("__repl_module");
symbol_table.push_function_scope("__repl_fn");
let container_scope_id = symbol_table.push_block_scope("__repl_fn_body");
fn_body_scope_id.replace(container_scope_id);
container_scope_id
}
}
fn compile_expression(
input: &str,
register_count: usize,
symbol_table: &mut SymbolTable,
types_table: &mut TypesTable,
fn_body_scope_id: &mut Option<usize>,
fn_local_variables: &HashMap<Rc<VariableSymbol>, Rc<RefCell<IrVariable>>>,
offset_counter: &mut OffsetCounter,
constants_table: &mut ConstantsTable,
) -> Result<Function, Vec<Diagnostic>> {
// parse
let mut expression = parse_expression(input)?;
let parse_result = parse_expression(input);
let mut expression = parse_result?;
// init scopes, if necessary
let container_scope = prepare_scopes(symbol_table, fn_body_scope_id);
let mut symbol_table = SymbolTable::new();
let mut types_table = TypesTable::new();
// inner scopes
expression.init_scopes(symbol_table, container_scope);
// "fake" scopes
symbol_table.push_module_scope("__repl_module");
let function_scope_id = symbol_table.push_function_scope("__repl_function");
expression.init_scopes(&mut symbol_table, function_scope_id);
symbol_table.pop_scope(); // function
symbol_table.pop_scope(); // module
// names
let diagnostics = expression.check_static_fn_local_names(&symbol_table);
if !diagnostics.is_empty() {
return Err(diagnostics);
}
// type check
expression.type_check(&symbol_table, types_table)?;
expression.type_check(&symbol_table, &mut types_table)?;
// synthesize a function
// init ir_builder
let mut ir_builder = IrBuilder::new();
// copy all previous declared variables to here so we preserve their stack offsets
for (key, value) in fn_local_variables {
ir_builder
.local_variables_mut()
.insert(key.clone(), value.clone());
}
let entry_block_id = ir_builder.new_block();
let maybe_ir_expression =
@ -223,75 +138,6 @@ fn compile_expression(
entry_block.clone(),
);
// spilled registers are put on the stack, so we need to add it to our stack size
ir_function.assign_registers(register_count, offset_counter);
Ok(ir_function.assemble(offset_counter.get_count(), constants_table))
}
fn compile_let_statement(
input: &str,
register_count: usize,
symbol_table: &mut SymbolTable,
body_scope_id: &mut Option<usize>,
offset_counter: &mut OffsetCounter,
local_variables: &mut HashMap<Rc<VariableSymbol>, Rc<RefCell<IrVariable>>>,
types_table: &mut TypesTable,
constants_table: &mut ConstantsTable,
) -> Result<Function, Vec<Diagnostic>> {
// parse
let mut let_statement = parse_let_statement(input)?;
// names
let container_scope_id = prepare_scopes(symbol_table, body_scope_id);
let_statement.init_scopes(symbol_table, container_scope_id);
let name_diagnostics = let_statement.analyze_static_fn_local_names(symbol_table);
if !name_diagnostics.is_empty() {
return Err(name_diagnostics);
}
// types
let_statement.type_check(&symbol_table, types_table)?;
// init the ir builder
let mut ir_builder = IrBuilder::new();
// put previous locals in ir_builder so expressions can find them
for (k, v) in local_variables.iter() {
ir_builder
.local_variables_mut()
.insert(k.clone(), v.clone());
}
// ir function
let entry_block_id = ir_builder.new_block();
let destination_ir_variable = let_statement.to_repl_ir(
&mut ir_builder,
symbol_table,
types_table,
offset_counter.next() as isize,
); // put it on top
// Now that we've translated to ir, we can add the new local variable in the IrBuilder to our
// record of them to be used for next loop iteration.
let variable_symbol = let_statement.get_destination_symbol(symbol_table);
local_variables.insert(variable_symbol, destination_ir_variable);
ir_builder.finish_block();
let entry_block = ir_builder.get_block(entry_block_id);
let mut ir_function = IrFunction::new(
"__repl".into(),
vec![],
&TypeInfo::Void,
entry_block.clone(),
);
// By here, the variables should all be assigned to their new or existing stack slots.
// Register allocation should only be required for temp variables.
ir_function.assign_registers(register_count, offset_counter);
Ok(ir_function.assemble(offset_counter.get_count(), constants_table))
let (_, stack_size) = ir_function.assign_registers(register_count);
Ok(ir_function.assemble(stack_size, constants_table))
}

19
dm/src/run.rs
View File

@ -6,13 +6,13 @@ use dm_std_lib::add_all_std_core;
use dmc_lib::constants_table::ConstantsTable;
use dmc_lib::diagnostic::Diagnostic;
use dmc_lib::intrinsics::{insert_intrinsic_symbols, insert_intrinsic_types};
use dmc_lib::offset_counter::OffsetCounter;
use dmc_lib::parser::parse_compilation_unit;
use dmc_lib::symbol_table::SymbolTable;
use dmc_lib::types_table::TypesTable;
use dvm_lib::vm::constant::{Constant, StringConstant};
use dvm_lib::vm::function::Function;
use dvm_lib::vm::{DvmContext, call};
use dvm_lib::vm::operand::Operand;
use dvm_lib::vm::{CallStack, DvmContext, call};
use std::path::PathBuf;
use std::rc::Rc;
@ -67,9 +67,8 @@ fn run(
let mut constants_table = ConstantsTable::new();
for ir_function in &mut ir_functions {
let mut offset_counter = OffsetCounter::new();
ir_function.assign_registers(register_count, &mut offset_counter);
let function = ir_function.assemble(offset_counter.get_count(), &mut constants_table);
let (_, stack_size) = ir_function.assign_registers(register_count);
let function = ir_function.assemble(stack_size, &mut constants_table);
functions.push(function);
}
@ -108,8 +107,16 @@ fn run(
);
}
let result = call(&dvm_context, "main", &vec![], register_count);
let mut registers: Vec<Operand> = vec![Operand::Null; register_count];
let mut call_stack = CallStack::new();
let result = call(
&dvm_context,
&mut registers,
&mut call_stack,
"main",
&vec![],
);
if let Some(value) = result {
println!("{}", value);
}

2
dmc-lib/src/ast/compilation_unit.rs
View File

@ -8,7 +8,7 @@ use crate::ir::ir_class::IrClass;
use crate::ir::ir_function::IrFunction;
use crate::symbol_table::SymbolTable;
use crate::types_table::TypesTable;
use crate::{diagnostics_result, handle_diagnostics};
use crate::{diagnostics_result, handle_diagnostics, ok_or_err_diagnostics};
pub struct CompilationUnit {
functions: Vec<Function>,

70
dmc-lib/src/ast/let_statement.rs
View File

@ -10,7 +10,6 @@ use crate::symbol::Symbol;
use crate::symbol::class_symbol::ClassSymbol;
use crate::symbol::variable_symbol::VariableSymbol;
use crate::symbol_table::SymbolTable;
use crate::type_info::TypeInfo;
use crate::types_table::TypesTable;
use std::cell::RefCell;
use std::rc::Rc;
@ -56,10 +55,6 @@ impl LetStatement {
self.initializer.init_scopes(symbol_table, container_scope);
}
pub fn scope_id(&self) -> usize {
self.scope_id.unwrap()
}
fn make_and_insert_variable_symbol(
&self,
symbol_table: &mut SymbolTable,
@ -136,7 +131,6 @@ impl LetStatement {
types_table: &mut TypesTable,
) -> Result<(), Vec<Diagnostic>> {
self.initializer.type_check(symbol_table, types_table)?;
// TODO: this is wrong. We need to check assignability
let initializer_type_info = self
.initializer
.type_info(symbol_table, types_table)
@ -150,34 +144,6 @@ impl LetStatement {
Ok(())
}
fn make_vr_variable(&self, builder: &mut IrBuilder, destination_type: &TypeInfo) -> IrVariable {
IrVariable::new_vr(
self.declared_name().into(),
builder.current_block().id(),
destination_type,
)
}
fn make_stack_variable(
&self,
builder: &mut IrBuilder,
destination_type: &TypeInfo,
offset: isize,
) -> IrVariable {
IrVariable::new_stack_with_offset(
self.declared_name().into(),
builder.current_block().id(),
destination_type,
offset,
)
}
pub fn get_destination_symbol(&self, symbol_table: &SymbolTable) -> Rc<VariableSymbol> {
symbol_table
.get_variable_symbol_owned(self.scope_id.unwrap(), &self.declared_name)
.unwrap()
}
pub fn to_ir(
&self,
builder: &mut IrBuilder,
@ -188,14 +154,20 @@ impl LetStatement {
.initializer
.to_ir_operation(builder, symbol_table, types_table);
let destination_symbol = self.get_destination_symbol(symbol_table);
let destination_symbol = symbol_table
.get_variable_symbol_owned(self.scope_id.unwrap(), &self.declared_name)
.unwrap();
let destination_type = types_table
.variable_types()
.get(&destination_symbol)
.unwrap();
let destination_vr_variable = self.make_vr_variable(builder, destination_type);
let destination_vr_variable = IrVariable::new_vr(
self.declared_name().into(),
builder.current_block().id(),
destination_type,
);
let as_rc = Rc::new(RefCell::new(destination_vr_variable));
let ir_assign = IrAssign::new(as_rc.clone(), init_operation);
@ -208,30 +180,4 @@ impl LetStatement {
.current_block_mut()
.add_statement(IrStatement::Assign(ir_assign));
}
pub fn to_repl_ir(
&self,
builder: &mut IrBuilder,
symbol_table: &SymbolTable,
types_table: &TypesTable,
destination_stack_offset: isize,
) -> Rc<RefCell<IrVariable>> {
let init_operation = self
.initializer
.to_ir_operation(builder, symbol_table, types_table);
let destination_symbol = self.get_destination_symbol(symbol_table);
let destination_type = types_table
.variable_types()
.get(&destination_symbol)
.unwrap();
let destination_stack_variable =
self.make_stack_variable(builder, destination_type, destination_stack_offset);
let as_rc = Rc::new(RefCell::new(destination_stack_variable));
let ir_assign = IrAssign::new(as_rc.clone(), init_operation);
// do not need to save variable to builder as a new one is created for each repl function
builder
.current_block_mut()
.add_statement(IrStatement::Assign(ir_assign));
as_rc
}
}

2
dmc-lib/src/diagnostic.rs
View File

@ -1,5 +1,3 @@
pub type Diagnostics = Vec<Diagnostic>;
#[derive(Debug)]
pub struct Diagnostic {
message: String,

2
dmc-lib/src/error_codes.rs
View File

@ -1,7 +1,5 @@
pub type ErrorCode = usize;
pub const LEXER_ERROR: ErrorCode = 1;
pub const PARSE_ERROR: ErrorCode = 2;
pub const SYMBOL_NOT_FOUND: ErrorCode = 13;
pub const SYMBOL_ALREADY_DECLARED: ErrorCode = 14;
pub const BINARY_INCOMPATIBLE_TYPES: ErrorCode = 15;

5
dmc-lib/src/ir/ir_assign.rs
View File

@ -2,9 +2,8 @@ use crate::constants_table::ConstantsTable;
use crate::ir::assemble::{Assemble, InstructionsBuilder};
use crate::ir::ir_operation::IrOperation;
use crate::ir::ir_variable::{IrVariable, IrVariableDescriptor, IrVrVariableDescriptor};
use crate::ir::register_allocation::VrUser;
use crate::ir::register_allocation::{OffsetCounter, VrUser};
use crate::ir::util::propagate_spills;
use crate::offset_counter::OffsetCounter;
use dvm_lib::instruction::Instruction;
use std::cell::RefCell;
use std::collections::{HashMap, HashSet};
@ -58,7 +57,7 @@ impl VrUser for IrAssign {
fn propagate_stack_offsets(&mut self, counter: &mut OffsetCounter) {
let mut borrowed_destination = self.destination.borrow_mut();
if let IrVariableDescriptor::Stack(stack_variable) = borrowed_destination.descriptor_mut() {
stack_variable.use_next_offset(counter);
stack_variable.set_offset(counter.next());
}
}
}

6
dmc-lib/src/ir/ir_block.rs
View File

@ -2,8 +2,7 @@ use crate::constants_table::ConstantsTable;
use crate::ir::assemble::{Assemble, InstructionsBuilder};
use crate::ir::ir_statement::IrStatement;
use crate::ir::ir_variable::IrVrVariableDescriptor;
use crate::ir::register_allocation::{HasVrUsers, VrUser};
use crate::offset_counter::OffsetCounter;
use crate::ir::register_allocation::{HasVrUsers, OffsetCounter, VrUser};
use std::cell::RefCell;
use std::collections::{HashMap, HashSet};
use std::fmt::{Display, Formatter};
@ -123,7 +122,6 @@ impl Display for IrBlock {
#[cfg(test)]
mod tests {
use crate::diagnostic::Diagnostic;
use crate::offset_counter::OffsetCounter;
use crate::parser::parse_compilation_unit;
use crate::symbol_table::SymbolTable;
use crate::types_table::TypesTable;
@ -157,7 +155,7 @@ mod tests {
.unwrap();
let mut main_ir = main.to_ir(&symbol_table, &types_table, None);
let register_assignments = main_ir.assign_registers(2, &mut OffsetCounter::new());
let (register_assignments, _) = main_ir.assign_registers(2);
assert_eq!(register_assignments.len(), 4);
Ok(())

10
dmc-lib/src/ir/ir_function.rs
View File

@ -4,7 +4,6 @@ use crate::ir::ir_block::IrBlock;
use crate::ir::ir_parameter::IrParameter;
use crate::ir::ir_variable::IrVrVariableDescriptor;
use crate::ir::register_allocation::HasVrUsers;
use crate::offset_counter::OffsetCounter;
use crate::type_info::TypeInfo;
use dvm_lib::vm::function::Function;
use std::cell::RefCell;
@ -37,14 +36,11 @@ impl IrFunction {
pub fn assign_registers(
&mut self,
register_count: usize,
offset_counter: &mut OffsetCounter,
) -> HashMap<IrVrVariableDescriptor, usize> {
self.entry
.borrow_mut()
.assign_registers(register_count, offset_counter)
) -> (HashMap<IrVrVariableDescriptor, usize>, isize) {
self.entry.borrow_mut().assign_registers(register_count)
}
pub fn assemble(&self, stack_size: usize, constants_table: &mut ConstantsTable) -> Function {
pub fn assemble(&self, stack_size: isize, constants_table: &mut ConstantsTable) -> Function {
let mut builder = InstructionsBuilder::new();
self.entry.borrow().assemble(&mut builder, constants_table);
let instructions = builder.take_instructions();

3
dmc-lib/src/ir/ir_statement.rs
View File

@ -5,8 +5,7 @@ use crate::ir::ir_call::IrCall;
use crate::ir::ir_return::IrReturn;
use crate::ir::ir_set_field::IrSetField;
use crate::ir::ir_variable::IrVrVariableDescriptor;
use crate::ir::register_allocation::VrUser;
use crate::offset_counter::OffsetCounter;
use crate::ir::register_allocation::{OffsetCounter, VrUser};
use std::collections::{HashMap, HashSet};
use std::fmt::{Display, Formatter};

18
dmc-lib/src/ir/ir_variable.rs
View File

@ -1,4 +1,3 @@
use crate::offset_counter::OffsetCounter;
use crate::type_info::TypeInfo;
use dvm_lib::instruction::Location;
use std::collections::HashSet;
@ -21,16 +20,9 @@ impl IrVariable {
}
}
pub fn new_stack_with_offset(
name: Rc<str>,
block_id: usize,
type_info: &TypeInfo,
offset: isize,
) -> Self {
let mut descriptor = IrStackVariableDescriptor::new(name, block_id);
descriptor.set_offset(offset);
pub fn new_stack(name: Rc<str>, block_id: usize, type_info: TypeInfo) -> Self {
Self {
descriptor: IrVariableDescriptor::Stack(descriptor),
descriptor: IrVariableDescriptor::Stack(IrStackVariableDescriptor::new(name, block_id)),
type_info: type_info.clone(),
}
}
@ -185,12 +177,6 @@ impl IrStackVariableDescriptor {
self.name.clone()
}
pub fn use_next_offset(&mut self, offset_counter: &mut OffsetCounter) {
if self.offset.is_none() {
self.offset = Some(offset_counter.next() as isize);
}
}
pub fn set_offset(&mut self, offset: isize) {
self.offset = Some(offset);
}

29
dmc-lib/src/ir/register_allocation.rs
View File

@ -1,5 +1,4 @@
use crate::ir::ir_variable::IrVrVariableDescriptor;
use crate::offset_counter::OffsetCounter;
use std::collections::{HashMap, HashSet};
pub type InterferenceGraph = HashMap<IrVrVariableDescriptor, HashSet<IrVrVariableDescriptor>>;
@ -105,8 +104,7 @@ pub trait HasVrUsers {
fn assign_registers(
&mut self,
register_count: usize,
offset_counter: &mut OffsetCounter,
) -> HashMap<IrVrVariableDescriptor, usize> {
) -> (HashMap<IrVrVariableDescriptor, usize>, isize) {
let mut spills: HashSet<IrVrVariableDescriptor> = HashSet::new();
loop {
let mut interference_graph = self.interference_graph();
@ -125,11 +123,12 @@ pub trait HasVrUsers {
vr_user.propagate_register_assignments(&registers);
}
// also set offsets
let mut offset_counter = OffsetCounter::new();
for vr_user in self.vr_users_mut() {
vr_user.propagate_stack_offsets(offset_counter);
vr_user.propagate_stack_offsets(&mut offset_counter);
}
return registers;
return (registers, offset_counter.get_count());
}
}
}
@ -158,6 +157,26 @@ pub trait VrUser {
}
}
pub struct OffsetCounter {
counter: isize,
}
impl OffsetCounter {
pub fn new() -> Self {
Self { counter: 0 }
}
pub fn next(&mut self) -> isize {
let offset = self.counter;
self.counter += 1;
offset
}
pub fn get_count(&self) -> isize {
self.counter
}
}
#[derive(Debug)]
struct WorkItem {
vr: IrVrVariableDescriptor,

241
dmc-lib/src/lexer.rs
View File

@ -34,133 +34,134 @@ impl<'a> Lexer<'a> {
}
}
let mut chars = chunk.chars();
let current = chars.next().unwrap(); // safe because we return None if chunk is empty above
let peek = chars.next();
let (end, kind) = match current {
'-' => {
if let Some(peek) = peek {
if peek == '>' {
(self.position + 2, TokenKind::RightArrow)
let token = if chunk.starts_with("->") {
Token::new(self.position, self.position + 2, TokenKind::RightArrow)
} else if chunk.starts_with("-") {
Token::new(self.position, self.position + 1, TokenKind::Minus)
} else if chunk.starts_with("(") {
Token::new(self.position, self.position + 1, TokenKind::LeftParentheses)
} else if chunk.starts_with(")") {
Token::new(
self.position,
self.position + 1,
TokenKind::RightParentheses,
)
} else if chunk.starts_with("*") {
Token::new(self.position, self.position + 1, TokenKind::Star)
} else if chunk.starts_with("/") {
Token::new(self.position, self.position + 1, TokenKind::Slash)
} else if chunk.starts_with("%") {
Token::new(self.position, self.position + 1, TokenKind::Modulo)
} else if chunk.starts_with("+") {
Token::new(self.position, self.position + 1, TokenKind::Plus)
} else if chunk.starts_with("&") {
Token::new(self.position, self.position + 1, TokenKind::Ampersand)
} else if chunk.starts_with("^") {
Token::new(self.position, self.position + 1, TokenKind::Caret)
} else if chunk.starts_with("|") {
Token::new(self.position, self.position + 1, TokenKind::Bar)
} else if chunk.starts_with("=") {
Token::new(self.position, self.position + 1, TokenKind::Equals)
} else if chunk.starts_with(",") {
Token::new(self.position, self.position + 1, TokenKind::Comma)
} else if chunk.starts_with(":") {
Token::new(self.position, self.position + 1, TokenKind::Colon)
} else if chunk.starts_with(".") {
Token::new(self.position, self.position + 1, TokenKind::Dot)
} else if chunk.starts_with("[") {
Token::new(self.position, self.position + 1, TokenKind::LeftSquare)
} else if chunk.starts_with("]") {
Token::new(self.position, self.position + 1, TokenKind::RightSquare)
} else if chunk.starts_with("<") {
Token::new(self.position, self.position + 1, TokenKind::Lt)
} else if chunk.starts_with(">") {
Token::new(self.position, self.position + 1, TokenKind::Gt)
} else {
// more than one char token
if chunk.starts_with(|c: char| c.is_ascii_digit()) {
// number literal
let mut end = self.position;
let mut whole_chars = chunk.chars();
while let Some(c) = whole_chars.next() {
if c.is_ascii_digit() {
end += 1;
} else {
(self.position + 1, TokenKind::Minus)
break;
}
} else {
(self.position + 1, TokenKind::Minus)
}
}
'(' => (self.position + 1, TokenKind::LeftParentheses),
')' => (self.position + 1, TokenKind::RightParentheses),
'[' => (self.position + 1, TokenKind::LeftSquare),
']' => (self.position + 1, TokenKind::RightSquare),
'<' => (self.position + 1, TokenKind::Lt),
'>' => (self.position + 1, TokenKind::Gt),
'*' => (self.position + 1, TokenKind::Star),
'/' => (self.position + 1, TokenKind::Slash),
'%' => (self.position + 1, TokenKind::Modulo),
'+' => (self.position + 1, TokenKind::Plus),
'&' => (self.position + 1, TokenKind::Ampersand),
'^' => (self.position + 1, TokenKind::Caret),
'|' => (self.position + 1, TokenKind::Bar),
'=' => (self.position + 1, TokenKind::Equals),
',' => (self.position + 1, TokenKind::Comma),
'.' => (self.position + 1, TokenKind::Dot),
':' => (self.position + 1, TokenKind::Colon),
_ => {
// more than one char token
if chunk.starts_with(|c: char| c.is_ascii_digit()) {
// number literal
let mut end = self.position;
let mut whole_chars = chunk.chars();
while let Some(c) = whole_chars.next() {
let mut fraction_chars = chunk.chars().skip(end - self.position);
if fraction_chars.next().map(|c| c == '.').unwrap_or(false) {
let mut found_fraction = false;
while let Some(c) = fraction_chars.next() {
if c.is_ascii_digit() {
end += 1;
} else {
break;
}
}
let mut fraction_chars = chunk.chars().skip(end - self.position);
if fraction_chars.next().map(|c| c == '.').unwrap_or(false) {
let mut found_fraction = false;
while let Some(c) = fraction_chars.next() {
if c.is_ascii_digit() {
end += 1;
if !found_fraction {
end += 1; // to account for decimal point
found_fraction = true;
}
} else {
break;
if !found_fraction {
end += 1; // to account for decimal point
found_fraction = true;
}
}
if found_fraction {
(end, TokenKind::DoubleLiteral)
} else {
(end, TokenKind::IntegerLiteral)
break;
}
}
if found_fraction {
Token::new(self.position, end, TokenKind::DoubleLiteral)
} else {
(end, TokenKind::IntegerLiteral)
Token::new(self.position, end, TokenKind::IntegerLiteral)
}
} else if chunk.starts_with("\"") {
// string literal
let mut end = self.position;
let mut terminated = false;
let mut chars = chunk.chars();
chars.next(); // skip opening quote
end += 1;
for char in chars {
end += 1;
if char == '"' {
terminated = true;
break;
}
}
if !terminated {
return Some(Err(LexerError::new(
self.position,
end,
LexerErrorKind::UnterminatedString,
)));
}
(end, TokenKind::String)
} else {
// keyword or identifier
let mut prefix = String::new();
for char in chunk.chars() {
if char.is_alphanumeric() || char == '_' {
prefix.push(char);
} else {
break;
}
}
if prefix.len() == 0 {
return Some(Err(LexerError::new(
self.position,
self.position + 1,
LexerErrorKind::UnrecognizedCharacter(chunk.chars().next().unwrap()),
)));
}
let token_kind = match prefix.as_str() {
"fn" => TokenKind::Fn,
"end" => TokenKind::End,
"let" => TokenKind::Let,
"extern" => TokenKind::Extern,
"class" => TokenKind::Class,
"self" => TokenKind::SelfKw,
"pub" => TokenKind::Public,
"mut" => TokenKind::Mut,
"ctor" => TokenKind::Ctor,
_ => TokenKind::Identifier,
};
(self.position + prefix.len(), token_kind)
Token::new(self.position, end, TokenKind::IntegerLiteral)
}
} else if chunk.starts_with("\"") {
// string literal
let mut end = self.position;
let mut terminated = false;
let mut chars = chunk.chars();
chars.next(); // skip opening quote
end += 1;
for char in chars {
end += 1;
if char == '"' {
terminated = true;
break;
}
}
if !terminated {
return Some(Err(LexerError::new(LexerErrorKind::UnterminatedString)));
}
Token::new(self.position, end, TokenKind::String)
} else {
// keyword or identifier
let mut prefix = String::new();
for char in chunk.chars() {
if char.is_alphanumeric() || char == '_' {
prefix.push(char);
} else {
break;
}
}
if prefix.len() == 0 {
return Some(Err(LexerError::new(LexerErrorKind::UnrecognizedCharacter(
chunk.chars().next().unwrap(),
))));
}
let token_kind = match prefix.as_str() {
"fn" => TokenKind::Fn,
"end" => TokenKind::End,
"let" => TokenKind::Let,
"extern" => TokenKind::Extern,
"class" => TokenKind::Class,
"self" => TokenKind::SelfKw,
"pub" => TokenKind::Public,
"mut" => TokenKind::Mut,
"ctor" => TokenKind::Ctor,
_ => TokenKind::Identifier,
};
Token::new(self.position, self.position + prefix.len(), token_kind)
}
};
let token = Token::new(self.position, end, kind);
self.position += token.end() - token.start();
Some(Ok(token))
}
@ -168,22 +169,12 @@ impl<'a> Lexer<'a> {
#[derive(Debug, Eq, PartialEq)]
pub struct LexerError {
start: usize,
end: usize,
kind: LexerErrorKind,
}
impl LexerError {
pub fn new(start: usize, end: usize, kind: LexerErrorKind) -> Self {
Self { start, end, kind }
}
pub fn start(&self) -> usize {
self.start
}
pub fn end(&self) -> usize {
self.end
pub fn new(kind: LexerErrorKind) -> Self {
Self { kind }
}
pub fn kind(&self) -> LexerErrorKind {

1
dmc-lib/src/lib.rs
View File

@ -5,7 +5,6 @@ pub mod error_codes;
pub mod intrinsics;
pub mod ir;
pub mod lexer;
pub mod offset_counter;
pub mod parser;
pub mod scope;
pub mod source_range;

23
dmc-lib/src/offset_counter.rs
View File

@ -1,23 +0,0 @@
pub struct OffsetCounter {
counter: usize,
}
impl OffsetCounter {
pub fn new() -> Self {
Self { counter: 0 }
}
pub fn with_base(base: usize) -> Self {
Self { counter: base }
}
pub fn next(&mut self) -> usize {
let offset = self.counter;
self.counter += 1;
offset
}
pub fn get_count(&self) -> usize {
self.counter
}
}

357
dmc-lib/src/parser.rs
View File

@ -19,34 +19,23 @@ use crate::ast::parameter::Parameter;
use crate::ast::statement::Statement;
use crate::ast::string_literal::StringLiteral;
use crate::ast::type_use::TypeUse;
use crate::diagnostic::{Diagnostic, Diagnostics};
use crate::error_codes::{LEXER_ERROR, PARSE_ERROR};
use crate::lexer::{Lexer, LexerErrorKind};
use crate::diagnostic::Diagnostic;
use crate::lexer::Lexer;
use crate::source_range::SourceRange;
use crate::token::{Token, TokenKind};
use crate::{handle_diagnostics, ok_or_err_diagnostics};
use std::str::FromStr;
pub type ParseResult<T> = Result<T, Diagnostics>;
pub fn parse_compilation_unit(input: &str) -> ParseResult<CompilationUnit> {
pub fn parse_compilation_unit(input: &str) -> Result<CompilationUnit, Vec<Diagnostic>> {
let mut parser = Parser::new(input);
parser.advance()?; // get started
parser.compilation_unit()
}
pub fn parse_expression(input: &str) -> ParseResult<Expression> {
pub fn parse_expression(input: &str) -> Result<Expression, Vec<Diagnostic>> {
let mut parser = Parser::new(input);
parser.advance()?; // get started
parser.advance(); // get started
parser.expression()
}
pub fn parse_let_statement(input: &str) -> ParseResult<LetStatement> {
let mut parser = Parser::new(input);
parser.advance()?;
parser.let_statement()
}
macro_rules! matches_expression_first {
( $token_kind : expr ) => {
matches!(
@ -107,116 +96,96 @@ impl<'a> Parser<'a> {
}
}
fn advance(&mut self) -> Result<(), Diagnostics> {
fn fetch(lexer: &mut Lexer) -> Result<Option<Token>, Diagnostics> {
let mut diagnostics = vec![];
let mut maybe_token: Option<Token> = None;
while let Some(lexer_result) = lexer.next() {
match lexer_result {
Ok(token) => {
maybe_token = Some(token);
break;
}
Err(lexer_error) => {
let diagnostic = match lexer_error.kind() {
LexerErrorKind::UnterminatedString => Diagnostic::new(
"Unterminated string literal.",
lexer_error.start(),
lexer_error.end(),
)
.with_error_code(LEXER_ERROR),
LexerErrorKind::UnrecognizedCharacter(c) => Diagnostic::new(
&format!("Unrecognized character: {}", c),
lexer_error.start(),
lexer_error.end(),
)
.with_error_code(LEXER_ERROR),
};
diagnostics.push(diagnostic);
}
}
}
ok_or_err_diagnostics!(maybe_token, diagnostics)
}
fn advance(&mut self) {
if self.lookahead.is_some() {
// we've advanced at least once
self.current = self.lookahead.take();
self.lookahead = fetch(&mut self.lexer)?;
Ok(())
self.lookahead = match self.lexer.next() {
None => None,
Some(result) => match result {
Ok(token) => Some(token),
Err(lexer_error) => {
panic!("{:?}", lexer_error);
}
},
}
} else if self.lookahead.is_none() && self.current.is_some() {
// we're on the last token
self.current = None;
Ok(())
} else {
// we've not yet advanced, so fetch both
self.current = fetch(&mut self.lexer)?;
self.lookahead = fetch(&mut self.lexer)?;
Ok(())
// current
match self.lexer.next() {
None => {}
Some(result) => match result {
Ok(token) => {
self.current = Some(token);
}
Err(lexer_error) => {
panic!("{:?}", lexer_error);
}
},
}
// lookahead
match self.lexer.next() {
None => {}
Some(result) => match result {
Ok(token) => {
self.lookahead = Some(token);
}
Err(lexer_error) => {
panic!("{:?}", lexer_error);
}
},
}
}
}
fn join_kinds(kinds: &[TokenKind]) -> String {
kinds
.iter()
.map(|kind| format!("{:?}", kind))
.collect::<Vec<_>>()
.join(", ")
}
fn get_expected_but_found(kinds: &[TokenKind], found: &Token) -> Diagnostic {
Diagnostic::new(
&format!("Unexpected token: {:?}.", found.kind()),
found.start(),
found.end(),
)
.with_error_code(PARSE_ERROR)
.with_primary_label_message(&format!("Expected {}.", Self::join_kinds(kinds)))
}
fn get_expected_but_found_eoi(kinds: &[TokenKind], position: usize) -> Diagnostic {
Diagnostic::new("Unexpected end-of-input.", position, position)
.with_error_code(PARSE_ERROR)
.with_primary_label_message(&format!("Expected {}.", Self::join_kinds(kinds)))
}
fn expect_advance(&mut self, token_kind: TokenKind) -> Result<Token, Diagnostics> {
fn expect_advance(&mut self, token_kind: TokenKind) -> Result<Token, Vec<Diagnostic>> {
match self.current.take() {
None => Err(vec![Self::get_expected_but_found_eoi(
&[token_kind],
self.input.len(),
)]),
None => Err(vec![
Diagnostic::new(
&format!("Expected {:?} but found end-of-input.", token_kind),
self.input.len(),
self.input.len(),
)
.with_reporter(file!(), line!()),
]),
Some(token) => {
if token.kind() == token_kind {
self.advance()?;
self.advance();
Ok(token)
} else {
self.advance_until(&[token_kind]);
Err(vec![Self::get_expected_but_found(&[token_kind], &token)])
Err(vec![
Diagnostic::new(
&format!("Expected {:?} but found {:?}", token_kind, token.kind()),
token.start(),
token.end(),
)
.with_reporter(file!(), line!()),
])
}
}
}
}
fn expect_immediately_after_advance(
fn expect_position_advance(
&mut self,
token_kind: TokenKind,
previous_token: &Token,
) -> Result<Token, Diagnostics> {
start_position: usize,
) -> Result<Token, Vec<Diagnostic>> {
let matched = self.expect_advance(token_kind)?;
if matched.start() == previous_token.end() {
if matched.start() == start_position {
Ok(matched)
} else {
Err(vec![
Diagnostic::new(
&format!(
"Expected {:?} immediately after previous token.",
token_kind
),
&format!("Expected {:?} but found {:?}", token_kind, matched.kind()),
matched.start(),
matched.end(),
)
.with_error_code(PARSE_ERROR),
.with_reporter(file!(), line!()),
])
}
}
@ -237,6 +206,13 @@ impl<'a> Parser<'a> {
}
}
fn peek_lookahead(&self, token_kind: TokenKind) -> bool {
match &self.lookahead {
None => panic!("Unexpected end of input."),
Some(token) => token.kind() == token_kind,
}
}
fn sample_input(&self, start: usize, end: usize) -> &'a str {
&self.input[start..end]
}
@ -245,40 +221,49 @@ impl<'a> Parser<'a> {
self.sample_input(token.start(), token.end())
}
fn compilation_unit(&mut self) -> Result<CompilationUnit, Vec<Diagnostic>> {
pub fn compilation_unit(&mut self) -> Result<CompilationUnit, Vec<Diagnostic>> {
let mut functions: Vec<Function> = vec![];
let mut extern_functions: Vec<ExternFunction> = vec![];
let mut classes: Vec<Class> = vec![];
let mut diagnostics = vec![];
self.advance(); // get started
while self.current.is_some() {
let current = self.get_current();
match current.kind() {
TokenKind::Fn | TokenKind::Extern | TokenKind::Class => {
handle_diagnostics!(
self.module_level_declaration(
&mut functions,
&mut extern_functions,
&mut classes
),
diagnostics
);
match self.module_level_declaration(
&mut functions,
&mut extern_functions,
&mut classes,
) {
Ok(_) => {}
Err(mut declaration_diagnostics) => {
diagnostics.append(&mut declaration_diagnostics)
}
}
}
_ => {
diagnostics.push(Self::get_expected_but_found(
&[TokenKind::Fn, TokenKind::Extern, TokenKind::Class],
current,
diagnostics.push(Diagnostic::new(
&format!(
"Expected any of {:?}; found {:?}",
[TokenKind::Fn, TokenKind::Extern, TokenKind::Class],
current.kind()
),
current.start(),
current.end(),
));
self.advance_until(&[TokenKind::Fn, TokenKind::Extern, TokenKind::Class]);
self.advance_until(&[TokenKind::Fn, TokenKind::Extern]);
}
}
}
ok_or_err_diagnostics!(
CompilationUnit::new(functions, extern_functions, classes),
diagnostics
)
if diagnostics.is_empty() {
Ok(CompilationUnit::new(functions, extern_functions, classes))
} else {
Err(diagnostics)
}
}
fn module_level_declaration(
@ -322,7 +307,7 @@ impl<'a> Parser<'a> {
fn function(&mut self) -> Result<Function, Vec<Diagnostic>> {
let is_public = if self.current.is_some() && self.peek_current(TokenKind::Public) {
self.advance()?; // pub
self.advance(); // pub
true
} else {
false
@ -371,17 +356,18 @@ impl<'a> Parser<'a> {
_ => {}
}
ok_or_err_diagnostics!(
Function::new(
if diagnostics.is_empty() {
Ok(Function::new(
self.token_text(&identifier_token),
SourceRange::new(identifier_token.start(), identifier_token.end()),
is_public,
parameters,
return_type,
statements,
),
diagnostics
)
))
} else {
Err(diagnostics)
}
}
fn extern_function(&mut self) -> Result<ExternFunction, Vec<Diagnostic>> {
@ -413,15 +399,16 @@ impl<'a> Parser<'a> {
let return_type = self.return_type()?;
ok_or_err_diagnostics!(
ExternFunction::new(
if diagnostics.is_empty() {
Ok(ExternFunction::new(
self.token_text(&identifier_token),
SourceRange::new(identifier_token.start(), identifier_token.end()),
maybe_parameters.unwrap(),
return_type,
),
diagnostics
)
))
} else {
Err(diagnostics)
}
}
fn class(&mut self) -> Result<Class, Vec<Diagnostic>> {
@ -472,17 +459,18 @@ impl<'a> Parser<'a> {
self.expect_advance(TokenKind::End)?;
ok_or_err_diagnostics!(
Class::new(
if diagnostics.is_empty() {
Ok(Class::new(
self.token_text(&identifier_token),
SourceRange::new(identifier_token.start(), identifier_token.end()),
generic_parameters,
maybe_constructor,
fields,
functions,
),
diagnostics
)
))
} else {
Err(diagnostics)
}
}
fn parameter_list(&mut self) -> Result<Vec<Parameter>, Vec<Diagnostic>> {
@ -499,15 +487,14 @@ impl<'a> Parser<'a> {
}
}
if self.current.is_some() && self.peek_current(TokenKind::Comma) {
match self.advance() {
Ok(_) => {}
Err(mut ds) => {
diagnostics.append(&mut ds);
}
};
self.advance();
}
}
ok_or_err_diagnostics!(parameters, diagnostics)
if diagnostics.is_empty() {
Ok(parameters)
} else {
Err(diagnostics)
}
}
fn parameter(&mut self) -> Result<Parameter, Vec<Diagnostic>> {
@ -531,7 +518,7 @@ impl<'a> Parser<'a> {
let current = self.get_current();
return match current.kind() {
TokenKind::LeftSquare => {
self.advance()?; // [
self.advance(); // [
let inner_type_use = self.type_use()?;
self.expect_advance(TokenKind::RightSquare)?;
todo!()
@ -540,7 +527,7 @@ impl<'a> Parser<'a> {
let identifier_token = self.expect_advance(TokenKind::Identifier)?;
let generic_arguments =
if self.current.is_some() && self.peek_current(TokenKind::Lt) {
self.advance()?; // <
self.advance(); // <
let generic_arguments = self.generic_arguments_list()?;
self.expect_advance(TokenKind::Gt)?; // >
generic_arguments
@ -553,14 +540,19 @@ impl<'a> Parser<'a> {
generic_arguments,
))
}
_ => Err(vec![Self::get_expected_but_found(
&[TokenKind::LeftSquare, TokenKind::Identifier],
current,
_ => Err(vec![Diagnostic::new(
&format!(
"Expected LeftSquare or Identifier; found: {:?}",
current.kind()
),
current.start(),
current.end(),
)]),
};
}
Err(vec![Self::get_expected_but_found_eoi(
&[TokenKind::LeftSquare, TokenKind::Identifier],
Err(vec![Diagnostic::new(
"Expected LeftSquare or Identifier; found end of input.",
self.input.len(),
self.input.len(),
)])
}
@ -570,7 +562,7 @@ impl<'a> Parser<'a> {
while self.current.is_some() && matches_type_use_first!(self.get_current().kind()) {
generic_arguments.push(self.type_use()?);
if self.current.is_some() && self.peek_current(TokenKind::Comma) {
self.advance()?; // comma
self.advance(); // comma
} else {
break;
}
@ -584,7 +576,7 @@ impl<'a> Parser<'a> {
while self.current.is_some() && self.peek_current(TokenKind::Identifier) {
parameters.push(self.generic_parameter()?);
if self.current.is_some() && self.peek_current(TokenKind::Plus) {
self.advance()?; // +
self.advance(); // +
} else {
break;
}
@ -597,11 +589,11 @@ impl<'a> Parser<'a> {
let identifier = self.expect_advance(TokenKind::Identifier)?;
let mut extends_list: Vec<TypeUse> = vec![];
if self.current.is_some() && self.peek_current(TokenKind::Colon) {
self.advance()?; // :
self.advance(); // :
while self.current.is_some() && matches_type_use_first!(self.get_current().kind()) {
extends_list.push(self.type_use()?);
if self.current.is_some() && self.peek_current(TokenKind::Comma) {
self.advance()?; // ,
self.advance(); // ,
} else {
break;
}
@ -631,25 +623,34 @@ impl<'a> Parser<'a> {
}
_ => {
let lookahead = self.lookahead.as_ref().unwrap();
return Err(vec![Self::get_expected_but_found(
&[TokenKind::Mut, TokenKind::Identifier, TokenKind::Fn],
lookahead,
return Err(vec![Diagnostic::new(
&format!(
"Expected any of {:?}; found {:?}",
[TokenKind::Mut, TokenKind::Identifier, TokenKind::Fn],
lookahead.kind()
),
lookahead.start(),
lookahead.end(),
)]);
}
}
Ok(())
} else {
let current = self.current.as_ref().unwrap();
Err(vec![Self::get_expected_but_found(
&[TokenKind::Mut, TokenKind::Identifier, TokenKind::Fn],
current,
Err(vec![Diagnostic::new(
&format!(
"Expected any of {:?}; found end-of-input.",
[TokenKind::Mut, TokenKind::Identifier, TokenKind::Fn]
),
current.end(),
current.end(),
)])
}
}
fn constructor(&mut self) -> Result<Constructor, Vec<Diagnostic>> {
let is_public = if self.current.is_some() && self.peek_current(TokenKind::Public) {
self.advance()?;
self.advance();
true
} else {
false
@ -699,14 +700,14 @@ impl<'a> Parser<'a> {
fn field(&mut self) -> Result<Field, Vec<Diagnostic>> {
let is_public = if self.current.is_some() && self.peek_current(TokenKind::Public) {
self.advance()?;
self.advance();
true
} else {
false
};
let is_mut = if self.current.is_some() && self.peek_current(TokenKind::Mut) {
self.advance()?;
self.advance();
true
} else {
false
@ -715,14 +716,14 @@ impl<'a> Parser<'a> {
let identifier = self.expect_advance(TokenKind::Identifier)?;
let declared_type = if self.current.is_some() && self.peek_current(TokenKind::Colon) {
self.advance()?; // colon
self.advance(); // colon
Some(self.type_use()?)
} else {
None
};
let initializer = if self.current.is_some() && self.peek_current(TokenKind::Equals) {
self.advance()?; // equals
self.advance(); // equals
Some(self.expression()?)
} else {
None
@ -750,7 +751,7 @@ impl<'a> Parser<'a> {
self.expect_advance(TokenKind::Let)?;
let is_mut = if self.current.is_some() && self.peek_current(TokenKind::Mut) {
self.advance()?;
self.advance();
true
} else {
false
@ -789,7 +790,7 @@ impl<'a> Parser<'a> {
fn bitwise_or_expression(&mut self) -> Result<Expression, Vec<Diagnostic>> {
let mut result = self.bitwise_xor_expression()?;
while self.current.is_some() && self.peek_current(TokenKind::Bar) {
self.advance()?; // |
self.advance(); // |
let rhs = self.bitwise_xor_expression()?;
let source_range =
SourceRange::new(result.source_range().start(), rhs.source_range().end());
@ -806,7 +807,7 @@ impl<'a> Parser<'a> {
fn bitwise_xor_expression(&mut self) -> Result<Expression, Vec<Diagnostic>> {
let mut result = self.bitwise_and_expression()?;
while self.current.is_some() && self.peek_current(TokenKind::Caret) {
self.advance()?; // ^
self.advance(); // ^
let rhs = self.bitwise_and_expression()?;
let source_range =
SourceRange::new(result.source_range().start(), rhs.source_range().end());
@ -823,7 +824,7 @@ impl<'a> Parser<'a> {
fn bitwise_and_expression(&mut self) -> Result<Expression, Vec<Diagnostic>> {
let mut result = self.shift_expression()?;
while self.current.is_some() && self.peek_current(TokenKind::Ampersand) {
self.advance()?; // &
self.advance(); // &
let rhs = self.shift_expression()?;
let source_range =
SourceRange::new(result.source_range().start(), rhs.source_range().end());
@ -843,9 +844,9 @@ impl<'a> Parser<'a> {
let current = self.get_current();
match current.kind() {
TokenKind::Lt => {
let previous_cloned = current.clone();
self.advance()?; // first <
self.expect_immediately_after_advance(TokenKind::Lt, &previous_cloned)?; // second <
let second_lt_start = current.start() + 1;
self.advance(); // first <
self.expect_position_advance(TokenKind::Lt, second_lt_start)?; // second <
let rhs = self.additive_expression()?;
let source_range =
SourceRange::new(result.source_range().start(), rhs.source_range().end());
@ -857,9 +858,9 @@ impl<'a> Parser<'a> {
));
}
TokenKind::Gt => {
let previous_cloned = current.clone();
self.advance()?; // first >
self.expect_immediately_after_advance(TokenKind::Gt, &previous_cloned)?; // second gt
let second_gt_start = current.start() + 1;
self.advance(); // first >
self.expect_position_advance(TokenKind::Gt, second_gt_start)?; // second gt
let rhs = self.additive_expression()?;
let source_range =
SourceRange::new(result.source_range().start(), rhs.source_range().end());
@ -882,7 +883,7 @@ impl<'a> Parser<'a> {
let current = self.get_current();
match current.kind() {
TokenKind::Plus => {
self.advance()?; // plus
self.advance(); // plus
let rhs = self.multiplicative_expression()?;
let source_range =
SourceRange::new(result.source_range().start(), rhs.source_range().end());
@ -894,7 +895,7 @@ impl<'a> Parser<'a> {
));
}
TokenKind::Minus => {
self.advance()?; // minus
self.advance(); // minus
let rhs = self.multiplicative_expression()?;
let source_range =
SourceRange::new(result.source_range().start(), rhs.source_range().end());
@ -917,7 +918,7 @@ impl<'a> Parser<'a> {
let current = self.get_current();
match current.kind() {
TokenKind::Star => {
self.advance()?; // multiply
self.advance(); // multiply
let rhs = self.prefix_expression()?;
let source_range =
SourceRange::new(result.source_range().start(), rhs.source_range().end());
@ -929,7 +930,7 @@ impl<'a> Parser<'a> {
));
}
TokenKind::Slash => {
self.advance()?; // slash
self.advance(); // slash
let rhs = self.prefix_expression()?;
let source_range =
SourceRange::new(result.source_range().start(), rhs.source_range().end());
@ -941,7 +942,7 @@ impl<'a> Parser<'a> {
))
}
TokenKind::Modulo => {
self.advance()?; // modulo
self.advance(); // modulo
let rhs = self.prefix_expression()?;
let source_range =
SourceRange::new(result.source_range().start(), rhs.source_range().end());
@ -966,7 +967,7 @@ impl<'a> Parser<'a> {
match current.kind() {
TokenKind::Minus => {
operator_tokens.push(current.clone()); // unfortunately necessary
self.advance()?;
self.advance();
}
_ => break,
}
@ -1009,7 +1010,7 @@ impl<'a> Parser<'a> {
TokenKind::IntegerLiteral => {
let raw = self.token_text(&current);
let source_range = SourceRange::new(current.start(), current.end());
self.advance()?;
self.advance();
Ok(Expression::Integer(IntegerLiteral::new(
i32::from_str(raw).unwrap(),
source_range,
@ -1018,7 +1019,7 @@ impl<'a> Parser<'a> {
TokenKind::DoubleLiteral => {
let raw = self.token_text(&current);
let source_range = SourceRange::new(current.start(), current.end());
self.advance()?;
self.advance();
Ok(Expression::Double(DoubleLiteral::new(
f64::from_str(raw).unwrap(),
source_range,
@ -1027,7 +1028,7 @@ impl<'a> Parser<'a> {
TokenKind::String => {
let with_quotes = self.token_text(&current);
let source_range = SourceRange::new(current.start(), current.end());
self.advance()?;
self.advance();
Ok(Expression::String(StringLiteral::new(
&with_quotes[1..with_quotes.len() - 1],
source_range,
@ -1036,7 +1037,7 @@ impl<'a> Parser<'a> {
TokenKind::Identifier => {
let declared_name = self.token_text(&current);
let source_range = SourceRange::new(current.start(), current.end());
self.advance()?;
self.advance();
Ok(Expression::Identifier(Identifier::new(
declared_name,
source_range,
@ -1064,7 +1065,7 @@ impl<'a> Parser<'a> {
let mut expressions = vec![];
expressions.push(self.expression()?);
while self.current.is_some() && self.peek_current(TokenKind::Comma) {
self.advance()?; // comma
self.advance(); // comma
expressions.push(self.expression()?);
}
Ok(expressions)

4
dmc-lib/src/symbol_table/mod.rs
View File

@ -89,10 +89,6 @@ impl SymbolTable {
self.current_scope_id.unwrap()
}
pub fn change_scope(&mut self, scope_id: usize) {
self.current_scope_id = Some(scope_id);
}
fn scope(&self, scope_id: usize) -> &Scope {
&self.scopes[scope_id]
}

6
dvm-lib/src/vm/function.rs
View File

@ -5,7 +5,7 @@ use std::rc::Rc;
pub struct Function {
name: Rc<str>,
parameter_count: usize,
stack_size: usize,
stack_size: isize,
instructions: Vec<Instruction>,
}
@ -13,7 +13,7 @@ impl Function {
pub fn new(
name: Rc<str>,
parameter_count: usize,
stack_size: usize,
stack_size: isize,
instructions: Vec<Instruction>,
) -> Self {
Self {
@ -36,7 +36,7 @@ impl Function {
self.parameter_count
}
pub fn stack_size(&self) -> usize {
pub fn stack_size(&self) -> isize {
self.stack_size
}

59
dvm-lib/src/vm/mod.rs
View File

@ -115,15 +115,15 @@ impl<'a> CallFrame<'a> {
self.ip += 1;
}
pub fn stack(&self) -> &[Operand] {
fn stack(&self) -> &[Operand] {
&self.stack
}
pub fn stack_mut(&mut self) -> &mut Vec<Operand> {
fn stack_mut(&mut self) -> &mut Vec<Operand> {
&mut self.stack
}
pub fn fp(&self) -> usize {
fn fp(&self) -> usize {
self.fp
}
@ -163,11 +163,11 @@ impl<'a> CallStack<'a> {
self.call_frames.pop().unwrap()
}
pub fn top(&self) -> &CallFrame<'a> {
fn top(&self) -> &CallFrame<'a> {
&self.call_frames[self.call_frames.len() - 1]
}
pub fn top_mut(&mut self) -> &mut CallFrame<'a> {
fn top_mut(&mut self) -> &mut CallFrame<'a> {
let call_frames_len = self.call_frames.len();
&mut self.call_frames[call_frames_len - 1] // dang borrow checker
}
@ -190,24 +190,18 @@ impl<'a> CallStack<'a> {
}
}
pub fn call(
context: &DvmContext,
function_name: &str,
arguments: &[Value],
register_count: usize,
) -> Option<Value> {
let mut call_stack = CallStack::new();
let mut registers = vec![Operand::Null; register_count];
prepare_for_instruction_loop(context, function_name, &mut call_stack, arguments);
loop_instructions(context, &mut registers, &mut call_stack)
}
pub fn prepare_for_instruction_loop<'a>(
pub fn call<'a>(
context: &'a DvmContext,
function_name: &str,
registers: &mut Vec<Operand>,
call_stack: &mut CallStack<'a>,
function_name: &str,
arguments: &[Value],
) {
) -> Option<Value> {
// check if DVM_DEBUG is enabled
let debug = std::env::var("DVM_DEBUG")
.map(|s| s.eq("1"))
.unwrap_or(false);
let function = context
.functions
.get(function_name)
@ -228,22 +222,12 @@ pub fn prepare_for_instruction_loop<'a>(
call_stack.top_mut().set_fp(arguments.len());
// ensure enough stack space
call_stack
.top_mut()
.stack_mut()
.resize(arguments.len() + function.stack_size(), Operand::Null);
}
pub fn loop_instructions<'a>(
context: &'a DvmContext,
registers: &mut Vec<Operand>,
call_stack: &mut CallStack<'a>,
) -> Option<Value> {
// check if DVM_DEBUG is enabled
let debug = std::env::var("DVM_DEBUG")
.map(|s| s.eq("1"))
.unwrap_or(false);
call_stack.top_mut().stack_mut().resize(
arguments.len() + (function.stack_size() as usize),
Operand::Null,
);
// container for final return value
let mut return_value: Option<Value> = None;
while call_stack.maybe_top().is_some()
@ -312,7 +296,7 @@ pub fn loop_instructions<'a>(
call_stack
.top_mut()
.stack_mut()
.resize(arg_count + function.stack_size(), Operand::Null);
.resize(arg_count + (function.stack_size() as usize), Operand::Null);
// set fp for callee frame
let callee_frame = call_stack.top_mut();
@ -706,9 +690,6 @@ pub fn loop_instructions<'a>(
// callee is the bottommost frame ("main" or whatever was called)
// set the return value
return_value = callee.take_return_value();
// push the frame back on the call stack for further inspection by caller
call_stack.push(callee);
return return_value;
}
// do not increment ip of the caller; this was done above BEFORE the call
}

19
e2e-tests/src/lib.rs
View File

@ -3,15 +3,15 @@ mod e2e_tests {
use dmc_lib::constants_table::ConstantsTable;
use dmc_lib::diagnostic::Diagnostic;
use dmc_lib::intrinsics::{insert_intrinsic_symbols, insert_intrinsic_types};
use dmc_lib::offset_counter::OffsetCounter;
use dmc_lib::parser::parse_compilation_unit;
use dmc_lib::symbol_table::SymbolTable;
use dmc_lib::types_table::TypesTable;
use dvm_lib::vm::class::Class;
use dvm_lib::vm::constant::{Constant, StringConstant};
use dvm_lib::vm::function::Function;
use dvm_lib::vm::operand::Operand;
use dvm_lib::vm::value::Value;
use dvm_lib::vm::{DvmContext, call};
use dvm_lib::vm::{CallStack, DvmContext, call};
use std::rc::Rc;
const REGISTER_COUNT: usize = 8;
@ -56,9 +56,7 @@ mod e2e_tests {
let mut constants_table = ConstantsTable::new();
for ir_function in &mut ir_functions {
let mut offset_counter = OffsetCounter::new();
ir_function.assign_registers(REGISTER_COUNT, &mut offset_counter);
let stack_size = offset_counter.get_count();
let (_, stack_size) = ir_function.assign_registers(REGISTER_COUNT);
let function = ir_function.assemble(stack_size, &mut constants_table);
functions.push(function);
}
@ -97,7 +95,16 @@ mod e2e_tests {
function_name: &str,
arguments: &[Value],
) -> Option<Value> {
call(&dvm_context, function_name, &arguments, REGISTER_COUNT)
let mut registers: Vec<Operand> = vec![Operand::Null; REGISTER_COUNT];
let mut call_stack = CallStack::new();
call(
&dvm_context,
&mut registers,
&mut call_stack,
function_name,
&arguments,
)
}
fn assert_result(input: &str, function_name: &str, arguments: &[Value], expected_value: Value) {